API reference

Modules API reference #

Heap allocation raw functions #

Memory allocated with these functions are taken into account by Redict key eviction algorithms and are reported in Redict memory usage information.

RedictModule_Alloc #

void *RedictModule_Alloc(size_t bytes);

Available since: Redict 7.3.0

Use like malloc(). Memory allocated with this function is reported in Redict INFO memory, used for keys eviction according to maxmemory settings and in general is taken into account as memory allocated by Redict. You should avoid using malloc(). This function panics if unable to allocate enough memory.

RedictModule_TryAlloc #

void *RedictModule_TryAlloc(size_t bytes);

Available since: Redict 7.3.0

Similar to RedictModule_Alloc, but returns NULL in case of allocation failure, instead of panicking.

RedictModule_Calloc #

void *RedictModule_Calloc(size_t nmemb, size_t size);

Available since: Redict 7.3.0

Use like calloc(). Memory allocated with this function is reported in Redict INFO memory, used for keys eviction according to maxmemory settings and in general is taken into account as memory allocated by Redict. You should avoid using calloc() directly.

RedictModule_TryCalloc #

void *RedictModule_TryCalloc(size_t nmemb, size_t size);

Available since: Redict 7.3.0

Similar to RedictModule_Calloc, but returns NULL in case of allocation failure, instead of panicking.

RedictModule_Realloc #

void* RedictModule_Realloc(void *ptr, size_t bytes);

Available since: Redict 7.3.0

Use like realloc() for memory obtained with RedictModule_Alloc().

RedictModule_TryRealloc #

void *RedictModule_TryRealloc(void *ptr, size_t bytes);

Available since: Redict 7.3.0

Similar to RedictModule_Realloc, but returns NULL in case of allocation failure, instead of panicking.

RedictModule_Free #

void RedictModule_Free(void *ptr);

Available since: Redict 7.3.0

Use like free() for memory obtained by RedictModule_Alloc() and RedictModule_Realloc(). However you should never try to free with RedictModule_Free() memory allocated with malloc() inside your module.

RedictModule_Strdup #

char *RedictModule_Strdup(const char *str);

Available since: Redict 7.3.0

Like strdup() but returns memory allocated with RedictModule_Alloc().

RedictModule_PoolAlloc #

void *RedictModule_PoolAlloc(RedictModuleCtx *ctx, size_t bytes);

Available since: Redict 7.3.0

Return heap allocated memory that will be freed automatically when the module callback function returns. Mostly suitable for small allocations that are short living and must be released when the callback returns anyway. The returned memory is aligned to the architecture word size if at least word size bytes are requested, otherwise it is just aligned to the next power of two, so for example a 3 bytes request is 4 bytes aligned while a 2 bytes request is 2 bytes aligned.

There is no realloc style function since when this is needed to use the pool allocator is not a good idea.

The function returns NULL if bytes is 0.

Commands API #

These functions are used to implement custom Redict commands.

For examples, see https://redis.io/topics/modules-intro.

RedictModule_IsKeysPositionRequest #

int RedictModule_IsKeysPositionRequest(RedictModuleCtx *ctx);

Available since: Redict 7.3.0

Return non-zero if a module command, that was declared with the flag “getkeys-api”, is called in a special way to get the keys positions and not to get executed. Otherwise zero is returned.

RedictModule_KeyAtPosWithFlags #

void RedictModule_KeyAtPosWithFlags(RedictModuleCtx *ctx, int pos, int flags);

Available since: Redict 7.3.0

When a module command is called in order to obtain the position of keys, since it was flagged as “getkeys-api” during the registration, the command implementation checks for this special call using the RedictModule_IsKeysPositionRequest() API and uses this function in order to report keys.

The supported flags are the ones used by RedictModule_SetCommandInfo, see REDICTMODULE_CMD_KEY_*.

The following is an example of how it could be used:

if (RedictModule_IsKeysPositionRequest(ctx)) {
    RedictModule_KeyAtPosWithFlags(ctx, 2, REDICTMODULE_CMD_KEY_RO | REDICTMODULE_CMD_KEY_ACCESS);
    RedictModule_KeyAtPosWithFlags(ctx, 1, REDICTMODULE_CMD_KEY_RW | REDICTMODULE_CMD_KEY_UPDATE | REDICTMODULE_CMD_KEY_ACCESS);
}

Note: in the example above the get keys API could have been handled by key-specs (preferred). Implementing the getkeys-api is required only when is it not possible to declare key-specs that cover all keys.

RedictModule_KeyAtPos #

void RedictModule_KeyAtPos(RedictModuleCtx *ctx, int pos);

Available since: Redict 7.3.0

This API existed before RedictModule_KeyAtPosWithFlags was added, now deprecated and can be used for compatibility with older versions, before key-specs and flags were introduced.

RedictModule_IsChannelsPositionRequest #

int RedictModule_IsChannelsPositionRequest(RedictModuleCtx *ctx);

Available since: Redict 7.3.0

Return non-zero if a module command, that was declared with the flag “getchannels-api”, is called in a special way to get the channel positions and not to get executed. Otherwise zero is returned.

RedictModule_ChannelAtPosWithFlags #

void RedictModule_ChannelAtPosWithFlags(RedictModuleCtx *ctx,
                                        int pos,
                                        int flags);

Available since: Redict 7.3.0

When a module command is called in order to obtain the position of channels, since it was flagged as “getchannels-api” during the registration, the command implementation checks for this special call using the RedictModule_IsChannelsPositionRequest() API and uses this function in order to report the channels.

The supported flags are:

  • REDICTMODULE_CMD_CHANNEL_SUBSCRIBE: This command will subscribe to the channel.
  • REDICTMODULE_CMD_CHANNEL_UNSUBSCRIBE: This command will unsubscribe from this channel.
  • REDICTMODULE_CMD_CHANNEL_PUBLISH: This command will publish to this channel.
  • REDICTMODULE_CMD_CHANNEL_PATTERN: Instead of acting on a specific channel, will act on any channel specified by the pattern. This is the same access used by the PSUBSCRIBE and PUNSUBSCRIBE commands available in Redict. Not intended to be used with PUBLISH permissions.

The following is an example of how it could be used:

if (RedictModule_IsChannelsPositionRequest(ctx)) {
    RedictModule_ChannelAtPosWithFlags(ctx, 1, REDICTMODULE_CMD_CHANNEL_SUBSCRIBE | REDICTMODULE_CMD_CHANNEL_PATTERN);
    RedictModule_ChannelAtPosWithFlags(ctx, 1, REDICTMODULE_CMD_CHANNEL_PUBLISH);
}

Note: One usage of declaring channels is for evaluating ACL permissions. In this context, unsubscribing is always allowed, so commands will only be checked against subscribe and publish permissions. This is preferred over using RedictModule_ACLCheckChannelPermissions, since it allows the ACLs to be checked before the command is executed.

RedictModule_CreateCommand #

int RedictModule_CreateCommand(RedictModuleCtx *ctx,
                               const char *name,
                               RedictModuleCmdFunc cmdfunc,
                               const char *strflags,
                               int firstkey,
                               int lastkey,
                               int keystep);

Available since: Redict 7.3.0

Register a new command in the Redict server, that will be handled by calling the function pointer ‘cmdfunc’ using the RedictModule calling convention.

The function returns REDICTMODULE_ERR in these cases:

  • If creation of module command is called outside the RedictModule_OnLoad.
  • The specified command is already busy.
  • The command name contains some chars that are not allowed.
  • A set of invalid flags were passed.

Otherwise REDICTMODULE_OK is returned and the new command is registered.

This function must be called during the initialization of the module inside the RedictModule_OnLoad() function. Calling this function outside of the initialization function is not defined.

The command function type is the following:

 int MyCommand_RedictCommand(RedictModuleCtx *ctx, RedictModuleString **argv, int argc);

And is supposed to always return REDICTMODULE_OK.

The set of flags ‘strflags’ specify the behavior of the command, and should be passed as a C string composed of space separated words, like for example “write deny-oom”. The set of flags are:

  • “write”: The command may modify the data set (it may also read from it).
  • “readonly”: The command returns data from keys but never writes.
  • “admin”: The command is an administrative command (may change replication or perform similar tasks).
  • “deny-oom”: The command may use additional memory and should be denied during out of memory conditions.
  • “deny-script”: Don’t allow this command in Lua scripts.
  • “allow-loading”: Allow this command while the server is loading data. Only commands not interacting with the data set should be allowed to run in this mode. If not sure don’t use this flag.
  • “pubsub”: The command publishes things on Pub/Sub channels.
  • “random”: The command may have different outputs even starting from the same input arguments and key values. Starting from Redis 7.0 this flag has been deprecated. Declaring a command as “random” can be done using command tips, see https://redis.io/topics/command-tips.
  • “allow-stale”: The command is allowed to run on slaves that don’t serve stale data. Don’t use if you don’t know what this means.
  • “no-monitor”: Don’t propagate the command on monitor. Use this if the command has sensitive data among the arguments.
  • “no-slowlog”: Don’t log this command in the slowlog. Use this if the command has sensitive data among the arguments.
  • “fast”: The command time complexity is not greater than O(log(N)) where N is the size of the collection or anything else representing the normal scalability issue with the command.
  • “getkeys-api”: The command implements the interface to return the arguments that are keys. Used when start/stop/step is not enough because of the command syntax.
  • “no-cluster”: The command should not register in Redict Cluster since is not designed to work with it because, for example, is unable to report the position of the keys, programmatically creates key names, or any other reason.
  • “no-auth”: This command can be run by an un-authenticated client. Normally this is used by a command that is used to authenticate a client.
  • “may-replicate”: This command may generate replication traffic, even though it’s not a write command.
  • “no-mandatory-keys”: All the keys this command may take are optional
  • “blocking”: The command has the potential to block the client.
  • “allow-busy”: Permit the command while the server is blocked either by a script or by a slow module command, see RedictModule_Yield.
  • “getchannels-api”: The command implements the interface to return the arguments that are channels.

The last three parameters specify which arguments of the new command are Redict keys. See https://redis.io/commands/command for more information.

  • firstkey: One-based index of the first argument that’s a key. Position 0 is always the command name itself. 0 for commands with no keys.
  • lastkey: One-based index of the last argument that’s a key. Negative numbers refer to counting backwards from the last argument (-1 means the last argument provided) 0 for commands with no keys.
  • keystep: Step between first and last key indexes. 0 for commands with no keys.

This information is used by ACL, Cluster and the COMMAND command.

NOTE: The scheme described above serves a limited purpose and can only be used to find keys that exist at constant indices. For non-trivial key arguments, you may pass 0,0,0 and use RedictModule_SetCommandInfo to set key specs using a more advanced scheme and use RedictModule_SetCommandACLCategories to set Redict ACL categories of the commands.

RedictModule_GetCommand #

RedictModuleCommand *RedictModule_GetCommand(RedictModuleCtx *ctx,
                                             const char *name);

Available since: Redict 7.3.0

Get an opaque structure, representing a module command, by command name. This structure is used in some of the command-related APIs.

NULL is returned in case of the following errors:

  • Command not found
  • The command is not a module command
  • The command doesn’t belong to the calling module

RedictModule_CreateSubcommand #

int RedictModule_CreateSubcommand(RedictModuleCommand *parent,
                                  const char *name,
                                  RedictModuleCmdFunc cmdfunc,
                                  const char *strflags,
                                  int firstkey,
                                  int lastkey,
                                  int keystep);

Available since: Redict 7.3.0

Very similar to RedictModule_CreateCommand except that it is used to create a subcommand, associated with another, container, command.

Example: If a module has a configuration command, MODULE.CONFIG, then GET and SET should be individual subcommands, while MODULE.CONFIG is a command, but should not be registered with a valid funcptr:

 if (RedictModule_CreateCommand(ctx,"module.config",NULL,"",0,0,0) == REDICTMODULE_ERR)
     return REDICTMODULE_ERR;

 RedictModuleCommand *parent = RedictModule_GetCommand(ctx,,"module.config");

 if (RedictModule_CreateSubcommand(parent,"set",cmd_config_set,"",0,0,0) == REDICTMODULE_ERR)
    return REDICTMODULE_ERR;

 if (RedictModule_CreateSubcommand(parent,"get",cmd_config_get,"",0,0,0) == REDICTMODULE_ERR)
    return REDICTMODULE_ERR;

Returns REDICTMODULE_OK on success and REDICTMODULE_ERR in case of the following errors:

  • Error while parsing strflags
  • Command is marked as no-cluster but cluster mode is enabled
  • parent is already a subcommand (we do not allow more than one level of command nesting)
  • parent is a command with an implementation (RedictModuleCmdFunc) (A parent command should be a pure container of subcommands)
  • parent already has a subcommand called name
  • Creating a subcommand is called outside of RedictModule_OnLoad.

RedictModule_AddACLCategory #

int RedictModule_AddACLCategory(RedictModuleCtx *ctx, const char *name);

Available since: Redict 7.3.0

RedictModule_AddACLCategory can be used to add new ACL command categories. Category names can only contain alphanumeric characters, underscores, or dashes. Categories can only be added during the RedictModule_OnLoad function. Once a category has been added, it can not be removed. Any module can register a command to any added categories using RedictModule_SetCommandACLCategories.

Returns:

  • REDICTMODULE_OK on successfully adding the new ACL category.
  • REDICTMODULE_ERR on failure.

On error the errno is set to:

  • EINVAL if the name contains invalid characters.
  • EBUSY if the category name already exists.
  • ENOMEM if the number of categories reached the max limit of 64 categories.

RedictModule_SetCommandACLCategories #

int RedictModule_SetCommandACLCategories(RedictModuleCommand *command,
                                         const char *aclflags);

Available since: Redict 7.3.0

RedictModule_SetCommandACLCategories can be used to set ACL categories to module commands and subcommands. The set of ACL categories should be passed as a space separated C string ‘aclflags’.

Example, the acl flags ‘write slow’ marks the command as part of the write and slow ACL categories.

On success REDICTMODULE_OK is returned. On error REDICTMODULE_ERR is returned.

This function can only be called during the RedictModule_OnLoad function. If called outside of this function, an error is returned.

RedictModule_SetCommandInfo #

int RedictModule_SetCommandInfo(RedictModuleCommand *command,
                                const RedictModuleCommandInfo *info);

Available since: Redict 7.3.0

Set additional command information.

Affects the output of COMMAND, COMMAND INFO and COMMAND DOCS, Cluster, ACL and is used to filter commands with the wrong number of arguments before the call reaches the module code.

This function can be called after creating a command using RedictModule_CreateCommand and fetching the command pointer using RedictModule_GetCommand. The information can only be set once for each command and has the following structure:

typedef struct RedictModuleCommandInfo {
    const RedictModuleCommandInfoVersion *version;
    const char *summary;
    const char *complexity;
    const char *since;
    RedictModuleCommandHistoryEntry *history;
    const char *tips;
    int arity;
    RedictModuleCommandKeySpec *key_specs;
    RedictModuleCommandArg *args;
} RedictModuleCommandInfo;

All fields except version are optional. Explanation of the fields:

  • version: This field enables compatibility with different Redict versions. Always set this field to REDICTMODULE_COMMAND_INFO_VERSION.

  • summary: A short description of the command (optional).

  • complexity: Complexity description (optional).

  • since: The version where the command was introduced (optional). Note: The version specified should be the module’s, not Redict version.

  • history: An array of RedictModuleCommandHistoryEntry (optional), which is a struct with the following fields:

      const char *since;
      const char *changes;
    

    since is a version string and changes is a string describing the changes. The array is terminated by a zeroed entry, i.e. an entry with both strings set to NULL.

  • tips: A string of space-separated tips regarding this command, meant for clients and proxies. See https://redis.io/topics/command-tips.

  • arity: Number of arguments, including the command name itself. A positive number specifies an exact number of arguments and a negative number specifies a minimum number of arguments, so use -N to say >= N. Redict validates a call before passing it to a module, so this can replace an arity check inside the module command implementation. A value of 0 (or an omitted arity field) is equivalent to -2 if the command has sub commands and -1 otherwise.

  • key_specs: An array of RedictModuleCommandKeySpec, terminated by an element memset to zero. This is a scheme that tries to describe the positions of key arguments better than the old RedictModule_CreateCommand arguments firstkey, lastkey, keystep and is needed if those three are not enough to describe the key positions. There are two steps to retrieve key positions: begin search (BS) in which index should find the first key and find keys (FK) which, relative to the output of BS, describes how can we will which arguments are keys. Additionally, there are key specific flags.

    Key-specs cause the triplet (firstkey, lastkey, keystep) given in RedictModule_CreateCommand to be recomputed, but it is still useful to provide these three parameters in RedictModule_CreateCommand, to better support old Redict versions where RedictModule_SetCommandInfo is not available.

    Note that key-specs don’t fully replace the “getkeys-api” (see RedictModule_CreateCommand, RedictModule_IsKeysPositionRequest and RedictModule_KeyAtPosWithFlags) so it may be a good idea to supply both key-specs and implement the getkeys-api.

    A key-spec has the following structure:

      typedef struct RedictModuleCommandKeySpec {
          const char *notes;
          uint64_t flags;
          RedictModuleKeySpecBeginSearchType begin_search_type;
          union {
              struct {
                  int pos;
              } index;
              struct {
                  const char *keyword;
                  int startfrom;
              } keyword;
          } bs;
          RedictModuleKeySpecFindKeysType find_keys_type;
          union {
              struct {
                  int lastkey;
                  int keystep;
                  int limit;
              } range;
              struct {
                  int keynumidx;
                  int firstkey;
                  int keystep;
              } keynum;
          } fk;
      } RedictModuleCommandKeySpec;
    

    Explanation of the fields of RedictModuleCommandKeySpec:

    • notes: Optional notes or clarifications about this key spec.

    • flags: A bitwise or of key-spec flags described below.

    • begin_search_type: This describes how the first key is discovered. There are two ways to determine the first key:

      • REDICTMODULE_KSPEC_BS_UNKNOWN: There is no way to tell where the key args start.
      • REDICTMODULE_KSPEC_BS_INDEX: Key args start at a constant index.
      • REDICTMODULE_KSPEC_BS_KEYWORD: Key args start just after a specific keyword.
    • bs: This is a union in which the index or keyword branch is used depending on the value of the begin_search_type field.

      • bs.index.pos: The index from which we start the search for keys. (REDICTMODULE_KSPEC_BS_INDEX only.)

      • bs.keyword.keyword: The keyword (string) that indicates the beginning of key arguments. (REDICTMODULE_KSPEC_BS_KEYWORD only.)

      • bs.keyword.startfrom: An index in argv from which to start searching. Can be negative, which means start search from the end, in reverse. Example: -2 means to start in reverse from the penultimate argument. (REDICTMODULE_KSPEC_BS_KEYWORD only.)

    • find_keys_type: After the “begin search”, this describes which arguments are keys. The strategies are:

      • REDICTMODULE_KSPEC_BS_UNKNOWN: There is no way to tell where the key args are located.
      • REDICTMODULE_KSPEC_FK_RANGE: Keys end at a specific index (or relative to the last argument).
      • REDICTMODULE_KSPEC_FK_KEYNUM: There’s an argument that contains the number of key args somewhere before the keys themselves.

      find_keys_type and fk can be omitted if this keyspec describes exactly one key.

    • fk: This is a union in which the range or keynum branch is used depending on the value of the find_keys_type field.

      • fk.range (for REDICTMODULE_KSPEC_FK_RANGE): A struct with the following fields:

        • lastkey: Index of the last key relative to the result of the begin search step. Can be negative, in which case it’s not relative. -1 indicates the last argument, -2 one before the last and so on.

        • keystep: How many arguments should we skip after finding a key, in order to find the next one?

        • limit: If lastkey is -1, we use limit to stop the search by a factor. 0 and 1 mean no limit. 2 means 1/2 of the remaining args, 3 means 1/3, and so on.

      • fk.keynum (for REDICTMODULE_KSPEC_FK_KEYNUM): A struct with the following fields:

        • keynumidx: Index of the argument containing the number of keys to come, relative to the result of the begin search step.

        • firstkey: Index of the fist key relative to the result of the begin search step. (Usually it’s just after keynumidx, in which case it should be set to keynumidx + 1.)

        • keystep: How many arguments should we skip after finding a key, in order to find the next one?

    Key-spec flags:

    The first four refer to what the command actually does with the value or metadata of the key, and not necessarily the user data or how it affects it. Each key-spec may must have exactly one of these. Any operation that’s not distinctly deletion, overwrite or read-only would be marked as RW.

    • REDICTMODULE_CMD_KEY_RO: Read-Only. Reads the value of the key, but doesn’t necessarily return it.

    • REDICTMODULE_CMD_KEY_RW: Read-Write. Modifies the data stored in the value of the key or its metadata.

    • REDICTMODULE_CMD_KEY_OW: Overwrite. Overwrites the data stored in the value of the key.

    • REDICTMODULE_CMD_KEY_RM: Deletes the key.

    The next four refer to user data inside the value of the key, not the metadata like LRU, type, cardinality. It refers to the logical operation on the user’s data (actual input strings or TTL), being used/returned/copied/changed. It doesn’t refer to modification or returning of metadata (like type, count, presence of data). ACCESS can be combined with one of the write operations INSERT, DELETE or UPDATE. Any write that’s not an INSERT or a DELETE would be UPDATE.

    • REDICTMODULE_CMD_KEY_ACCESS: Returns, copies or uses the user data from the value of the key.

    • REDICTMODULE_CMD_KEY_UPDATE: Updates data to the value, new value may depend on the old value.

    • REDICTMODULE_CMD_KEY_INSERT: Adds data to the value with no chance of modification or deletion of existing data.

    • REDICTMODULE_CMD_KEY_DELETE: Explicitly deletes some content from the value of the key.

    Other flags:

    • REDICTMODULE_CMD_KEY_NOT_KEY: The key is not actually a key, but should be routed in cluster mode as if it was a key.

    • REDICTMODULE_CMD_KEY_INCOMPLETE: The keyspec might not point out all the keys it should cover.

    • REDICTMODULE_CMD_KEY_VARIABLE_FLAGS: Some keys might have different flags depending on arguments.

  • args: An array of RedictModuleCommandArg, terminated by an element memset to zero. RedictModuleCommandArg is a structure with at the fields described below.

      typedef struct RedictModuleCommandArg {
          const char *name;
          RedictModuleCommandArgType type;
          int key_spec_index;
          const char *token;
          const char *summary;
          const char *since;
          int flags;
          struct RedictModuleCommandArg *subargs;
      } RedictModuleCommandArg;
    

    Explanation of the fields:

    • name: Name of the argument.

    • type: The type of the argument. See below for details. The types REDICTMODULE_ARG_TYPE_ONEOF and REDICTMODULE_ARG_TYPE_BLOCK require an argument to have sub-arguments, i.e. subargs.

    • key_spec_index: If the type is REDICTMODULE_ARG_TYPE_KEY you must provide the index of the key-spec associated with this argument. See key_specs above. If the argument is not a key, you may specify -1.

    • token: The token preceding the argument (optional). Example: the argument seconds in SET has a token EX. If the argument consists of only a token (for example NX in SET) the type should be REDICTMODULE_ARG_TYPE_PURE_TOKEN and value should be NULL.

    • summary: A short description of the argument (optional).

    • since: The first version which included this argument (optional).

    • flags: A bitwise or of the macros REDICTMODULE_CMD_ARG_*. See below.

    • value: The display-value of the argument. This string is what should be displayed when creating the command syntax from the output of COMMAND. If token is not NULL, it should also be displayed.

    Explanation of RedictModuleCommandArgType:

    • REDICTMODULE_ARG_TYPE_STRING: String argument.
    • REDICTMODULE_ARG_TYPE_INTEGER: Integer argument.
    • REDICTMODULE_ARG_TYPE_DOUBLE: Double-precision float argument.
    • REDICTMODULE_ARG_TYPE_KEY: String argument representing a keyname.
    • REDICTMODULE_ARG_TYPE_PATTERN: String, but regex pattern.
    • REDICTMODULE_ARG_TYPE_UNIX_TIME: Integer, but Unix timestamp.
    • REDICTMODULE_ARG_TYPE_PURE_TOKEN: Argument doesn’t have a placeholder. It’s just a token without a value. Example: the KEEPTTL option of the SET command.
    • REDICTMODULE_ARG_TYPE_ONEOF: Used when the user can choose only one of a few sub-arguments. Requires subargs. Example: the NX and XX options of SET.
    • REDICTMODULE_ARG_TYPE_BLOCK: Used when one wants to group together several sub-arguments, usually to apply something on all of them, like making the entire group “optional”. Requires subargs. Example: the LIMIT offset count parameters in ZRANGE.

    Explanation of the command argument flags:

    • REDICTMODULE_CMD_ARG_OPTIONAL: The argument is optional (like GET in the SET command).
    • REDICTMODULE_CMD_ARG_MULTIPLE: The argument may repeat itself (like key in DEL).
    • REDICTMODULE_CMD_ARG_MULTIPLE_TOKEN: The argument may repeat itself, and so does its token (like GET pattern in SORT).

On success REDICTMODULE_OK is returned. On error REDICTMODULE_ERR is returned and errno is set to EINVAL if invalid info was provided or EEXIST if info has already been set. If the info is invalid, a warning is logged explaining which part of the info is invalid and why.

Module information and time measurement #

RedictModule_IsModuleNameBusy #

int RedictModule_IsModuleNameBusy(const char *name);

Available since: Redict 7.3.0

Return non-zero if the module name is busy. Otherwise zero is returned.

RedictModule_Milliseconds #

mstime_t RedictModule_Milliseconds(void);

Available since: Redict 7.3.0

Return the current UNIX time in milliseconds.

RedictModule_MonotonicMicroseconds #

uint64_t RedictModule_MonotonicMicroseconds(void);

Available since: Redict 7.3.0

Return counter of micro-seconds relative to an arbitrary point in time.

RedictModule_Microseconds #

ustime_t RedictModule_Microseconds(void);

Available since: Redict 7.3.0

Return the current UNIX time in microseconds

RedictModule_CachedMicroseconds #

ustime_t RedictModule_CachedMicroseconds(void);

Available since: Redict 7.3.0

Return the cached UNIX time in microseconds. It is updated in the server cron job and before executing a command. It is useful for complex call stacks, such as a command causing a key space notification, causing a module to execute a RedictModule_Call, causing another notification, etc. It makes sense that all this callbacks would use the same clock.

RedictModule_BlockedClientMeasureTimeStart #

int RedictModule_BlockedClientMeasureTimeStart(RedictModuleBlockedClient *bc);

Available since: Redict 7.3.0

Mark a point in time that will be used as the start time to calculate the elapsed execution time when RedictModule_BlockedClientMeasureTimeEnd() is called. Within the same command, you can call multiple times RedictModule_BlockedClientMeasureTimeStart() and RedictModule_BlockedClientMeasureTimeEnd() to accumulate independent time intervals to the background duration. This method always return REDICTMODULE_OK.

This function is not thread safe, If used in module thread and blocked callback (possibly main thread) simultaneously, it’s recommended to protect them with lock owned by caller instead of GIL.

RedictModule_BlockedClientMeasureTimeEnd #

int RedictModule_BlockedClientMeasureTimeEnd(RedictModuleBlockedClient *bc);

Available since: Redict 7.3.0

Mark a point in time that will be used as the end time to calculate the elapsed execution time. On success REDICTMODULE_OK is returned. This method only returns REDICTMODULE_ERR if no start time was previously defined ( meaning RedictModule_BlockedClientMeasureTimeStart was not called ).

This function is not thread safe, If used in module thread and blocked callback (possibly main thread) simultaneously, it’s recommended to protect them with lock owned by caller instead of GIL.

RedictModule_Yield #

void RedictModule_Yield(RedictModuleCtx *ctx,
                        int flags,
                        const char *busy_reply);

Available since: Redict 7.3.0

This API allows modules to let Redict process background tasks, and some commands during long blocking execution of a module command. The module can call this API periodically. The flags is a bit mask of these:

  • REDICTMODULE_YIELD_FLAG_NONE: No special flags, can perform some background operations, but not process client commands.
  • REDICTMODULE_YIELD_FLAG_CLIENTS: Redict can also process client commands.

The busy_reply argument is optional, and can be used to control the verbose error string after the -BUSY error code.

When the REDICTMODULE_YIELD_FLAG_CLIENTS is used, Redict will only start processing client commands after the time defined by the busy-reply-threshold config, in which case Redict will start rejecting most commands with -BUSY error, but allow the ones marked with the allow-busy flag to be executed. This API can also be used in thread safe context (while locked), and during loading (in the rdb_load callback, in which case it’ll reject commands with the -LOADING error)

RedictModule_SetModuleOptions #

void RedictModule_SetModuleOptions(RedictModuleCtx *ctx, int options);

Available since: Redict 7.3.0

Set flags defining capabilities or behavior bit flags.

REDICTMODULE_OPTIONS_HANDLE_IO_ERRORS: Generally, modules don’t need to bother with this, as the process will just terminate if a read error happens, however, setting this flag would allow repl-diskless-load to work if enabled. The module should use RedictModule_IsIOError after reads, before using the data that was read, and in case of error, propagate it upwards, and also be able to release the partially populated value and all it’s allocations.

REDICTMODULE_OPTION_NO_IMPLICIT_SIGNAL_MODIFIED: See RedictModule_SignalModifiedKey().

REDICTMODULE_OPTIONS_HANDLE_REPL_ASYNC_LOAD: Setting this flag indicates module awareness of diskless async replication (repl-diskless-load=swapdb) and that redict could be serving reads during replication instead of blocking with LOADING status.

REDICTMODULE_OPTIONS_ALLOW_NESTED_KEYSPACE_NOTIFICATIONS: Declare that the module wants to get nested key-space notifications. By default, Redict will not fire key-space notifications that happened inside a key-space notification callback. This flag allows to change this behavior and fire nested key-space notifications. Notice: if enabled, the module should protected itself from infinite recursion.

RedictModule_SignalModifiedKey #

int RedictModule_SignalModifiedKey(RedictModuleCtx *ctx,
                                   RedictModuleString *keyname);

Available since: Redict 7.3.0

Signals that the key is modified from user’s perspective (i.e. invalidate WATCH and client side caching).

This is done automatically when a key opened for writing is closed, unless the option REDICTMODULE_OPTION_NO_IMPLICIT_SIGNAL_MODIFIED has been set using RedictModule_SetModuleOptions().

Automatic memory management for modules #

RedictModule_AutoMemory #

void RedictModule_AutoMemory(RedictModuleCtx *ctx);

Available since: Redict 7.3.0

Enable automatic memory management.

The function must be called as the first function of a command implementation that wants to use automatic memory.

When enabled, automatic memory management tracks and automatically frees keys, call replies and Redict string objects once the command returns. In most cases this eliminates the need of calling the following functions:

  1. RedictModule_CloseKey()
  2. RedictModule_FreeCallReply()
  3. RedictModule_FreeString()

These functions can still be used with automatic memory management enabled, to optimize loops that make numerous allocations for example.

String objects APIs #

RedictModule_CreateString #

RedictModuleString *RedictModule_CreateString(RedictModuleCtx *ctx,
                                              const char *ptr,
                                              size_t len);

Available since: Redict 7.3.0

Create a new module string object. The returned string must be freed with RedictModule_FreeString(), unless automatic memory is enabled.

The string is created by copying the len bytes starting at ptr. No reference is retained to the passed buffer.

The module context ‘ctx’ is optional and may be NULL if you want to create a string out of the context scope. However in that case, the automatic memory management will not be available, and the string memory must be managed manually.

RedictModule_CreateStringPrintf #

RedictModuleString *RedictModule_CreateStringPrintf(RedictModuleCtx *ctx,
                                                    const char *fmt,
                                                    ...);

Available since: Redict 7.3.0

Create a new module string object from a printf format and arguments. The returned string must be freed with RedictModule_FreeString(), unless automatic memory is enabled.

The string is created using the sds formatter function sdscatvprintf().

The passed context ‘ctx’ may be NULL if necessary, see the RedictModule_CreateString() documentation for more info.

RedictModule_CreateStringFromLongLong #

RedictModuleString *RedictModule_CreateStringFromLongLong(RedictModuleCtx *ctx,
                                                          long long ll);

Available since: Redict 7.3.0

Like RedictModule_CreateString(), but creates a string starting from a long long integer instead of taking a buffer and its length.

The returned string must be released with RedictModule_FreeString() or by enabling automatic memory management.

The passed context ‘ctx’ may be NULL if necessary, see the RedictModule_CreateString() documentation for more info.

RedictModule_CreateStringFromULongLong #

RedictModuleString *RedictModule_CreateStringFromULongLong(RedictModuleCtx *ctx,
                                                           unsigned long long ull);

Available since: Redict 7.3.0

Like RedictModule_CreateString(), but creates a string starting from a unsigned long long integer instead of taking a buffer and its length.

The returned string must be released with RedictModule_FreeString() or by enabling automatic memory management.

The passed context ‘ctx’ may be NULL if necessary, see the RedictModule_CreateString() documentation for more info.

RedictModule_CreateStringFromDouble #

RedictModuleString *RedictModule_CreateStringFromDouble(RedictModuleCtx *ctx,
                                                        double d);

Available since: Redict 7.3.0

Like RedictModule_CreateString(), but creates a string starting from a double instead of taking a buffer and its length.

The returned string must be released with RedictModule_FreeString() or by enabling automatic memory management.

RedictModule_CreateStringFromLongDouble #

RedictModuleString *RedictModule_CreateStringFromLongDouble(RedictModuleCtx *ctx,
                                                            long double ld,
                                                            int humanfriendly);

Available since: Redict 7.3.0

Like RedictModule_CreateString(), but creates a string starting from a long double.

The returned string must be released with RedictModule_FreeString() or by enabling automatic memory management.

The passed context ‘ctx’ may be NULL if necessary, see the RedictModule_CreateString() documentation for more info.

RedictModule_CreateStringFromString #

RedictModuleString *RedictModule_CreateStringFromString(RedictModuleCtx *ctx,
                                                        const RedictModuleString *str);

Available since: Redict 7.3.0

Like RedictModule_CreateString(), but creates a string starting from another RedictModuleString.

The returned string must be released with RedictModule_FreeString() or by enabling automatic memory management.

The passed context ‘ctx’ may be NULL if necessary, see the RedictModule_CreateString() documentation for more info.

RedictModule_CreateStringFromStreamID #

RedictModuleString *RedictModule_CreateStringFromStreamID(RedictModuleCtx *ctx,
                                                          const RedictModuleStreamID *id);

Available since: Redict 7.3.0

Creates a string from a stream ID. The returned string must be released with RedictModule_FreeString(), unless automatic memory is enabled.

The passed context ctx may be NULL if necessary. See the RedictModule_CreateString() documentation for more info.

RedictModule_FreeString #

void RedictModule_FreeString(RedictModuleCtx *ctx, RedictModuleString *str);

Available since: Redict 7.3.0

Free a module string object obtained with one of the Redict modules API calls that return new string objects.

It is possible to call this function even when automatic memory management is enabled. In that case the string will be released ASAP and removed from the pool of string to release at the end.

If the string was created with a NULL context ‘ctx’, it is also possible to pass ctx as NULL when releasing the string (but passing a context will not create any issue). Strings created with a context should be freed also passing the context, so if you want to free a string out of context later, make sure to create it using a NULL context.

This API is not thread safe, access to these retained strings (if they originated from a client command arguments) must be done with GIL locked.

RedictModule_RetainString #

void RedictModule_RetainString(RedictModuleCtx *ctx, RedictModuleString *str);

Available since: Redict 7.3.0

Every call to this function, will make the string ‘str’ requiring an additional call to RedictModule_FreeString() in order to really free the string. Note that the automatic freeing of the string obtained enabling modules automatic memory management counts for one RedictModule_FreeString() call (it is just executed automatically).

Normally you want to call this function when, at the same time the following conditions are true:

  1. You have automatic memory management enabled.
  2. You want to create string objects.
  3. Those string objects you create need to live after the callback function(for example a command implementation) creating them returns.

Usually you want this in order to store the created string object into your own data structure, for example when implementing a new data type.

Note that when memory management is turned off, you don’t need any call to RetainString() since creating a string will always result into a string that lives after the callback function returns, if no FreeString() call is performed.

It is possible to call this function with a NULL context.

When strings are going to be retained for an extended duration, it is good practice to also call RedictModule_TrimStringAllocation() in order to optimize memory usage.

Threaded modules that reference retained strings from other threads must explicitly trim the allocation as soon as the string is retained. Not doing so may result with automatic trimming which is not thread safe.

This API is not thread safe, access to these retained strings (if they originated from a client command arguments) must be done with GIL locked.

RedictModule_HoldString #

RedictModuleString* RedictModule_HoldString(RedictModuleCtx *ctx,
                                            RedictModuleString *str);

Available since: Redict 7.3.0

This function can be used instead of RedictModule_RetainString(). The main difference between the two is that this function will always succeed, whereas RedictModule_RetainString() may fail because of an assertion.

The function returns a pointer to RedictModuleString, which is owned by the caller. It requires a call to RedictModule_FreeString() to free the string when automatic memory management is disabled for the context. When automatic memory management is enabled, you can either call RedictModule_FreeString() or let the automation free it.

This function is more efficient than RedictModule_CreateStringFromString() because whenever possible, it avoids copying the underlying RedictModuleString. The disadvantage of using this function is that it might not be possible to use RedictModule_StringAppendBuffer() on the returned RedictModuleString.

It is possible to call this function with a NULL context.

When strings are going to be held for an extended duration, it is good practice to also call RedictModule_TrimStringAllocation() in order to optimize memory usage.

Threaded modules that reference held strings from other threads must explicitly trim the allocation as soon as the string is held. Not doing so may result with automatic trimming which is not thread safe.

This API is not thread safe, access to these retained strings (if they originated from a client command arguments) must be done with GIL locked.

RedictModule_StringPtrLen #

const char *RedictModule_StringPtrLen(const RedictModuleString *str,
                                      size_t *len);

Available since: Redict 7.3.0

Given a string module object, this function returns the string pointer and length of the string. The returned pointer and length should only be used for read only accesses and never modified.

RedictModule_StringToLongLong #

int RedictModule_StringToLongLong(const RedictModuleString *str,
                                  long long *ll);

Available since: Redict 7.3.0

Convert the string into a long long integer, storing it at *ll. Returns REDICTMODULE_OK on success. If the string can’t be parsed as a valid, strict long long (no spaces before/after), REDICTMODULE_ERR is returned.

RedictModule_StringToULongLong #

int RedictModule_StringToULongLong(const RedictModuleString *str,
                                   unsigned long long *ull);

Available since: Redict 7.3.0

Convert the string into a unsigned long long integer, storing it at *ull. Returns REDICTMODULE_OK on success. If the string can’t be parsed as a valid, strict unsigned long long (no spaces before/after), REDICTMODULE_ERR is returned.

RedictModule_StringToDouble #

int RedictModule_StringToDouble(const RedictModuleString *str, double *d);

Available since: Redict 7.3.0

Convert the string into a double, storing it at *d. Returns REDICTMODULE_OK on success or REDICTMODULE_ERR if the string is not a valid string representation of a double value.

RedictModule_StringToLongDouble #

int RedictModule_StringToLongDouble(const RedictModuleString *str,
                                    long double *ld);

Available since: Redict 7.3.0

Convert the string into a long double, storing it at *ld. Returns REDICTMODULE_OK on success or REDICTMODULE_ERR if the string is not a valid string representation of a double value.

RedictModule_StringToStreamID #

int RedictModule_StringToStreamID(const RedictModuleString *str,
                                  RedictModuleStreamID *id);

Available since: Redict 7.3.0

Convert the string into a stream ID, storing it at *id. Returns REDICTMODULE_OK on success and returns REDICTMODULE_ERR if the string is not a valid string representation of a stream ID. The special IDs “+” and “-” are allowed.

RedictModule_StringCompare #

int RedictModule_StringCompare(const RedictModuleString *a,
                               const RedictModuleString *b);

Available since: Redict 7.3.0

Compare two string objects, returning -1, 0 or 1 respectively if a < b, a == b, a > b. Strings are compared byte by byte as two binary blobs without any encoding care / collation attempt.

RedictModule_StringAppendBuffer #

int RedictModule_StringAppendBuffer(RedictModuleCtx *ctx,
                                    RedictModuleString *str,
                                    const char *buf,
                                    size_t len);

Available since: Redict 7.3.0

Append the specified buffer to the string ‘str’. The string must be a string created by the user that is referenced only a single time, otherwise REDICTMODULE_ERR is returned and the operation is not performed.

RedictModule_TrimStringAllocation #

void RedictModule_TrimStringAllocation(RedictModuleString *str);

Available since: Redict 7.3.0

Trim possible excess memory allocated for a RedictModuleString.

Sometimes a RedictModuleString may have more memory allocated for it than required, typically for argv arguments that were constructed from network buffers. This function optimizes such strings by reallocating their memory, which is useful for strings that are not short lived but retained for an extended duration.

This operation is not thread safe and should only be called when no concurrent access to the string is guaranteed. Using it for an argv string in a module command before the string is potentially available to other threads is generally safe.

Currently, Redict may also automatically trim retained strings when a module command returns. However, doing this explicitly should still be a preferred option:

  1. Future versions of Redict may abandon auto-trimming.
  2. Auto-trimming as currently implemented is not thread safe. A background thread manipulating a recently retained string may end up in a race condition with the auto-trim, which could result with data corruption.

Reply APIs #

These functions are used for sending replies to the client.

Most functions always return REDICTMODULE_OK so you can use it with ‘return’ in order to return from the command implementation with:

if (... some condition ...)
    return RedictModule_ReplyWithLongLong(ctx,mycount);

Reply with collection functions #

After starting a collection reply, the module must make calls to other ReplyWith* style functions in order to emit the elements of the collection. Collection types include: Array, Map, Set and Attribute.

When producing collections with a number of elements that is not known beforehand, the function can be called with a special flag REDICTMODULE_POSTPONED_LEN (REDICTMODULE_POSTPONED_ARRAY_LEN in the past), and the actual number of elements can be later set with RedictModule_ReplySet*Length() call (which will set the latest “open” count if there are multiple ones).

RedictModule_WrongArity #

int RedictModule_WrongArity(RedictModuleCtx *ctx);

Available since: Redict 7.3.0

Send an error about the number of arguments given to the command, citing the command name in the error message. Returns REDICTMODULE_OK.

Example:

if (argc != 3) return RedictModule_WrongArity(ctx);

RedictModule_ReplyWithLongLong #

int RedictModule_ReplyWithLongLong(RedictModuleCtx *ctx, long long ll);

Available since: Redict 7.3.0

Send an integer reply to the client, with the specified long long value. The function always returns REDICTMODULE_OK.

RedictModule_ReplyWithError #

int RedictModule_ReplyWithError(RedictModuleCtx *ctx, const char *err);

Available since: Redict 7.3.0

Reply with the error ’err’.

Note that ’err’ must contain all the error, including the initial error code. The function only provides the initial “-”, so the usage is, for example:

RedictModule_ReplyWithError(ctx,"ERR Wrong Type");

and not just:

RedictModule_ReplyWithError(ctx,"Wrong Type");

The function always returns REDICTMODULE_OK.

RedictModule_ReplyWithErrorFormat #

int RedictModule_ReplyWithErrorFormat(RedictModuleCtx *ctx,
                                      const char *fmt,
                                      ...);

Available since: Redict 7.3.0

Reply with the error create from a printf format and arguments.

Note that ‘fmt’ must contain all the error, including the initial error code. The function only provides the initial “-”, so the usage is, for example:

RedictModule_ReplyWithErrorFormat(ctx,"ERR Wrong Type: %s",type);

and not just:

RedictModule_ReplyWithErrorFormat(ctx,"Wrong Type: %s",type);

The function always returns REDICTMODULE_OK.

RedictModule_ReplyWithSimpleString #

int RedictModule_ReplyWithSimpleString(RedictModuleCtx *ctx, const char *msg);

Available since: Redict 7.3.0

Reply with a simple string (+... \r\n in RESP protocol). This replies are suitable only when sending a small non-binary string with small overhead, like “OK” or similar replies.

The function always returns REDICTMODULE_OK.

RedictModule_ReplyWithArray #

int RedictModule_ReplyWithArray(RedictModuleCtx *ctx, long len);

Available since: Redict 7.3.0

Reply with an array type of ’len’ elements.

After starting an array reply, the module must make len calls to other ReplyWith* style functions in order to emit the elements of the array. See Reply APIs section for more details.

Use RedictModule_ReplySetArrayLength() to set deferred length.

The function always returns REDICTMODULE_OK.

RedictModule_ReplyWithMap #

int RedictModule_ReplyWithMap(RedictModuleCtx *ctx, long len);

Available since: Redict 7.3.0

Reply with a RESP3 Map type of ’len’ pairs. Visit https://github.com/antirez/RESP3/blob/master/spec.md for more info about RESP3.

After starting a map reply, the module must make len*2 calls to other ReplyWith* style functions in order to emit the elements of the map. See Reply APIs section for more details.

If the connected client is using RESP2, the reply will be converted to a flat array.

Use RedictModule_ReplySetMapLength() to set deferred length.

The function always returns REDICTMODULE_OK.

RedictModule_ReplyWithSet #

int RedictModule_ReplyWithSet(RedictModuleCtx *ctx, long len);

Available since: Redict 7.3.0

Reply with a RESP3 Set type of ’len’ elements. Visit https://github.com/antirez/RESP3/blob/master/spec.md for more info about RESP3.

After starting a set reply, the module must make len calls to other ReplyWith* style functions in order to emit the elements of the set. See Reply APIs section for more details.

If the connected client is using RESP2, the reply will be converted to an array type.

Use RedictModule_ReplySetSetLength() to set deferred length.

The function always returns REDICTMODULE_OK.

RedictModule_ReplyWithAttribute #

int RedictModule_ReplyWithAttribute(RedictModuleCtx *ctx, long len);

Available since: Redict 7.3.0

Add attributes (metadata) to the reply. Should be done before adding the actual reply. see https://github.com/antirez/RESP3/blob/master/spec.md#attribute-type

After starting an attribute’s reply, the module must make len*2 calls to other ReplyWith* style functions in order to emit the elements of the attribute map. See Reply APIs section for more details.

Use RedictModule_ReplySetAttributeLength() to set deferred length.

Not supported by RESP2 and will return REDICTMODULE_ERR, otherwise the function always returns REDICTMODULE_OK.

RedictModule_ReplyWithNullArray #

int RedictModule_ReplyWithNullArray(RedictModuleCtx *ctx);

Available since: Redict 7.3.0

Reply to the client with a null array, simply null in RESP3, null array in RESP2.

Note: In RESP3 there’s no difference between Null reply and NullArray reply, so to prevent ambiguity it’s better to avoid using this API and use RedictModule_ReplyWithNull instead.

The function always returns REDICTMODULE_OK.

RedictModule_ReplyWithEmptyArray #

int RedictModule_ReplyWithEmptyArray(RedictModuleCtx *ctx);

Available since: Redict 7.3.0

Reply to the client with an empty array.

The function always returns REDICTMODULE_OK.

RedictModule_ReplySetArrayLength #

void RedictModule_ReplySetArrayLength(RedictModuleCtx *ctx, long len);

Available since: Redict 7.3.0

When RedictModule_ReplyWithArray() is used with the argument REDICTMODULE_POSTPONED_LEN, because we don’t know beforehand the number of items we are going to output as elements of the array, this function will take care to set the array length.

Since it is possible to have multiple array replies pending with unknown length, this function guarantees to always set the latest array length that was created in a postponed way.

For example in order to output an array like [1,[10,20,30]] we could write:

 RedictModule_ReplyWithArray(ctx,REDICTMODULE_POSTPONED_LEN);
 RedictModule_ReplyWithLongLong(ctx,1);
 RedictModule_ReplyWithArray(ctx,REDICTMODULE_POSTPONED_LEN);
 RedictModule_ReplyWithLongLong(ctx,10);
 RedictModule_ReplyWithLongLong(ctx,20);
 RedictModule_ReplyWithLongLong(ctx,30);
 RedictModule_ReplySetArrayLength(ctx,3); // Set len of 10,20,30 array.
 RedictModule_ReplySetArrayLength(ctx,2); // Set len of top array

Note that in the above example there is no reason to postpone the array length, since we produce a fixed number of elements, but in the practice the code may use an iterator or other ways of creating the output so that is not easy to calculate in advance the number of elements.

RedictModule_ReplySetMapLength #

void RedictModule_ReplySetMapLength(RedictModuleCtx *ctx, long len);

Available since: Redict 7.3.0

Very similar to RedictModule_ReplySetArrayLength except len should exactly half of the number of ReplyWith* functions called in the context of the map. Visit https://github.com/antirez/RESP3/blob/master/spec.md for more info about RESP3.

RedictModule_ReplySetSetLength #

void RedictModule_ReplySetSetLength(RedictModuleCtx *ctx, long len);

Available since: Redict 7.3.0

Very similar to RedictModule_ReplySetArrayLength Visit https://github.com/antirez/RESP3/blob/master/spec.md for more info about RESP3.

RedictModule_ReplySetAttributeLength #

void RedictModule_ReplySetAttributeLength(RedictModuleCtx *ctx, long len);

Available since: Redict 7.3.0

Very similar to RedictModule_ReplySetMapLength Visit https://github.com/antirez/RESP3/blob/master/spec.md for more info about RESP3.

Must not be called if RedictModule_ReplyWithAttribute returned an error.

RedictModule_ReplyWithStringBuffer #

int RedictModule_ReplyWithStringBuffer(RedictModuleCtx *ctx,
                                       const char *buf,
                                       size_t len);

Available since: Redict 7.3.0

Reply with a bulk string, taking in input a C buffer pointer and length.

The function always returns REDICTMODULE_OK.

RedictModule_ReplyWithCString #

int RedictModule_ReplyWithCString(RedictModuleCtx *ctx, const char *buf);

Available since: Redict 7.3.0

Reply with a bulk string, taking in input a C buffer pointer that is assumed to be null-terminated.

The function always returns REDICTMODULE_OK.

RedictModule_ReplyWithString #

int RedictModule_ReplyWithString(RedictModuleCtx *ctx,
                                 RedictModuleString *str);

Available since: Redict 7.3.0

Reply with a bulk string, taking in input a RedictModuleString object.

The function always returns REDICTMODULE_OK.

RedictModule_ReplyWithEmptyString #

int RedictModule_ReplyWithEmptyString(RedictModuleCtx *ctx);

Available since: Redict 7.3.0

Reply with an empty string.

The function always returns REDICTMODULE_OK.

RedictModule_ReplyWithVerbatimStringType #

int RedictModule_ReplyWithVerbatimStringType(RedictModuleCtx *ctx,
                                             const char *buf,
                                             size_t len,
                                             const char *ext);

Available since: Redict 7.3.0

Reply with a binary safe string, which should not be escaped or filtered taking in input a C buffer pointer, length and a 3 character type/extension.

The function always returns REDICTMODULE_OK.

RedictModule_ReplyWithVerbatimString #

int RedictModule_ReplyWithVerbatimString(RedictModuleCtx *ctx,
                                         const char *buf,
                                         size_t len);

Available since: Redict 7.3.0

Reply with a binary safe string, which should not be escaped or filtered taking in input a C buffer pointer and length.

The function always returns REDICTMODULE_OK.

RedictModule_ReplyWithNull #

int RedictModule_ReplyWithNull(RedictModuleCtx *ctx);

Available since: Redict 7.3.0

Reply to the client with a NULL.

The function always returns REDICTMODULE_OK.

RedictModule_ReplyWithBool #

int RedictModule_ReplyWithBool(RedictModuleCtx *ctx, int b);

Available since: Redict 7.3.0

Reply with a RESP3 Boolean type. Visit https://github.com/antirez/RESP3/blob/master/spec.md for more info about RESP3.

In RESP3, this is boolean type In RESP2, it’s a string response of “1” and “0” for true and false respectively.

The function always returns REDICTMODULE_OK.

RedictModule_ReplyWithCallReply #

int RedictModule_ReplyWithCallReply(RedictModuleCtx *ctx,
                                    RedictModuleCallReply *reply);

Available since: Redict 7.3.0

Reply exactly what a Redict command returned us with RedictModule_Call(). This function is useful when we use RedictModule_Call() in order to execute some command, as we want to reply to the client exactly the same reply we obtained by the command.

Return:

  • REDICTMODULE_OK on success.
  • REDICTMODULE_ERR if the given reply is in RESP3 format but the client expects RESP2. In case of an error, it’s the module writer responsibility to translate the reply to RESP2 (or handle it differently by returning an error). Notice that for module writer convenience, it is possible to pass 0 as a parameter to the fmt argument of RedictModule_Call so that the RedictModuleCallReply will return in the same protocol (RESP2 or RESP3) as set in the current client’s context.

RedictModule_ReplyWithDouble #

int RedictModule_ReplyWithDouble(RedictModuleCtx *ctx, double d);

Available since: Redict 7.3.0

Reply with a RESP3 Double type. Visit https://github.com/antirez/RESP3/blob/master/spec.md for more info about RESP3.

Send a string reply obtained converting the double ’d’ into a bulk string. This function is basically equivalent to converting a double into a string into a C buffer, and then calling the function RedictModule_ReplyWithStringBuffer() with the buffer and length.

In RESP3 the string is tagged as a double, while in RESP2 it’s just a plain string that the user will have to parse.

The function always returns REDICTMODULE_OK.

RedictModule_ReplyWithBigNumber #

int RedictModule_ReplyWithBigNumber(RedictModuleCtx *ctx,
                                    const char *bignum,
                                    size_t len);

Available since: Redict 7.3.0

Reply with a RESP3 BigNumber type. Visit https://github.com/antirez/RESP3/blob/master/spec.md for more info about RESP3.

In RESP3, this is a string of length len that is tagged as a BigNumber, however, it’s up to the caller to ensure that it’s a valid BigNumber. In RESP2, this is just a plain bulk string response.

The function always returns REDICTMODULE_OK.

RedictModule_ReplyWithLongDouble #

int RedictModule_ReplyWithLongDouble(RedictModuleCtx *ctx, long double ld);

Available since: Redict 7.3.0

Send a string reply obtained converting the long double ’ld’ into a bulk string. This function is basically equivalent to converting a long double into a string into a C buffer, and then calling the function RedictModule_ReplyWithStringBuffer() with the buffer and length. The double string uses human readable formatting (see addReplyHumanLongDouble in networking.c).

The function always returns REDICTMODULE_OK.

Commands replication API #

RedictModule_Replicate #

int RedictModule_Replicate(RedictModuleCtx *ctx,
                           const char *cmdname,
                           const char *fmt,
                           ...);

Available since: Redict 7.3.0

Replicate the specified command and arguments to slaves and AOF, as effect of execution of the calling command implementation.

The replicated commands are always wrapped into the MULTI/EXEC that contains all the commands replicated in a given module command execution. However the commands replicated with RedictModule_Call() are the first items, the ones replicated with RedictModule_Replicate() will all follow before the EXEC.

Modules should try to use one interface or the other.

This command follows exactly the same interface of RedictModule_Call(), so a set of format specifiers must be passed, followed by arguments matching the provided format specifiers.

Please refer to RedictModule_Call() for more information.

Using the special “A” and “R” modifiers, the caller can exclude either the AOF or the replicas from the propagation of the specified command. Otherwise, by default, the command will be propagated in both channels.

Note about calling this function from a thread safe context: #

Normally when you call this function from the callback implementing a module command, or any other callback provided by the Redict Module API, Redict will accumulate all the calls to this function in the context of the callback, and will propagate all the commands wrapped in a MULTI/EXEC transaction. However when calling this function from a threaded safe context that can live an undefined amount of time, and can be locked/unlocked in at will, the behavior is different: MULTI/EXEC wrapper is not emitted and the command specified is inserted in the AOF and replication stream immediately.

Return value #

The command returns REDICTMODULE_ERR if the format specifiers are invalid or the command name does not belong to a known command.

RedictModule_ReplicateVerbatim #

int RedictModule_ReplicateVerbatim(RedictModuleCtx *ctx);

Available since: Redict 7.3.0

This function will replicate the command exactly as it was invoked by the client. Note that this function will not wrap the command into a MULTI/EXEC stanza, so it should not be mixed with other replication commands.

Basically this form of replication is useful when you want to propagate the command to the slaves and AOF file exactly as it was called, since the command can just be re-executed to deterministically re-create the new state starting from the old one.

The function always returns REDICTMODULE_OK.

DB and Key APIs – Generic API #

RedictModule_GetClientId #

unsigned long long RedictModule_GetClientId(RedictModuleCtx *ctx);

Available since: Redict 7.3.0

Return the ID of the current client calling the currently active module command. The returned ID has a few guarantees:

  1. The ID is different for each different client, so if the same client executes a module command multiple times, it can be recognized as having the same ID, otherwise the ID will be different.
  2. The ID increases monotonically. Clients connecting to the server later are guaranteed to get IDs greater than any past ID previously seen.

Valid IDs are from 1 to 2^64 - 1. If 0 is returned it means there is no way to fetch the ID in the context the function was currently called.

After obtaining the ID, it is possible to check if the command execution is actually happening in the context of AOF loading, using this macro:

 if (RedictModule_IsAOFClient(RedictModule_GetClientId(ctx)) {
     // Handle it differently.
 }

RedictModule_GetClientUserNameById #

RedictModuleString *RedictModule_GetClientUserNameById(RedictModuleCtx *ctx,
                                                       uint64_t id);

Available since: Redict 7.3.0

Return the ACL user name used by the client with the specified client ID. Client ID can be obtained with RedictModule_GetClientId() API. If the client does not exist, NULL is returned and errno is set to ENOENT. If the client isn’t using an ACL user, NULL is returned and errno is set to ENOTSUP

RedictModule_GetClientInfoById #

int RedictModule_GetClientInfoById(void *ci, uint64_t id);

Available since: Redict 7.3.0

Return information about the client with the specified ID (that was previously obtained via the RedictModule_GetClientId() API). If the client exists, REDICTMODULE_OK is returned, otherwise REDICTMODULE_ERR is returned.

When the client exist and the ci pointer is not NULL, but points to a structure of type RedictModuleClientInfoV1, previously initialized with the correct REDICTMODULE_CLIENTINFO_INITIALIZER_V1, the structure is populated with the following fields:

 uint64_t flags;         // REDICTMODULE_CLIENTINFO_FLAG_*
 uint64_t id;            // Client ID
 char addr[46];          // IPv4 or IPv6 address.
 uint16_t port;          // TCP port.
 uint16_t db;            // Selected DB.

Note: the client ID is useless in the context of this call, since we already know, however the same structure could be used in other contexts where we don’t know the client ID, yet the same structure is returned.

With flags having the following meaning:

REDICTMODULE_CLIENTINFO_FLAG_SSL          Client using SSL connection.
REDICTMODULE_CLIENTINFO_FLAG_PUBSUB       Client in Pub/Sub mode.
REDICTMODULE_CLIENTINFO_FLAG_BLOCKED      Client blocked in command.
REDICTMODULE_CLIENTINFO_FLAG_TRACKING     Client with keys tracking on.
REDICTMODULE_CLIENTINFO_FLAG_UNIXSOCKET   Client using unix domain socket.
REDICTMODULE_CLIENTINFO_FLAG_MULTI        Client in MULTI state.

However passing NULL is a way to just check if the client exists in case we are not interested in any additional information.

This is the correct usage when we want the client info structure returned:

 RedictModuleClientInfo ci = REDICTMODULE_CLIENTINFO_INITIALIZER;
 int retval = RedictModule_GetClientInfoById(&ci,client_id);
 if (retval == REDICTMODULE_OK) {
     printf("Address: %s\n", ci.addr);
 }

RedictModule_GetClientNameById #

RedictModuleString *RedictModule_GetClientNameById(RedictModuleCtx *ctx,
                                                   uint64_t id);

Available since: Redict 7.3.0

Returns the name of the client connection with the given ID.

If the client ID does not exist or if the client has no name associated with it, NULL is returned.

RedictModule_SetClientNameById #

int RedictModule_SetClientNameById(uint64_t id, RedictModuleString *name);

Available since: Redict 7.3.0

Sets the name of the client with the given ID. This is equivalent to the client calling CLIENT SETNAME name.

Returns REDICTMODULE_OK on success. On failure, REDICTMODULE_ERR is returned and errno is set as follows:

  • ENOENT if the client does not exist
  • EINVAL if the name contains invalid characters

RedictModule_PublishMessage #

int RedictModule_PublishMessage(RedictModuleCtx *ctx,
                                RedictModuleString *channel,
                                RedictModuleString *message);

Available since: Redict 7.3.0

Publish a message to subscribers (see PUBLISH command).

RedictModule_PublishMessageShard #

int RedictModule_PublishMessageShard(RedictModuleCtx *ctx,
                                     RedictModuleString *channel,
                                     RedictModuleString *message);

Available since: Redict 7.3.0

Publish a message to shard-subscribers (see SPUBLISH command).

RedictModule_GetSelectedDb #

int RedictModule_GetSelectedDb(RedictModuleCtx *ctx);

Available since: Redict 7.3.0

Return the currently selected DB.

RedictModule_GetContextFlags #

int RedictModule_GetContextFlags(RedictModuleCtx *ctx);

Available since: Redict 7.3.0

Return the current context’s flags. The flags provide information on the current request context (whether the client is a Lua script or in a MULTI), and about the Redict instance in general, i.e replication and persistence.

It is possible to call this function even with a NULL context, however in this case the following flags will not be reported:

  • LUA, MULTI, REPLICATED, DIRTY (see below for more info).

Available flags and their meaning:

  • REDICTMODULE_CTX_FLAGS_LUA: The command is running in a Lua script

  • REDICTMODULE_CTX_FLAGS_MULTI: The command is running inside a transaction

  • REDICTMODULE_CTX_FLAGS_REPLICATED: The command was sent over the replication link by the MASTER

  • REDICTMODULE_CTX_FLAGS_MASTER: The Redict instance is a master

  • REDICTMODULE_CTX_FLAGS_SLAVE: The Redict instance is a slave

  • REDICTMODULE_CTX_FLAGS_READONLY: The Redict instance is read-only

  • REDICTMODULE_CTX_FLAGS_CLUSTER: The Redict instance is in cluster mode

  • REDICTMODULE_CTX_FLAGS_AOF: The Redict instance has AOF enabled

  • REDICTMODULE_CTX_FLAGS_RDB: The instance has RDB enabled

  • REDICTMODULE_CTX_FLAGS_MAXMEMORY: The instance has Maxmemory set

  • REDICTMODULE_CTX_FLAGS_EVICT: Maxmemory is set and has an eviction policy that may delete keys

  • REDICTMODULE_CTX_FLAGS_OOM: Redict is out of memory according to the maxmemory setting.

  • REDICTMODULE_CTX_FLAGS_OOM_WARNING: Less than 25% of memory remains before reaching the maxmemory level.

  • REDICTMODULE_CTX_FLAGS_LOADING: Server is loading RDB/AOF

  • REDICTMODULE_CTX_FLAGS_REPLICA_IS_STALE: No active link with the master.

  • REDICTMODULE_CTX_FLAGS_REPLICA_IS_CONNECTING: The replica is trying to connect with the master.

  • REDICTMODULE_CTX_FLAGS_REPLICA_IS_TRANSFERRING: Master -> Replica RDB transfer is in progress.

  • REDICTMODULE_CTX_FLAGS_REPLICA_IS_ONLINE: The replica has an active link with its master. This is the contrary of STALE state.

  • REDICTMODULE_CTX_FLAGS_ACTIVE_CHILD: There is currently some background process active (RDB, AUX or module).

  • REDICTMODULE_CTX_FLAGS_MULTI_DIRTY: The next EXEC will fail due to dirty CAS (touched keys).

  • REDICTMODULE_CTX_FLAGS_IS_CHILD: Redict is currently running inside background child process.

  • REDICTMODULE_CTX_FLAGS_RESP3: Indicate the that client attached to this context is using RESP3.

  • REDICTMODULE_CTX_FLAGS_SERVER_STARTUP: The Redict instance is starting

RedictModule_AvoidReplicaTraffic #

int RedictModule_AvoidReplicaTraffic(void);

Available since: Redict 7.3.0

Returns true if a client sent the CLIENT PAUSE command to the server or if Redict Cluster does a manual failover, pausing the clients. This is needed when we have a master with replicas, and want to write, without adding further data to the replication channel, that the replicas replication offset, match the one of the master. When this happens, it is safe to failover the master without data loss.

However modules may generate traffic by calling RedictModule_Call() with the “!” flag, or by calling RedictModule_Replicate(), in a context outside commands execution, for instance in timeout callbacks, threads safe contexts, and so forth. When modules will generate too much traffic, it will be hard for the master and replicas offset to match, because there is more data to send in the replication channel.

So modules may want to try to avoid very heavy background work that has the effect of creating data to the replication channel, when this function returns true. This is mostly useful for modules that have background garbage collection tasks, or that do writes and replicate such writes periodically in timer callbacks or other periodic callbacks.

RedictModule_SelectDb #

int RedictModule_SelectDb(RedictModuleCtx *ctx, int newid);

Available since: Redict 7.3.0

Change the currently selected DB. Returns an error if the id is out of range.

Note that the client will retain the currently selected DB even after the Redict command implemented by the module calling this function returns.

If the module command wishes to change something in a different DB and returns back to the original one, it should call RedictModule_GetSelectedDb() before in order to restore the old DB number before returning.

RedictModule_KeyExists #

int RedictModule_KeyExists(RedictModuleCtx *ctx, robj *keyname);

Available since: Redict 7.3.0

Check if a key exists, without affecting its last access time.

This is equivalent to calling RedictModule_OpenKey with the mode REDICTMODULE_READ | REDICTMODULE_OPEN_KEY_NOTOUCH, then checking if NULL was returned and, if not, calling RedictModule_CloseKey on the opened key.

RedictModule_OpenKey #

RedictModuleKey *RedictModule_OpenKey(RedictModuleCtx *ctx,
                                      robj *keyname,
                                      int mode);

Available since: Redict 7.3.0

Return a handle representing a Redict key, so that it is possible to call other APIs with the key handle as argument to perform operations on the key.

The return value is the handle representing the key, that must be closed with RedictModule_CloseKey().

If the key does not exist and REDICTMODULE_WRITE mode is requested, the handle is still returned, since it is possible to perform operations on a yet not existing key (that will be created, for example, after a list push operation). If the mode is just REDICTMODULE_READ instead, and the key does not exist, NULL is returned. However it is still safe to call RedictModule_CloseKey() and RedictModule_KeyType() on a NULL value.

Extra flags that can be pass to the API under the mode argument:

  • REDICTMODULE_OPEN_KEY_NOTOUCH - Avoid touching the LRU/LFU of the key when opened.
  • REDICTMODULE_OPEN_KEY_NONOTIFY - Don’t trigger keyspace event on key misses.
  • REDICTMODULE_OPEN_KEY_NOSTATS - Don’t update keyspace hits/misses counters.
  • REDICTMODULE_OPEN_KEY_NOEXPIRE - Avoid deleting lazy expired keys.
  • REDICTMODULE_OPEN_KEY_NOEFFECTS - Avoid any effects from fetching the key.

RedictModule_GetOpenKeyModesAll #

int RedictModule_GetOpenKeyModesAll(void);

Available since: Redict 7.3.0

Returns the full OpenKey modes mask, using the return value the module can check if a certain set of OpenKey modes are supported by the redict server version in use. Example:

   int supportedMode = RedictModule_GetOpenKeyModesAll();
   if (supportedMode & REDICTMODULE_OPEN_KEY_NOTOUCH) {
         // REDICTMODULE_OPEN_KEY_NOTOUCH is supported
   } else{
         // REDICTMODULE_OPEN_KEY_NOTOUCH is not supported
   }

RedictModule_CloseKey #

void RedictModule_CloseKey(RedictModuleKey *key);

Available since: Redict 7.3.0

Close a key handle.

RedictModule_KeyType #

int RedictModule_KeyType(RedictModuleKey *key);

Available since: Redict 7.3.0

Return the type of the key. If the key pointer is NULL then REDICTMODULE_KEYTYPE_EMPTY is returned.

RedictModule_ValueLength #

size_t RedictModule_ValueLength(RedictModuleKey *key);

Available since: Redict 7.3.0

Return the length of the value associated with the key. For strings this is the length of the string. For all the other types is the number of elements (just counting keys for hashes).

If the key pointer is NULL or the key is empty, zero is returned.

RedictModule_DeleteKey #

int RedictModule_DeleteKey(RedictModuleKey *key);

Available since: Redict 7.3.0

If the key is open for writing, remove it, and setup the key to accept new writes as an empty key (that will be created on demand). On success REDICTMODULE_OK is returned. If the key is not open for writing REDICTMODULE_ERR is returned.

RedictModule_UnlinkKey #

int RedictModule_UnlinkKey(RedictModuleKey *key);

Available since: Redict 7.3.0

If the key is open for writing, unlink it (that is delete it in a non-blocking way, not reclaiming memory immediately) and setup the key to accept new writes as an empty key (that will be created on demand). On success REDICTMODULE_OK is returned. If the key is not open for writing REDICTMODULE_ERR is returned.

RedictModule_GetExpire #

mstime_t RedictModule_GetExpire(RedictModuleKey *key);

Available since: Redict 7.3.0

Return the key expire value, as milliseconds of remaining TTL. If no TTL is associated with the key or if the key is empty, REDICTMODULE_NO_EXPIRE is returned.

RedictModule_SetExpire #

int RedictModule_SetExpire(RedictModuleKey *key, mstime_t expire);

Available since: Redict 7.3.0

Set a new expire for the key. If the special expire REDICTMODULE_NO_EXPIRE is set, the expire is cancelled if there was one (the same as the PERSIST command).

Note that the expire must be provided as a positive integer representing the number of milliseconds of TTL the key should have.

The function returns REDICTMODULE_OK on success or REDICTMODULE_ERR if the key was not open for writing or is an empty key.

RedictModule_GetAbsExpire #

mstime_t RedictModule_GetAbsExpire(RedictModuleKey *key);

Available since: Redict 7.3.0

Return the key expire value, as absolute Unix timestamp. If no TTL is associated with the key or if the key is empty, REDICTMODULE_NO_EXPIRE is returned.

RedictModule_SetAbsExpire #

int RedictModule_SetAbsExpire(RedictModuleKey *key, mstime_t expire);

Available since: Redict 7.3.0

Set a new expire for the key. If the special expire REDICTMODULE_NO_EXPIRE is set, the expire is cancelled if there was one (the same as the PERSIST command).

Note that the expire must be provided as a positive integer representing the absolute Unix timestamp the key should have.

The function returns REDICTMODULE_OK on success or REDICTMODULE_ERR if the key was not open for writing or is an empty key.

RedictModule_ResetDataset #

void RedictModule_ResetDataset(int restart_aof, int async);

Available since: Redict 7.3.0

Performs similar operation to FLUSHALL, and optionally start a new AOF file (if enabled) If restart_aof is true, you must make sure the command that triggered this call is not propagated to the AOF file. When async is set to true, db contents will be freed by a background thread.

RedictModule_DbSize #

unsigned long long RedictModule_DbSize(RedictModuleCtx *ctx);

Available since: Redict 7.3.0

Returns the number of keys in the current db.

RedictModule_RandomKey #

RedictModuleString *RedictModule_RandomKey(RedictModuleCtx *ctx);

Available since: Redict 7.3.0

Returns a name of a random key, or NULL if current db is empty.

RedictModule_GetKeyNameFromOptCtx #

const RedictModuleString *RedictModule_GetKeyNameFromOptCtx(RedictModuleKeyOptCtx *ctx);

Available since: Redict 7.3.0

Returns the name of the key currently being processed.

RedictModule_GetToKeyNameFromOptCtx #

const RedictModuleString *RedictModule_GetToKeyNameFromOptCtx(RedictModuleKeyOptCtx *ctx);

Available since: Redict 7.3.0

Returns the name of the target key currently being processed.

RedictModule_GetDbIdFromOptCtx #

int RedictModule_GetDbIdFromOptCtx(RedictModuleKeyOptCtx *ctx);

Available since: Redict 7.3.0

Returns the dbid currently being processed.

RedictModule_GetToDbIdFromOptCtx #

int RedictModule_GetToDbIdFromOptCtx(RedictModuleKeyOptCtx *ctx);

Available since: Redict 7.3.0

Returns the target dbid currently being processed.

Key API for String type #

See also RedictModule_ValueLength(), which returns the length of a string.

RedictModule_StringSet #

int RedictModule_StringSet(RedictModuleKey *key, RedictModuleString *str);

Available since: Redict 7.3.0

If the key is open for writing, set the specified string ‘str’ as the value of the key, deleting the old value if any. On success REDICTMODULE_OK is returned. If the key is not open for writing or there is an active iterator, REDICTMODULE_ERR is returned.

RedictModule_StringDMA #

char *RedictModule_StringDMA(RedictModuleKey *key, size_t *len, int mode);

Available since: Redict 7.3.0

Prepare the key associated string value for DMA access, and returns a pointer and size (by reference), that the user can use to read or modify the string in-place accessing it directly via pointer.

The ‘mode’ is composed by bitwise OR-ing the following flags:

REDICTMODULE_READ -- Read access
REDICTMODULE_WRITE -- Write access

If the DMA is not requested for writing, the pointer returned should only be accessed in a read-only fashion.

On error (wrong type) NULL is returned.

DMA access rules:

  1. No other key writing function should be called since the moment the pointer is obtained, for all the time we want to use DMA access to read or modify the string.

  2. Each time RedictModule_StringTruncate() is called, to continue with the DMA access, RedictModule_StringDMA() should be called again to re-obtain a new pointer and length.

  3. If the returned pointer is not NULL, but the length is zero, no byte can be touched (the string is empty, or the key itself is empty) so a RedictModule_StringTruncate() call should be used if there is to enlarge the string, and later call StringDMA() again to get the pointer.

RedictModule_StringTruncate #

int RedictModule_StringTruncate(RedictModuleKey *key, size_t newlen);

Available since: Redict 7.3.0

If the key is open for writing and is of string type, resize it, padding with zero bytes if the new length is greater than the old one.

After this call, RedictModule_StringDMA() must be called again to continue DMA access with the new pointer.

The function returns REDICTMODULE_OK on success, and REDICTMODULE_ERR on error, that is, the key is not open for writing, is not a string or resizing for more than 512 MB is requested.

If the key is empty, a string key is created with the new string value unless the new length value requested is zero.

Key API for List type #

Many of the list functions access elements by index. Since a list is in essence a doubly-linked list, accessing elements by index is generally an O(N) operation. However, if elements are accessed sequentially or with indices close together, the functions are optimized to seek the index from the previous index, rather than seeking from the ends of the list.

This enables iteration to be done efficiently using a simple for loop:

long n = RedictModule_ValueLength(key);
for (long i = 0; i < n; i++) {
    RedictModuleString *elem = RedictModule_ListGet(key, i);
    // Do stuff...
}

Note that after modifying a list using RedictModule_ListPop, RedictModule_ListSet or RedictModule_ListInsert, the internal iterator is invalidated so the next operation will require a linear seek.

Modifying a list in any another way, for example using RedictModule_Call(), while a key is open will confuse the internal iterator and may cause trouble if the key is used after such modifications. The key must be reopened in this case.

See also RedictModule_ValueLength(), which returns the length of a list.

RedictModule_ListPush #

int RedictModule_ListPush(RedictModuleKey *key,
                          int where,
                          RedictModuleString *ele);

Available since: Redict 7.3.0

Push an element into a list, on head or tail depending on ‘where’ argument (REDICTMODULE_LIST_HEAD or REDICTMODULE_LIST_TAIL). If the key refers to an empty key opened for writing, the key is created. On success, REDICTMODULE_OK is returned. On failure, REDICTMODULE_ERR is returned and errno is set as follows:

  • EINVAL if key or ele is NULL.
  • ENOTSUP if the key is of another type than list.
  • EBADF if the key is not opened for writing.

Note: Before Redis 7.0, errno was not set by this function.

RedictModule_ListPop #

RedictModuleString *RedictModule_ListPop(RedictModuleKey *key, int where);

Available since: Redict 7.3.0

Pop an element from the list, and returns it as a module string object that the user should be free with RedictModule_FreeString() or by enabling automatic memory. The where argument specifies if the element should be popped from the beginning or the end of the list (REDICTMODULE_LIST_HEAD or REDICTMODULE_LIST_TAIL). On failure, the command returns NULL and sets errno as follows:

  • EINVAL if key is NULL.
  • ENOTSUP if the key is empty or of another type than list.
  • EBADF if the key is not opened for writing.

Note: Before Redis 7.0, errno was not set by this function.

RedictModule_ListGet #

RedictModuleString *RedictModule_ListGet(RedictModuleKey *key, long index);

Available since: Redict 7.3.0

Returns the element at index index in the list stored at key, like the LINDEX command. The element should be free’d using RedictModule_FreeString() or using automatic memory management.

The index is zero-based, so 0 means the first element, 1 the second element and so on. Negative indices can be used to designate elements starting at the tail of the list. Here, -1 means the last element, -2 means the penultimate and so forth.

When no value is found at the given key and index, NULL is returned and errno is set as follows:

  • EINVAL if key is NULL.
  • ENOTSUP if the key is not a list.
  • EBADF if the key is not opened for reading.
  • EDOM if the index is not a valid index in the list.

RedictModule_ListSet #

int RedictModule_ListSet(RedictModuleKey *key,
                         long index,
                         RedictModuleString *value);

Available since: Redict 7.3.0

Replaces the element at index index in the list stored at key.

The index is zero-based, so 0 means the first element, 1 the second element and so on. Negative indices can be used to designate elements starting at the tail of the list. Here, -1 means the last element, -2 means the penultimate and so forth.

On success, REDICTMODULE_OK is returned. On failure, REDICTMODULE_ERR is returned and errno is set as follows:

  • EINVAL if key or value is NULL.
  • ENOTSUP if the key is not a list.
  • EBADF if the key is not opened for writing.
  • EDOM if the index is not a valid index in the list.

RedictModule_ListInsert #

int RedictModule_ListInsert(RedictModuleKey *key,
                            long index,
                            RedictModuleString *value);

Available since: Redict 7.3.0

Inserts an element at the given index.

The index is zero-based, so 0 means the first element, 1 the second element and so on. Negative indices can be used to designate elements starting at the tail of the list. Here, -1 means the last element, -2 means the penultimate and so forth. The index is the element’s index after inserting it.

On success, REDICTMODULE_OK is returned. On failure, REDICTMODULE_ERR is returned and errno is set as follows:

  • EINVAL if key or value is NULL.
  • ENOTSUP if the key of another type than list.
  • EBADF if the key is not opened for writing.
  • EDOM if the index is not a valid index in the list.

RedictModule_ListDelete #

int RedictModule_ListDelete(RedictModuleKey *key, long index);

Available since: Redict 7.3.0

Removes an element at the given index. The index is 0-based. A negative index can also be used, counting from the end of the list.

On success, REDICTMODULE_OK is returned. On failure, REDICTMODULE_ERR is returned and errno is set as follows:

  • EINVAL if key or value is NULL.
  • ENOTSUP if the key is not a list.
  • EBADF if the key is not opened for writing.
  • EDOM if the index is not a valid index in the list.

Key API for Sorted Set type #

See also RedictModule_ValueLength(), which returns the length of a sorted set.

RedictModule_ZsetAdd #

int RedictModule_ZsetAdd(RedictModuleKey *key,
                         double score,
                         RedictModuleString *ele,
                         int *flagsptr);

Available since: Redict 7.3.0

Add a new element into a sorted set, with the specified ‘score’. If the element already exists, the score is updated.

A new sorted set is created at value if the key is an empty open key setup for writing.

Additional flags can be passed to the function via a pointer, the flags are both used to receive input and to communicate state when the function returns. ‘flagsptr’ can be NULL if no special flags are used.

The input flags are:

REDICTMODULE_ZADD_XX: Element must already exist. Do nothing otherwise.
REDICTMODULE_ZADD_NX: Element must not exist. Do nothing otherwise.
REDICTMODULE_ZADD_GT: If element exists, new score must be greater than the current score.
                     Do nothing otherwise. Can optionally be combined with XX.
REDICTMODULE_ZADD_LT: If element exists, new score must be less than the current score.
                     Do nothing otherwise. Can optionally be combined with XX.

The output flags are:

REDICTMODULE_ZADD_ADDED: The new element was added to the sorted set.
REDICTMODULE_ZADD_UPDATED: The score of the element was updated.
REDICTMODULE_ZADD_NOP: No operation was performed because XX or NX flags.

On success the function returns REDICTMODULE_OK. On the following errors REDICTMODULE_ERR is returned:

  • The key was not opened for writing.
  • The key is of the wrong type.
  • ‘score’ double value is not a number (NaN).

RedictModule_ZsetIncrby #

int RedictModule_ZsetIncrby(RedictModuleKey *key,
                            double score,
                            RedictModuleString *ele,
                            int *flagsptr,
                            double *newscore);

Available since: Redict 7.3.0

This function works exactly like RedictModule_ZsetAdd(), but instead of setting a new score, the score of the existing element is incremented, or if the element does not already exist, it is added assuming the old score was zero.

The input and output flags, and the return value, have the same exact meaning, with the only difference that this function will return REDICTMODULE_ERR even when ‘score’ is a valid double number, but adding it to the existing score results into a NaN (not a number) condition.

This function has an additional field ’newscore’, if not NULL is filled with the new score of the element after the increment, if no error is returned.

RedictModule_ZsetRem #

int RedictModule_ZsetRem(RedictModuleKey *key,
                         RedictModuleString *ele,
                         int *deleted);

Available since: Redict 7.3.0

Remove the specified element from the sorted set. The function returns REDICTMODULE_OK on success, and REDICTMODULE_ERR on one of the following conditions:

  • The key was not opened for writing.
  • The key is of the wrong type.

The return value does NOT indicate the fact the element was really removed (since it existed) or not, just if the function was executed with success.

In order to know if the element was removed, the additional argument ‘deleted’ must be passed, that populates the integer by reference setting it to 1 or 0 depending on the outcome of the operation. The ‘deleted’ argument can be NULL if the caller is not interested to know if the element was really removed.

Empty keys will be handled correctly by doing nothing.

RedictModule_ZsetScore #

int RedictModule_ZsetScore(RedictModuleKey *key,
                           RedictModuleString *ele,
                           double *score);

Available since: Redict 7.3.0

On success retrieve the double score associated at the sorted set element ’ele’ and returns REDICTMODULE_OK. Otherwise REDICTMODULE_ERR is returned to signal one of the following conditions:

  • There is no such element ’ele’ in the sorted set.
  • The key is not a sorted set.
  • The key is an open empty key.

Key API for Sorted Set iterator #

RedictModule_ZsetRangeStop #

void RedictModule_ZsetRangeStop(RedictModuleKey *key);

Available since: Redict 7.3.0

Stop a sorted set iteration.

RedictModule_ZsetRangeEndReached #

int RedictModule_ZsetRangeEndReached(RedictModuleKey *key);

Available since: Redict 7.3.0

Return the “End of range” flag value to signal the end of the iteration.

RedictModule_ZsetFirstInScoreRange #

int RedictModule_ZsetFirstInScoreRange(RedictModuleKey *key,
                                       double min,
                                       double max,
                                       int minex,
                                       int maxex);

Available since: Redict 7.3.0

Setup a sorted set iterator seeking the first element in the specified range. Returns REDICTMODULE_OK if the iterator was correctly initialized otherwise REDICTMODULE_ERR is returned in the following conditions:

  1. The value stored at key is not a sorted set or the key is empty.

The range is specified according to the two double values ‘min’ and ‘max’. Both can be infinite using the following two macros:

  • REDICTMODULE_POSITIVE_INFINITE for positive infinite value
  • REDICTMODULE_NEGATIVE_INFINITE for negative infinite value

‘minex’ and ‘maxex’ parameters, if true, respectively setup a range where the min and max value are exclusive (not included) instead of inclusive.

RedictModule_ZsetLastInScoreRange #

int RedictModule_ZsetLastInScoreRange(RedictModuleKey *key,
                                      double min,
                                      double max,
                                      int minex,
                                      int maxex);

Available since: Redict 7.3.0

Exactly like RedictModule_ZsetFirstInScoreRange() but the last element of the range is selected for the start of the iteration instead.

RedictModule_ZsetFirstInLexRange #

int RedictModule_ZsetFirstInLexRange(RedictModuleKey *key,
                                     RedictModuleString *min,
                                     RedictModuleString *max);

Available since: Redict 7.3.0

Setup a sorted set iterator seeking the first element in the specified lexicographical range. Returns REDICTMODULE_OK if the iterator was correctly initialized otherwise REDICTMODULE_ERR is returned in the following conditions:

  1. The value stored at key is not a sorted set or the key is empty.
  2. The lexicographical range ‘min’ and ‘max’ format is invalid.

‘min’ and ‘max’ should be provided as two RedictModuleString objects in the same format as the parameters passed to the ZRANGEBYLEX command. The function does not take ownership of the objects, so they can be released ASAP after the iterator is setup.

RedictModule_ZsetLastInLexRange #

int RedictModule_ZsetLastInLexRange(RedictModuleKey *key,
                                    RedictModuleString *min,
                                    RedictModuleString *max);

Available since: Redict 7.3.0

Exactly like RedictModule_ZsetFirstInLexRange() but the last element of the range is selected for the start of the iteration instead.

RedictModule_ZsetRangeCurrentElement #

RedictModuleString *RedictModule_ZsetRangeCurrentElement(RedictModuleKey *key,
                                                         double *score);

Available since: Redict 7.3.0

Return the current sorted set element of an active sorted set iterator or NULL if the range specified in the iterator does not include any element.

RedictModule_ZsetRangeNext #

int RedictModule_ZsetRangeNext(RedictModuleKey *key);

Available since: Redict 7.3.0

Go to the next element of the sorted set iterator. Returns 1 if there was a next element, 0 if we are already at the latest element or the range does not include any item at all.

RedictModule_ZsetRangePrev #

int RedictModule_ZsetRangePrev(RedictModuleKey *key);

Available since: Redict 7.3.0

Go to the previous element of the sorted set iterator. Returns 1 if there was a previous element, 0 if we are already at the first element or the range does not include any item at all.

Key API for Hash type #

See also RedictModule_ValueLength(), which returns the number of fields in a hash.

RedictModule_HashSet #

int RedictModule_HashSet(RedictModuleKey *key, int flags, ...);

Available since: Redict 7.3.0

Set the field of the specified hash field to the specified value. If the key is an empty key open for writing, it is created with an empty hash value, in order to set the specified field.

The function is variadic and the user must specify pairs of field names and values, both as RedictModuleString pointers (unless the CFIELD option is set, see later). At the end of the field/value-ptr pairs, NULL must be specified as last argument to signal the end of the arguments in the variadic function.

Example to set the hash argv[1] to the value argv[2]:

 RedictModule_HashSet(key,REDICTMODULE_HASH_NONE,argv[1],argv[2],NULL);

The function can also be used in order to delete fields (if they exist) by setting them to the specified value of REDICTMODULE_HASH_DELETE:

 RedictModule_HashSet(key,REDICTMODULE_HASH_NONE,argv[1],
                     REDICTMODULE_HASH_DELETE,NULL);

The behavior of the command changes with the specified flags, that can be set to REDICTMODULE_HASH_NONE if no special behavior is needed.

REDICTMODULE_HASH_NX: The operation is performed only if the field was not
                     already existing in the hash.
REDICTMODULE_HASH_XX: The operation is performed only if the field was
                     already existing, so that a new value could be
                     associated to an existing filed, but no new fields
                     are created.
REDICTMODULE_HASH_CFIELDS: The field names passed are null terminated C
                          strings instead of RedictModuleString objects.
REDICTMODULE_HASH_COUNT_ALL: Include the number of inserted fields in the
                            returned number, in addition to the number of
                            updated and deleted fields. (Added in Redis
                            6.2.)

Unless NX is specified, the command overwrites the old field value with the new one.

When using REDICTMODULE_HASH_CFIELDS, field names are reported using normal C strings, so for example to delete the field “foo” the following code can be used:

 RedictModule_HashSet(key,REDICTMODULE_HASH_CFIELDS,"foo",
                     REDICTMODULE_HASH_DELETE,NULL);

Return value:

The number of fields existing in the hash prior to the call, which have been updated (its old value has been replaced by a new value) or deleted. If the flag REDICTMODULE_HASH_COUNT_ALL is set, inserted fields not previously existing in the hash are also counted.

If the return value is zero, errno is set (since Redis 6.2) as follows:

  • EINVAL if any unknown flags are set or if key is NULL.
  • ENOTSUP if the key is associated with a non Hash value.
  • EBADF if the key was not opened for writing.
  • ENOENT if no fields were counted as described under Return value above. This is not actually an error. The return value can be zero if all fields were just created and the COUNT_ALL flag was unset, or if changes were held back due to the NX and XX flags.

NOTICE: The return value semantics of this function are very different between Redis 6.2 and older versions. Modules that use it should determine the Redict version and handle it accordingly.

RedictModule_HashGet #

int RedictModule_HashGet(RedictModuleKey *key, int flags, ...);

Available since: Redict 7.3.0

Get fields from a hash value. This function is called using a variable number of arguments, alternating a field name (as a RedictModuleString pointer) with a pointer to a RedictModuleString pointer, that is set to the value of the field if the field exists, or NULL if the field does not exist. At the end of the field/value-ptr pairs, NULL must be specified as last argument to signal the end of the arguments in the variadic function.

This is an example usage:

 RedictModuleString *first, *second;
 RedictModule_HashGet(mykey,REDICTMODULE_HASH_NONE,argv[1],&first,
                     argv[2],&second,NULL);

As with RedictModule_HashSet() the behavior of the command can be specified passing flags different than REDICTMODULE_HASH_NONE:

REDICTMODULE_HASH_CFIELDS: field names as null terminated C strings.

REDICTMODULE_HASH_EXISTS: instead of setting the value of the field expecting a RedictModuleString pointer to pointer, the function just reports if the field exists or not and expects an integer pointer as the second element of each pair.

Example of REDICTMODULE_HASH_CFIELDS:

 RedictModuleString *username, *hashedpass;
 RedictModule_HashGet(mykey,REDICTMODULE_HASH_CFIELDS,"username",&username,"hp",&hashedpass, NULL);

Example of REDICTMODULE_HASH_EXISTS:

 int exists;
 RedictModule_HashGet(mykey,REDICTMODULE_HASH_EXISTS,argv[1],&exists,NULL);

The function returns REDICTMODULE_OK on success and REDICTMODULE_ERR if the key is not a hash value.

Memory management:

The returned RedictModuleString objects should be released with RedictModule_FreeString(), or by enabling automatic memory management.

Key API for Stream type #

For an introduction to streams, see https://redis.io/topics/streams-intro.

The type RedictModuleStreamID, which is used in stream functions, is a struct with two 64-bit fields and is defined as

typedef struct RedictModuleStreamID {
    uint64_t ms;
    uint64_t seq;
} RedictModuleStreamID;

See also RedictModule_ValueLength(), which returns the length of a stream, and the conversion functions RedictModule_StringToStreamID() and RedictModule_CreateStringFromStreamID().

RedictModule_StreamAdd #

int RedictModule_StreamAdd(RedictModuleKey *key,
                           int flags,
                           RedictModuleStreamID *id,
                           RedictModuleString **argv,
                           long numfields);

Available since: Redict 7.3.0

Adds an entry to a stream. Like XADD without trimming.

  • key: The key where the stream is (or will be) stored
  • flags: A bit field of
    • REDICTMODULE_STREAM_ADD_AUTOID: Assign a stream ID automatically, like * in the XADD command.
  • id: If the AUTOID flag is set, this is where the assigned ID is returned. Can be NULL if AUTOID is set, if you don’t care to receive the ID. If AUTOID is not set, this is the requested ID.
  • argv: A pointer to an array of size numfields * 2 containing the fields and values.
  • numfields: The number of field-value pairs in argv.

Returns REDICTMODULE_OK if an entry has been added. On failure, REDICTMODULE_ERR is returned and errno is set as follows:

  • EINVAL if called with invalid arguments
  • ENOTSUP if the key refers to a value of a type other than stream
  • EBADF if the key was not opened for writing
  • EDOM if the given ID was 0-0 or not greater than all other IDs in the stream (only if the AUTOID flag is unset)
  • EFBIG if the stream has reached the last possible ID
  • ERANGE if the elements are too large to be stored.

RedictModule_StreamDelete #

int RedictModule_StreamDelete(RedictModuleKey *key, RedictModuleStreamID *id);

Available since: Redict 7.3.0

Deletes an entry from a stream.

  • key: A key opened for writing, with no stream iterator started.
  • id: The stream ID of the entry to delete.

Returns REDICTMODULE_OK on success. On failure, REDICTMODULE_ERR is returned and errno is set as follows:

  • EINVAL if called with invalid arguments
  • ENOTSUP if the key refers to a value of a type other than stream or if the key is empty
  • EBADF if the key was not opened for writing or if a stream iterator is associated with the key
  • ENOENT if no entry with the given stream ID exists

See also RedictModule_StreamIteratorDelete() for deleting the current entry while iterating using a stream iterator.

RedictModule_StreamIteratorStart #

int RedictModule_StreamIteratorStart(RedictModuleKey *key,
                                     int flags,
                                     RedictModuleStreamID *start,
                                     RedictModuleStreamID *end);

Available since: Redict 7.3.0

Sets up a stream iterator.

  • key: The stream key opened for reading using RedictModule_OpenKey().
  • flags:
    • REDICTMODULE_STREAM_ITERATOR_EXCLUSIVE: Don’t include start and end in the iterated range.
    • REDICTMODULE_STREAM_ITERATOR_REVERSE: Iterate in reverse order, starting from the end of the range.
  • start: The lower bound of the range. Use NULL for the beginning of the stream.
  • end: The upper bound of the range. Use NULL for the end of the stream.

Returns REDICTMODULE_OK on success. On failure, REDICTMODULE_ERR is returned and errno is set as follows:

  • EINVAL if called with invalid arguments
  • ENOTSUP if the key refers to a value of a type other than stream or if the key is empty
  • EBADF if the key was not opened for writing or if a stream iterator is already associated with the key
  • EDOM if start or end is outside the valid range

Returns REDICTMODULE_OK on success and REDICTMODULE_ERR if the key doesn’t refer to a stream or if invalid arguments were given.

The stream IDs are retrieved using RedictModule_StreamIteratorNextID() and for each stream ID, the fields and values are retrieved using RedictModule_StreamIteratorNextField(). The iterator is freed by calling RedictModule_StreamIteratorStop().

Example (error handling omitted):

RedictModule_StreamIteratorStart(key, 0, startid_ptr, endid_ptr);
RedictModuleStreamID id;
long numfields;
while (RedictModule_StreamIteratorNextID(key, &id, &numfields) ==
       REDICTMODULE_OK) {
    RedictModuleString *field, *value;
    while (RedictModule_StreamIteratorNextField(key, &field, &value) ==
           REDICTMODULE_OK) {
        //
        // ... Do stuff ...
        //
        RedictModule_FreeString(ctx, field);
        RedictModule_FreeString(ctx, value);
    }
}
RedictModule_StreamIteratorStop(key);

RedictModule_StreamIteratorStop #

int RedictModule_StreamIteratorStop(RedictModuleKey *key);

Available since: Redict 7.3.0

Stops a stream iterator created using RedictModule_StreamIteratorStart() and reclaims its memory.

Returns REDICTMODULE_OK on success. On failure, REDICTMODULE_ERR is returned and errno is set as follows:

  • EINVAL if called with a NULL key
  • ENOTSUP if the key refers to a value of a type other than stream or if the key is empty
  • EBADF if the key was not opened for writing or if no stream iterator is associated with the key

RedictModule_StreamIteratorNextID #

int RedictModule_StreamIteratorNextID(RedictModuleKey *key,
                                      RedictModuleStreamID *id,
                                      long *numfields);

Available since: Redict 7.3.0

Finds the next stream entry and returns its stream ID and the number of fields.

  • key: Key for which a stream iterator has been started using RedictModule_StreamIteratorStart().
  • id: The stream ID returned. NULL if you don’t care.
  • numfields: The number of fields in the found stream entry. NULL if you don’t care.

Returns REDICTMODULE_OK and sets *id and *numfields if an entry was found. On failure, REDICTMODULE_ERR is returned and errno is set as follows:

  • EINVAL if called with a NULL key
  • ENOTSUP if the key refers to a value of a type other than stream or if the key is empty
  • EBADF if no stream iterator is associated with the key
  • ENOENT if there are no more entries in the range of the iterator

In practice, if RedictModule_StreamIteratorNextID() is called after a successful call to RedictModule_StreamIteratorStart() and with the same key, it is safe to assume that an REDICTMODULE_ERR return value means that there are no more entries.

Use RedictModule_StreamIteratorNextField() to retrieve the fields and values. See the example at RedictModule_StreamIteratorStart().

RedictModule_StreamIteratorNextField #

int RedictModule_StreamIteratorNextField(RedictModuleKey *key,
                                         RedictModuleString **field_ptr,
                                         RedictModuleString **value_ptr);

Available since: Redict 7.3.0

Retrieves the next field of the current stream ID and its corresponding value in a stream iteration. This function should be called repeatedly after calling RedictModule_StreamIteratorNextID() to fetch each field-value pair.

  • key: Key where a stream iterator has been started.
  • field_ptr: This is where the field is returned.
  • value_ptr: This is where the value is returned.

Returns REDICTMODULE_OK and points *field_ptr and *value_ptr to freshly allocated RedictModuleString objects. The string objects are freed automatically when the callback finishes if automatic memory is enabled. On failure, REDICTMODULE_ERR is returned and errno is set as follows:

  • EINVAL if called with a NULL key
  • ENOTSUP if the key refers to a value of a type other than stream or if the key is empty
  • EBADF if no stream iterator is associated with the key
  • ENOENT if there are no more fields in the current stream entry

In practice, if RedictModule_StreamIteratorNextField() is called after a successful call to RedictModule_StreamIteratorNextID() and with the same key, it is safe to assume that an REDICTMODULE_ERR return value means that there are no more fields.

See the example at RedictModule_StreamIteratorStart().

RedictModule_StreamIteratorDelete #

int RedictModule_StreamIteratorDelete(RedictModuleKey *key);

Available since: Redict 7.3.0

Deletes the current stream entry while iterating.

This function can be called after RedictModule_StreamIteratorNextID() or after any calls to RedictModule_StreamIteratorNextField().

Returns REDICTMODULE_OK on success. On failure, REDICTMODULE_ERR is returned and errno is set as follows:

  • EINVAL if key is NULL
  • ENOTSUP if the key is empty or is of another type than stream
  • EBADF if the key is not opened for writing, if no iterator has been started
  • ENOENT if the iterator has no current stream entry

RedictModule_StreamTrimByLength #

long long RedictModule_StreamTrimByLength(RedictModuleKey *key,
                                          int flags,
                                          long long length);

Available since: Redict 7.3.0

Trim a stream by length, similar to XTRIM with MAXLEN.

  • key: Key opened for writing.
  • flags: A bitfield of
    • REDICTMODULE_STREAM_TRIM_APPROX: Trim less if it improves performance, like XTRIM with ~.
  • length: The number of stream entries to keep after trimming.

Returns the number of entries deleted. On failure, a negative value is returned and errno is set as follows:

  • EINVAL if called with invalid arguments
  • ENOTSUP if the key is empty or of a type other than stream
  • EBADF if the key is not opened for writing

RedictModule_StreamTrimByID #

long long RedictModule_StreamTrimByID(RedictModuleKey *key,
                                      int flags,
                                      RedictModuleStreamID *id);

Available since: Redict 7.3.0

Trim a stream by ID, similar to XTRIM with MINID.

  • key: Key opened for writing.
  • flags: A bitfield of
    • REDICTMODULE_STREAM_TRIM_APPROX: Trim less if it improves performance, like XTRIM with ~.
  • id: The smallest stream ID to keep after trimming.

Returns the number of entries deleted. On failure, a negative value is returned and errno is set as follows:

  • EINVAL if called with invalid arguments
  • ENOTSUP if the key is empty or of a type other than stream
  • EBADF if the key is not opened for writing

Calling Redict commands from modules #

RedictModule_Call() sends a command to Redict. The remaining functions handle the reply.

RedictModule_FreeCallReply #

void RedictModule_FreeCallReply(RedictModuleCallReply *reply);

Available since: Redict 7.3.0

Free a Call reply and all the nested replies it contains if it’s an array.

RedictModule_CallReplyType #

int RedictModule_CallReplyType(RedictModuleCallReply *reply);

Available since: Redict 7.3.0

Return the reply type as one of the following:

  • REDICTMODULE_REPLY_UNKNOWN
  • REDICTMODULE_REPLY_STRING
  • REDICTMODULE_REPLY_ERROR
  • REDICTMODULE_REPLY_INTEGER
  • REDICTMODULE_REPLY_ARRAY
  • REDICTMODULE_REPLY_NULL
  • REDICTMODULE_REPLY_MAP
  • REDICTMODULE_REPLY_SET
  • REDICTMODULE_REPLY_BOOL
  • REDICTMODULE_REPLY_DOUBLE
  • REDICTMODULE_REPLY_BIG_NUMBER
  • REDICTMODULE_REPLY_VERBATIM_STRING
  • REDICTMODULE_REPLY_ATTRIBUTE
  • REDICTMODULE_REPLY_PROMISE

RedictModule_CallReplyLength #

size_t RedictModule_CallReplyLength(RedictModuleCallReply *reply);

Available since: Redict 7.3.0

Return the reply type length, where applicable.

RedictModule_CallReplyArrayElement #

RedictModuleCallReply *RedictModule_CallReplyArrayElement(RedictModuleCallReply *reply,
                                                          size_t idx);

Available since: Redict 7.3.0

Return the ‘idx’-th nested call reply element of an array reply, or NULL if the reply type is wrong or the index is out of range.

RedictModule_CallReplyInteger #

long long RedictModule_CallReplyInteger(RedictModuleCallReply *reply);

Available since: Redict 7.3.0

Return the long long of an integer reply.

RedictModule_CallReplyDouble #

double RedictModule_CallReplyDouble(RedictModuleCallReply *reply);

Available since: Redict 7.3.0

Return the double value of a double reply.

RedictModule_CallReplyBigNumber #

const char *RedictModule_CallReplyBigNumber(RedictModuleCallReply *reply,
                                            size_t *len);

Available since: Redict 7.3.0

Return the big number value of a big number reply.

RedictModule_CallReplyVerbatim #

const char *RedictModule_CallReplyVerbatim(RedictModuleCallReply *reply,
                                           size_t *len,
                                           const char **format);

Available since: Redict 7.3.0

Return the value of a verbatim string reply, An optional output argument can be given to get verbatim reply format.

RedictModule_CallReplyBool #

int RedictModule_CallReplyBool(RedictModuleCallReply *reply);

Available since: Redict 7.3.0

Return the Boolean value of a Boolean reply.

RedictModule_CallReplySetElement #

RedictModuleCallReply *RedictModule_CallReplySetElement(RedictModuleCallReply *reply,
                                                        size_t idx);

Available since: Redict 7.3.0

Return the ‘idx’-th nested call reply element of a set reply, or NULL if the reply type is wrong or the index is out of range.

RedictModule_CallReplyMapElement #

int RedictModule_CallReplyMapElement(RedictModuleCallReply *reply,
                                     size_t idx,
                                     RedictModuleCallReply **key,
                                     RedictModuleCallReply **val);

Available since: Redict 7.3.0

Retrieve the ‘idx’-th key and value of a map reply.

Returns:

  • REDICTMODULE_OK on success.
  • REDICTMODULE_ERR if idx out of range or if the reply type is wrong.

The key and value arguments are used to return by reference, and may be NULL if not required.

RedictModule_CallReplyAttribute #

RedictModuleCallReply *RedictModule_CallReplyAttribute(RedictModuleCallReply *reply);

Available since: Redict 7.3.0

Return the attribute of the given reply, or NULL if no attribute exists.

RedictModule_CallReplyAttributeElement #

int RedictModule_CallReplyAttributeElement(RedictModuleCallReply *reply,
                                           size_t idx,
                                           RedictModuleCallReply **key,
                                           RedictModuleCallReply **val);

Available since: Redict 7.3.0

Retrieve the ‘idx’-th key and value of an attribute reply.

Returns:

  • REDICTMODULE_OK on success.
  • REDICTMODULE_ERR if idx out of range or if the reply type is wrong.

The key and value arguments are used to return by reference, and may be NULL if not required.

RedictModule_CallReplyPromiseSetUnblockHandler #

void RedictModule_CallReplyPromiseSetUnblockHandler(RedictModuleCallReply *reply,
                                                    RedictModuleOnUnblocked on_unblock,
                                                    void *private_data);

Available since: Redict 7.3.0

Set unblock handler (callback and private data) on the given promise RedictModuleCallReply. The given reply must be of promise type (REDICTMODULE_REPLY_PROMISE).

RedictModule_CallReplyPromiseAbort #

int RedictModule_CallReplyPromiseAbort(RedictModuleCallReply *reply,
                                       void **private_data);

Available since: Redict 7.3.0

Abort the execution of a given promise RedictModuleCallReply. return REDMODULE_OK in case the abort was done successfully and REDICTMODULE_ERR if its not possible to abort the execution (execution already finished). In case the execution was aborted (REDMODULE_OK was returned), the private_data out parameter will be set with the value of the private data that was given on ‘RedictModule_CallReplyPromiseSetUnblockHandler’ so the caller will be able to release the private data.

If the execution was aborted successfully, it is promised that the unblock handler will not be called. That said, it is possible that the abort operation will successes but the operation will still continue. This can happened if, for example, a module implements some blocking command and does not respect the disconnect callback. For pure Redict commands this can not happened.

RedictModule_CallReplyStringPtr #

const char *RedictModule_CallReplyStringPtr(RedictModuleCallReply *reply,
                                            size_t *len);

Available since: Redict 7.3.0

Return the pointer and length of a string or error reply.

RedictModule_CreateStringFromCallReply #

RedictModuleString *RedictModule_CreateStringFromCallReply(RedictModuleCallReply *reply);

Available since: Redict 7.3.0

Return a new string object from a call reply of type string, error or integer. Otherwise (wrong reply type) return NULL.

RedictModule_SetContextUser #

void RedictModule_SetContextUser(RedictModuleCtx *ctx,
                                 const RedictModuleUser *user);

Available since: Redict 7.3.0

Modifies the user that RedictModule_Call will use (e.g. for ACL checks)

RedictModule_Call #

RedictModuleCallReply *RedictModule_Call(RedictModuleCtx *ctx,
                                         const char *cmdname,
                                         const char *fmt,
                                         ...);

Available since: Redict 7.3.0

Exported API to call any Redict command from modules.

  • cmdname: The Redict command to call.

  • fmt: A format specifier string for the command’s arguments. Each of the arguments should be specified by a valid type specification. The format specifier can also contain the modifiers !, A, 3 and R which don’t have a corresponding argument.

    • b – The argument is a buffer and is immediately followed by another argument that is the buffer’s length.

    • c – The argument is a pointer to a plain C string (null-terminated).

    • l – The argument is a long long integer.

    • s – The argument is a RedictModuleString.

    • v – The argument(s) is a vector of RedictModuleString.

    • ! – Sends the Redict command and its arguments to replicas and AOF.

    • A – Suppress AOF propagation, send only to replicas (requires !).

    • R – Suppress replicas propagation, send only to AOF (requires !).

    • 3 – Return a RESP3 reply. This will change the command reply. e.g., HGETALL returns a map instead of a flat array.

    • 0 – Return the reply in auto mode, i.e. the reply format will be the same as the client attached to the given RedictModuleCtx. This will probably used when you want to pass the reply directly to the client.

    • C – Run a command as the user attached to the context. User is either attached automatically via the client that directly issued the command and created the context or via RedictModule_SetContextUser. If the context is not directly created by an issued command (such as a background context and no user was set on it via RedictModule_SetContextUser, RedictModule_Call will fail. Checks if the command can be executed according to ACL rules and causes the command to run as the determined user, so that any future user dependent activity, such as ACL checks within scripts will proceed as expected. Otherwise, the command will run as the Redict unrestricted user.

    • S – Run the command in a script mode, this means that it will raise an error if a command which are not allowed inside a script (flagged with the deny-script flag) is invoked (like SHUTDOWN). In addition, on script mode, write commands are not allowed if there are not enough good replicas (as configured with min-replicas-to-write) or when the server is unable to persist to the disk.

    • W – Do not allow to run any write command (flagged with the write flag).

    • M – Do not allow deny-oom flagged commands when over the memory limit.

    • E – Return error as RedictModuleCallReply. If there is an error before invoking the command, the error is returned using errno mechanism. This flag allows to get the error also as an error CallReply with relevant error message.

    • ‘D’ – A “Dry Run” mode. Return before executing the underlying call(). If everything succeeded, it will return with a NULL, otherwise it will return with a CallReply object denoting the error, as if it was called with the ‘E’ code.

    • ‘K’ – Allow running blocking commands. If enabled and the command gets blocked, a special REDICTMODULE_REPLY_PROMISE will be returned. This reply type indicates that the command was blocked and the reply will be given asynchronously. The module can use this reply object to set a handler which will be called when the command gets unblocked using RedictModule_CallReplyPromiseSetUnblockHandler. The handler must be set immediately after the command invocation (without releasing the Redict lock in between). If the handler is not set, the blocking command will still continue its execution but the reply will be ignored (fire and forget), notice that this is dangerous in case of role change, as explained below. The module can use RedictModule_CallReplyPromiseAbort to abort the command invocation if it was not yet finished (see RedictModule_CallReplyPromiseAbort documentation for more details). It is also the module’s responsibility to abort the execution on role change, either by using server event (to get notified when the instance becomes a replica) or relying on the disconnect callback of the original client. Failing to do so can result in a write operation on a replica. Unlike other call replies, promise call reply must be freed while the Redict GIL is locked. Notice that on unblocking, the only promise is that the unblock handler will be called, If the blocking RedictModule_Call caused the module to also block some real client (using RedictModule_BlockClient), it is the module responsibility to unblock this client on the unblock handler. On the unblock handler it is only allowed to perform the following: * Calling additional Redict commands using RedictModule_Call * Open keys using RedictModule_OpenKey * Replicate data to the replica or AOF

         Specifically, it is not allowed to call any Redict module API which are client related such as:
         * RedictModule_Reply* API's
         * RedictModule_BlockClient
         * RedictModule_GetCurrentUserName
      
  • : The actual arguments to the Redict command.

On success a RedictModuleCallReply object is returned, otherwise NULL is returned and errno is set to the following values:

  • EBADF: wrong format specifier.
  • EINVAL: wrong command arity.
  • ENOENT: command does not exist.
  • EPERM: operation in Cluster instance with key in non local slot.
  • EROFS: operation in Cluster instance when a write command is sent in a readonly state.
  • ENETDOWN: operation in Cluster instance when cluster is down.
  • ENOTSUP: No ACL user for the specified module context
  • EACCES: Command cannot be executed, according to ACL rules
  • ENOSPC: Write or deny-oom command is not allowed
  • ESPIPE: Command not allowed on script mode

Example code fragment:

 reply = RedictModule_Call(ctx,"INCRBY","sc",argv[1],"10");
 if (RedictModule_CallReplyType(reply) == REDICTMODULE_REPLY_INTEGER) {
   long long myval = RedictModule_CallReplyInteger(reply);
   // Do something with myval.
 }

This API is documented here: https://redis.io/topics/modules-intro

RedictModule_CallReplyProto #

const char *RedictModule_CallReplyProto(RedictModuleCallReply *reply,
                                        size_t *len);

Available since: Redict 7.3.0

Return a pointer, and a length, to the protocol returned by the command that returned the reply object.

Modules data types #

When String DMA or using existing data structures is not enough, it is possible to create new data types from scratch and export them to Redict. The module must provide a set of callbacks for handling the new values exported (for example in order to provide RDB saving/loading, AOF rewrite, and so forth). In this section we define this API.

RedictModule_CreateDataType #

moduleType *RedictModule_CreateDataType(RedictModuleCtx *ctx,
                                        const char *name,
                                        int encver,
                                        void *typemethods_ptr);

Available since: Redict 7.3.0

Register a new data type exported by the module. The parameters are the following. Please for in depth documentation check the modules API documentation, especially https://redis.io/topics/modules-native-types.

  • name: A 9 characters data type name that MUST be unique in the Redict Modules ecosystem. Be creative… and there will be no collisions. Use the charset A-Z a-z 9-0, plus the two “-_” characters. A good idea is to use, for example <typename>-<vendor>. For example “tree-AntZ” may mean “Tree data structure by @antirez”. To use both lower case and upper case letters helps in order to prevent collisions.

  • encver: Encoding version, which is, the version of the serialization that a module used in order to persist data. As long as the “name” matches, the RDB loading will be dispatched to the type callbacks whatever ’encver’ is used, however the module can understand if the encoding it must load are of an older version of the module. For example the module “tree-AntZ” initially used encver=0. Later after an upgrade, it started to serialize data in a different format and to register the type with encver=1. However this module may still load old data produced by an older version if the rdb_load callback is able to check the encver value and act accordingly. The encver must be a positive value between 0 and 1023.

  • typemethods_ptr is a pointer to a RedictModuleTypeMethods structure that should be populated with the methods callbacks and structure version, like in the following example:

      RedictModuleTypeMethods tm = {
          .version = REDICTMODULE_TYPE_METHOD_VERSION,
          .rdb_load = myType_RDBLoadCallBack,
          .rdb_save = myType_RDBSaveCallBack,
          .aof_rewrite = myType_AOFRewriteCallBack,
          .free = myType_FreeCallBack,
    
          // Optional fields
          .digest = myType_DigestCallBack,
          .mem_usage = myType_MemUsageCallBack,
          .aux_load = myType_AuxRDBLoadCallBack,
          .aux_save = myType_AuxRDBSaveCallBack,
          .free_effort = myType_FreeEffortCallBack,
          .unlink = myType_UnlinkCallBack,
          .copy = myType_CopyCallback,
          .defrag = myType_DefragCallback
    
          // Enhanced optional fields
          .mem_usage2 = myType_MemUsageCallBack2,
          .free_effort2 = myType_FreeEffortCallBack2,
          .unlink2 = myType_UnlinkCallBack2,
          .copy2 = myType_CopyCallback2,
      }
    
  • rdb_load: A callback function pointer that loads data from RDB files.

  • rdb_save: A callback function pointer that saves data to RDB files.

  • aof_rewrite: A callback function pointer that rewrites data as commands.

  • digest: A callback function pointer that is used for DEBUG DIGEST.

  • free: A callback function pointer that can free a type value.

  • aux_save: A callback function pointer that saves out of keyspace data to RDB files. ‘when’ argument is either REDICTMODULE_AUX_BEFORE_RDB or REDICTMODULE_AUX_AFTER_RDB.

  • aux_load: A callback function pointer that loads out of keyspace data from RDB files. Similar to aux_save, returns REDICTMODULE_OK on success, and ERR otherwise.

  • free_effort: A callback function pointer that used to determine whether the module’s memory needs to be lazy reclaimed. The module should return the complexity involved by freeing the value. for example: how many pointers are gonna be freed. Note that if it returns 0, we’ll always do an async free.

  • unlink: A callback function pointer that used to notifies the module that the key has been removed from the DB by redict, and may soon be freed by a background thread. Note that it won’t be called on FLUSHALL/FLUSHDB (both sync and async), and the module can use the RedictModuleEvent_FlushDB to hook into that.

  • copy: A callback function pointer that is used to make a copy of the specified key. The module is expected to perform a deep copy of the specified value and return it. In addition, hints about the names of the source and destination keys is provided. A NULL return value is considered an error and the copy operation fails. Note: if the target key exists and is being overwritten, the copy callback will be called first, followed by a free callback to the value that is being replaced.

  • defrag: A callback function pointer that is used to request the module to defrag a key. The module should then iterate pointers and call the relevant RedictModule_Defrag*() functions to defragment pointers or complex types. The module should continue iterating as long as RedictModule_DefragShouldStop() returns a zero value, and return a zero value if finished or non-zero value if more work is left to be done. If more work needs to be done, RedictModule_DefragCursorSet() and RedictModule_DefragCursorGet() can be used to track this work across different calls. Normally, the defrag mechanism invokes the callback without a time limit, so RedictModule_DefragShouldStop() always returns zero. The “late defrag” mechanism which has a time limit and provides cursor support is used only for keys that are determined to have significant internal complexity. To determine this, the defrag mechanism uses the free_effort callback and the ‘active-defrag-max-scan-fields’ config directive. NOTE: The value is passed as a void** and the function is expected to update the pointer if the top-level value pointer is defragmented and consequently changes.

  • mem_usage2: Similar to mem_usage, but provides the RedictModuleKeyOptCtx parameter so that meta information such as key name and db id can be obtained, and the sample_size for size estimation (see MEMORY USAGE command).

  • free_effort2: Similar to free_effort, but provides the RedictModuleKeyOptCtx parameter so that meta information such as key name and db id can be obtained.

  • unlink2: Similar to unlink, but provides the RedictModuleKeyOptCtx parameter so that meta information such as key name and db id can be obtained.

  • copy2: Similar to copy, but provides the RedictModuleKeyOptCtx parameter so that meta information such as key names and db ids can be obtained.

  • aux_save2: Similar to aux_save, but with small semantic change, if the module saves nothing on this callback then no data about this aux field will be written to the RDB and it will be possible to load the RDB even if the module is not loaded.

Note: the module name “AAAAAAAAA” is reserved and produces an error, it happens to be pretty lame as well.

If RedictModule_CreateDataType() is called outside of RedictModule_OnLoad() function, there is already a module registering a type with the same name, or if the module name or encver is invalid, NULL is returned. Otherwise the new type is registered into Redict, and a reference of type RedictModuleType is returned: the caller of the function should store this reference into a global variable to make future use of it in the modules type API, since a single module may register multiple types. Example code fragment:

 static RedictModuleType *BalancedTreeType;

 int RedictModule_OnLoad(RedictModuleCtx *ctx) {
     // some code here ...
     BalancedTreeType = RedictModule_CreateDataType(...);
 }

RedictModule_ModuleTypeSetValue #

int RedictModule_ModuleTypeSetValue(RedictModuleKey *key,
                                    moduleType *mt,
                                    void *value);

Available since: Redict 7.3.0

If the key is open for writing, set the specified module type object as the value of the key, deleting the old value if any. On success REDICTMODULE_OK is returned. If the key is not open for writing or there is an active iterator, REDICTMODULE_ERR is returned.

RedictModule_ModuleTypeGetType #

moduleType *RedictModule_ModuleTypeGetType(RedictModuleKey *key);

Available since: Redict 7.3.0

Assuming RedictModule_KeyType() returned REDICTMODULE_KEYTYPE_MODULE on the key, returns the module type pointer of the value stored at key.

If the key is NULL, is not associated with a module type, or is empty, then NULL is returned instead.

RedictModule_ModuleTypeGetValue #

void *RedictModule_ModuleTypeGetValue(RedictModuleKey *key);

Available since: Redict 7.3.0

Assuming RedictModule_KeyType() returned REDICTMODULE_KEYTYPE_MODULE on the key, returns the module type low-level value stored at key, as it was set by the user via RedictModule_ModuleTypeSetValue().

If the key is NULL, is not associated with a module type, or is empty, then NULL is returned instead.

RDB loading and saving functions #

RedictModule_IsIOError #

int RedictModule_IsIOError(RedictModuleIO *io);

Available since: Redict 7.3.0

Returns true if any previous IO API failed. for Load* APIs the REDICTMODULE_OPTIONS_HANDLE_IO_ERRORS flag must be set with RedictModule_SetModuleOptions first.

RedictModule_SaveUnsigned #

void RedictModule_SaveUnsigned(RedictModuleIO *io, uint64_t value);

Available since: Redict 7.3.0

Save an unsigned 64 bit value into the RDB file. This function should only be called in the context of the rdb_save method of modules implementing new data types.

RedictModule_LoadUnsigned #

uint64_t RedictModule_LoadUnsigned(RedictModuleIO *io);

Available since: Redict 7.3.0

Load an unsigned 64 bit value from the RDB file. This function should only be called in the context of the rdb_load method of modules implementing new data types.

RedictModule_SaveSigned #

void RedictModule_SaveSigned(RedictModuleIO *io, int64_t value);

Available since: Redict 7.3.0

Like RedictModule_SaveUnsigned() but for signed 64 bit values.

RedictModule_LoadSigned #

int64_t RedictModule_LoadSigned(RedictModuleIO *io);

Available since: Redict 7.3.0

Like RedictModule_LoadUnsigned() but for signed 64 bit values.

RedictModule_SaveString #

void RedictModule_SaveString(RedictModuleIO *io, RedictModuleString *s);

Available since: Redict 7.3.0

In the context of the rdb_save method of a module type, saves a string into the RDB file taking as input a RedictModuleString.

The string can be later loaded with RedictModule_LoadString() or other Load family functions expecting a serialized string inside the RDB file.

RedictModule_SaveStringBuffer #

void RedictModule_SaveStringBuffer(RedictModuleIO *io,
                                   const char *str,
                                   size_t len);

Available since: Redict 7.3.0

Like RedictModule_SaveString() but takes a raw C pointer and length as input.

RedictModule_LoadString #

RedictModuleString *RedictModule_LoadString(RedictModuleIO *io);

Available since: Redict 7.3.0

In the context of the rdb_load method of a module data type, loads a string from the RDB file, that was previously saved with RedictModule_SaveString() functions family.

The returned string is a newly allocated RedictModuleString object, and the user should at some point free it with a call to RedictModule_FreeString().

If the data structure does not store strings as RedictModuleString objects, the similar function RedictModule_LoadStringBuffer() could be used instead.

RedictModule_LoadStringBuffer #

char *RedictModule_LoadStringBuffer(RedictModuleIO *io, size_t *lenptr);

Available since: Redict 7.3.0

Like RedictModule_LoadString() but returns a heap allocated string that was allocated with RedictModule_Alloc(), and can be resized or freed with RedictModule_Realloc() or RedictModule_Free().

The size of the string is stored at ‘*lenptr’ if not NULL. The returned string is not automatically NULL terminated, it is loaded exactly as it was stored inside the RDB file.

RedictModule_SaveDouble #

void RedictModule_SaveDouble(RedictModuleIO *io, double value);

Available since: Redict 7.3.0

In the context of the rdb_save method of a module data type, saves a double value to the RDB file. The double can be a valid number, a NaN or infinity. It is possible to load back the value with RedictModule_LoadDouble().

RedictModule_LoadDouble #

double RedictModule_LoadDouble(RedictModuleIO *io);

Available since: Redict 7.3.0

In the context of the rdb_save method of a module data type, loads back the double value saved by RedictModule_SaveDouble().

RedictModule_SaveFloat #

void RedictModule_SaveFloat(RedictModuleIO *io, float value);

Available since: Redict 7.3.0

In the context of the rdb_save method of a module data type, saves a float value to the RDB file. The float can be a valid number, a NaN or infinity. It is possible to load back the value with RedictModule_LoadFloat().

RedictModule_LoadFloat #

float RedictModule_LoadFloat(RedictModuleIO *io);

Available since: Redict 7.3.0

In the context of the rdb_save method of a module data type, loads back the float value saved by RedictModule_SaveFloat().

RedictModule_SaveLongDouble #

void RedictModule_SaveLongDouble(RedictModuleIO *io, long double value);

Available since: Redict 7.3.0

In the context of the rdb_save method of a module data type, saves a long double value to the RDB file. The double can be a valid number, a NaN or infinity. It is possible to load back the value with RedictModule_LoadLongDouble().

RedictModule_LoadLongDouble #

long double RedictModule_LoadLongDouble(RedictModuleIO *io);

Available since: Redict 7.3.0

In the context of the rdb_save method of a module data type, loads back the long double value saved by RedictModule_SaveLongDouble().

Key digest API (DEBUG DIGEST interface for modules types) #

RedictModule_DigestAddStringBuffer #

void RedictModule_DigestAddStringBuffer(RedictModuleDigest *md,
                                        const char *ele,
                                        size_t len);

Available since: Redict 7.3.0

Add a new element to the digest. This function can be called multiple times one element after the other, for all the elements that constitute a given data structure. The function call must be followed by the call to RedictModule_DigestEndSequence eventually, when all the elements that are always in a given order are added. See the Redict Modules data types documentation for more info. However this is a quick example that uses Redict data types as an example.

To add a sequence of unordered elements (for example in the case of a Redict Set), the pattern to use is:

foreach element {
    AddElement(element);
    EndSequence();
}

Because Sets are not ordered, so every element added has a position that does not depend from the other. However if instead our elements are ordered in pairs, like field-value pairs of a Hash, then one should use:

foreach key,value {
    AddElement(key);
    AddElement(value);
    EndSequence();
}

Because the key and value will be always in the above order, while instead the single key-value pairs, can appear in any position into a Redict hash.

A list of ordered elements would be implemented with:

foreach element {
    AddElement(element);
}
EndSequence();

RedictModule_DigestAddLongLong #

void RedictModule_DigestAddLongLong(RedictModuleDigest *md, long long ll);

Available since: Redict 7.3.0

Like RedictModule_DigestAddStringBuffer() but takes a long long as input that gets converted into a string before adding it to the digest.

RedictModule_DigestEndSequence #

void RedictModule_DigestEndSequence(RedictModuleDigest *md);

Available since: Redict 7.3.0

See the documentation for RedictModule_DigestAddElement().

RedictModule_LoadDataTypeFromStringEncver #

void *RedictModule_LoadDataTypeFromStringEncver(const RedictModuleString *str,
                                                const moduleType *mt,
                                                int encver);

Available since: Redict 7.3.0

Decode a serialized representation of a module data type ‘mt’, in a specific encoding version ’encver’ from string ‘str’ and return a newly allocated value, or NULL if decoding failed.

This call basically reuses the ‘rdb_load’ callback which module data types implement in order to allow a module to arbitrarily serialize/de-serialize keys, similar to how the Redict ‘DUMP’ and ‘RESTORE’ commands are implemented.

Modules should generally use the REDICTMODULE_OPTIONS_HANDLE_IO_ERRORS flag and make sure the de-serialization code properly checks and handles IO errors (freeing allocated buffers and returning a NULL).

If this is NOT done, Redict will handle corrupted (or just truncated) serialized data by producing an error message and terminating the process.

RedictModule_LoadDataTypeFromString #

void *RedictModule_LoadDataTypeFromString(const RedictModuleString *str,
                                          const moduleType *mt);

Available since: Redict 7.3.0

Similar to RedictModule_LoadDataTypeFromStringEncver, original version of the API, kept for backward compatibility.

RedictModule_SaveDataTypeToString #

RedictModuleString *RedictModule_SaveDataTypeToString(RedictModuleCtx *ctx,
                                                      void *data,
                                                      const moduleType *mt);

Available since: Redict 7.3.0

Encode a module data type ‘mt’ value ‘data’ into serialized form, and return it as a newly allocated RedictModuleString.

This call basically reuses the ‘rdb_save’ callback which module data types implement in order to allow a module to arbitrarily serialize/de-serialize keys, similar to how the Redict ‘DUMP’ and ‘RESTORE’ commands are implemented.

RedictModule_GetKeyNameFromDigest #

const RedictModuleString *RedictModule_GetKeyNameFromDigest(RedictModuleDigest *dig);

Available since: Redict 7.3.0

Returns the name of the key currently being processed.

RedictModule_GetDbIdFromDigest #

int RedictModule_GetDbIdFromDigest(RedictModuleDigest *dig);

Available since: Redict 7.3.0

Returns the database id of the key currently being processed.

AOF API for modules data types #

RedictModule_EmitAOF #

void RedictModule_EmitAOF(RedictModuleIO *io,
                          const char *cmdname,
                          const char *fmt,
                          ...);

Available since: Redict 7.3.0

Emits a command into the AOF during the AOF rewriting process. This function is only called in the context of the aof_rewrite method of data types exported by a module. The command works exactly like RedictModule_Call() in the way the parameters are passed, but it does not return anything as the error handling is performed by Redict itself.

IO context handling #

RedictModule_GetKeyNameFromIO #

const RedictModuleString *RedictModule_GetKeyNameFromIO(RedictModuleIO *io);

Available since: Redict 7.3.0

Returns the name of the key currently being processed. There is no guarantee that the key name is always available, so this may return NULL.

RedictModule_GetKeyNameFromModuleKey #

const RedictModuleString *RedictModule_GetKeyNameFromModuleKey(RedictModuleKey *key);

Available since: Redict 7.3.0

Returns a RedictModuleString with the name of the key from RedictModuleKey.

RedictModule_GetDbIdFromModuleKey #

int RedictModule_GetDbIdFromModuleKey(RedictModuleKey *key);

Available since: Redict 7.3.0

Returns a database id of the key from RedictModuleKey.

RedictModule_GetDbIdFromIO #

int RedictModule_GetDbIdFromIO(RedictModuleIO *io);

Available since: Redict 7.3.0

Returns the database id of the key currently being processed. There is no guarantee that this info is always available, so this may return -1.

Logging #

RedictModule_Log #

void RedictModule_Log(RedictModuleCtx *ctx,
                      const char *levelstr,
                      const char *fmt,
                      ...);

Available since: Redict 7.3.0

Produces a log message to the standard Redict log, the format accepts printf-alike specifiers, while level is a string describing the log level to use when emitting the log, and must be one of the following:

  • “debug” (REDICTMODULE_LOGLEVEL_DEBUG)
  • “verbose” (REDICTMODULE_LOGLEVEL_VERBOSE)
  • “notice” (REDICTMODULE_LOGLEVEL_NOTICE)
  • “warning” (REDICTMODULE_LOGLEVEL_WARNING)

If the specified log level is invalid, verbose is used by default. There is a fixed limit to the length of the log line this function is able to emit, this limit is not specified but is guaranteed to be more than a few lines of text.

The ctx argument may be NULL if cannot be provided in the context of the caller for instance threads or callbacks, in which case a generic “module” will be used instead of the module name.

RedictModule_LogIOError #

void RedictModule_LogIOError(RedictModuleIO *io,
                             const char *levelstr,
                             const char *fmt,
                             ...);

Available since: Redict 7.3.0

Log errors from RDB / AOF serialization callbacks.

This function should be used when a callback is returning a critical error to the caller since cannot load or save the data for some critical reason.

RedictModule__Assert #

void RedictModule__Assert(const char *estr, const char *file, int line);

Available since: Redict 7.3.0

Redict-like assert function.

The macro RedictModule_Assert(expression) is recommended, rather than calling this function directly.

A failed assertion will shut down the server and produce logging information that looks identical to information generated by Redict itself.

RedictModule_LatencyAddSample #

void RedictModule_LatencyAddSample(const char *event, mstime_t latency);

Available since: Redict 7.3.0

Allows adding event to the latency monitor to be observed by the LATENCY command. The call is skipped if the latency is smaller than the configured latency-monitor-threshold.

Blocking clients from modules #

For a guide about blocking commands in modules, see https://redis.io/topics/modules-blocking-ops.

RedictModule_RegisterAuthCallback #

void RedictModule_RegisterAuthCallback(RedictModuleCtx *ctx,
                                       RedictModuleAuthCallback cb);

Available since: Redict 7.3.0

This API registers a callback to execute in addition to normal password based authentication. Multiple callbacks can be registered across different modules. When a Module is unloaded, all the auth callbacks registered by it are unregistered. The callbacks are attempted (in the order of most recently registered first) when the AUTH/HELLO (with AUTH field provided) commands are called. The callbacks will be called with a module context along with a username and a password, and are expected to take one of the following actions: (1) Authenticate - Use the RedictModule_AuthenticateClient* API and return REDICTMODULE_AUTH_HANDLED. This will immediately end the auth chain as successful and add the OK reply. (2) Deny Authentication - Return REDICTMODULE_AUTH_HANDLED without authenticating or blocking the client. Optionally, err can be set to a custom error message and err will be automatically freed by the server. This will immediately end the auth chain as unsuccessful and add the ERR reply. (3) Block a client on authentication - Use the RedictModule_BlockClientOnAuth API and return REDICTMODULE_AUTH_HANDLED. Here, the client will be blocked until the RedictModule_UnblockClient API is used which will trigger the auth reply callback (provided through the RedictModule_BlockClientOnAuth). In this reply callback, the Module should authenticate, deny or skip handling authentication. (4) Skip handling Authentication - Return REDICTMODULE_AUTH_NOT_HANDLED without blocking the client. This will allow the engine to attempt the next module auth callback. If none of the callbacks authenticate or deny auth, then password based auth is attempted and will authenticate or add failure logs and reply to the clients accordingly.

Note: If a client is disconnected while it was in the middle of blocking module auth, that occurrence of the AUTH or HELLO command will not be tracked in the INFO command stats.

The following is an example of how non-blocking module based authentication can be used:

 int auth_cb(RedictModuleCtx *ctx, RedictModuleString *username, RedictModuleString *password, RedictModuleString **err) {
     const char *user = RedictModule_StringPtrLen(username, NULL);
     const char *pwd = RedictModule_StringPtrLen(password, NULL);
     if (!strcmp(user,"foo") && !strcmp(pwd,"valid_password")) {
         RedictModule_AuthenticateClientWithACLUser(ctx, "foo", 3, NULL, NULL, NULL);
         return REDICTMODULE_AUTH_HANDLED;
     }

     else if (!strcmp(user,"foo") && !strcmp(pwd,"wrong_password")) {
         RedictModuleString *log = RedictModule_CreateString(ctx, "Module Auth", 11);
         RedictModule_ACLAddLogEntryByUserName(ctx, username, log, REDICTMODULE_ACL_LOG_AUTH);
         RedictModule_FreeString(ctx, log);
         const char *err_msg = "Auth denied by Misc Module.";
         *err = RedictModule_CreateString(ctx, err_msg, strlen(err_msg));
         return REDICTMODULE_AUTH_HANDLED;
     }
     return REDICTMODULE_AUTH_NOT_HANDLED;
  }

 int RedictModule_OnLoad(RedictModuleCtx *ctx, RedictModuleString **argv, int argc) {
     if (RedictModule_Init(ctx,"authmodule",1,REDICTMODULE_APIVER_1)== REDICTMODULE_ERR)
         return REDICTMODULE_ERR;
     RedictModule_RegisterAuthCallback(ctx, auth_cb);
     return REDICTMODULE_OK;
 }

RedictModule_BlockClient #

RedictModuleBlockedClient *RedictModule_BlockClient(RedictModuleCtx *ctx,
                                                    RedictModuleCmdFunc reply_callback,
                                                    ;

Available since: Redict 7.3.0

Block a client in the context of a blocking command, returning a handle which will be used, later, in order to unblock the client with a call to RedictModule_UnblockClient(). The arguments specify callback functions and a timeout after which the client is unblocked.

The callbacks are called in the following contexts:

reply_callback:   called after a successful RedictModule_UnblockClient()
                  call in order to reply to the client and unblock it.

timeout_callback: called when the timeout is reached or if `CLIENT UNBLOCK`
                  is invoked, in order to send an error to the client.

free_privdata:    called in order to free the private data that is passed
                  by RedictModule_UnblockClient() call.

Note: RedictModule_UnblockClient should be called for every blocked client, even if client was killed, timed-out or disconnected. Failing to do so will result in memory leaks.

There are some cases where RedictModule_BlockClient() cannot be used:

  1. If the client is a Lua script.
  2. If the client is executing a MULTI block.

In these cases, a call to RedictModule_BlockClient() will not block the client, but instead produce a specific error reply.

A module that registers a timeout_callback function can also be unblocked using the CLIENT UNBLOCK command, which will trigger the timeout callback. If a callback function is not registered, then the blocked client will be treated as if it is not in a blocked state and CLIENT UNBLOCK will return a zero value.

Measuring background time: By default the time spent in the blocked command is not account for the total command duration. To include such time you should use RedictModule_BlockedClientMeasureTimeStart() and RedictModule_BlockedClientMeasureTimeEnd() one, or multiple times within the blocking command background work.

RedictModule_BlockClientOnAuth #

RedictModuleBlockedClient *RedictModule_BlockClientOnAuth(RedictModuleCtx *ctx,
                                                          RedictModuleAuthCallback reply_callback,
                                                          ;

Available since: Redict 7.3.0

Block the current client for module authentication in the background. If module auth is not in progress on the client, the API returns NULL. Otherwise, the client is blocked and the RedictModule_BlockedClient is returned similar to the RedictModule_BlockClient API. Note: Only use this API from the context of a module auth callback.

RedictModule_BlockClientGetPrivateData #

void *RedictModule_BlockClientGetPrivateData(RedictModuleBlockedClient *blocked_client);

Available since: Redict 7.3.0

Get the private data that was previusely set on a blocked client

RedictModule_BlockClientSetPrivateData #

void RedictModule_BlockClientSetPrivateData(RedictModuleBlockedClient *blocked_client,
                                            void *private_data);

Available since: Redict 7.3.0

Set private data on a blocked client

RedictModule_BlockClientOnKeys #

RedictModuleBlockedClient *RedictModule_BlockClientOnKeys(RedictModuleCtx *ctx,
                                                          RedictModuleCmdFunc reply_callback,
                                                          ;

Available since: Redict 7.3.0

This call is similar to RedictModule_BlockClient(), however in this case we don’t just block the client, but also ask Redict to unblock it automatically once certain keys become “ready”, that is, contain more data.

Basically this is similar to what a typical Redict command usually does, like BLPOP or BZPOPMAX: the client blocks if it cannot be served ASAP, and later when the key receives new data (a list push for instance), the client is unblocked and served.

However in the case of this module API, when the client is unblocked?

  1. If you block on a key of a type that has blocking operations associated, like a list, a sorted set, a stream, and so forth, the client may be unblocked once the relevant key is targeted by an operation that normally unblocks the native blocking operations for that type. So if we block on a list key, an RPUSH command may unblock our client and so forth.
  2. If you are implementing your native data type, or if you want to add new unblocking conditions in addition to “1”, you can call the modules API RedictModule_SignalKeyAsReady().

Anyway we can’t be sure if the client should be unblocked just because the key is signaled as ready: for instance a successive operation may change the key, or a client in queue before this one can be served, modifying the key as well and making it empty again. So when a client is blocked with RedictModule_BlockClientOnKeys() the reply callback is not called after RedictModule_UnblockClient() is called, but every time a key is signaled as ready: if the reply callback can serve the client, it returns REDICTMODULE_OK and the client is unblocked, otherwise it will return REDICTMODULE_ERR and we’ll try again later.

The reply callback can access the key that was signaled as ready by calling the API RedictModule_GetBlockedClientReadyKey(), that returns just the string name of the key as a RedictModuleString object.

Thanks to this system we can setup complex blocking scenarios, like unblocking a client only if a list contains at least 5 items or other more fancy logics.

Note that another difference with RedictModule_BlockClient(), is that here we pass the private data directly when blocking the client: it will be accessible later in the reply callback. Normally when blocking with RedictModule_BlockClient() the private data to reply to the client is passed when calling RedictModule_UnblockClient() but here the unblocking is performed by Redict itself, so we need to have some private data before hand. The private data is used to store any information about the specific unblocking operation that you are implementing. Such information will be freed using the free_privdata callback provided by the user.

However the reply callback will be able to access the argument vector of the command, so the private data is often not needed.

Note: Under normal circumstances RedictModule_UnblockClient should not be called for clients that are blocked on keys (Either the key will become ready or a timeout will occur). If for some reason you do want to call RedictModule_UnblockClient it is possible: Client will be handled as if it were timed-out (You must implement the timeout callback in that case).

RedictModule_BlockClientOnKeysWithFlags #

RedictModuleBlockedClient *RedictModule_BlockClientOnKeysWithFlags(RedictModuleCtx *ctx,
                                                                   RedictModuleCmdFunc reply_callback,
                                                                   ;

Available since: Redict 7.3.0

Same as RedictModule_BlockClientOnKeys, but can take REDICTMODULE_BLOCK_* flags Can be either REDICTMODULE_BLOCK_UNBLOCK_DEFAULT, which means default behavior (same as calling RedictModule_BlockClientOnKeys)

The flags is a bit mask of these:

  • REDICTMODULE_BLOCK_UNBLOCK_DELETED: The clients should to be awakened in case any of keys are deleted. Mostly useful for commands that require the key to exist (like XREADGROUP)

RedictModule_SignalKeyAsReady #

void RedictModule_SignalKeyAsReady(RedictModuleCtx *ctx,
                                   RedictModuleString *key);

Available since: Redict 7.3.0

This function is used in order to potentially unblock a client blocked on keys with RedictModule_BlockClientOnKeys(). When this function is called, all the clients blocked for this key will get their reply_callback called.

RedictModule_UnblockClient #

int RedictModule_UnblockClient(RedictModuleBlockedClient *bc, void *privdata);

Available since: Redict 7.3.0

Unblock a client blocked by RedictModule_BlockedClient. This will trigger the reply callbacks to be called in order to reply to the client. The ‘privdata’ argument will be accessible by the reply callback, so the caller of this function can pass any value that is needed in order to actually reply to the client.

A common usage for ‘privdata’ is a thread that computes something that needs to be passed to the client, included but not limited some slow to compute reply or some reply obtained via networking.

Note 1: this function can be called from threads spawned by the module.

Note 2: when we unblock a client that is blocked for keys using the API RedictModule_BlockClientOnKeys(), the privdata argument here is not used. Unblocking a client that was blocked for keys using this API will still require the client to get some reply, so the function will use the “timeout” handler in order to do so (The privdata provided in RedictModule_BlockClientOnKeys() is accessible from the timeout callback via RedictModule_GetBlockedClientPrivateData).

RedictModule_AbortBlock #

int RedictModule_AbortBlock(RedictModuleBlockedClient *bc);

Available since: Redict 7.3.0

Abort a blocked client blocking operation: the client will be unblocked without firing any callback.

RedictModule_SetDisconnectCallback #

void RedictModule_SetDisconnectCallback(RedictModuleBlockedClient *bc,
                                        RedictModuleDisconnectFunc callback);

Available since: Redict 7.3.0

Set a callback that will be called if a blocked client disconnects before the module has a chance to call RedictModule_UnblockClient()

Usually what you want to do there, is to cleanup your module state so that you can call RedictModule_UnblockClient() safely, otherwise the client will remain blocked forever if the timeout is large.

Notes:

  1. It is not safe to call Reply* family functions here, it is also useless since the client is gone.

  2. This callback is not called if the client disconnects because of a timeout. In such a case, the client is unblocked automatically and the timeout callback is called.

RedictModule_IsBlockedReplyRequest #

int RedictModule_IsBlockedReplyRequest(RedictModuleCtx *ctx);

Available since: Redict 7.3.0

Return non-zero if a module command was called in order to fill the reply for a blocked client.

RedictModule_IsBlockedTimeoutRequest #

int RedictModule_IsBlockedTimeoutRequest(RedictModuleCtx *ctx);

Available since: Redict 7.3.0

Return non-zero if a module command was called in order to fill the reply for a blocked client that timed out.

RedictModule_GetBlockedClientPrivateData #

void *RedictModule_GetBlockedClientPrivateData(RedictModuleCtx *ctx);

Available since: Redict 7.3.0

Get the private data set by RedictModule_UnblockClient()

RedictModule_GetBlockedClientReadyKey #

RedictModuleString *RedictModule_GetBlockedClientReadyKey(RedictModuleCtx *ctx);

Available since: Redict 7.3.0

Get the key that is ready when the reply callback is called in the context of a client blocked by RedictModule_BlockClientOnKeys().

RedictModule_GetBlockedClientHandle #

RedictModuleBlockedClient *RedictModule_GetBlockedClientHandle(RedictModuleCtx *ctx);

Available since: Redict 7.3.0

Get the blocked client associated with a given context. This is useful in the reply and timeout callbacks of blocked clients, before sometimes the module has the blocked client handle references around, and wants to cleanup it.

RedictModule_BlockedClientDisconnected #

int RedictModule_BlockedClientDisconnected(RedictModuleCtx *ctx);

Available since: Redict 7.3.0

Return true if when the free callback of a blocked client is called, the reason for the client to be unblocked is that it disconnected while it was blocked.

Thread Safe Contexts #

RedictModule_GetThreadSafeContext #

RedictModuleCtx *RedictModule_GetThreadSafeContext(RedictModuleBlockedClient *bc);

Available since: Redict 7.3.0

Return a context which can be used inside threads to make Redict context calls with certain modules APIs. If ‘bc’ is not NULL then the module will be bound to a blocked client, and it will be possible to use the RedictModule_Reply* family of functions to accumulate a reply for when the client will be unblocked. Otherwise the thread safe context will be detached by a specific client.

To call non-reply APIs, the thread safe context must be prepared with:

RedictModule_ThreadSafeContextLock(ctx);
... make your call here ...
RedictModule_ThreadSafeContextUnlock(ctx);

This is not needed when using RedictModule_Reply* functions, assuming that a blocked client was used when the context was created, otherwise no RedictModule_Reply* call should be made at all.

NOTE: If you’re creating a detached thread safe context (bc is NULL), consider using RedictModule_GetDetachedThreadSafeContext which will also retain the module ID and thus be more useful for logging.

RedictModule_GetDetachedThreadSafeContext #

RedictModuleCtx *RedictModule_GetDetachedThreadSafeContext(RedictModuleCtx *ctx);

Available since: Redict 7.3.0

Return a detached thread safe context that is not associated with any specific blocked client, but is associated with the module’s context.

This is useful for modules that wish to hold a global context over a long term, for purposes such as logging.

RedictModule_FreeThreadSafeContext #

void RedictModule_FreeThreadSafeContext(RedictModuleCtx *ctx);

Available since: Redict 7.3.0

Release a thread safe context.

RedictModule_ThreadSafeContextLock #

void RedictModule_ThreadSafeContextLock(RedictModuleCtx *ctx);

Available since: Redict 7.3.0

Acquire the server lock before executing a thread safe API call. This is not needed for RedictModule_Reply* calls when there is a blocked client connected to the thread safe context.

RedictModule_ThreadSafeContextTryLock #

int RedictModule_ThreadSafeContextTryLock(RedictModuleCtx *ctx);

Available since: Redict 7.3.0

Similar to RedictModule_ThreadSafeContextLock but this function would not block if the server lock is already acquired.

If successful (lock acquired) REDICTMODULE_OK is returned, otherwise REDICTMODULE_ERR is returned and errno is set accordingly.

RedictModule_ThreadSafeContextUnlock #

void RedictModule_ThreadSafeContextUnlock(RedictModuleCtx *ctx);

Available since: Redict 7.3.0

Release the server lock after a thread safe API call was executed.

Module Keyspace Notifications API #

RedictModule_SubscribeToKeyspaceEvents #

int RedictModule_SubscribeToKeyspaceEvents(RedictModuleCtx *ctx,
                                           int types,
                                           RedictModuleNotificationFunc callback);

Available since: Redict 7.3.0

Subscribe to keyspace notifications. This is a low-level version of the keyspace-notifications API. A module can register callbacks to be notified when keyspace events occur.

Notification events are filtered by their type (string events, set events, etc), and the subscriber callback receives only events that match a specific mask of event types.

When subscribing to notifications with RedictModule_SubscribeToKeyspaceEvents the module must provide an event type-mask, denoting the events the subscriber is interested in. This can be an ORed mask of any of the following flags:

  • REDICTMODULE_NOTIFY_GENERIC: Generic commands like DEL, EXPIRE, RENAME
  • REDICTMODULE_NOTIFY_STRING: String events
  • REDICTMODULE_NOTIFY_LIST: List events
  • REDICTMODULE_NOTIFY_SET: Set events
  • REDICTMODULE_NOTIFY_HASH: Hash events
  • REDICTMODULE_NOTIFY_ZSET: Sorted Set events
  • REDICTMODULE_NOTIFY_EXPIRED: Expiration events
  • REDICTMODULE_NOTIFY_EVICTED: Eviction events
  • REDICTMODULE_NOTIFY_STREAM: Stream events
  • REDICTMODULE_NOTIFY_MODULE: Module types events
  • REDICTMODULE_NOTIFY_KEYMISS: Key-miss events Notice, key-miss event is the only type of event that is fired from within a read command. Performing RedictModule_Call with a write command from within this notification is wrong and discourage. It will cause the read command that trigger the event to be replicated to the AOF/Replica.
  • REDICTMODULE_NOTIFY_ALL: All events (Excluding REDICTMODULE_NOTIFY_KEYMISS)
  • REDICTMODULE_NOTIFY_LOADED: A special notification available only for modules, indicates that the key was loaded from persistence. Notice, when this event fires, the given key can not be retained, use RedictModule_CreateStringFromString instead.

We do not distinguish between key events and keyspace events, and it is up to the module to filter the actions taken based on the key.

The subscriber signature is:

int (*RedictModuleNotificationFunc) (RedictModuleCtx *ctx, int type,
                                    const char *event,
                                    RedictModuleString *key);

type is the event type bit, that must match the mask given at registration time. The event string is the actual command being executed, and key is the relevant Redict key.

Notification callback gets executed with a redict context that can not be used to send anything to the client, and has the db number where the event occurred as its selected db number.

Notice that it is not necessary to enable notifications in redict.conf for module notifications to work.

Warning: the notification callbacks are performed in a synchronous manner, so notification callbacks must to be fast, or they would slow Redict down. If you need to take long actions, use threads to offload them.

Moreover, the fact that the notification is executed synchronously means that the notification code will be executed in the middle on Redict logic (commands logic, eviction, expire). Changing the key space while the logic runs is dangerous and discouraged. In order to react to key space events with write actions, please refer to RedictModule_AddPostNotificationJob.

See https://redis.io/topics/notifications for more information.

RedictModule_AddPostNotificationJob #

int RedictModule_AddPostNotificationJob(RedictModuleCtx *ctx,
                                        RedictModulePostNotificationJobFunc callback,
                                        void *privdata,
                                        void (*free_privdata)(void*));

Available since: Redict 7.3.0

When running inside a key space notification callback, it is dangerous and highly discouraged to perform any write operation (See RedictModule_SubscribeToKeyspaceEvents). In order to still perform write actions in this scenario, Redict provides RedictModule_AddPostNotificationJob API. The API allows to register a job callback which Redict will call when the following condition are promised to be fulfilled:

  1. It is safe to perform any write operation.
  2. The job will be called atomically along side the key space notification.

Notice, one job might trigger key space notifications that will trigger more jobs. This raises a concerns of entering an infinite loops, we consider infinite loops as a logical bug that need to be fixed in the module, an attempt to protect against infinite loops by halting the execution could result in violation of the feature correctness and so Redict will make no attempt to protect the module from infinite loops.

free_pd’ can be NULL and in such case will not be used.

Return REDICTMODULE_OK on success and REDICTMODULE_ERR if was called while loading data from disk (AOF or RDB) or if the instance is a readonly replica.

RedictModule_GetNotifyKeyspaceEvents #

int RedictModule_GetNotifyKeyspaceEvents(void);

Available since: Redict 7.3.0

Get the configured bitmap of notify-keyspace-events (Could be used for additional filtering in RedictModuleNotificationFunc)

RedictModule_NotifyKeyspaceEvent #

int RedictModule_NotifyKeyspaceEvent(RedictModuleCtx *ctx,
                                     int type,
                                     const char *event,
                                     RedictModuleString *key);

Available since: Redict 7.3.0

Expose notifyKeyspaceEvent to modules

Modules Cluster API #

RedictModule_RegisterClusterMessageReceiver #

void RedictModule_RegisterClusterMessageReceiver(RedictModuleCtx *ctx,
                                                 uint8_t type,
                                                 RedictModuleClusterMessageReceiver callback);

Available since: Redict 7.3.0

Register a callback receiver for cluster messages of type ’type’. If there was already a registered callback, this will replace the callback function with the one provided, otherwise if the callback is set to NULL and there is already a callback for this function, the callback is unregistered (so this API call is also used in order to delete the receiver).

RedictModule_SendClusterMessage #

int RedictModule_SendClusterMessage(RedictModuleCtx *ctx,
                                    const char *target_id,
                                    uint8_t type,
                                    const char *msg,
                                    uint32_t len);

Available since: Redict 7.3.0

Send a message to all the nodes in the cluster if target is NULL, otherwise at the specified target, which is a REDICTMODULE_NODE_ID_LEN bytes node ID, as returned by the receiver callback or by the nodes iteration functions.

The function returns REDICTMODULE_OK if the message was successfully sent, otherwise if the node is not connected or such node ID does not map to any known cluster node, REDICTMODULE_ERR is returned.

RedictModule_GetClusterNodesList #

char **RedictModule_GetClusterNodesList(RedictModuleCtx *ctx,
                                        size_t *numnodes);

Available since: Redict 7.3.0

Return an array of string pointers, each string pointer points to a cluster node ID of exactly REDICTMODULE_NODE_ID_LEN bytes (without any null term). The number of returned node IDs is stored into *numnodes. However if this function is called by a module not running an a Redict instance with Redict Cluster enabled, NULL is returned instead.

The IDs returned can be used with RedictModule_GetClusterNodeInfo() in order to get more information about single node.

The array returned by this function must be freed using the function RedictModule_FreeClusterNodesList().

Example:

size_t count, j;
char **ids = RedictModule_GetClusterNodesList(ctx,&count);
for (j = 0; j < count; j++) {
    RedictModule_Log(ctx,"notice","Node %.*s",
        REDICTMODULE_NODE_ID_LEN,ids[j]);
}
RedictModule_FreeClusterNodesList(ids);

RedictModule_FreeClusterNodesList #

void RedictModule_FreeClusterNodesList(char **ids);

Available since: Redict 7.3.0

Free the node list obtained with RedictModule_GetClusterNodesList.

RedictModule_GetMyClusterID #

const char *RedictModule_GetMyClusterID(void);

Available since: Redict 7.3.0

Return this node ID (REDICTMODULE_CLUSTER_ID_LEN bytes) or NULL if the cluster is disabled.

RedictModule_GetClusterSize #

size_t RedictModule_GetClusterSize(void);

Available since: Redict 7.3.0

Return the number of nodes in the cluster, regardless of their state (handshake, noaddress, …) so that the number of active nodes may actually be smaller, but not greater than this number. If the instance is not in cluster mode, zero is returned.

RedictModule_GetClusterNodeInfo #

int RedictModule_GetClusterNodeInfo(RedictModuleCtx *ctx,
                                    const char *id,
                                    char *ip,
                                    char *master_id,
                                    int *port,
                                    int *flags);

Available since: Redict 7.3.0

Populate the specified info for the node having as ID the specified ‘id’, then returns REDICTMODULE_OK. Otherwise if the format of node ID is invalid or the node ID does not exist from the POV of this local node, REDICTMODULE_ERR is returned.

The arguments ip, master_id, port and flags can be NULL in case we don’t need to populate back certain info. If an ip and master_id (only populated if the instance is a slave) are specified, they point to buffers holding at least REDICTMODULE_NODE_ID_LEN bytes. The strings written back as ip and master_id are not null terminated.

The list of flags reported is the following:

  • REDICTMODULE_NODE_MYSELF: This node
  • REDICTMODULE_NODE_MASTER: The node is a master
  • REDICTMODULE_NODE_SLAVE: The node is a replica
  • REDICTMODULE_NODE_PFAIL: We see the node as failing
  • REDICTMODULE_NODE_FAIL: The cluster agrees the node is failing
  • REDICTMODULE_NODE_NOFAILOVER: The slave is configured to never failover

RedictModule_SetClusterFlags #

void RedictModule_SetClusterFlags(RedictModuleCtx *ctx, uint64_t flags);

Available since: Redict 7.3.0

Set Redict Cluster flags in order to change the normal behavior of Redict Cluster, especially with the goal of disabling certain functions. This is useful for modules that use the Cluster API in order to create a different distributed system, but still want to use the Redict Cluster message bus. Flags that can be set:

  • CLUSTER_MODULE_FLAG_NO_FAILOVER
  • CLUSTER_MODULE_FLAG_NO_REDIRECTION

With the following effects:

  • NO_FAILOVER: prevent Redict Cluster slaves from failing over a dead master. Also disables the replica migration feature.

  • NO_REDIRECTION: Every node will accept any key, without trying to perform partitioning according to the Redict Cluster algorithm. Slots information will still be propagated across the cluster, but without effect.

RedictModule_ClusterKeySlot #

unsigned int RedictModule_ClusterKeySlot(RedictModuleString *key);

Available since: Redict 7.3.0

Returns the cluster slot of a key, similar to the CLUSTER KEYSLOT command. This function works even if cluster mode is not enabled.

RedictModule_ClusterCanonicalKeyNameInSlot #

const char *RedictModule_ClusterCanonicalKeyNameInSlot(unsigned int slot);

Available since: Redict 7.3.0

Returns a short string that can be used as a key or as a hash tag in a key, such that the key maps to the given cluster slot. Returns NULL if slot is not a valid slot.

Modules Timers API #

Module timers are a high precision “green timers” abstraction where every module can register even millions of timers without problems, even if the actual event loop will just have a single timer that is used to awake the module timers subsystem in order to process the next event.

All the timers are stored into a radix tree, ordered by expire time, when the main Redict event loop timer callback is called, we try to process all the timers already expired one after the other. Then we re-enter the event loop registering a timer that will expire when the next to process module timer will expire.

Every time the list of active timers drops to zero, we unregister the main event loop timer, so that there is no overhead when such feature is not used.

RedictModule_CreateTimer #

RedictModuleTimerID RedictModule_CreateTimer(RedictModuleCtx *ctx,
                                             mstime_t period,
                                             RedictModuleTimerProc callback,
                                             void *data);

Available since: Redict 7.3.0

Create a new timer that will fire after period milliseconds, and will call the specified function using data as argument. The returned timer ID can be used to get information from the timer or to stop it before it fires. Note that for the common use case of a repeating timer (Re-registration of the timer inside the RedictModuleTimerProc callback) it matters when this API is called: If it is called at the beginning of ‘callback’ it means the event will triggered every ‘period’. If it is called at the end of ‘callback’ it means there will ‘period’ milliseconds gaps between events. (If the time it takes to execute ‘callback’ is negligible the two statements above mean the same)

RedictModule_StopTimer #

int RedictModule_StopTimer(RedictModuleCtx *ctx,
                           RedictModuleTimerID id,
                           void **data);

Available since: Redict 7.3.0

Stop a timer, returns REDICTMODULE_OK if the timer was found, belonged to the calling module, and was stopped, otherwise REDICTMODULE_ERR is returned. If not NULL, the data pointer is set to the value of the data argument when the timer was created.

RedictModule_GetTimerInfo #

int RedictModule_GetTimerInfo(RedictModuleCtx *ctx,
                              RedictModuleTimerID id,
                              uint64_t *remaining,
                              void **data);

Available since: Redict 7.3.0

Obtain information about a timer: its remaining time before firing (in milliseconds), and the private data pointer associated with the timer. If the timer specified does not exist or belongs to a different module no information is returned and the function returns REDICTMODULE_ERR, otherwise REDICTMODULE_OK is returned. The arguments remaining or data can be NULL if the caller does not need certain information.

Modules EventLoop API #

RedictModule_EventLoopAdd #

int RedictModule_EventLoopAdd(int fd,
                              int mask,
                              RedictModuleEventLoopFunc func,
                              void *user_data);

Available since: Redict 7.3.0

Add a pipe / socket event to the event loop.

  • mask must be one of the following values:

    • REDICTMODULE_EVENTLOOP_READABLE
    • REDICTMODULE_EVENTLOOP_WRITABLE
    • REDICTMODULE_EVENTLOOP_READABLE | REDICTMODULE_EVENTLOOP_WRITABLE

On success REDICTMODULE_OK is returned, otherwise REDICTMODULE_ERR is returned and errno is set to the following values:

  • ERANGE: fd is negative or higher than maxclients Redict config.
  • EINVAL: callback is NULL or mask value is invalid.

errno might take other values in case of an internal error.

Example:

void onReadable(int fd, void *user_data, int mask) {
    char buf[32];
    int bytes = read(fd,buf,sizeof(buf));
    printf("Read %d bytes \n", bytes);
}
RedictModule_EventLoopAdd(fd, REDICTMODULE_EVENTLOOP_READABLE, onReadable, NULL);

RedictModule_EventLoopDel #

int RedictModule_EventLoopDel(int fd, int mask);

Available since: Redict 7.3.0

Delete a pipe / socket event from the event loop.

  • mask must be one of the following values:

    • REDICTMODULE_EVENTLOOP_READABLE
    • REDICTMODULE_EVENTLOOP_WRITABLE
    • REDICTMODULE_EVENTLOOP_READABLE | REDICTMODULE_EVENTLOOP_WRITABLE

On success REDICTMODULE_OK is returned, otherwise REDICTMODULE_ERR is returned and errno is set to the following values:

  • ERANGE: fd is negative or higher than maxclients Redict config.
  • EINVAL: mask value is invalid.

RedictModule_EventLoopAddOneShot #

int RedictModule_EventLoopAddOneShot(RedictModuleEventLoopOneShotFunc func,
                                     void *user_data);

Available since: Redict 7.3.0

This function can be called from other threads to trigger callback on Redict main thread. On success REDICTMODULE_OK is returned. If func is NULL REDICTMODULE_ERR is returned and errno is set to EINVAL.

Modules ACL API #

Implements a hook into the authentication and authorization within Redict.

RedictModule_CreateModuleUser #

RedictModuleUser *RedictModule_CreateModuleUser(const char *name);

Available since: Redict 7.3.0

Creates a Redict ACL user that the module can use to authenticate a client. After obtaining the user, the module should set what such user can do using the RedictModule_SetUserACL() function. Once configured, the user can be used in order to authenticate a connection, with the specified ACL rules, using the RedictModule_AuthClientWithUser() function.

Note that:

  • Users created here are not listed by the ACL command.
  • Users created here are not checked for duplicated name, so it’s up to the module calling this function to take care of not creating users with the same name.
  • The created user can be used to authenticate multiple Redict connections.

The caller can later free the user using the function RedictModule_FreeModuleUser(). When this function is called, if there are still clients authenticated with this user, they are disconnected. The function to free the user should only be used when the caller really wants to invalidate the user to define a new one with different capabilities.

RedictModule_FreeModuleUser #

int RedictModule_FreeModuleUser(RedictModuleUser *user);

Available since: Redict 7.3.0

Frees a given user and disconnects all of the clients that have been authenticated with it. See RedictModule_CreateModuleUser for detailed usage.

RedictModule_SetModuleUserACL #

int RedictModule_SetModuleUserACL(RedictModuleUser *user, const char* acl);

Available since: Redict 7.3.0

Sets the permissions of a user created through the redict module interface. The syntax is the same as ACL SETUSER, so refer to the documentation in acl.c for more information. See RedictModule_CreateModuleUser for detailed usage.

Returns REDICTMODULE_OK on success and REDICTMODULE_ERR on failure and will set an errno describing why the operation failed.

RedictModule_SetModuleUserACLString #

int RedictModule_SetModuleUserACLString(RedictModuleCtx *ctx,
                                        RedictModuleUser *user,
                                        const char *acl,
                                        RedictModuleString **error);

Available since: Redict 7.3.0

Sets the permission of a user with a complete ACL string, such as one would use on the redict ACL SETUSER command line API. This differs from RedictModule_SetModuleUserACL, which only takes single ACL operations at a time.

Returns REDICTMODULE_OK on success and REDICTMODULE_ERR on failure if a RedictModuleString is provided in error, a string describing the error will be returned

RedictModule_GetModuleUserACLString #

RedictModuleString *RedictModule_GetModuleUserACLString(RedictModuleUser *user);

Available since: Redict 7.3.0

Get the ACL string for a given user Returns a RedictModuleString

RedictModule_GetCurrentUserName #

RedictModuleString *RedictModule_GetCurrentUserName(RedictModuleCtx *ctx);

Available since: Redict 7.3.0

Retrieve the user name of the client connection behind the current context. The user name can be used later, in order to get a RedictModuleUser. See more information in RedictModule_GetModuleUserFromUserName.

The returned string must be released with RedictModule_FreeString() or by enabling automatic memory management.

RedictModule_GetModuleUserFromUserName #

RedictModuleUser *RedictModule_GetModuleUserFromUserName(RedictModuleString *name);

Available since: Redict 7.3.0

A RedictModuleUser can be used to check if command, key or channel can be executed or accessed according to the ACLs rules associated with that user. When a Module wants to do ACL checks on a general ACL user (not created by RedictModule_CreateModuleUser), it can get the RedictModuleUser from this API, based on the user name retrieved by RedictModule_GetCurrentUserName.

Since a general ACL user can be deleted at any time, this RedictModuleUser should be used only in the context where this function was called. In order to do ACL checks out of that context, the Module can store the user name, and call this API at any other context.

Returns NULL if the user is disabled or the user does not exist. The caller should later free the user using the function RedictModule_FreeModuleUser().

RedictModule_ACLCheckCommandPermissions #

int RedictModule_ACLCheckCommandPermissions(RedictModuleUser *user,
                                            RedictModuleString **argv,
                                            int argc);

Available since: Redict 7.3.0

Checks if the command can be executed by the user, according to the ACLs associated with it.

On success a REDICTMODULE_OK is returned, otherwise REDICTMODULE_ERR is returned and errno is set to the following values:

  • ENOENT: Specified command does not exist.
  • EACCES: Command cannot be executed, according to ACL rules

RedictModule_ACLCheckKeyPermissions #

int RedictModule_ACLCheckKeyPermissions(RedictModuleUser *user,
                                        RedictModuleString *key,
                                        int flags);

Available since: Redict 7.3.0

Check if the key can be accessed by the user according to the ACLs attached to the user and the flags representing the key access. The flags are the same that are used in the keyspec for logical operations. These flags are documented in RedictModule_SetCommandInfo as the REDICTMODULE_CMD_KEY_ACCESS, REDICTMODULE_CMD_KEY_UPDATE, REDICTMODULE_CMD_KEY_INSERT, and REDICTMODULE_CMD_KEY_DELETE flags.

If no flags are supplied, the user is still required to have some access to the key for this command to return successfully.

If the user is able to access the key then REDICTMODULE_OK is returned, otherwise REDICTMODULE_ERR is returned and errno is set to one of the following values:

  • EINVAL: The provided flags are invalid.
  • EACCESS: The user does not have permission to access the key.

RedictModule_ACLCheckChannelPermissions #

int RedictModule_ACLCheckChannelPermissions(RedictModuleUser *user,
                                            RedictModuleString *ch,
                                            int flags);

Available since: Redict 7.3.0

Check if the pubsub channel can be accessed by the user based off of the given access flags. See RedictModule_ChannelAtPosWithFlags for more information about the possible flags that can be passed in.

If the user is able to access the pubsub channel then REDICTMODULE_OK is returned, otherwise REDICTMODULE_ERR is returned and errno is set to one of the following values:

  • EINVAL: The provided flags are invalid.
  • EACCESS: The user does not have permission to access the pubsub channel.

RedictModule_ACLAddLogEntry #

int RedictModule_ACLAddLogEntry(RedictModuleCtx *ctx,
                                RedictModuleUser *user,
                                RedictModuleString *object,
                                RedictModuleACLLogEntryReason reason);

Available since: Redict 7.3.0

Adds a new entry in the ACL log. Returns REDICTMODULE_OK on success and REDICTMODULE_ERR on error.

For more information about ACL log, please refer to https://redis.io/commands/acl-log

RedictModule_ACLAddLogEntryByUserName #

int RedictModule_ACLAddLogEntryByUserName(RedictModuleCtx *ctx,
                                          RedictModuleString *username,
                                          RedictModuleString *object,
                                          RedictModuleACLLogEntryReason reason);

Available since: Redict 7.3.0

Adds a new entry in the ACL log with the username RedictModuleString provided. Returns REDICTMODULE_OK on success and REDICTMODULE_ERR on error.

For more information about ACL log, please refer to https://redis.io/commands/acl-log

RedictModule_AuthenticateClientWithUser #

int RedictModule_AuthenticateClientWithUser(RedictModuleCtx *ctx,
                                            RedictModuleUser *module_user,
                                            RedictModuleUserChangedFunc callback,
                                            void *privdata,
                                            uint64_t *client_id);

Available since: Redict 7.3.0

Authenticate the current context’s user with the provided redict acl user. Returns REDICTMODULE_ERR if the user is disabled.

See authenticateClientWithUser for information about callback, client_id, and general usage for authentication.

RedictModule_AuthenticateClientWithACLUser #

int RedictModule_AuthenticateClientWithACLUser(RedictModuleCtx *ctx,
                                               const char *name,
                                               size_t len,
                                               RedictModuleUserChangedFunc callback,
                                               void *privdata,
                                               uint64_t *client_id);

Available since: Redict 7.3.0

Authenticate the current context’s user with the provided redict acl user. Returns REDICTMODULE_ERR if the user is disabled or the user does not exist.

See authenticateClientWithUser for information about callback, client_id, and general usage for authentication.

RedictModule_DeauthenticateAndCloseClient #

int RedictModule_DeauthenticateAndCloseClient(RedictModuleCtx *ctx,
                                              uint64_t client_id);

Available since: Redict 7.3.0

Deauthenticate and close the client. The client resources will not be immediately freed, but will be cleaned up in a background job. This is the recommended way to deauthenticate a client since most clients can’t handle users becoming deauthenticated. Returns REDICTMODULE_ERR when the client doesn’t exist and REDICTMODULE_OK when the operation was successful.

The client ID is returned from the RedictModule_AuthenticateClientWithUser and RedictModule_AuthenticateClientWithACLUser APIs, but can be obtained through the CLIENT api or through server events.

This function is not thread safe, and must be executed within the context of a command or thread safe context.

RedictModule_RedactClientCommandArgument #

int RedictModule_RedactClientCommandArgument(RedictModuleCtx *ctx, int pos);

Available since: Redict 7.3.0

Redact the client command argument specified at the given position. Redacted arguments are obfuscated in user facing commands such as SLOWLOG or MONITOR, as well as never being written to server logs. This command may be called multiple times on the same position.

Note that the command name, position 0, can not be redacted.

Returns REDICTMODULE_OK if the argument was redacted and REDICTMODULE_ERR if there was an invalid parameter passed in or the position is outside the client argument range.

RedictModule_GetClientCertificate #

RedictModuleString *RedictModule_GetClientCertificate(RedictModuleCtx *ctx,
                                                      uint64_t client_id);

Available since: Redict 7.3.0

Return the X.509 client-side certificate used by the client to authenticate this connection.

The return value is an allocated RedictModuleString that is a X.509 certificate encoded in PEM (Base64) format. It should be freed (or auto-freed) by the caller.

A NULL value is returned in the following conditions:

  • Connection ID does not exist
  • Connection is not a TLS connection
  • Connection is a TLS connection but no client certificate was used

Modules Dictionary API #

Implements a sorted dictionary (actually backed by a radix tree) with the usual get / set / del / num-items API, together with an iterator capable of going back and forth.

RedictModule_CreateDict #

RedictModuleDict *RedictModule_CreateDict(RedictModuleCtx *ctx);

Available since: Redict 7.3.0

Create a new dictionary. The ‘ctx’ pointer can be the current module context or NULL, depending on what you want. Please follow the following rules:

  1. Use a NULL context if you plan to retain a reference to this dictionary that will survive the time of the module callback where you created it.
  2. Use a NULL context if no context is available at the time you are creating the dictionary (of course…).
  3. However use the current callback context as ‘ctx’ argument if the dictionary time to live is just limited to the callback scope. In this case, if enabled, you can enjoy the automatic memory management that will reclaim the dictionary memory, as well as the strings returned by the Next / Prev dictionary iterator calls.

RedictModule_FreeDict #

void RedictModule_FreeDict(RedictModuleCtx *ctx, RedictModuleDict *d);

Available since: Redict 7.3.0

Free a dictionary created with RedictModule_CreateDict(). You need to pass the context pointer ‘ctx’ only if the dictionary was created using the context instead of passing NULL.

RedictModule_DictSize #

uint64_t RedictModule_DictSize(RedictModuleDict *d);

Available since: Redict 7.3.0

Return the size of the dictionary (number of keys).

RedictModule_DictSetC #

int RedictModule_DictSetC(RedictModuleDict *d,
                          void *key,
                          size_t keylen,
                          void *ptr);

Available since: Redict 7.3.0

Store the specified key into the dictionary, setting its value to the pointer ‘ptr’. If the key was added with success, since it did not already exist, REDICTMODULE_OK is returned. Otherwise if the key already exists the function returns REDICTMODULE_ERR.

RedictModule_DictReplaceC #

int RedictModule_DictReplaceC(RedictModuleDict *d,
                              void *key,
                              size_t keylen,
                              void *ptr);

Available since: Redict 7.3.0

Like RedictModule_DictSetC() but will replace the key with the new value if the key already exists.

RedictModule_DictSet #

int RedictModule_DictSet(RedictModuleDict *d,
                         RedictModuleString *key,
                         void *ptr);

Available since: Redict 7.3.0

Like RedictModule_DictSetC() but takes the key as a RedictModuleString.

RedictModule_DictReplace #

int RedictModule_DictReplace(RedictModuleDict *d,
                             RedictModuleString *key,
                             void *ptr);

Available since: Redict 7.3.0

Like RedictModule_DictReplaceC() but takes the key as a RedictModuleString.

RedictModule_DictGetC #

void *RedictModule_DictGetC(RedictModuleDict *d,
                            void *key,
                            size_t keylen,
                            int *nokey);

Available since: Redict 7.3.0

Return the value stored at the specified key. The function returns NULL both in the case the key does not exist, or if you actually stored NULL at key. So, optionally, if the ’nokey’ pointer is not NULL, it will be set by reference to 1 if the key does not exist, or to 0 if the key exists.

RedictModule_DictGet #

void *RedictModule_DictGet(RedictModuleDict *d,
                           RedictModuleString *key,
                           int *nokey);

Available since: Redict 7.3.0

Like RedictModule_DictGetC() but takes the key as a RedictModuleString.

RedictModule_DictDelC #

int RedictModule_DictDelC(RedictModuleDict *d,
                          void *key,
                          size_t keylen,
                          void *oldval);

Available since: Redict 7.3.0

Remove the specified key from the dictionary, returning REDICTMODULE_OK if the key was found and deleted, or REDICTMODULE_ERR if instead there was no such key in the dictionary. When the operation is successful, if ‘oldval’ is not NULL, then ‘*oldval’ is set to the value stored at the key before it was deleted. Using this feature it is possible to get a pointer to the value (for instance in order to release it), without having to call RedictModule_DictGet() before deleting the key.

RedictModule_DictDel #

int RedictModule_DictDel(RedictModuleDict *d,
                         RedictModuleString *key,
                         void *oldval);

Available since: Redict 7.3.0

Like RedictModule_DictDelC() but gets the key as a RedictModuleString.

RedictModule_DictIteratorStartC #

RedictModuleDictIter *RedictModule_DictIteratorStartC(RedictModuleDict *d,
                                                      const char *op,
                                                      void *key,
                                                      size_t keylen);

Available since: Redict 7.3.0

Return an iterator, setup in order to start iterating from the specified key by applying the operator ‘op’, which is just a string specifying the comparison operator to use in order to seek the first element. The operators available are:

  • ^ – Seek the first (lexicographically smaller) key.
  • $ – Seek the last (lexicographically bigger) key.
  • > – Seek the first element greater than the specified key.
  • >= – Seek the first element greater or equal than the specified key.
  • < – Seek the first element smaller than the specified key.
  • <= – Seek the first element smaller or equal than the specified key.
  • == – Seek the first element matching exactly the specified key.

Note that for ^ and $ the passed key is not used, and the user may just pass NULL with a length of 0.

If the element to start the iteration cannot be seeked based on the key and operator passed, RedictModule_DictNext() / Prev() will just return REDICTMODULE_ERR at the first call, otherwise they’ll produce elements.

RedictModule_DictIteratorStart #

RedictModuleDictIter *RedictModule_DictIteratorStart(RedictModuleDict *d,
                                                     const char *op,
                                                     RedictModuleString *key);

Available since: Redict 7.3.0

Exactly like RedictModule_DictIteratorStartC, but the key is passed as a RedictModuleString.

RedictModule_DictIteratorStop #

void RedictModule_DictIteratorStop(RedictModuleDictIter *di);

Available since: Redict 7.3.0

Release the iterator created with RedictModule_DictIteratorStart(). This call is mandatory otherwise a memory leak is introduced in the module.

RedictModule_DictIteratorReseekC #

int RedictModule_DictIteratorReseekC(RedictModuleDictIter *di,
                                     const char *op,
                                     void *key,
                                     size_t keylen);

Available since: Redict 7.3.0

After its creation with RedictModule_DictIteratorStart(), it is possible to change the currently selected element of the iterator by using this API call. The result based on the operator and key is exactly like the function RedictModule_DictIteratorStart(), however in this case the return value is just REDICTMODULE_OK in case the seeked element was found, or REDICTMODULE_ERR in case it was not possible to seek the specified element. It is possible to reseek an iterator as many times as you want.

RedictModule_DictIteratorReseek #

int RedictModule_DictIteratorReseek(RedictModuleDictIter *di,
                                    const char *op,
                                    RedictModuleString *key);

Available since: Redict 7.3.0

Like RedictModule_DictIteratorReseekC() but takes the key as a RedictModuleString.

RedictModule_DictNextC #

void *RedictModule_DictNextC(RedictModuleDictIter *di,
                             size_t *keylen,
                             void **dataptr);

Available since: Redict 7.3.0

Return the current item of the dictionary iterator di and steps to the next element. If the iterator already yield the last element and there are no other elements to return, NULL is returned, otherwise a pointer to a string representing the key is provided, and the *keylen length is set by reference (if keylen is not NULL). The *dataptr, if not NULL is set to the value of the pointer stored at the returned key as auxiliary data (as set by the RedictModule_DictSet API).

Usage example:

 ... create the iterator here ...
 char *key;
 void *data;
 while((key = RedictModule_DictNextC(iter,&keylen,&data)) != NULL) {
     printf("%.*s %p\n", (int)keylen, key, data);
 }

The returned pointer is of type void because sometimes it makes sense to cast it to a char* sometimes to an unsigned char* depending on the fact it contains or not binary data, so this API ends being more comfortable to use.

The validity of the returned pointer is until the next call to the next/prev iterator step. Also the pointer is no longer valid once the iterator is released.

RedictModule_DictPrevC #

void *RedictModule_DictPrevC(RedictModuleDictIter *di,
                             size_t *keylen,
                             void **dataptr);

Available since: Redict 7.3.0

This function is exactly like RedictModule_DictNext() but after returning the currently selected element in the iterator, it selects the previous element (lexicographically smaller) instead of the next one.

RedictModule_DictNext #

RedictModuleString *RedictModule_DictNext(RedictModuleCtx *ctx,
                                          RedictModuleDictIter *di,
                                          void **dataptr);

Available since: Redict 7.3.0

Like RedictModuleNextC(), but instead of returning an internally allocated buffer and key length, it returns directly a module string object allocated in the specified context ‘ctx’ (that may be NULL exactly like for the main API RedictModule_CreateString).

The returned string object should be deallocated after use, either manually or by using a context that has automatic memory management active.

RedictModule_DictPrev #

RedictModuleString *RedictModule_DictPrev(RedictModuleCtx *ctx,
                                          RedictModuleDictIter *di,
                                          void **dataptr);

Available since: Redict 7.3.0

Like RedictModule_DictNext() but after returning the currently selected element in the iterator, it selects the previous element (lexicographically smaller) instead of the next one.

RedictModule_DictCompareC #

int RedictModule_DictCompareC(RedictModuleDictIter *di,
                              const char *op,
                              void *key,
                              size_t keylen);

Available since: Redict 7.3.0

Compare the element currently pointed by the iterator to the specified element given by key/keylen, according to the operator ‘op’ (the set of valid operators are the same valid for RedictModule_DictIteratorStart). If the comparison is successful the command returns REDICTMODULE_OK otherwise REDICTMODULE_ERR is returned.

This is useful when we want to just emit a lexicographical range, so in the loop, as we iterate elements, we can also check if we are still on range.

The function return REDICTMODULE_ERR if the iterator reached the end of elements condition as well.

RedictModule_DictCompare #

int RedictModule_DictCompare(RedictModuleDictIter *di,
                             const char *op,
                             RedictModuleString *key);

Available since: Redict 7.3.0

Like RedictModule_DictCompareC but gets the key to compare with the current iterator key as a RedictModuleString.

Modules Info fields #

RedictModule_InfoAddSection #

int RedictModule_InfoAddSection(RedictModuleInfoCtx *ctx, const char *name);

Available since: Redict 7.3.0

Used to start a new section, before adding any fields. the section name will be prefixed by <modulename>_ and must only include A-Z,a-z,0-9. NULL or empty string indicates the default section (only <modulename>) is used. When return value is REDICTMODULE_ERR, the section should and will be skipped.

RedictModule_InfoBeginDictField #

int RedictModule_InfoBeginDictField(RedictModuleInfoCtx *ctx,
                                    const char *name);

Available since: Redict 7.3.0

Starts a dict field, similar to the ones in INFO KEYSPACE. Use normal RedictModule_InfoAddField* functions to add the items to this field, and terminate with RedictModule_InfoEndDictField.

RedictModule_InfoEndDictField #

int RedictModule_InfoEndDictField(RedictModuleInfoCtx *ctx);

Available since: Redict 7.3.0

Ends a dict field, see RedictModule_InfoBeginDictField

RedictModule_InfoAddFieldString #

int RedictModule_InfoAddFieldString(RedictModuleInfoCtx *ctx,
                                    const char *field,
                                    RedictModuleString *value);

Available since: Redict 7.3.0

Used by RedictModuleInfoFunc to add info fields. Each field will be automatically prefixed by <modulename>_. Field names or values must not include \r\n or :.

RedictModule_InfoAddFieldCString #

int RedictModule_InfoAddFieldCString(RedictModuleInfoCtx *ctx,
                                     const char *field,
                                     const char *value);

Available since: Redict 7.3.0

See RedictModule_InfoAddFieldString().

RedictModule_InfoAddFieldDouble #

int RedictModule_InfoAddFieldDouble(RedictModuleInfoCtx *ctx,
                                    const char *field,
                                    double value);

Available since: Redict 7.3.0

See RedictModule_InfoAddFieldString().

RedictModule_InfoAddFieldLongLong #

int RedictModule_InfoAddFieldLongLong(RedictModuleInfoCtx *ctx,
                                      const char *field,
                                      long long value);

Available since: Redict 7.3.0

See RedictModule_InfoAddFieldString().

RedictModule_InfoAddFieldULongLong #

int RedictModule_InfoAddFieldULongLong(RedictModuleInfoCtx *ctx,
                                       const char *field,
                                       unsigned long long value);

Available since: Redict 7.3.0

See RedictModule_InfoAddFieldString().

RedictModule_RegisterInfoFunc #

int RedictModule_RegisterInfoFunc(RedictModuleCtx *ctx,
                                  RedictModuleInfoFunc cb);

Available since: Redict 7.3.0

Registers callback for the INFO command. The callback should add INFO fields by calling the RedictModule_InfoAddField*() functions.

RedictModule_GetServerInfo #

RedictModuleServerInfoData *RedictModule_GetServerInfo(RedictModuleCtx *ctx,
                                                       const char *section);

Available since: Redict 7.3.0

Get information about the server similar to the one that returns from the INFO command. This function takes an optional ‘section’ argument that may be NULL. The return value holds the output and can be used with RedictModule_ServerInfoGetField and alike to get the individual fields. When done, it needs to be freed with RedictModule_FreeServerInfo or with the automatic memory management mechanism if enabled.

RedictModule_FreeServerInfo #

void RedictModule_FreeServerInfo(RedictModuleCtx *ctx,
                                 RedictModuleServerInfoData *data);

Available since: Redict 7.3.0

Free data created with RedictModule_GetServerInfo(). You need to pass the context pointer ‘ctx’ only if the dictionary was created using the context instead of passing NULL.

RedictModule_ServerInfoGetField #

RedictModuleString *RedictModule_ServerInfoGetField(RedictModuleCtx *ctx,
                                                    RedictModuleServerInfoData *data,
                                                    const char* field);

Available since: Redict 7.3.0

Get the value of a field from data collected with RedictModule_GetServerInfo(). You need to pass the context pointer ‘ctx’ only if you want to use auto memory mechanism to release the returned string. Return value will be NULL if the field was not found.

RedictModule_ServerInfoGetFieldC #

const char *RedictModule_ServerInfoGetFieldC(RedictModuleServerInfoData *data,
                                             const char* field);

Available since: Redict 7.3.0

Similar to RedictModule_ServerInfoGetField, but returns a char* which should not be freed but the caller.

RedictModule_ServerInfoGetFieldSigned #

long long RedictModule_ServerInfoGetFieldSigned(RedictModuleServerInfoData *data,
                                                const char* field,
                                                int *out_err);

Available since: Redict 7.3.0

Get the value of a field from data collected with RedictModule_GetServerInfo(). If the field is not found, or is not numerical or out of range, return value will be 0, and the optional out_err argument will be set to REDICTMODULE_ERR.

RedictModule_ServerInfoGetFieldUnsigned #

unsigned long long RedictModule_ServerInfoGetFieldUnsigned(RedictModuleServerInfoData *data,
                                                           const char* field,
                                                           int *out_err);

Available since: Redict 7.3.0

Get the value of a field from data collected with RedictModule_GetServerInfo(). If the field is not found, or is not numerical or out of range, return value will be 0, and the optional out_err argument will be set to REDICTMODULE_ERR.

RedictModule_ServerInfoGetFieldDouble #

double RedictModule_ServerInfoGetFieldDouble(RedictModuleServerInfoData *data,
                                             const char* field,
                                             int *out_err);

Available since: Redict 7.3.0

Get the value of a field from data collected with RedictModule_GetServerInfo(). If the field is not found, or is not a double, return value will be 0, and the optional out_err argument will be set to REDICTMODULE_ERR.

Modules utility APIs #

RedictModule_GetRandomBytes #

void RedictModule_GetRandomBytes(unsigned char *dst, size_t len);

Available since: Redict 7.3.0

Return random bytes using SHA1 in counter mode with a /dev/urandom initialized seed. This function is fast so can be used to generate many bytes without any effect on the operating system entropy pool. Currently this function is not thread safe.

RedictModule_GetRandomHexChars #

void RedictModule_GetRandomHexChars(char *dst, size_t len);

Available since: Redict 7.3.0

Like RedictModule_GetRandomBytes() but instead of setting the string to random bytes the string is set to random characters in the in the hex charset [0-9a-f].

Modules API exporting / importing #

RedictModule_ExportSharedAPI #

int RedictModule_ExportSharedAPI(RedictModuleCtx *ctx,
                                 const char *apiname,
                                 void *func);

Available since: Redict 7.3.0

This function is called by a module in order to export some API with a given name. Other modules will be able to use this API by calling the symmetrical function RedictModule_GetSharedAPI() and casting the return value to the right function pointer.

The function will return REDICTMODULE_OK if the name is not already taken, otherwise REDICTMODULE_ERR will be returned and no operation will be performed.

IMPORTANT: the apiname argument should be a string literal with static lifetime. The API relies on the fact that it will always be valid in the future.

RedictModule_GetSharedAPI #

void *RedictModule_GetSharedAPI(RedictModuleCtx *ctx, const char *apiname);

Available since: Redict 7.3.0

Request an exported API pointer. The return value is just a void pointer that the caller of this function will be required to cast to the right function pointer, so this is a private contract between modules.

If the requested API is not available then NULL is returned. Because modules can be loaded at different times with different order, this function calls should be put inside some module generic API registering step, that is called every time a module attempts to execute a command that requires external APIs: if some API cannot be resolved, the command should return an error.

Here is an example:

int ... myCommandImplementation(void) {
   if (getExternalAPIs() == 0) {
        reply with an error here if we cannot have the APIs
   }
   // Use the API:
   myFunctionPointer(foo);
}

And the function registerAPI() is:

int getExternalAPIs(void) {
    static int api_loaded = 0;
    if (api_loaded != 0) return 1; // APIs already resolved.

    myFunctionPointer = RedictModule_GetSharedAPI("...");
    if (myFunctionPointer == NULL) return 0;

    return 1;
}

Module Command Filter API #

RedictModule_RegisterCommandFilter #

RedictModuleCommandFilter *RedictModule_RegisterCommandFilter(RedictModuleCtx *ctx,
                                                              RedictModuleCommandFilterFunc callback,
                                                              int flags);

Available since: Redict 7.3.0

Register a new command filter function.

Command filtering makes it possible for modules to extend Redict by plugging into the execution flow of all commands.

A registered filter gets called before Redict executes any command. This includes both core Redict commands and commands registered by any module. The filter applies in all execution paths including:

  1. Invocation by a client.
  2. Invocation through RedictModule_Call() by any module.
  3. Invocation through Lua redict.call().
  4. Replication of a command from a master.

The filter executes in a special filter context, which is different and more limited than a RedictModuleCtx. Because the filter affects any command, it must be implemented in a very efficient way to reduce the performance impact on Redict. All Redict Module API calls that require a valid context (such as RedictModule_Call(), RedictModule_OpenKey(), etc.) are not supported in a filter context.

The RedictModuleCommandFilterCtx can be used to inspect or modify the executed command and its arguments. As the filter executes before Redict begins processing the command, any change will affect the way the command is processed. For example, a module can override Redict commands this way:

  1. Register a MODULE.SET command which implements an extended version of the Redict SET command.
  2. Register a command filter which detects invocation of SET on a specific pattern of keys. Once detected, the filter will replace the first argument from SET to MODULE.SET.
  3. When filter execution is complete, Redict considers the new command name and therefore executes the module’s own command.

Note that in the above use case, if MODULE.SET itself uses RedictModule_Call() the filter will be applied on that call as well. If that is not desired, the REDICTMODULE_CMDFILTER_NOSELF flag can be set when registering the filter.

The REDICTMODULE_CMDFILTER_NOSELF flag prevents execution flows that originate from the module’s own RedictModule_Call() from reaching the filter. This flag is effective for all execution flows, including nested ones, as long as the execution begins from the module’s command context or a thread-safe context that is associated with a blocking command.

Detached thread-safe contexts are not associated with the module and cannot be protected by this flag.

If multiple filters are registered (by the same or different modules), they are executed in the order of registration.

RedictModule_UnregisterCommandFilter #

int RedictModule_UnregisterCommandFilter(RedictModuleCtx *ctx,
                                         RedictModuleCommandFilter *filter);

Available since: Redict 7.3.0

Unregister a command filter.

RedictModule_CommandFilterArgsCount #

int RedictModule_CommandFilterArgsCount(RedictModuleCommandFilterCtx *fctx);

Available since: Redict 7.3.0

Return the number of arguments a filtered command has. The number of arguments include the command itself.

RedictModule_CommandFilterArgGet #

RedictModuleString *RedictModule_CommandFilterArgGet(RedictModuleCommandFilterCtx *fctx,
                                                     int pos);

Available since: Redict 7.3.0

Return the specified command argument. The first argument (position 0) is the command itself, and the rest are user-provided args.

RedictModule_CommandFilterArgInsert #

int RedictModule_CommandFilterArgInsert(RedictModuleCommandFilterCtx *fctx,
                                        int pos,
                                        RedictModuleString *arg);

Available since: Redict 7.3.0

Modify the filtered command by inserting a new argument at the specified position. The specified RedictModuleString argument may be used by Redict after the filter context is destroyed, so it must not be auto-memory allocated, freed or used elsewhere.

RedictModule_CommandFilterArgReplace #

int RedictModule_CommandFilterArgReplace(RedictModuleCommandFilterCtx *fctx,
                                         int pos,
                                         RedictModuleString *arg);

Available since: Redict 7.3.0

Modify the filtered command by replacing an existing argument with a new one. The specified RedictModuleString argument may be used by Redict after the filter context is destroyed, so it must not be auto-memory allocated, freed or used elsewhere.

RedictModule_CommandFilterArgDelete #

int RedictModule_CommandFilterArgDelete(RedictModuleCommandFilterCtx *fctx,
                                        int pos);

Available since: Redict 7.3.0

Modify the filtered command by deleting an argument at the specified position.

RedictModule_CommandFilterGetClientId #

unsigned long long RedictModule_CommandFilterGetClientId(RedictModuleCommandFilterCtx *fctx);

Available since: Redict 7.3.0

Get Client ID for client that issued the command we are filtering

RedictModule_MallocSize #

size_t RedictModule_MallocSize(void* ptr);

Available since: Redict 7.3.0

For a given pointer allocated via RedictModule_Alloc() or RedictModule_Realloc(), return the amount of memory allocated for it. Note that this may be different (larger) than the memory we allocated with the allocation calls, since sometimes the underlying allocator will allocate more memory.

RedictModule_MallocUsableSize #

size_t RedictModule_MallocUsableSize(void *ptr);

Available since: Redict 7.3.0

Similar to RedictModule_MallocSize, the difference is that RedictModule_MallocUsableSize returns the usable size of memory by the module.

RedictModule_MallocSizeString #

size_t RedictModule_MallocSizeString(RedictModuleString* str);

Available since: Redict 7.3.0

Same as RedictModule_MallocSize, except it works on RedictModuleString pointers.

RedictModule_MallocSizeDict #

size_t RedictModule_MallocSizeDict(RedictModuleDict* dict);

Available since: Redict 7.3.0

Same as RedictModule_MallocSize, except it works on RedictModuleDict pointers. Note that the returned value is only the overhead of the underlying structures, it does not include the allocation size of the keys and values.

RedictModule_GetUsedMemoryRatio #

float RedictModule_GetUsedMemoryRatio(void);

Available since: Redict 7.3.0

Return the a number between 0 to 1 indicating the amount of memory currently used, relative to the Redict “maxmemory” configuration.

  • 0 - No memory limit configured.
  • Between 0 and 1 - The percentage of the memory used normalized in 0-1 range.
  • Exactly 1 - Memory limit reached.
  • Greater 1 - More memory used than the configured limit.

Scanning keyspace and hashes #

RedictModule_ScanCursorCreate #

RedictModuleScanCursor *RedictModule_ScanCursorCreate(void);

Available since: Redict 7.3.0

Create a new cursor to be used with RedictModule_Scan

RedictModule_ScanCursorRestart #

void RedictModule_ScanCursorRestart(RedictModuleScanCursor *cursor);

Available since: Redict 7.3.0

Restart an existing cursor. The keys will be rescanned.

RedictModule_ScanCursorDestroy #

void RedictModule_ScanCursorDestroy(RedictModuleScanCursor *cursor);

Available since: Redict 7.3.0

Destroy the cursor struct.

RedictModule_Scan #

int RedictModule_Scan(RedictModuleCtx *ctx,
                      RedictModuleScanCursor *cursor,
                      RedictModuleScanCB fn,
                      void *privdata);

Available since: Redict 7.3.0

Scan API that allows a module to scan all the keys and value in the selected db.

Callback for scan implementation.

void scan_callback(RedictModuleCtx *ctx, RedictModuleString *keyname,
                   RedictModuleKey *key, void *privdata);
  • ctx: the redict module context provided to for the scan.
  • keyname: owned by the caller and need to be retained if used after this function.
  • key: holds info on the key and value, it is provided as best effort, in some cases it might be NULL, in which case the user should (can) use RedictModule_OpenKey() (and CloseKey too). when it is provided, it is owned by the caller and will be free when the callback returns.
  • privdata: the user data provided to RedictModule_Scan().

The way it should be used:

 RedictModuleScanCursor *c = RedictModule_ScanCursorCreate();
 while(RedictModule_Scan(ctx, c, callback, privateData));
 RedictModule_ScanCursorDestroy(c);

It is also possible to use this API from another thread while the lock is acquired during the actual call to RedictModule_Scan:

 RedictModuleScanCursor *c = RedictModule_ScanCursorCreate();
 RedictModule_ThreadSafeContextLock(ctx);
 while(RedictModule_Scan(ctx, c, callback, privateData)){
     RedictModule_ThreadSafeContextUnlock(ctx);
     // do some background job
     RedictModule_ThreadSafeContextLock(ctx);
 }
 RedictModule_ScanCursorDestroy(c);

The function will return 1 if there are more elements to scan and 0 otherwise, possibly setting errno if the call failed.

It is also possible to restart an existing cursor using RedictModule_ScanCursorRestart.

IMPORTANT: This API is very similar to the Redict SCAN command from the point of view of the guarantees it provides. This means that the API may report duplicated keys, but guarantees to report at least one time every key that was there from the start to the end of the scanning process.

NOTE: If you do database changes within the callback, you should be aware that the internal state of the database may change. For instance it is safe to delete or modify the current key, but may not be safe to delete any other key. Moreover playing with the Redict keyspace while iterating may have the effect of returning more duplicates. A safe pattern is to store the keys names you want to modify elsewhere, and perform the actions on the keys later when the iteration is complete. However this can cost a lot of memory, so it may make sense to just operate on the current key when possible during the iteration, given that this is safe.

RedictModule_ScanKey #

int RedictModule_ScanKey(RedictModuleKey *key,
                         RedictModuleScanCursor *cursor,
                         RedictModuleScanKeyCB fn,
                         void *privdata);

Available since: Redict 7.3.0

Scan api that allows a module to scan the elements in a hash, set or sorted set key

Callback for scan implementation.

void scan_callback(RedictModuleKey *key, RedictModuleString* field, RedictModuleString* value, void *privdata);
  • key - the redict key context provided to for the scan.
  • field - field name, owned by the caller and need to be retained if used after this function.
  • value - value string or NULL for set type, owned by the caller and need to be retained if used after this function.
  • privdata - the user data provided to RedictModule_ScanKey.

The way it should be used:

 RedictModuleScanCursor *c = RedictModule_ScanCursorCreate();
 RedictModuleKey *key = RedictModule_OpenKey(...)
 while(RedictModule_ScanKey(key, c, callback, privateData));
 RedictModule_CloseKey(key);
 RedictModule_ScanCursorDestroy(c);

It is also possible to use this API from another thread while the lock is acquired during the actual call to RedictModule_ScanKey, and re-opening the key each time:

 RedictModuleScanCursor *c = RedictModule_ScanCursorCreate();
 RedictModule_ThreadSafeContextLock(ctx);
 RedictModuleKey *key = RedictModule_OpenKey(...)
 while(RedictModule_ScanKey(ctx, c, callback, privateData)){
     RedictModule_CloseKey(key);
     RedictModule_ThreadSafeContextUnlock(ctx);
     // do some background job
     RedictModule_ThreadSafeContextLock(ctx);
     RedictModuleKey *key = RedictModule_OpenKey(...)
 }
 RedictModule_CloseKey(key);
 RedictModule_ScanCursorDestroy(c);

The function will return 1 if there are more elements to scan and 0 otherwise, possibly setting errno if the call failed. It is also possible to restart an existing cursor using RedictModule_ScanCursorRestart.

NOTE: Certain operations are unsafe while iterating the object. For instance while the API guarantees to return at least one time all the elements that are present in the data structure consistently from the start to the end of the iteration (see HSCAN and similar commands documentation), the more you play with the elements, the more duplicates you may get. In general deleting the current element of the data structure is safe, while removing the key you are iterating is not safe.

Module fork API #

RedictModule_Fork #

int RedictModule_Fork(RedictModuleForkDoneHandler cb, void *user_data);

Available since: Redict 7.3.0

Create a background child process with the current frozen snapshot of the main process where you can do some processing in the background without affecting / freezing the traffic and no need for threads and GIL locking. Note that Redict allows for only one concurrent fork. When the child wants to exit, it should call RedictModule_ExitFromChild. If the parent wants to kill the child it should call RedictModule_KillForkChild The done handler callback will be executed on the parent process when the child existed (but not when killed) Return: -1 on failure, on success the parent process will get a positive PID of the child, and the child process will get 0.

RedictModule_SendChildHeartbeat #

void RedictModule_SendChildHeartbeat(double progress);

Available since: Redict 7.3.0

The module is advised to call this function from the fork child once in a while, so that it can report progress and COW memory to the parent which will be reported in INFO. The progress argument should between 0 and 1, or -1 when not available.

RedictModule_ExitFromChild #

int RedictModule_ExitFromChild(int retcode);

Available since: Redict 7.3.0

Call from the child process when you want to terminate it. retcode will be provided to the done handler executed on the parent process.

RedictModule_KillForkChild #

int RedictModule_KillForkChild(int child_pid);

Available since: Redict 7.3.0

Can be used to kill the forked child process from the parent process. child_pid would be the return value of RedictModule_Fork.

Server hooks implementation #

RedictModule_SubscribeToServerEvent #

int RedictModule_SubscribeToServerEvent(RedictModuleCtx *ctx,
                                        RedictModuleEvent event,
                                        RedictModuleEventCallback callback);

Available since: Redict 7.3.0

Register to be notified, via a callback, when the specified server event happens. The callback is called with the event as argument, and an additional argument which is a void pointer and should be cased to a specific type that is event-specific (but many events will just use NULL since they do not have additional information to pass to the callback).

If the callback is NULL and there was a previous subscription, the module will be unsubscribed. If there was a previous subscription and the callback is not null, the old callback will be replaced with the new one.

The callback must be of this type:

int (*RedictModuleEventCallback)(RedictModuleCtx *ctx,
                                RedictModuleEvent eid,
                                uint64_t subevent,
                                void *data);

The ‘ctx’ is a normal Redict module context that the callback can use in order to call other modules APIs. The ’eid’ is the event itself, this is only useful in the case the module subscribed to multiple events: using the ‘id’ field of this structure it is possible to check if the event is one of the events we registered with this callback. The ‘subevent’ field depends on the event that fired.

Finally the ‘data’ pointer may be populated, only for certain events, with more relevant data.

Here is a list of events you can use as ’eid’ and related sub events:

  • RedictModuleEvent_ReplicationRoleChanged:

    This event is called when the instance switches from master to replica or the other way around, however the event is also called when the replica remains a replica but starts to replicate with a different master.

    The following sub events are available:

    • REDICTMODULE_SUBEVENT_REPLROLECHANGED_NOW_MASTER
    • REDICTMODULE_SUBEVENT_REPLROLECHANGED_NOW_REPLICA

    The ‘data’ field can be casted by the callback to a RedictModuleReplicationInfo structure with the following fields:

      int master; // true if master, false if replica
      char *masterhost; // master instance hostname for NOW_REPLICA
      int masterport; // master instance port for NOW_REPLICA
      char *replid1; // Main replication ID
      char *replid2; // Secondary replication ID
      uint64_t repl1_offset; // Main replication offset
      uint64_t repl2_offset; // Offset of replid2 validity
    
  • RedictModuleEvent_Persistence

    This event is called when RDB saving or AOF rewriting starts and ends. The following sub events are available:

    • REDICTMODULE_SUBEVENT_PERSISTENCE_RDB_START
    • REDICTMODULE_SUBEVENT_PERSISTENCE_AOF_START
    • REDICTMODULE_SUBEVENT_PERSISTENCE_SYNC_RDB_START
    • REDICTMODULE_SUBEVENT_PERSISTENCE_SYNC_AOF_START
    • REDICTMODULE_SUBEVENT_PERSISTENCE_ENDED
    • REDICTMODULE_SUBEVENT_PERSISTENCE_FAILED

    The above events are triggered not just when the user calls the relevant commands like BGSAVE, but also when a saving operation or AOF rewriting occurs because of internal server triggers. The SYNC_RDB_START sub events are happening in the foreground due to SAVE command, FLUSHALL, or server shutdown, and the other RDB and AOF sub events are executed in a background fork child, so any action the module takes can only affect the generated AOF or RDB, but will not be reflected in the parent process and affect connected clients and commands. Also note that the AOF_START sub event may end up saving RDB content in case of an AOF with rdb-preamble.

  • RedictModuleEvent_FlushDB

    The FLUSHALL, FLUSHDB or an internal flush (for instance because of replication, after the replica synchronization) happened. The following sub events are available:

    • REDICTMODULE_SUBEVENT_FLUSHDB_START
    • REDICTMODULE_SUBEVENT_FLUSHDB_END

    The data pointer can be casted to a RedictModuleFlushInfo structure with the following fields:

      int32_t async;  // True if the flush is done in a thread.
                      // See for instance FLUSHALL ASYNC.
                      // In this case the END callback is invoked
                      // immediately after the database is put
                      // in the free list of the thread.
      int32_t dbnum;  // Flushed database number, -1 for all the DBs
                      // in the case of the FLUSHALL operation.
    

    The start event is called before the operation is initiated, thus allowing the callback to call DBSIZE or other operation on the yet-to-free keyspace.

  • RedictModuleEvent_Loading

    Called on loading operations: at startup when the server is started, but also after a first synchronization when the replica is loading the RDB file from the master. The following sub events are available:

    • REDICTMODULE_SUBEVENT_LOADING_RDB_START
    • REDICTMODULE_SUBEVENT_LOADING_AOF_START
    • REDICTMODULE_SUBEVENT_LOADING_REPL_START
    • REDICTMODULE_SUBEVENT_LOADING_ENDED
    • REDICTMODULE_SUBEVENT_LOADING_FAILED

    Note that AOF loading may start with an RDB data in case of rdb-preamble, in which case you’ll only receive an AOF_START event.

  • RedictModuleEvent_ClientChange

    Called when a client connects or disconnects. The data pointer can be casted to a RedictModuleClientInfo structure, documented in RedictModule_GetClientInfoById(). The following sub events are available:

    • REDICTMODULE_SUBEVENT_CLIENT_CHANGE_CONNECTED
    • REDICTMODULE_SUBEVENT_CLIENT_CHANGE_DISCONNECTED
  • RedictModuleEvent_Shutdown

    The server is shutting down. No subevents are available.

  • RedictModuleEvent_ReplicaChange

    This event is called when the instance (that can be both a master or a replica) get a new online replica, or lose a replica since it gets disconnected. The following sub events are available:

    • REDICTMODULE_SUBEVENT_REPLICA_CHANGE_ONLINE
    • REDICTMODULE_SUBEVENT_REPLICA_CHANGE_OFFLINE

    No additional information is available so far: future versions of Redict will have an API in order to enumerate the replicas connected and their state.

  • RedictModuleEvent_CronLoop

    This event is called every time Redict calls the serverCron() function in order to do certain bookkeeping. Modules that are required to do operations from time to time may use this callback. Normally Redict calls this function 10 times per second, but this changes depending on the “hz” configuration. No sub events are available.

    The data pointer can be casted to a RedictModuleCronLoop structure with the following fields:

      int32_t hz;  // Approximate number of events per second.
    
  • RedictModuleEvent_MasterLinkChange

    This is called for replicas in order to notify when the replication link becomes functional (up) with our master, or when it goes down. Note that the link is not considered up when we just connected to the master, but only if the replication is happening correctly. The following sub events are available:

    • REDICTMODULE_SUBEVENT_MASTER_LINK_UP
    • REDICTMODULE_SUBEVENT_MASTER_LINK_DOWN
  • RedictModuleEvent_ModuleChange

    This event is called when a new module is loaded or one is unloaded. The following sub events are available:

    • REDICTMODULE_SUBEVENT_MODULE_LOADED
    • REDICTMODULE_SUBEVENT_MODULE_UNLOADED

    The data pointer can be casted to a RedictModuleModuleChange structure with the following fields:

      const char* module_name;  // Name of module loaded or unloaded.
      int32_t module_version;  // Module version.
    
  • RedictModuleEvent_LoadingProgress

    This event is called repeatedly called while an RDB or AOF file is being loaded. The following sub events are available:

    • REDICTMODULE_SUBEVENT_LOADING_PROGRESS_RDB
    • REDICTMODULE_SUBEVENT_LOADING_PROGRESS_AOF

    The data pointer can be casted to a RedictModuleLoadingProgress structure with the following fields:

      int32_t hz;  // Approximate number of events per second.
      int32_t progress;  // Approximate progress between 0 and 1024,
                         // or -1 if unknown.
    
  • RedictModuleEvent_SwapDB

    This event is called when a SWAPDB command has been successfully Executed. For this event call currently there is no subevents available.

    The data pointer can be casted to a RedictModuleSwapDbInfo structure with the following fields:

      int32_t dbnum_first;    // Swap Db first dbnum
      int32_t dbnum_second;   // Swap Db second dbnum
    
  • RedictModuleEvent_ReplBackup

    WARNING: Replication Backup events are deprecated since Redis 7.0 and are never fired. See RedictModuleEvent_ReplAsyncLoad for understanding how Async Replication Loading events are now triggered when repl-diskless-load is set to swapdb.

    Called when repl-diskless-load config is set to swapdb, And redict needs to backup the current database for the possibility to be restored later. A module with global data and maybe with aux_load and aux_save callbacks may need to use this notification to backup / restore / discard its globals. The following sub events are available:

    • REDICTMODULE_SUBEVENT_REPL_BACKUP_CREATE
    • REDICTMODULE_SUBEVENT_REPL_BACKUP_RESTORE
    • REDICTMODULE_SUBEVENT_REPL_BACKUP_DISCARD
  • RedictModuleEvent_ReplAsyncLoad

    Called when repl-diskless-load config is set to swapdb and a replication with a master of same data set history (matching replication ID) occurs. In which case redict serves current data set while loading new database in memory from socket. Modules must have declared they support this mechanism in order to activate it, through REDICTMODULE_OPTIONS_HANDLE_REPL_ASYNC_LOAD flag. The following sub events are available:

    • REDICTMODULE_SUBEVENT_REPL_ASYNC_LOAD_STARTED
    • REDICTMODULE_SUBEVENT_REPL_ASYNC_LOAD_ABORTED
    • REDICTMODULE_SUBEVENT_REPL_ASYNC_LOAD_COMPLETED
  • RedictModuleEvent_ForkChild

    Called when a fork child (AOFRW, RDBSAVE, module fork…) is born/dies The following sub events are available:

    • REDICTMODULE_SUBEVENT_FORK_CHILD_BORN
    • REDICTMODULE_SUBEVENT_FORK_CHILD_DIED
  • RedictModuleEvent_EventLoop

    Called on each event loop iteration, once just before the event loop goes to sleep or just after it wakes up. The following sub events are available:

    • REDICTMODULE_SUBEVENT_EVENTLOOP_BEFORE_SLEEP
    • REDICTMODULE_SUBEVENT_EVENTLOOP_AFTER_SLEEP
  • RedictModule_Event_Config

    Called when a configuration event happens The following sub events are available:

    • REDICTMODULE_SUBEVENT_CONFIG_CHANGE

    The data pointer can be casted to a RedictModuleConfigChange structure with the following fields:

      const char **config_names; // An array of C string pointers containing the
                                 // name of each modified configuration item
      uint32_t num_changes;      // The number of elements in the config_names array
    
  • RedictModule_Event_Key

    Called when a key is removed from the keyspace. We can’t modify any key in the event. The following sub events are available:

    • REDICTMODULE_SUBEVENT_KEY_DELETED
    • REDICTMODULE_SUBEVENT_KEY_EXPIRED
    • REDICTMODULE_SUBEVENT_KEY_EVICTED
    • REDICTMODULE_SUBEVENT_KEY_OVERWRITTEN

    The data pointer can be casted to a RedictModuleKeyInfo structure with the following fields:

      RedictModuleKey *key;    // Key name
    

The function returns REDICTMODULE_OK if the module was successfully subscribed for the specified event. If the API is called from a wrong context or unsupported event is given then REDICTMODULE_ERR is returned.

RedictModule_IsSubEventSupported #

int RedictModule_IsSubEventSupported(RedictModuleEvent event,
                                     int64_t subevent);

Available since: Redict 7.3.0

For a given server event and subevent, return zero if the subevent is not supported and non-zero otherwise.

Module Configurations API #

RedictModule_RegisterStringConfig #

int RedictModule_RegisterStringConfig(RedictModuleCtx *ctx,
                                      const char *name,
                                      const char *default_val,
                                      unsigned int flags,
                                      RedictModuleConfigGetStringFunc getfn,
                                      RedictModuleConfigSetStringFunc setfn,
                                      RedictModuleConfigApplyFunc applyfn,
                                      void *privdata);

Available since: Redict 7.3.0

Create a string config that Redict users can interact with via the Redict config file, CONFIG SET, CONFIG GET, and CONFIG REWRITE commands.

The actual config value is owned by the module, and the getfn, setfn and optional applyfn callbacks that are provided to Redict in order to access or manipulate the value. The getfn callback retrieves the value from the module, while the setfn callback provides a value to be stored into the module config. The optional applyfn callback is called after a CONFIG SET command modified one or more configs using the setfn callback and can be used to atomically apply a config after several configs were changed together. If there are multiple configs with applyfn callbacks set by a single CONFIG SET command, they will be deduplicated if their applyfn function and privdata pointers are identical, and the callback will only be run once. Both the setfn and applyfn can return an error if the provided value is invalid or cannot be used. The config also declares a type for the value that is validated by Redict and provided to the module. The config system provides the following types:

  • Redict String: Binary safe string data.
  • Enum: One of a finite number of string tokens, provided during registration.
  • Numeric: 64 bit signed integer, which also supports min and max values.
  • Bool: Yes or no value.

The setfn callback is expected to return REDICTMODULE_OK when the value is successfully applied. It can also return REDICTMODULE_ERR if the value can’t be applied, and the *err pointer can be set with a RedictModuleString error message to provide to the client. This RedictModuleString will be freed by redict after returning from the set callback.

All configs are registered with a name, a type, a default value, private data that is made available in the callbacks, as well as several flags that modify the behavior of the config. The name must only contain alphanumeric characters or dashes. The supported flags are:

  • REDICTMODULE_CONFIG_DEFAULT: The default flags for a config. This creates a config that can be modified after startup.
  • REDICTMODULE_CONFIG_IMMUTABLE: This config can only be provided loading time.
  • REDICTMODULE_CONFIG_SENSITIVE: The value stored in this config is redacted from all logging.
  • REDICTMODULE_CONFIG_HIDDEN: The name is hidden from CONFIG GET with pattern matching.
  • REDICTMODULE_CONFIG_PROTECTED: This config will be only be modifiable based off the value of enable-protected-configs.
  • REDICTMODULE_CONFIG_DENY_LOADING: This config is not modifiable while the server is loading data.
  • REDICTMODULE_CONFIG_MEMORY: For numeric configs, this config will convert data unit notations into their byte equivalent.
  • REDICTMODULE_CONFIG_BITFLAGS: For enum configs, this config will allow multiple entries to be combined as bit flags.

Default values are used on startup to set the value if it is not provided via the config file or command line. Default values are also used to compare to on a config rewrite.

Notes:

  1. On string config sets that the string passed to the set callback will be freed after execution and the module must retain it.
  2. On string config gets the string will not be consumed and will be valid after execution.

Example implementation:

RedictModuleString *strval;
int adjustable = 1;
RedictModuleString *getStringConfigCommand(const char *name, void *privdata) {
    return strval;
}

int setStringConfigCommand(const char *name, RedictModuleString *new, void *privdata, RedictModuleString **err) {
   if (adjustable) {
       RedictModule_Free(strval);
       RedictModule_RetainString(NULL, new);
       strval = new;
       return REDICTMODULE_OK;
   }
   *err = RedictModule_CreateString(NULL, "Not adjustable.", 15);
   return REDICTMODULE_ERR;
}
...
RedictModule_RegisterStringConfig(ctx, "string", NULL, REDICTMODULE_CONFIG_DEFAULT, getStringConfigCommand, setStringConfigCommand, NULL, NULL);

If the registration fails, REDICTMODULE_ERR is returned and one of the following errno is set:

  • EBUSY: Registering the Config outside of RedictModule_OnLoad.
  • EINVAL: The provided flags are invalid for the registration or the name of the config contains invalid characters.
  • EALREADY: The provided configuration name is already used.

RedictModule_RegisterBoolConfig #

int RedictModule_RegisterBoolConfig(RedictModuleCtx *ctx,
                                    const char *name,
                                    int default_val,
                                    unsigned int flags,
                                    RedictModuleConfigGetBoolFunc getfn,
                                    RedictModuleConfigSetBoolFunc setfn,
                                    RedictModuleConfigApplyFunc applyfn,
                                    void *privdata);

Available since: Redict 7.3.0

Create a bool config that server clients can interact with via the CONFIG SET, CONFIG GET, and CONFIG REWRITE commands. See RedictModule_RegisterStringConfig for detailed information about configs.

RedictModule_RegisterEnumConfig #

int RedictModule_RegisterEnumConfig(RedictModuleCtx *ctx,
                                    const char *name,
                                    int default_val,
                                    unsigned int flags,
                                    const char **enum_values,
                                    const int *int_values,
                                    int num_enum_vals,
                                    RedictModuleConfigGetEnumFunc getfn,
                                    RedictModuleConfigSetEnumFunc setfn,
                                    RedictModuleConfigApplyFunc applyfn,
                                    void *privdata);

Available since: Redict 7.3.0

Create an enum config that server clients can interact with via the CONFIG SET, CONFIG GET, and CONFIG REWRITE commands. Enum configs are a set of string tokens to corresponding integer values, where the string value is exposed to Redict clients but the value passed Redict and the module is the integer value. These values are defined in enum_values, an array of null-terminated c strings, and int_vals, an array of enum values who has an index partner in enum_values. Example Implementation: const char *enum_vals[3] = {“first”, “second”, “third”}; const int int_vals[3] = {0, 2, 4}; int enum_val = 0;

 int getEnumConfigCommand(const char *name, void *privdata) {
     return enum_val;
 }

 int setEnumConfigCommand(const char *name, int val, void *privdata, const char **err) {
     enum_val = val;
     return REDICTMODULE_OK;
 }
 ...
 RedictModule_RegisterEnumConfig(ctx, "enum", 0, REDICTMODULE_CONFIG_DEFAULT, enum_vals, int_vals, 3, getEnumConfigCommand, setEnumConfigCommand, NULL, NULL);

Note that you can use REDICTMODULE_CONFIG_BITFLAGS so that multiple enum string can be combined into one integer as bit flags, in which case you may want to sort your enums so that the preferred combinations are present first.

See RedictModule_RegisterStringConfig for detailed general information about configs.

RedictModule_RegisterNumericConfig #

int RedictModule_RegisterNumericConfig(RedictModuleCtx *ctx,
                                       const char *name,
                                       long long default_val,
                                       unsigned int flags,
                                       long long min,
                                       long long max,
                                       RedictModuleConfigGetNumericFunc getfn,
                                       RedictModuleConfigSetNumericFunc setfn,
                                       RedictModuleConfigApplyFunc applyfn,
                                       void *privdata);

Available since: Redict 7.3.0

Create an integer config that server clients can interact with via the CONFIG SET, CONFIG GET, and CONFIG REWRITE commands. See RedictModule_RegisterStringConfig for detailed information about configs.

RedictModule_LoadConfigs #

int RedictModule_LoadConfigs(RedictModuleCtx *ctx);

Available since: Redict 7.3.0

Applies all pending configurations on the module load. This should be called after all of the configurations have been registered for the module inside of RedictModule_OnLoad. This will return REDICTMODULE_ERR if it is called outside RedictModule_OnLoad. This API needs to be called when configurations are provided in either MODULE LOADEX or provided as startup arguments.

RDB load/save API #

RedictModule_RdbStreamCreateFromFile #

RedictModuleRdbStream *RedictModule_RdbStreamCreateFromFile(const char *filename);

Available since: Redict 7.3.0

Create a stream object to save/load RDB to/from a file.

This function returns a pointer to RedictModuleRdbStream which is owned by the caller. It requires a call to RedictModule_RdbStreamFree() to free the object.

RedictModule_RdbStreamFree #

void RedictModule_RdbStreamFree(RedictModuleRdbStream *stream);

Available since: Redict 7.3.0

Release an RDB stream object.

RedictModule_RdbLoad #

int RedictModule_RdbLoad(RedictModuleCtx *ctx,
                         RedictModuleRdbStream *stream,
                         int flags);

Available since: Redict 7.3.0

Load RDB file from the stream. Dataset will be cleared first and then RDB file will be loaded.

flags must be zero. This parameter is for future use.

On success REDICTMODULE_OK is returned, otherwise REDICTMODULE_ERR is returned and errno is set accordingly.

Example:

RedictModuleRdbStream *s = RedictModule_RdbStreamCreateFromFile("exp.rdb");
RedictModule_RdbLoad(ctx, s, 0);
RedictModule_RdbStreamFree(s);

RedictModule_RdbSave #

int RedictModule_RdbSave(RedictModuleCtx *ctx,
                         RedictModuleRdbStream *stream,
                         int flags);

Available since: Redict 7.3.0

Save dataset to the RDB stream.

flags must be zero. This parameter is for future use.

On success REDICTMODULE_OK is returned, otherwise REDICTMODULE_ERR is returned and errno is set accordingly.

Example:

RedictModuleRdbStream *s = RedictModule_RdbStreamCreateFromFile("exp.rdb");
RedictModule_RdbSave(ctx, s, 0);
RedictModule_RdbStreamFree(s);

Key eviction API #

RedictModule_SetLRU #

int RedictModule_SetLRU(RedictModuleKey *key, mstime_t lru_idle);

Available since: Redict 7.3.0

Set the key last access time for LRU based eviction. not relevant if the servers’s maxmemory policy is LFU based. Value is idle time in milliseconds. returns REDICTMODULE_OK if the LRU was updated, REDICTMODULE_ERR otherwise.

RedictModule_GetLRU #

int RedictModule_GetLRU(RedictModuleKey *key, mstime_t *lru_idle);

Available since: Redict 7.3.0

Gets the key last access time. Value is idletime in milliseconds or -1 if the server’s eviction policy is LFU based. returns REDICTMODULE_OK if when key is valid.

RedictModule_SetLFU #

int RedictModule_SetLFU(RedictModuleKey *key, long long lfu_freq);

Available since: Redict 7.3.0

Set the key access frequency. only relevant if the server’s maxmemory policy is LFU based. The frequency is a logarithmic counter that provides an indication of the access frequencyonly (must be <= 255). returns REDICTMODULE_OK if the LFU was updated, REDICTMODULE_ERR otherwise.

RedictModule_GetLFU #

int RedictModule_GetLFU(RedictModuleKey *key, long long *lfu_freq);

Available since: Redict 7.3.0

Gets the key access frequency or -1 if the server’s eviction policy is not LFU based. returns REDICTMODULE_OK if when key is valid.

Miscellaneous APIs #

RedictModule_GetModuleOptionsAll #

int RedictModule_GetModuleOptionsAll(void);

Available since: Redict 7.3.0

Returns the full module options flags mask, using the return value the module can check if a certain set of module options are supported by the redict server version in use. Example:

   int supportedFlags = RedictModule_GetModuleOptionsAll();
   if (supportedFlags & REDICTMODULE_OPTIONS_ALLOW_NESTED_KEYSPACE_NOTIFICATIONS) {
         // REDICTMODULE_OPTIONS_ALLOW_NESTED_KEYSPACE_NOTIFICATIONS is supported
   } else{
         // REDICTMODULE_OPTIONS_ALLOW_NESTED_KEYSPACE_NOTIFICATIONS is not supported
   }

RedictModule_GetContextFlagsAll #

int RedictModule_GetContextFlagsAll(void);

Available since: Redict 7.3.0

Returns the full ContextFlags mask, using the return value the module can check if a certain set of flags are supported by the redict server version in use. Example:

   int supportedFlags = RedictModule_GetContextFlagsAll();
   if (supportedFlags & REDICTMODULE_CTX_FLAGS_MULTI) {
         // REDICTMODULE_CTX_FLAGS_MULTI is supported
   } else{
         // REDICTMODULE_CTX_FLAGS_MULTI is not supported
   }

RedictModule_GetKeyspaceNotificationFlagsAll #

int RedictModule_GetKeyspaceNotificationFlagsAll(void);

Available since: Redict 7.3.0

Returns the full KeyspaceNotification mask, using the return value the module can check if a certain set of flags are supported by the redict server version in use. Example:

   int supportedFlags = RedictModule_GetKeyspaceNotificationFlagsAll();
   if (supportedFlags & REDICTMODULE_NOTIFY_LOADED) {
         // REDICTMODULE_NOTIFY_LOADED is supported
   } else{
         // REDICTMODULE_NOTIFY_LOADED is not supported
   }

RedictModule_GetServerVersion #

int RedictModule_GetServerVersion(void);

Available since: Redict 7.3.0

Return the redict version in format of 0x00MMmmpp. Example for 6.0.7 the return value will be 0x00060007.

RedictModule_GetTypeMethodVersion #

int RedictModule_GetTypeMethodVersion(void);

Available since: Redict 7.3.0

Return the current redict-server runtime value of REDICTMODULE_TYPE_METHOD_VERSION. You can use that when calling RedictModule_CreateDataType to know which fields of RedictModuleTypeMethods are gonna be supported and which will be ignored.

RedictModule_ModuleTypeReplaceValue #

int RedictModule_ModuleTypeReplaceValue(RedictModuleKey *key,
                                        moduleType *mt,
                                        void *new_value,
                                        void **old_value);

Available since: Redict 7.3.0

Replace the value assigned to a module type.

The key must be open for writing, have an existing value, and have a moduleType that matches the one specified by the caller.

Unlike RedictModule_ModuleTypeSetValue() which will free the old value, this function simply swaps the old value with the new value.

The function returns REDICTMODULE_OK on success, REDICTMODULE_ERR on errors such as:

  1. Key is not opened for writing.
  2. Key is not a module data type key.
  3. Key is a module datatype other than ‘mt’.

If old_value is non-NULL, the old value is returned by reference.

RedictModule_GetCommandKeysWithFlags #

int *RedictModule_GetCommandKeysWithFlags(RedictModuleCtx *ctx,
                                          RedictModuleString **argv,
                                          int argc,
                                          int *num_keys,
                                          int **out_flags);

Available since: Redict 7.3.0

For a specified command, parse its arguments and return an array that contains the indexes of all key name arguments. This function is essentially a more efficient way to do COMMAND GETKEYS.

The out_flags argument is optional, and can be set to NULL. When provided it is filled with REDICTMODULE_CMD_KEY_ flags in matching indexes with the key indexes of the returned array.

A NULL return value indicates the specified command has no keys, or an error condition. Error conditions are indicated by setting errno as follows:

  • ENOENT: Specified command does not exist.
  • EINVAL: Invalid command arity specified.

NOTE: The returned array is not a Redict Module object so it does not get automatically freed even when auto-memory is used. The caller must explicitly call RedictModule_Free() to free it, same as the out_flags pointer if used.

RedictModule_GetCommandKeys #

int *RedictModule_GetCommandKeys(RedictModuleCtx *ctx,
                                 RedictModuleString **argv,
                                 int argc,
                                 int *num_keys);

Available since: Redict 7.3.0

Identical to RedictModule_GetCommandKeysWithFlags when flags are not needed.

RedictModule_GetCurrentCommandName #

const char *RedictModule_GetCurrentCommandName(RedictModuleCtx *ctx);

Available since: Redict 7.3.0

Return the name of the command currently running

Defrag API #

RedictModule_RegisterDefragFunc #

int RedictModule_RegisterDefragFunc(RedictModuleCtx *ctx,
                                    RedictModuleDefragFunc cb);

Available since: Redict 7.3.0

Register a defrag callback for global data, i.e. anything that the module may allocate that is not tied to a specific data type.

RedictModule_DefragShouldStop #

int RedictModule_DefragShouldStop(RedictModuleDefragCtx *ctx);

Available since: Redict 7.3.0

When the data type defrag callback iterates complex structures, this function should be called periodically. A zero (false) return indicates the callback may continue its work. A non-zero value (true) indicates it should stop.

When stopped, the callback may use RedictModule_DefragCursorSet() to store its position so it can later use RedictModule_DefragCursorGet() to resume defragging.

When stopped and more work is left to be done, the callback should return 1. Otherwise, it should return 0.

NOTE: Modules should consider the frequency in which this function is called, so it generally makes sense to do small batches of work in between calls.

RedictModule_DefragCursorSet #

int RedictModule_DefragCursorSet(RedictModuleDefragCtx *ctx,
                                 unsigned long cursor);

Available since: Redict 7.3.0

Store an arbitrary cursor value for future re-use.

This should only be called if RedictModule_DefragShouldStop() has returned a non-zero value and the defrag callback is about to exit without fully iterating its data type.

This behavior is reserved to cases where late defrag is performed. Late defrag is selected for keys that implement the free_effort callback and return a free_effort value that is larger than the defrag ‘active-defrag-max-scan-fields’ configuration directive.

Smaller keys, keys that do not implement free_effort or the global defrag callback are not called in late-defrag mode. In those cases, a call to this function will return REDICTMODULE_ERR.

The cursor may be used by the module to represent some progress into the module’s data type. Modules may also store additional cursor-related information locally and use the cursor as a flag that indicates when traversal of a new key begins. This is possible because the API makes a guarantee that concurrent defragmentation of multiple keys will not be performed.

RedictModule_DefragCursorGet #

int RedictModule_DefragCursorGet(RedictModuleDefragCtx *ctx,
                                 unsigned long *cursor);

Available since: Redict 7.3.0

Fetch a cursor value that has been previously stored using RedictModule_DefragCursorSet().

If not called for a late defrag operation, REDICTMODULE_ERR will be returned and the cursor should be ignored. See RedictModule_DefragCursorSet() for more details on defrag cursors.

RedictModule_DefragAlloc #

void *RedictModule_DefragAlloc(RedictModuleDefragCtx *ctx, void *ptr);

Available since: Redict 7.3.0

Defrag a memory allocation previously allocated by RedictModule_Alloc, RedictModule_Calloc, etc. The defragmentation process involves allocating a new memory block and copying the contents to it, like realloc().

If defragmentation was not necessary, NULL is returned and the operation has no other effect.

If a non-NULL value is returned, the caller should use the new pointer instead of the old one and update any reference to the old pointer, which must not be used again.

RedictModule_DefragRedictModuleString #

RedictModuleString *RedictModule_DefragRedictModuleString(RedictModuleDefragCtx *ctx,
                                                          RedictModuleString *str);

Available since: Redict 7.3.0

Defrag a RedictModuleString previously allocated by RedictModule_Alloc, RedictModule_Calloc, etc. See RedictModule_DefragAlloc() for more information on how the defragmentation process works.

NOTE: It is only possible to defrag strings that have a single reference. Typically this means strings retained with RedictModule_RetainString or RedictModule_HoldString may not be defragmentable. One exception is command argvs which, if retained by the module, will end up with a single reference (because the reference on the Redict side is dropped as soon as the command callback returns).

RedictModule_GetKeyNameFromDefragCtx #

const RedictModuleString *RedictModule_GetKeyNameFromDefragCtx(RedictModuleDefragCtx *ctx);

Available since: Redict 7.3.0

Returns the name of the key currently being processed. There is no guarantee that the key name is always available, so this may return NULL.

RedictModule_GetDbIdFromDefragCtx #

int RedictModule_GetDbIdFromDefragCtx(RedictModuleDefragCtx *ctx);

Available since: Redict 7.3.0

Returns the database id of the key currently being processed. There is no guarantee that this info is always available, so this may return -1.

Function index #

xtID)

Redict logo courtesy of @janWilejan, CC-BY-SA-4.0. Download SVG ⤑

Portions of this website courtesy of Salvatore Sanfilippo, CC-BY-SA-4.0.