RediSeach Full Command Documentation

FT.CREATE

Format:

  1.   FT.CREATE {index} 
  2.     [NOOFFSETS] [NOFIELDS]
  3.     [STOPWORDS {num} {stopword} ...]
  4.     SCHEMA {field} [TEXT [NOSTEM] [WEIGHT {weight}] | NUMERIC | GEO] [SORTABLE] [NOINDEX] ...

Description:

Creates an index with the given spec. The index name will be used in all the key names so keep it short!

Note on field number limits

RediSearch supports up to 1024 fields per schema, out of which at most 128 can be TEXT fields.

On 32 bit builds, at most 64 fields can be TEXT fields.

Notice that the more fields you have, the larger your index will be, as each additional 8 fields require one extra byte per index record to encode.

You can always use the NOFIELDS option and not encode field information into the index, for saving space, if you do not need filtering by text fields. This will still allow filtering by numeric and geo fields.

Parameters:

  • index : the index name to create. If it exists the old spec will be overwritten

  • NOOFFSETS : If set, we do not store term offsets for documents (saves memory, does not allow exact searches or highlighting). Implies NOHL

  • NOHL : Conserves storage space and memory by disabling highlighting support. If set, we do not store corresponding byte offsets for term positions. NOHL is also implied by NOOFFSETS .

  • NOFIELDS : If set, we do not store field bits for each term. Saves memory, does not allow filtering by specific fields.

  • NOFREQS : If set, we avoid saving the term frequencies in the index. This saves memory but does not allow sorting based on the frequencies of a given term within the document.

  • STOPWORDS : If set, we set the index with a custom stopword list, to be ignored during indexing and search time. {num} is the number of stopwords, followed by a list of stopword arguments exactly the length of {num}.

    If not set, we take the default list of stopwords.

    If {num} is set to 0, the index will not have stopwords.

  • SCHEMA {field} {options…} : After the SCHEMA keyword we define the index fields. They can be numeric, textual or geographical. For textual fields we optionally specify a weight. The default weight is 1.0.

    Field Options

    • SORTABLE

      Numeric or text field can have the optional SORTABLE argument that allows the user to later sort the results by the value of this field (this adds memory overhead so do not declare it on large text fields).

    • NOSTEM

      Text fields can have the NOSTEM argument which will disable stemming when indexing its values. This may be ideal for things like proper names.

    • NOINDEX

      Fields can have the NOINDEX option, which means they will not be indexed. This is useful in conjunction with SORTABLE , to create fields whose update using PARTIAL will not cause full reindexing of the document. If a field has NOINDEX and doesn’t have SORTABLE, it will just be ignored by the index.

Complexity

O(1)

Returns:

OK or an error

FT.ADD

Format:

  1. FT.ADD {index} {docId} {score} 
  2.   [NOSAVE]
  3.   [REPLACE [PARTIAL]]
  4.   [LANGUAGE {language}] 
  5.   [PAYLOAD {payload}]
  6.   FIELDS {field} {value} [{field} {value}...]

Description

Add a documet to the index.

Parameters:

  • index : The Fulltext index name. The index must be first created with FT.CREATE

  • docId : The document’s id that will be returned from searches. Note that the same docId cannot be added twice to the same index

  • score : The document’s rank based on the user’s ranking. This must be between 0.0 and 1.0. If you don’t have a score just set it to 1

  • NOSAVE : If set to true, we will not save the actual document in the index and only index it.

  • REPLACE : If set, we will do an UPSERT style insertion - and ete an older version of the document if it exists.

  • PARTIAL (only applicable with REPLACE): If set, you do not have to specify all fields for reindexing. Fields not given to the command will be loaded from the current version of the document. Also, if only non indexable fields, score or payload are set - we do not do a full reindexing of the document, and this will be a lot faster.

  • FIELDS : Following the FIELDS specifier, we are looking for pairs of {field} {value} to be indexed. Each field will be scored based on the index spec given in FT.CREATE. Passing fields that are not in the index spec will make them be stored as part of the document, or ignored if NOSAVE is set

  • PAYLOAD {payload} : Optionally set a binary safe payload string to the document, that can be evaluated at query time by a custom scoring function, or retrieved to the client.

  • LANGUAGE language : If set, we use a stemmer for the supplied langauge during indexing. Defaults to English. If an unsupported language is sent, the command returns an error. The supported languages are:

    “arabic”, “danish”, “dutch”, “english”, “finnish”, “french”, “german”, “hungarian”, “italian”, “norwegian”, “portuguese”, “romanian”, “russian”, “spanish”, “swedish”, “tamil”, “turkish” “chinese”

If indexing a chinese language document, you must set the language to chinese in order for the chinese characters to be tokenized properly.

Adding Chinese Documents

When adding Chinese-language documents, LANGUAGE chinese should be set in order for the indexer to properly tokenize the terms. If the default language is used then search terms will be extracted based on punctuation characters and whitespace. The Chinese language tokenizer makes use of a segmentation algorithm (via Friso ) which segments texts and checks it against a predefined dictionary. See Stemming for more information.

Complexity

O(n), where n is the number of tokens in the document

Returns

OK on success, or an error if something went wrong.

FT.ADD with REPLACE and PARTIAL

By default, FT.ADD does not allow updating the document, and will fail if it already exists in the index.

However, updating the document is possible with the REPLACE and REPLACE PARTIAL options.

REPLACE : On its own, sets the document to the new values, and reindexes it. Any fields not given will not be loaded from the current version of the document.

REPLACE PARTIAL : When both arguments are used, we can update just part of the document fields, and the rest will be loaded before reindexing. Not only that, but if only the score, payload and non-indexed fields (using NOINDEX) are updated, we will not actually reindex the document, just update its metadata internally, which is a lot faster and does not create index garbage.

FT.ADDHASH

Format

  1.  FT.ADDHASH {index} {docId} {score} [LANGUAGE language] [REPLACE]

Description

Add a documet to the index from an existing HASH key in Redis.

Parameters:

  • index : The Fulltext index name. The index must be first created with FT.CREATE

  • docId : The document’s id. This has to be an existing HASH key in redis that will hold the fields the index needs.

  • score : The document’s rank based on the user’s ranking. This must be between 0.0 and 1.0. If you don’t have a score just set it to 1

  • REPLACE : If set, we will do an UPSERT style insertion - and delete an older version of the document if it exists.

  • LANGUAGE language : If set, we use a stemmer for the supplied langauge during indexing. Defaults to English. If an unsupported language is sent, the command returns an error. The supported languages are:

“arabic”, “danish”, “dutch”, “english”, “finnish”, “french”, “german”, “hungarian”, “italian”, “norwegian”, “portuguese”, “romanian”, “russian”, “spanish”, “swedish”, “tamil”, “turkish”

Complexity

O(n), where n is the number of tokens in the document

Returns

OK on success, or an error if something went wrong.

FT.INFO

Format

  1. FT.INFO {index}

Description

Return information and statistics on the index. Returned values include:

  • Number of documents.
  • Number of distinct terms.
  • Average bytes per record.
  • Size and capacity of the index buffers.

Example:

  1. 127.0.0.1:6379> ft.info wik{0}
  2.  1) index_name
  3.  2) wikipedia
  4.  3) fields
  5.  4) 1) 1) title
  6.        2) type
  7.        3) FULLTEXT
  8.        4) weight
  9.        5) "1"
  10.     2) 1) body
  11.        2) type
  12.        3) FULLTEXT
  13.        4) weight
  14.        5) "1"
  15.  5) num_docs
  16.  6) "502694"
  17.  7) num_terms
  18.  8) "439158"
  19.  9) num_records
  20. 10) "8098583"
  21. 11) inverted_sz_mb
  22. 12) "45.58
  23. 13) inverted_cap_mb
  24. 14) "56.61
  25. 15) inverted_cap_ovh
  26. 16) "0.19
  27. 17) offset_vectors_sz_mb
  28. 18) "9.27
  29. 19) skip_index_size_mb
  30. 20) "7.35
  31. 21) score_index_size_mb
  32. 22) "30.8
  33. 23) records_per_doc_avg
  34. 24) "16.1
  35. 25) bytes_per_record_avg
  36. 26) "5.90
  37. 27) offsets_per_term_avg
  38. 28) "1.20
  39. 29) offset_bits_per_record_avg
  40. 30) "8.00

Parameters

  • index : The Fulltext index name. The index must be first created with FT.CREATE

Complexity

O(1)

Returns

Array Response. A nested array of keys and values.

FT.SEARCH

Format

  1. FT.SEARCH {index} {query} [NOCONTENT] [VERBATIM] [NOSTOPWORDS] [WITHSCORES] [WITHPAYLOADS] [WITHSORTKEYS]
  2.   [FILTER {numeric_field} {min} {max}] ...
  3.   [GEOFILTER {geo_field} {lon} {lat} {raius} m|km|mi|ft]
  4.   [INKEYS {num} {key} ... ]
  5.   [INFIELDS {num} {field} ... ]
  6.   [RETURN {num} {field} ... ]
  7.   [SUMMARIZE [FIELDS {num} {field} ... ] [FRAGS {num}] [LEN {fragsize}] [SEPARATOR {separator}]]
  8.   [HIGHLIGHT [FIELDS {num} {field} ... ] [TAGS {open} {close}]]
  9.   [SLOP {slop}] [INORDER]
  10.   [LANGUAGE {language}]
  11.   [EXPANDER {expander}]
  12.   [SCORER {scorer}]
  13.   [PAYLOAD {payload}]
  14.   [SORTBY {field} [ASC|DESC]]
  15.   [LIMIT offset num]

Description

Search the index with a textual query, returning either documents or just ids.

Parameters

  • index : The Fulltext index name. The index must be first created with FT.CREATE
  • query : the text query to search. If it’s more than a single word, put it in quotes. See below for documentation on query syntax.
  • NOCONTENT : If it appears after the query, we only return the document ids and not the content. This is useful if rediseach is only an index on an external document collection
  • RETURN {num} {field} … : Use this keyword to limit which fields from the document are returned. num is the number of fields following the keyword. If num is 0, it acts like NOCONTENT .
  • SUMMARIZE … : Use this option to return only the sections of the field which contain the matched text. See Highlighting for more detailts
  • HIGHLIGHT … : Use this option to format occurrences of matched text. See Highligting for more details
  • LIMIT first num : If the parameters appear after the query, we limit the results to the offset and number of results given. The default is 0 10
  • INFIELDS {num} {field} … : If set, filter the results to ones appearing only in specific fields of the document, like title or url. num is the number of specified field arguments
  • INKEYS {num} {field} … : If set, we limit the result to a given set of keys specified in the list. the first argument must be the length of the list, and greater than zero. Non existent keys are ignored - unless all the keys are non existent.
  • SLOP {slop} : If set, we allow a maximum of N intervening number of unmatched offsets between phrase terms. (i.e the slop for exact phrases is 0)
  • INORDER : If set, and usually used in conjunction with SLOP, we make sure the query terms appear in the same order in the document as in the query, regardless of the offsets between them.
  • FILTER numeric_field min max : If set, and numeric_field is defined as a numeric field in FT.CREATE, we will limit results to those having numeric values ranging between min and max. min and max follow ZRANGE syntax, and can be -inf , +inf and use ( for exclusive ranges. Multiple numeric filters for different fields are supported in one query.
  • GEOFILTER {geo_field} {lon} {lat} {raius} m|km|mi|ft : If set, we filter the results to a given radius from lon and lat. Radius is given as a number and units. See GEORADIUS for more details.
  • NOSTOPWORDS : If set, we do not filter stopwords from the query.
  • WITHSCORES : If set, we also return the relative internal score of each document. this can be used to merge results from multiple instances
  • WITHSORTKEYS : Only relevant in conjunction with SORTBY . Returns the value of the sorting key, right after the id and score and /or payload if requested. This is usually not needed by users, and exists for distributed search coordination purposes.
  • VERBATIM : if set, we do not try to use stemming for query expansion but search the query terms verbatim.
  • LANGUAGE {language} : If set, we use a stemmer for the supplied langauge during search for query expansion. If querying documents in Chinese, this should be set to chinese in order to properly tokenize the query terms. Defaults to English. If an unsupported language is sent, the command returns an error. See FT.ADD for the list of languages.
  • EXPANDER {expander} : If set, we will use a custom query expander instead of the stemmer. See Extensions .
  • SCORER {scorer} : If set, we will use a custom scoring function defined by the user. See Extensions .
  • PAYLOAD {payload} : Add an arbitrary, binary safe payload that will be exposed to custom scoring functions. See Extensions .
  • WITHPAYLOADS : If set, we retrieve optional document payloads (see FT.ADD). the payloads follow the document id, and if WITHSCORES was set, follow the scores.
  • SORTBY {field} [ASC|DESC] : If specified, and field is a sortable field , the results are ordered by the value of this field. This applies to both text and numeric fields.

Complexity

O(n) for single word queries (though for popular words we save a cache of the top 50 results).

Complexity for complex queries changes, but in general it’s proportional to the number of words and the number of intersection points between them.

Returns

Array reply, where the first element is the total number of results, and then pairs of document id, and a nested array of field/value.

If NOCONTENT was given, we return an array where the first element is the total number of results, and the rest of the members are document ids.

FT.EXPLAIN

Format

  1. FT.EXPLAIN {index} {query}

Description

Return the execution plan for a complex query

Example:

  1. $ redis-cli --raw
  3. 127.0.0.1:6379> FT.EXPLAIN rd "(foo bar)|(hello world) @date:[100 200]|@date:[500 +inf]"
  4. INTERSECT {
  5.   UNION {
  6.     INTERSECT {
  7.       foo
  8.       bar
  9.     }
  10.     INTERSECT {
  11.       hello
  12.       world
  13.     }
  14.   }
  15.   UNION {
  16.     NUMERIC {100.000000 <= x <= 200.000000}
  17.     NUMERIC {500.000000 <= x <= inf}
  18.   }
  19. }

Parameters

  • index : The Fulltext index name. The index must be first created with FT.CREATE
  • query : The query string, as if sent to FT.SEARCH

Complexity

O(1)

Returns

String Response. A string representing the execution plan (see above example).

Note : You should use redis-cli --raw to properly read line-breaks in the returned response.

FT.DEL

Format

  1. FT.DEL {index} {doc_id} [DD]

Description

Delete a document from the index. Returns 1 if the document was in the index, or 0 if not.

After deletion, the document can be re-added to the index. It will get a different internal id and will be a new document from the index’s POV.

FT.DEL does not delete the actual document By default!

Since RediSearch regards documents as separate entities to the index, and allows things like adding existing documents or indexing without saving the document - by default FT.DEL only deletes the reference to the document from the index, not the actual Redis HASH key where the document is stored.

Specifying DD (Delete Document) after the document ID, will make RediSearch also delete the actual document if it is in the index .

Alternatively, you can just send an extra DEL {doc_id} to redis and delete the document directly. You can run both of them in a MULTI transaction.

Parameters

  • index : The Fulltext index name. The index must be first created with FT.CREATE
  • doc_id : the id of the document to be deleted. It does not actually delete the HASH key in which the document is stored. Use DEL to do that manually if needed.

Complexity

O(1)

Returns

Integer Reply: 1 if the document was deleted, 0 if not.

FT.GET

Format

  1. FT.GET {index} {doc id}

Description

Returns the full contents of a document.

Currently it is equivalent to HGETALL, but this is future proof and will allow us to change the internal representation of documents inside redis in the future. In addition, it allows simpler implementation of fetching documents in clustered mode.

If the document does not exist or is not a HASH object, we reutrn a NULL reply

Parameters

  • index : The Fulltext index name. The index must be first created with FT.CREATE
  • documentId : The id of the document as inserted to the index

Returns

Array Reply: Key-value pairs of field names and values of the document

FT.MGET

Format

  1. FT.GET {index} {docId} ...

Description

Returns the full contents of multiple documents. Currently it is equivalent to calling multiple HGETALL commands, although faster. This command is also future proof, and will allow us to change the internal representation of documents inside redis in the future. In addition, it allows simpler implementation of fetching documents in clustered mode.

We return an array with exactly the same number of elements as the number of keys sent to the command.

Each element in turn is an array of key-value pairs representing the document.

If a document is not found or is not a valid HASH object, its place in the parent array is filled with a Null reply object.

Parameters

  • index : The Fulltext index name. The index must be first created with FT.CREATE
  • documentIds : The ids of the requested documents as inserted to the index

Returns

Array Reply: An array with exactly the same number of elements as the number of keys sent to the command. Each element in it is either an array representing the document, or Null if it was not found.

FT.DROP

Format

  1. FT.DROP {index} [KEEPDOCS]

Description

Deletes all the keys associated with the index.

By default DROP deletes the document hashes as well, but adding the KEEPDOCS option keeps the documents in place, ready for re-indexing.

If no other data is on the redis instance, this is equivalent to FLUSHDB, apart from the fact that the index specification is not deleted.

Parameters

  • index : The Fulltext index name. The index must be first created with FT.CREATE
  • KEEPDOCS : IF set, the drop operation will not delete the actual document hashes.

Returns

Status Reply: OK on success.

FT.TAGVALS

Format

  1. FT.TAGVALS {index} {field_name}

Description

Return the distinct tags indexed in a Tag field .

This is useful if your tag field indexes things like cities, categories, etc.

Limitations

There is no paging or sorting, the tags are not alphabetically sorted.

This command only operates on Tag fields .

The strings return lower-cased and stripped of whitespaces, but otherwise unchanged.

Parameters

  • index : The Fulltext index name. The index must be first created with FT.CREATE
  • filed_name : The name of a Tag file defined in the schema.

Returns

Array Reply: All the distinct tags in the tag index.

Complexity

O(n), n being the cardinality of the tag field.

FT.SUGADD

Format

  1. FT.SUGADD {key} {string} {score} [INCR] [PAYLOAD {payload}]

Description

Add a suggestion string to an auto-complete suggestion dictionary. This is disconnected from the index definitions, and leaves creating and updating suggestino dictionaries to the user.

Parameters

  • key : the suggestion dictionary key.
  • string : the suggestion string we index
  • score : a floating point number of the suggestion string’s weight
  • INCR : if set, we increment the existing entry of the suggestion by the given score, instead of replacing the score. This is useful for updating the dictionary based on user queries in real time
  • PAYLOAD {payload} : If set, we save an extra payload with the suggestion, that can be fetched by adding the WITHPAYLOADS argument to FT.SUGGET .

Returns:

Integer Reply: the current size of the suggestion dictionary.

FT.SUGGET

Format

  1. FT.SUGGET {key} {prefix} [FUZZY] [WITHPAYLOADS] [MAX num]

Description

Get completion suggestions for a prefix

Parameters:

  • key : the suggestion dictionary key.
  • prefix : the prefix to complete on
  • FUZZY : if set,we do a fuzzy prefix search, including prefixes at levenshtein distance of 1 from the prefix sent
  • MAX num : If set, we limit the results to a maximum of num . ( Note : The default is 5, and the number cannot be greater than 10).
  • WITHSCORES : If set, we also return the score of each suggestion. this can be used to merge results from multiple instances
  • WITHPAYLOADS : If set, we return optional payloads saved along with the suggestions. If no payload is present for an entry, we return a Null Reply.

Returns:

Array Reply: a list of the top suggestions matching the prefix, optionally with score after each entry

FT.SUGDEL

Format

  1. FT.SUGDEL {key} {string}

Description

Delete a string from a suggestion index.

Parameters

  • key : the suggestion dictionary key.
  • string : the string to delete

Returns:

Integer Reply: 1 if the string was found and deleted, 0 otherwise.

FT.SUGLEN

Format

  1. FT.SUGLEN {key}

Description

Get the size of an autoc-complete suggestion dictionary

Parameters

  • key : the suggestion dictionary key.

Returns:

Integer Reply: the current size of the suggestion dictionary.

FT.OPTIMIZE DEPRECATED

Format

  1. FT.OPTIMIZE {index}

Description

This command is deprecated. Index optimizations are done by the internal garbage collector in the background. Client libraries should not implement this command, and remove it if they haven’t already.