Full-Text Module

From BaseX Documentation
Revision as of 14:37, 27 February 2020 by CG (talk | contribs)
Jump to navigation Jump to search

This XQuery Module extends the W3C Full Text Recommendation with some useful functions: The index can be directly accessed, fulltext results can be marked with additional elements, or the relevant parts can be extracted. Moreover, the score value, which is generated by the contains text expression, can be explicitly requested from items.

Conventions

All functions and errors in this module are assigned to the http://basex.org/modules/ft namespace, which is statically bound to the ft prefix.

Functions

ft:search

Signatures ft:search($db as xs:string, $terms as item()*) as text()*
ft:search($db as xs:string, $terms as item()*, $options as map(*)?) as text()*
Summary Returns all text nodes from the full-text index of the database $db that contain the specified $terms.
The options used for tokenizing the input and building the full-text will also be applied to the search terms. As an example, if the index terms have been stemmed, the search string will be stemmed as well.

The $options argument can be used to control full-text processing. The following options are supported (the introduction on Full-Text processing gives you equivalent expressions in the XQuery Full-Text notation):

  • mode: determines the mode how tokens are searched. Allowed values are any, any word, all, all words, and phrase. any is the default search mode.
  • fuzzy: turns fuzzy querying on or off. Allowed values are true and false. By default, fuzzy querying is turned off.
  • wildcards: turns wildcard querying on or off. Allowed values are true and false. By default, wildcard querying is turned off.
  • ordered: requires that all tokens occur in the order in which they are specified. Allowed values are true and false. The default is false.
  • content: specifies that the matched tokens need to occur at the beginning or end of a searched string, or need to cover the entire string. Allowed values are start, end, and entire. By default, the option is turned off.
  • scope: defines the scope in which tokens must be located. The option has following sub options:
    • same: can be set to true or false. It specifies if tokens need to occur in the same or different units.
    • unit: can be sentence or paragraph. It specifies the unit for finding tokens.
  • window: sets up a window in which all tokens must be located. By default, the option is turned off. It has following sub options:
    • size: specifies the size of the window in terms of units.
    • unit: can be sentences, sentences or paragraphs. The default is words.
  • distance: specifies the distance in which tokens must occur. By default, the option is turned off. It has following sub options:
    • min: specifies the minimum distance in terms of units. The default is 0.
    • max: specifies the maximum distance in terms of units. The default is .
    • unit: can be words, sentences or paragraphs. The default is words.
Errors db:open: The addressed database does not exist or could not be opened.
db:no-index: the index is not available.
options: the fuzzy and wildcard option cannot be both specified.
Examples
  • ft:search("DB", "QUERY"): Return all text nodes of the database DB that contain the term QUERY.
  • Return all text nodes of the database DB that contain the numbers 2010 and 2020:
    ft:search("DB", ("2010", "2020"), map { 'mode': 'all' })
  • Return text nodes that contain the terms A and B in a distance of at most 5 words:

<syntaxhighlight lang="xquery"> ft:search("db", ("A", "B"), map {

 "mode": "all words",
 "distance": map {
   "max": "5",
   "unit": "words"
 }

}) </syntaxhighlight>

  • Iterate over three databases and return all elements containing terms similar to Hello World in the text nodes:

<syntaxhighlight lang="xquery"> let $terms := "Hello Worlds" let $fuzzy := true() for $db in 1 to 3 let $dbname := 'DB' || $db return ft:search($dbname, $terms, map { 'fuzzy': $fuzzy })/.. </syntaxhighlight>

ft:contains

Signatures ft:contains($input as item()*, $terms as item()*) as xs:boolean
ft:contains($input as item()*, $terms as item()*, $options as map(*)?) as xs:boolean
Summary Checks if the specified $input items contain the specified $terms.
The function does the same as the Full-Text expression contains text, but options can be specified more dynamically. The $options are the same as for ft:search, and the following ones in addition:
  • case: determines how character case is processed. Allowed values are insensitive, sensitive, upper and lower. By default, search is case insensitive.
  • diacritics: determines how diacritical characters are processed. Allowed values are insensitive and sensitive. By default, search is diacritical insensitive.
  • stemming: determines is tokens are stemmed. Allowed values are true and false. By default, stemming is turned off.
  • language: determines the language. This option is relevant for stemming tokens. All language codes are supported. The default language is en.
Errors options: specified options are conflicting.
Examples
  • Checks if jack or john occurs in the input string John Doe:

<syntaxhighlight lang="xquery"> ft:contains("John Doe", ("jack", "john"), map { "mode": "any" }) </syntaxhighlight>

  • Calls the function with stemming turned on and off:

<syntaxhighlight lang="xquery"> (true(), false()) ! ft:contains("Häuser", "Haus", map { 'stemming': ., 'language':'de' }) </syntaxhighlight>

ft:mark

Signatures ft:mark($nodes as node()*) as node()*
ft:mark($nodes as node()*, $name as xs:string) as node()*
Summary Puts a marker element around the resulting $nodes of a full-text index request.
The default name of the marker element is mark. An alternative name can be chosen via the optional $name argument.
Please note that:
  • the full-text expression that computes the token positions must be specified as argument of the ft:mark() function, as all position information is lost in subsequent processing steps. You may need to specify more than one full-text expression if you want to use the function in a FLWOR expression, as shown in Example 2.
  • the XML node to be transformed must be an internal "database" node. The transform expression can be used to apply the method to a main-memory fragment, as shown in Example 3.
Examples Example 1: The following query returns <XML><mark>hello</mark> world</XML>, if one text node of the database DB has the value "hello world":

<syntaxhighlight lang="xquery"> ft:mark(db:open('DB')//*[text() contains text 'hello']) </syntaxhighlight> Example 2: The following expression loops through the first ten full-text results and marks the results in a second expression: <syntaxhighlight lang="xquery"> let $start := 1 let $end := 10 let $term := 'welcome' for $ft in (db:open('DB')//*[text() contains text { $term }])[position() = $start to $end] return element hit {

 ft:mark($ft[text() contains text { $term }])

} </syntaxhighlight> Example 3: The following expression returns <p><b>word</b></p>: <syntaxhighlight lang="xquery"> copy $p := <p>word</p> modify () return ft:mark($p[text() contains text 'word'], 'b')</syntaxhighlight>

ft:extract

Signatures ft:extract($nodes as node()*) as node()*
ft:extract($nodes as node()*, $name as xs:string) as node()*
ft:extract($nodes as node()*, $name as xs:string, $length as xs:integer) as node()*
Summary Extracts and returns relevant parts of full-text results. It puts a marker element around the resulting $nodes of a full-text index request and chops irrelevant sections of the result.
The default element name of the marker element is mark. An alternative element name can be chosen via the optional $name argument.
The default length of the returned text is 150 characters. An alternative length can be specified via the optional $length argument. Note that the effective text length may differ from the specified text due to formatting and readibility issues.
For more details on this function, please have a look at ft:mark.
Examples
  • The following query may return <XML>...<b>hello</b>...<XML> if a text node of the database DB contains the string "hello world":

<syntaxhighlight lang="xquery"> ft:extract(db:open('DB')//*[text() contains text 'hello'], 'b', 1) </syntaxhighlight>

ft:count

Signatures ft:count($nodes as node()*) as xs:integer
Summary Returns the number of occurrences of the search terms specified in a full-text expression.
Examples
  • ft:count(//*[text() contains text 'QUERY']) returns the xs:integer value 2 if a document contains two occurrences of the string "QUERY".

ft:score

Signatures ft:score($item as item()*) as xs:double*
Summary Returns the score values (0.0 - 1.0) that have been attached to the specified items. 0 is returned a value if no score was attached.
Examples
  • ft:score('a' contains text 'a') returns the xs:double value 1.

ft:tokens

Signatures ft:tokens($db as xs:string) as element(value)*
ft:tokens($db as xs:string, $prefix as xs:string) as element(value)*
Summary Returns all full-text tokens stored in the index of the database $db, along with their numbers of occurrences.
If $prefix is specified, the returned nodes will be refined to the strings starting with that prefix. The prefix will be tokenized according to the full-text used for creating the index.
Errors db:open: The addressed database does not exist or could not be opened.
db:no-index: the full-text index is not available.
Examples Returns the number of occurrences for a single, specific index entry:

<syntaxhighlight lang="xquery"> let $term := ft:tokenize($term) return number(ft:tokens('db', $term)[. = $term]/@count) </syntaxhighlight>

ft:tokenize

Signatures ft:tokenize($string as xs:string?) as xs:string*
ft:tokenize($string as xs:string?, $options as map(*)?) as xs:string*
Summary Tokenizes the given $string, using the current default full-text options or the $options specified as second argument, and returns a sequence with the tokenized string. The following options are available:
  • case: determines how character case is processed. Allowed values are insensitive, sensitive, upper and lower. By default, search is case insensitive.
  • diacritics: determines how diacritical characters are processed. Allowed values are insensitive and sensitive. By default, search is diacritical insensitive.
  • stemming: determines is tokens are stemmed. Allowed values are true and false. By default, stemming is turned off.
  • language: determines the language. This option is relevant for stemming tokens. All language codes are supported. The default language is en.

The $options argument can be used to control full-text processing.

Examples
  • ft:tokenize("No Doubt") returns the two strings no and doubt.
  • ft:tokenize("École", map { 'diacritics': 'sensitive' }) returns the string école.
  • declare ft-option using stemming; ft:tokenize("GIFTS") returns a single string gift.

ft:normalize

Signatures ft:normalize($string as xs:string?) as xs:string
ft:normalize($string as xs:string?, $options as map(*)?) as xs:string
Summary Normalizes the given $string, using the current default full-text options or the $options specified as second argument. The function expects the same arguments as ft:tokenize.
Examples
  • ft:tokenize("Häuser am Meer", map { 'case': 'sensitive' }) returns the string Hauser am Meer.

Errors

Code Description
options Both wildcards and fuzzy search have been specified as search options.

Changelog

Version 9.1
Version 9.0
  • Updated: error codes updated; errors now use the module namespace
Version 8.0
Version 7.8
Version 7.7
  • Updated: the functions no longer accept Database Nodes as reference. Instead, the name of a database must now be specified.
Version 7.2
  • Updated: ft:search (second argument generalized, third parameter added)
Version 7.1