Note
This documentation is for a prior release of Kinetica. For the latest documentation, click here.
The SQL FILTER_BY_STRING function, as well as the /filter/bystring endpoint, enables the use of several string-based search modes, including full text search via the search mode. The syntax and restrictions for this search follow, including examples of applying different search criteria.
In order for a full text search to be applied to a column, the following requirements must be met:
- Full text search must be enabled by setting enable_text_search to true
in the
/opt/gpudb/core/etc/gpudb.confconfiguration file; see Text Search for details. - The column must be a string type and have the
text_search column property applied.
The following effective types under the string base type are
text-searchable:
- char1 - char256
- date
- datetime
- ipv4
- string
- time
- uuid
Individual Term Match
To search for one or more search terms, add the terms to a space-separated list.
| Search Text | Result |
|---|---|
| perfect union | matches any record containing perfect, union, or both |
Exact Phrase Match
To search for an exact phrase, double quote the entire search text.
| Search Text | Result |
|---|---|
| "Perfect Union" | matches the exact phrase Perfect Union |
Boolean Operations
Valid operators include NOT, AND, OR, & parentheses (). An OR is assumed if no operator is specified.
| Search Text | Result |
|---|---|
| justice AND tranquility | matches records containing both justice and tranquility |
| justice OR tranquility | matches records containing either justice or tranquility (or both) Note This is the default behavior, so justice OR tranquility is equivalent to justice tranquility |
| justice NOT tranquility | matches records containing justice that don't contain tranquility |
| (justice tranquility honesty) AND peace | matches records containing peace that also contain any one or more of justice, tranquility, & honesty |
Wildcard Match
Wildcard matches can be specified with either single-character or multi-character wildcards, or a combination of the two:
- ? - Exactly one character
- * - Zero or more characters
| Search Text | Result |
|---|---|
| est*ish* | matches any records containing a word that starts with est, followed by any number of characters, followed by is, and ending with any number of additional characters; this would match records containing establish, establishable, and establishment |
| est???is? | matches any records containing a word that starts with est, followed by exactly three characters, followed by is, and ending with any single character; this would match records containing establish, but not establishable or establishment |
| est*abl? | matches any records containing a word that starts with est, followed by any number of characters, followed by abl, and ending with any single character; this would match records containing establishable, but not establish or establishment |
Fuzzy Match
Fuzzy matches can be specified by appending a tilde (~) to the end of the fuzzy match term. A decimal number between 0 (inclusive) and 1 (exclusive) can be used to specify the minimum relative similarity of the match, with 0 indicating the fuzziest match possible, and values closer to 1 indicating an increasingly stricter match. The default match threshold is .5.
| Search Text | Result |
|---|---|
| rear~ | matches records containing rear, fear, bear, or read |
| rear~.8 | matches records containing rear, but not fear, bear, or read |
Proximity Match
To match two words within a specified distance of each other, quote the two terms, separated by a space, and append a tilde (~) followed by the distance in number of words.
| Search Text | Result |
|---|---|
| "Union Tranquility"~10 | matches records containing Union and Tranquility within 10 words of each other |
Term Relevance
To increase the relevance of a matched search term versus other given terms, append a caret (^) followed by a positive numeric boosting factor for that term. A quote-enclosed phrase can also have its relevance increased. The default factor is 1.
| Search Text | Result |
|---|---|
| we have fun^5 | matches records containing any of we, have, or fun, but giving a greater match relevance score to records containing fun |
Required, Optional, & Prohibited Matches
Within the context of a group of terms, any of the terms can be marked as required or prohibited:
- + - when prepended to a term, requires that the term be matched
- - - when prepended to a term, prohibits matches on the term
- <no marker> - when neither marker is prepended to a term, the term is optional
| Search Text | Result |
|---|---|
| we -never have +fun | matches records containing fun, optionally containing we or have, and not containing never |
Range Matches
Inclusive [<term1> TO <term2>] and exclusive {<term1> TO <term2>} matches are supported. Matches where one side is inclusive and the other exclusive are not supported.
| Search Text | Result |
|---|---|
| [100 TO 200] | matches records containing any numbers that lexicographically sort between 100 and 200, including 100 and 200 Note This is a lexicographical comparison, so a search of [100 TO 200] would match 2, 20, or 1AZ |
| {alpha TO beta} | matches records containing words that lexicographically sort between alpha and beta, not including either alpha or beta |
Escaping Special Characters
Special characters are escaped with a backslash (\). Special characters include:
+ - ^ " | ~ * ? : | \ && || ! | ( ) { } [ ] |
Restrictions
The first character of a search term cannot be a wildcard (* or ?)
Search terms cannot be any of the following:
a
an
and
are
as
at
be
but
by
for
if
in
into
is
it
no
not
of
on
or
such
that
the
their
then
these
they
this
to
was
will
with
