babelfish-for-postgresql / babelfish_extensions / 8700893968

76%

BABEL_5_X_DEV: 77%

Build Type

push

github

Committed by web-flow

Commit Message

Support LIKE operator for AI collations (Latin characters) (#2412)

This PR supports the LIKE operator for Accent Insensitive collations based on Latin (All the collations ending *CS_AI and
*CI_AI which are also known as non-deterministic collations). The native Postgres does not support LIKE operator for
non-deterministic collations, so we have made some changes in the Babelfish extensions side to parse the incoming
node. We will try to identify the left and right node of LIKE operator and process them with the help of remove_accents_internal function using following transformation. (This remove_accents_internal nothing but will remove accents from input string.)

pattern LIKE text COLLATE CI_AI
<->
remove_accents_internal(pattern) as u_p LIKE remove_accents_internal(text) as u_t COLLATE CI_AS
<->
(u_p ILIKE u_t COLLATE CS_AS) AND 
(u_t BETWEEN unaccent(pattern_const) AND (unaccent(pattern_const) || 0xffff)) COLLATE CS_AS)

If we have not handled any kind of node explicitly, then we will throw error. The remove_accents_internal take help of ICU provided APIs to obtain unaccented version of input string with transformation rule.

Authored-by: Shameem Ahmed shmeeh@amazon.com
Co-Authored-by: Dipesh Dhameliya dddhamel@amazon.com
Signed-off-by: Shameem Ahmed shmeeh@amazon.com
Issues Resolved

[BABEL-4791]

Run Details

114 of 131 new or added lines in 1 file covered. (87.02%)

3 existing lines in 1 file now uncovered.

41186 of 56784 relevant lines covered (72.53%)

65104.82 hits per line