-- Upgrade pg_kazsearch 0.1.0 -> 2.2.0 -- -- 0.1.0 installs only mapped word-class tokens, so numbers, URLs and emails -- were dropped from every tsvector. 2.2.0 adds `simple` mappings for them. -- The template/dictionary/configuration objects are otherwise unchanged -- (stemmer behavior lives in the shared library, which is replaced on -- upgrade). -- -- NOTE: tsvectors built before this upgrade do not contain numeric/url -- tokens; regenerate stored/generated tsvector columns to pick them up. -- Idempotent: dev builds of "0.1.0" from newer sources may already carry -- these mappings, and ADD MAPPING has no IF NOT EXISTS. ALTER TEXT SEARCH CONFIGURATION kazakh_cfg DROP MAPPING IF EXISTS FOR numword, numhword, hword_numpart, int, uint, float, sfloat, version, email, url, url_path, host, file; ALTER TEXT SEARCH CONFIGURATION kazakh_cfg ADD MAPPING FOR numword, numhword, hword_numpart, int, uint, float, sfloat, version, email, url, url_path, host, file WITH simple;