heimrichhannot / contao-utils-bundle / 38693780554

84%

push

github

Sort search vector matches by relevance (#111)

* sort search vector matches by relevance

The words of an alternative may be found anywhere in a record, so a search
for "Dresdner Stollen" also matched a record mentioning both words in
unrelated sentences. Sorted alphabetically with a limit, such matches pushed
out the relevant ones.

orderByRelevance() and createRelevance() score a word at the start of a
word, a whole word and all words of an alternative in a row, using only LIKE
patterns.

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>

* add pull request link to changelog

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>

* score the words of quoted phrases and repeated words in the relevance

The relevance was built from the filter terms, which are deduplicated and
keep a quoted phrase as a single term. So "haus an haus am see" rewarded
the shortened sequence "haus an am see", and a quoted phrase scored like a
single word without its bonus for the words in a row.

Tokenizer::parseQueryWords() returns the words of each alternative in the
order of the query, with repetitions and quoted phrases split into words.
The relevance scores each distinct word and the full sequence.

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>

* parse only the representation of the search each method needs

Filter methods only need the filter terms and relevance methods only the
words, so the search is no longer tokenized and stemmed twice per call.

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>

---------

Co-authored-by: Claude Opus 5.5 <noreply@anthropic.com>

55 of 55 new or added lines in 2 files covered. (100.0%)

1625 of 1926 relevant lines covered (84.37%)

10.78 hits per line