fix(analytics): replace regexp_split_to_table with UNNEST+STRING_TO_ARRAY and fix regex escapes

This commit is contained in:
MythEclipse
2026-06-03 19:55:43 +07:00
parent 4e12394168
commit cbee35c348
@@ -323,15 +323,15 @@ export class AnalyticsRepository {
LOWER(TRIM(BOTH '.,!?;:\"()[]{}' FROM word)) AS word, LOWER(TRIM(BOTH '.,!?;:\"()[]{}' FROM word)) AS word,
ai_moderation_score ai_moderation_score
FROM messages, FROM messages,
LATERAL regexp_split_to_table(content, E'\\\\s+') AS word LATERAL UNNEST(STRING_TO_ARRAY(content, ' ')) AS word
${filter.where} ${filter.where}
AND content IS NOT NULL AND content IS NOT NULL
AND content != '' AND content != ''
AND LENGTH(TRIM(BOTH '.,!?;:\"()[]{}' FROM word)) >= 3 AND LENGTH(TRIM(BOTH '.,!?;:\"()[]{}' FROM word)) >= 3
AND word !~ '^<.+:\\d+>$' AND word !~ '^<.+:\\d+>$'
AND word !~ '^https?://' AND word !~ '^https?://'
AND word !~ '^discord\\.(gg|app|com)' AND word !~ '^discord\.(gg|app|com)'
AND word !~ '^cdn\\.discord' AND word !~ '^cdn\.discord'
) )
SELECT SELECT
word AS topic, word AS topic,