From dd9ba9cb9712e58c7a8f0b429d55f938e5f26a4b Mon Sep 17 00:00:00 2001 From: Jaifroid Date: Mon, 26 Jul 2021 12:22:06 +0100 Subject: [PATCH] Find duplicates more efficiently Former-commit-id: 94bf9313afa799b4b14c7106f1a2a05037f30fe6 [formerly 8dfec4c5e2eb471cd2dc463a632568045cd76b6c [formerly ec1ddcec929eaec26f44828357e00d507f6a0017]] Former-commit-id: 53342fcc51dbcecb79692a504b0f626048d8d69a Former-commit-id: cf821052e0d34a0441e3153d846a719cf4159f7d --- www/js/lib/kiwixServe.js | 4 ++-- 1 file changed, 2 insertions(+), 2 deletions(-) diff --git a/www/js/lib/kiwixServe.js b/www/js/lib/kiwixServe.js index 2973924c..f16dfdd1 100644 --- a/www/js/lib/kiwixServe.js +++ b/www/js/lib/kiwixServe.js @@ -780,10 +780,10 @@ define([], function () { if (/^(?:[^._]+\.stack(?:exchange|overflow)|askubuntu|stackapps|stackoverflow|superuser|serverfault)/mi.test(fromDoc)) { subList = fromDoc.replace(/^(?:.+(stackoverflow)|[^"]+"([^.]+)).+[\r\n]/img, '$1$2\n'); } else { - subList = fromDoc.replace(/^[^_]+_[^_]+_((?:[^_]|_(?!maxi|mini|nopic|\d\d\d\d))+).+[\r\n]*/mg, '$1\n'); + subList = fromDoc.replace(/^[^>]+>[^_]+_[^_]+_((?:[^_]|_(?!maxi|mini|nopic|\d\d\d\d))+).+[\r\n]*/mg, '$1\n'); } //Delete recurrences - subList = subList.replace(/\b([\w_-]+\n)(?=[\s\S]*\b\1\n?)/g, ''); + subList = subList.replace(/^([\w_-]+)$[\r\n]*(?=^\1$)/gm, ''); //Remove 'all' subList = subList.replace(/^all$/mi, ''); var subArray = subList.match(/^.+$/mg);