fuzzy matching rules: more exact rules to fix replay of youtube and v…

…imeo content (part of work for webrecorder/browsertrix-crawler#4) bump version to 2.5.3
webrecorder · Jan 27, 2021 · 9611847 · 9611847
1 parent cae03c0
commit 9611847
Show file tree

Hide file tree

Showing 3 changed files with 6 additions and 5 deletions.
diff --git a/dist/sw.js b/dist/sw.js
diff --git a/package.json b/package.json
@@ -1,6 +1,6 @@
 {
   "name": "@webrecorder/wabac",
-  "version": "2.5.2",
+  "version": "2.5.3",
   "main": "index.js",
   "license": "AGPL-3.0-or-later",
   "dependencies": {

diff --git a/src/fuzzymatcher.js b/src/fuzzymatcher.js
@@ -28,10 +28,11 @@ const DEFAULT_RULES =
   {
     "match": /(twimg.com\/profile_images\/[^/]+\/[^_]+)_([\w]+\.[\w]+)/,
     "replace": "$1=_args=$2",
-    "split": "_"
+    "splitLast": "_",
+    "last": true
   },
   {
-   "match": /(youtube\.com\/embed\/[^?]+)[?].*/i,
+   "match": /^https?:\/\/(youtube\.com\/embed\/[^?]+)[?].*/i,
    "replace": "$1"
   },
   {
@@ -130,7 +131,7 @@ class FuzzyMatcher {;
     }
 
     const split = rule && rule.split || "?";
-    const inx = reqUrl.lastIndexOf(split);
+    const inx = rule.splitLast ? reqUrl.lastIndexOf(split) : reqUrl.indexOf(split);
     const prefix = inx > 0 ? reqUrl.slice(0, inx + split.length) : reqUrl;
 
     return {prefix, rule, fuzzyCanonUrl};