fuzzymatching: only use default extension based first-non-query match… (

#110)

* fuzzymatching: only use default extension based first-non-query match fuzzy matching if <=2 results, otherwise use the more extensive levdist based matching
fuzzymatching: add rule for youtube profile
deps: bump wombat 3.4.6 for style url unrewriting
part of fix for #108

package.json

@@ -8,7 +8,7 @@
     "@peculiar/asn1-ecc": "^2.3.4",
     "@peculiar/asn1-schema": "^2.3.3",
     "@peculiar/x509": "^1.9.2",
-    "@webrecorder/wombat": "^3.4.4",
+    "@webrecorder/wombat": "^3.4.6",
     "auto-js-ipfs": "^2.1.0",
     "base64-js": "^1.5.1",
     "brotli": "^1.3.3",

src/fuzzymatcher.js

@@ -35,9 +35,14 @@ const DEFAULT_RULES =
     "split": "_",
     "splitLast": true
   },
+  // YouTube
   {
     "match": /^https?:\/\/(?:www\.)?(youtube\.com\/embed\/[^?]+)[?].*/i,
-    "replace": "$1",
+    "replace": "$1"
+  },
+  {
+    "match": /^(https?:\/\/(?:www\.)?)(youtube\.com\/@[^?]+)[?].*/i,
+    "fuzzyCanonReplace": "$1$2"
   },
   {
     "match": /\/\/(?:www\.)?youtube(?:-nocookie)?\.com\/(get_video_info)/i,
@@ -122,7 +127,8 @@ const DEFAULT_RULES =
   },
   {
     "match": /(\.(?:js|webm|mp4|gif|jpg|png|css|json|m3u8))\?.*/i,
-    "replace": "$1"
+    "replace": "$1",
+    "maxResults": 2
   }
 ];
 
@@ -190,7 +196,9 @@ class FuzzyMatcher {
       return null;
     }
 
-    if (matchedRule && matchedRule.replace !== undefined && matchedRule.match !== undefined) {
+    if (matchedRule && matchedRule.replace !== undefined && matchedRule.match !== undefined &&
+      // if limit exists, only apply if < limit results
+      (!matchedRule.maxResults || results.length <= matchedRule.maxResults)) {
       const match = matchedRule.match;
       const replace = matchedRule.replace;
       const fuzzyReqUrl = reqUrl.replace(match, replace);

yarn.lock

@@ -403,10 +403,10 @@
   resolved "https://registry.yarnpkg.com/@webpack-cli/serve/-/serve-1.5.2.tgz#ea584b637ff63c5a477f6f21604b5a205b72c9ec"
   integrity sha512-vgJ5OLWadI8aKjDlOH3rb+dYyPd2GTZuQC/Tihjct6F9GpXGZINo3Y/IVuZVTM1eDQB+/AOsjPUWH/WySDaXvw==
 
-"@webrecorder/wombat@^3.4.4":
-  version "3.4.4"
-  resolved "https://registry.yarnpkg.com/@webrecorder/wombat/-/wombat-3.4.4.tgz#f01105f06eaa52f72b1ad7ecfe8290c5f5fa4ee9"
-  integrity sha512-2n7fG4YVs4kg1iTX84iNsV8sOrQe/Gce/1mguLdnTrNp9mW843bkKOaSPvuFjM8sa7pr3a9faVbvA3nBZZ9R3Q==
+"@webrecorder/wombat@^3.4.6":
+  version "3.4.6"
+  resolved "https://registry.yarnpkg.com/@webrecorder/wombat/-/wombat-3.4.6.tgz#48c8a03b1d47789fd799f6d9d49ec1ca6c610a1f"
+  integrity sha512-YMGKT78B6nDOIV8irIvIPtcWGOa44/FxFXN6LqeUM8WwJrvLL3YI+pXfKYd7GYJFULNqir3Ao+nSNNrjy9PtJg==
   dependencies:
     warcio "^2.0.1"