#332 Regexp rules shortcut

AdguardTeam · Nov 18, 2016 · 07735b8 · 07735b8
1 parent a0656b1
commit 07735b8
Show file tree

Hide file tree

Showing 2 changed files with 67 additions and 4 deletions.
diff --git a/Extension/lib/filter/rules/url-filter-rule.js b/Extension/lib/filter/rules/url-filter-rule.js
@@ -64,6 +64,9 @@ var UrlFilterRule = exports.UrlFilterRule = function (rule, filterId) {
         if (!regexp) {
             throw 'Illegal regexp rule';
         }
+
+        // Extract shortcut from regexp rule
+        this.shortcut = extractRegexpShortcut(urlRuleText);
     } else {
         // Searching for shortcut
         this.shortcut = findShortcut(urlRuleText);
@@ -403,10 +406,15 @@ function parseRuleDomain(ruleText, parseOptions) {
     }
 }
 
-// Searches for the shortcut of this url mask.
-// Shortcut is the longest part of the mask without special characters:
-// *,^,|. If not found anything with the length greater or equal to 8 characters -
-// shortcut is not used.
+/**
+ * Searches for the shortcut of this url mask.
+ * Shortcut is the longest part of the mask without special characters:
+ * *,^,|. If not found anything with the length greater or equal to 8 characters -
+ * shortcut is not used.
+ *
+ * @param urlmask
+ * @returns {string}
+ */
 function findShortcut(urlmask) {
     var longest = "";
     var parts = urlmask.split(/[*^|]/);
@@ -419,6 +427,49 @@ function findShortcut(urlmask) {
     return longest.toLowerCase();
 }
 
+/**
+ * Extracts a shortcut from a regexp rule.
+ *
+ * @param ruleText
+ * @returns {*}
+ */
+function extractRegexpShortcut(ruleText) {
+
+    // Get the regexp text
+    var reText = ruleText.match(/\/(.*)\/(\$.*)?/)[1];
+
+    var specialCharacter = "...";
+
+    if (reText.indexOf('(?') >= 0 || reText.indexOf('(!?') >= 0) {
+        // Do not mess with complex expressions which use lookahead
+        return null;
+    }
+
+    // (Dirty) prepend specialCharacter for the following replace calls to work properly
+    reText = specialCharacter + reText;
+
+    // Strip all types of brackets
+    reText = reText.replace(/[^\\]\(.*[^\\]\)/, specialCharacter);
+    reText = reText.replace(/[^\\]\[.*[^\\]\]/, specialCharacter);
+    reText = reText.replace(/[^\\]\{.*[^\\]\}/, specialCharacter);
+
+    // Strip some special characters
+    reText = reText.replace(/[^\\]\\[a-zA-Z]/, specialCharacter);
+
+    // Split by special characters
+    var parts = reText.split(/[\\^$*+?.()|[\]{}]/);
+    var token = "";
+    var iParts = parts.length;
+    while (iParts--) {
+        var part = parts[iParts];
+        if (part.length > token.length) {
+            token = part;
+        }
+    }
+
+    return token;
+}
+
 /**
  * Parse rule text
  * @param ruleText

diff --git a/Extension/tests/url-filter/test-url-filter.js b/Extension/tests/url-filter/test-url-filter.js
@@ -253,4 +253,16 @@ QUnit.test("Complex regexp rule", function(assert) {
     assert.notOk(rule.isFiltered("http://connect.facebook.net/blahblah.js", true, RequestTypes.SCRIPT));
     assert.notOk(rule.isFiltered("https://ajax.cloudflare.com/blahblah.js", true, RequestTypes.SCRIPT));
     assert.notOk(rule.isFiltered("https://www.google-analytics.com/blahblah.js", true, RequestTypes.SCRIPT));
+});
+
+QUnit.test("Regexp rules shortcuts", function(assert) {
+    assert.equal(new UrlFilterRule('/quang%20cao/').shortcut, 'quang%20cao');
+    assert.equal(new UrlFilterRule('/YanAds/').shortcut, 'YanAds');
+    assert.equal(new UrlFilterRule('/^http://m\.autohome\.com\.cn\/[a-z0-9]{32}\//$domain=m.autohome.com.cn').shortcut, 'autohome');
+    assert.equal(new UrlFilterRule('/cdsbData_gal/bannerFile/$image,domain=mybogo.net|zipbogo.net	').shortcut, 'cdsbData_gal/bannerFile');
+    assert.equal(new UrlFilterRule('/http:\/\/rustorka.com\/[a-z]+\.js/$domain=rustorka.com').shortcut, 'http://rustorka');
+    assert.equal(new UrlFilterRule('/^http://www\.iqiyi\.com\/common\/flashplayer\/[0-9]{8}/[0-9a-z]{32}.swf/$domain=iqiyi.com').shortcut, 'com/common/flashplayer');
+    assert.equal(new UrlFilterRule('/ulightbox/$domain=hdkinomax.com|tvfru.net').shortcut, 'ulightbox');
+    assert.equal(new UrlFilterRule('/\.sharesix\.com/.*[a-zA-Z0-9]{4}/$script').shortcut, 'sharesix');
+    assert.equal(new UrlFilterRule('/serial_adv_files/$image,domain=xn--80aacbuczbw9a6a.xn--p1ai|куражбамбей.рф').shortcut, 'serial_adv_files');
 });