Issue 5345 - Whitelist $elemhide and $generichide domains where possible

lib/abp2blocklist.js

 /*
  * This file is part of Adblock Plus <https://adblockplus.org/>,
  * Copyright (C) 2006-2017 eyeo GmbH
  *
  * Adblock Plus is free software: you can redistribute it and/or modify
  * it under the terms of the GNU General Public License version 3 as
  * published by the Free Software Foundation.
  *
  * Adblock Plus is distributed in the hope that it will be useful,
  * but WITHOUT ANY WARRANTY; without even the implied warranty of
@@ skipping 60 matching lines @@
 
   for (let name of list)
   {
     if (name.length > suffixLength && name.slice(-suffixLength) == "." + domain)
       subdomains.push(name.slice(0, -suffixLength));
   }
 
   return subdomains;
 }
 
+function extractFilterDomains(filters)
+{
+  let domains = new Set();
+  for (let filter of filters)
+  {
+    let parsed = parseFilterRegexpSource(filter.regexpSource);
+    if (parsed.justHostname)
+      domains.add(parsed.hostname);
+  }
+  return domains;
+}
+
 function convertElemHideFilter(filter, elemhideSelectorExceptions)
 {
   let included = [];
   let excluded = [];
   let rules = [];
 
   parseDomains(filter.domains, included, excluded);
 
   if (excluded.length == 0 && !(filter.selector in elemhideSelectorExceptions))
     return {matchDomains: included.map(matchDomain), selector: filter.selector};
@@ skipping 35 matching lines @@
     // If we're currently inside the hostname we have to be careful not to
     // escape any characters until after we have converted it to punycode.
     if (hostnameStart != null && !hostnameFinished)
     {
       let endingChar = (c == "*" || c == "^" ||
                         c == "?" || c == "/" || c == "|");
       if (!endingChar && i != lastIndex)
         continue;
 
       hostname = punycode.toASCII(
-        text.substring(hostnameStart, endingChar ? i : i + 1)
+        text.substring(hostnameStart, endingChar ? i : i + 1).toLowerCase()
       );
       hostnameFinished = justHostname = true;
       regexp.push(escapeRegExp(hostname));
       if (!endingChar)
         break;
     }
 
     switch (c)
     {
       case "*":
@@ skipping 246 matching lines @@
   {
     newSelector.push(selector.substring(i, pos.start));
     newSelector.push('[id=', selector.substring(pos.start + 1, pos.end), ']');
     i = pos.end;
   }
   newSelector.push(selector.substring(i));
 
   return newSelector.join("");
 }
 
-function addCSSRules(rules, selectors, matchDomain)
+function addCSSRules(rules, selectors, matchDomain, exceptionDomains)
 {
+  let unlessDomain = exceptionDomains.size > 0 ? [] : null;

[kzar, 2017/07/11 12:20:03]

Nit: This seems like overkill since `[]` is falsey anyway.

[Manish Jethani, 2017/07/11 16:28:39]

[] evaluates to true. If you mean checking the length property, we could do
that, but then I'm just wondering why since it's inside the loop.

[kzar, 2017/07/11 16:30:36]

Whoops, you're right.

+
+  exceptionDomains.forEach(name => unlessDomain.push("*" + name));
+
   while (selectors.length)
   {
     let selector = selectors.splice(0, selectorLimit).join(", ");
 
     // As of Safari 9.0 element IDs are matched as lowercase. We work around
     // this by converting to the attribute format [id="elementID"]
     selector = convertIDSelectorsToAttributeSelectors(selector);
 
-    rules.push({
+    let rule = {
       trigger: {"url-filter": matchDomain,
                 "url-filter-is-case-sensitive": true},
       action: {type: "css-display-none",
                selector: selector}
-    });
+    };
+
+    if (unlessDomain)
+      rule.trigger["unless-domain"] = unlessDomain;
+
+    rules.push(rule);
   }
 }
 
 let ContentBlockerList =
 /**
  * Create a new Adblock Plus filter to content blocker list converter
  *
  * @constructor
  */
 exports.ContentBlockerList = function ()
@@ skipping 76 matching lines @@
     {
       for (let matchDomain of result.matchDomains)
       {
         let group = groupedElemhideFilters.get(matchDomain) || [];
         group.push(result.selector);
         groupedElemhideFilters.set(matchDomain, group);
       }
     }
   }
 
-  addCSSRules(rules, genericSelectors, "^https?://");
+  // Separate out the element hiding exceptions that have only a hostname part
+  // from the rest. This allows us to implement a workaround for issue #5345
+  // (WebKit bug #167423), but as a bonus it also reduces the number of
+  // generated rules. The downside is that the exception will only apply to the
+  // top-level document, not to iframes. We have to live with this until the
+  // WebKit bug is fixed in all supported versions of Safari.
+  // https://bugs.webkit.org/show_bug.cgi?id=167423
+  //
+  // Note that as a result of this workaround we end up with a huge rule set in
+  // terms of the amount of memory used. This can cause Node.js to throw
+  // "JavaScript heap out of memory". To avoid this, call Node.js with
+  // --max_old_space_size=4096
+  let elemhideExceptionDomains = extractFilterDomains(this.elemhideExceptions);
 
-  // Right after the generic element hiding filters, add the exceptions that
-  // should apply only to those filters.
-  for (let filter of this.generichideExceptions)
-    convertFilterAddRules(rules, filter, "ignore-previous-rules", false);
+  let genericSelectorExceptionDomains =
+    extractFilterDomains(this.generichideExceptions);
+  elemhideExceptionDomains.forEach(name =>
+  {
+    genericSelectorExceptionDomains.add(name);
+  });
+
+  addCSSRules(rules, genericSelectors, "^https?://",
+              genericSelectorExceptionDomains);
 
   groupedElemhideFilters.forEach((selectors, matchDomain) =>
   {
-    addCSSRules(rules, selectors, matchDomain);
+    addCSSRules(rules, selectors, matchDomain, elemhideExceptionDomains);
   });
 
-  for (let filter of this.elemhideExceptions)
-    convertFilterAddRules(rules, filter, "ignore-previous-rules", false);
-
   let requestFilterExceptionDomains = [];
   for (let filter of this.genericblockExceptions)
   {
     let parsed = parseFilterRegexpSource(filter.regexpSource);
     if (parsed.hostname)
       requestFilterExceptionDomains.push(parsed.hostname);
   }
 
   for (let filter of this.requestFilters)
   {
     convertFilterAddRules(rules, filter, "block", true,
                           requestFilterExceptionDomains);
   }
 
   for (let filter of this.requestExceptions)
     convertFilterAddRules(rules, filter, "ignore-previous-rules", true);
 
   return rules.filter(rule => !hasNonASCI(rule));
 };