Issue 3775 - fix saving of requests

lib/crawler.js

Index: lib/crawler.js
diff --git a/lib/crawler.js b/lib/crawler.js
index fae18c76edffa77382910c4ef55a45167635d02b..3445b2f5424fd846e71c3ffe3cb6fe964fc78329 100644
--- a/lib/crawler.js
+++ b/lib/crawler.js
@@ -20,11 +20,8 @@ function abprequire(module)
   return result.exports;
 }
 
-let {Policy} = abprequire("contentPolicy");
 let {RequestNotifier} = abprequire("requestNotifier");
-let {Utils} = abprequire("utils");
 
-let dataForTab = new WeakMap();
 
 /**
  * Creates a pool of tabs and allocates them to tasks on request.
@@ -193,31 +190,6 @@ WindowCloser.prototype = {
 };
 
 /**
- * Retrieves crawler results associated with a particular content window.
- *
- * @param {Window} window
- *    Content window to retrieve crawler results for
- * @result {Object}
- *    Crawler results or undefined if the window wasn't created by the crawler.
- */
-function getDataForWindow(window)
-{
-  let topWindow = window.top;
-  if (!topWindow.document)
-    throw new Error("No document associated with the node's top window");
-  let tabbrowser = Utils.getChromeWindow(topWindow).getBrowser();
-  if (!tabbrowser)
-    throw new Error("Unable to get a tabbrowser reference from the window");
-  let browser = tabbrowser.getBrowserForDocument(topWindow.document);
-  if (!browser)
-    throw new Error("Unable to get browser for the content window");
-  let tab = tabbrowser.getTabForBrowser(browser);
-  if (!tab)
-    throw new Error("Unable to get tab for the browser");
-  return dataForTab.get(tab);
-};
-
-/**
  * Starts the crawling session. The crawler opens each URL in a tab and stores
  * the results.
  *
@@ -232,11 +204,6 @@ function getDataForWindow(window)
  */
 function run(window, urls, timeout, maxtabs, targetURL, onDone)
 {
-  let requestNotifier = new RequestNotifier(null, function() {});
-
-  let origProcessNode = Policy.processNode;
-  Policy.processNode = processNodeReplacement.bind(null, origProcessNode, requestNotifier);
-
   let tabAllocator = new TabAllocator(window.getBrowser(), maxtabs);
   let loadListener = new LoadListener(window.getBrowser(), timeout);
   let running = 0;
@@ -246,8 +213,6 @@ function run(window, urls, timeout, maxtabs, targetURL, onDone)
     running--;
     if (running <= 0)
     {
-      Policy.processNode = origProcessNode;
-      requestNotifier.shutdown();
       loadListener.stop();
       windowCloser.stop();
       onDone();
@@ -294,12 +259,20 @@ exports.run = run;
 function* crawl_url(url, tabAllocator, loadListener)
 {
   let tab = yield tabAllocator.getTab();
-  let result = {url: url};
-
-  dataForTab.set(tab, result);
+  let result = {url, requests: []};
+  let requestNotifier;
   try
   {
     result.startTime = Date.now();
+    requestNotifier = new RequestNotifier(tab.linkedBrowser.outerWindowID,
+      function(entry, scanComplete)
+    {
+      if (!entry)
+        return;
+      let {type: contentType, location, filter} = entry;
+      result.requests.push({location, contentType, filter});
+    });
+
     tab.linkedBrowser.loadURI(url, null, null);
     [result.channelStatus, result.headers] = yield loadListener.waitForLoad(tab);
     result.endTime = Date.now();
@@ -331,6 +304,8 @@ function* crawl_url(url, tabAllocator, loadListener)
   }
   finally
   {
+    if (requestNotifier)
+      requestNotifier.shutdown();
     tabAllocator.releaseTab(tab);
   }
   return result;
@@ -345,63 +320,3 @@ function reportException(e)
   Cu.reportError(e);
   dump(e + "\n" + stack + "\n");
 }
-
-/**
- * Wrapper for the Policy.processNode() function in ABP. Calls the original
- * function and records all the data.
- *
- * @param {Function} origProcessNode
- *      The original processNode function.
- * @param {RequestNotifier} requestNotifier
- *      The crawler's RequestNotifier object instance.
- * @param {nsIDOMWindow} wnd
- * @param {nsIDOMElement} node
- * @param {Number} contentType
- * @param {nsIURI} location
- * @param  {Boolean} collapse
- * @return {Boolean}
- */
-function processNodeReplacement(origProcessNode, requestNotifier, wnd, node, contentType, location, collapse)
-{
-  let filters = [];
-  let origListener = requestNotifier.listener;
-  requestNotifier.listener = function(window, node, entry)
-  {
-    if (entry.filter)
-      filters.push(entry.filter.text);
-  };
-
-  /*
-   * Call the original processNode. If the original throws, then we will too, so this is outside a try clause.
-   */
-  let result;
-  try
-  {
-    result = origProcessNode(wnd, node, contentType, location, collapse);
-  }
-  finally
-  {
-    requestNotifier.listener = origListener;
-  }
-
-  try
-  {
-    let data = getDataForWindow(wnd);
-    if (data)
-    {
-      if (!("requests" in data))
-        data.requests = [];
-      data.requests.push({
-        contentType: contentType,
-        location: (contentType == Policy.type.ELEMHIDE ? location.text : location.spec),
-        blocked: result != Ci.nsIContentPolicy.ACCEPT,
-        filters: filters
-      });
-    }
-  }
-  catch (e)
-  {
-    reportException(e);
-  }
-  return result;
-};