Fix tab URL misattribution in learning listeners.

By checking the request's document URL when available.
EFForg · Oct 21, 2018 · 11c8da7 · 11c8da7
1 parent d8aedad
commit 11c8da7
Show file tree

Hide file tree

Showing 3 changed files with 72 additions and 63 deletions.
diff --git a/src/js/heuristicblocking.js b/src/js/heuristicblocking.js
@@ -114,6 +114,14 @@ HeuristicBlocker.prototype = {
 
     let tab_origin = tabOrigins[details.tabId];
 
+    // we may no longer be on the page the request is coming from
+    const request_doc_host = utils.getDocumentHostForRequest(details),
+      request_doc_origin = window.getBaseDomain(request_doc_host),
+      misattribution = request_doc_origin && request_doc_origin != tab_origin;
+    if (misattribution) {
+      tab_origin = request_doc_origin;
+    }
+
     // ignore first-party requests
     if (!tab_origin || request_origin == tab_origin) {
       return {};

diff --git a/src/js/utils.js b/src/js/utils.js
@@ -344,15 +344,75 @@ function getHostFromDomainInput(input) {
   return uri.host;
 }
 
+/**
+ * Gets the hostname for a given request's top-level document.
+ *
+ * The request's document may be different from the current top-level document
+ * loaded in tab as requests can come out of order:
+ *
+ * - "main_frame" requests usually but not always mark a boundary
+ *   (navigating to another site while the current page is still loading)
+ * - sometimes there is no "main_frame" request
+ *   (service worker pages in Firefox)
+ *
+ * @param {Object} details chrome.webRequest request details object
+ *
+ * @return {String} the hostname for the request's top-level document
+ */
+function getDocumentHostForRequest(details) {
+  let host, url;
+
+  // Firefox 54+
+  if (details.hasOwnProperty("documentUrl")) {
+    if (details.type == "main_frame") {
+      // the top-level document itself
+      url = details.url;
+    } else if (details.hasOwnProperty("frameAncestors")) {
+      // Firefox 58+
+      if (details.frameAncestors.length) {
+        // inside a frame
+        url = details.frameAncestors[details.frameAncestors.length - 1].url;
+      } else {
+        // inside the top-level document
+        url = details.documentUrl;
+      }
+    } else {
+      // TODO Firefox 54-57 or a service worker request
+      if (details.documentUrl.endsWith("/sw.js")) {
+        url = details.documentUrl;
+      }
+    }
+
+  // Chrome 63+
+  } else if (details.hasOwnProperty("initiator")) {
+    if (details.initiator && details.initiator != "null") {
+      if (details.type == "main_frame") {
+        url = details.url;
+      } else if (details.parentFrameId == -1 || details.type == "sub_frame" && details.parentFrameId === 0) {
+        // TODO can only rely on initiator for main frame resources:
+        // https://crbug.com/838242#c17
+        url = details.initiator + '/';
+      }
+    }
+  }
+
+  if (url) {
+    host = window.extractHostFromURL(url);
+  }
+
+  return host;
+}
+
 /************************************** exports */
 var exports = {
   arrayBufferToBase64,
   estimateMaxEntropy,
   explodeSubdomains,
+  getDocumentHostForRequest,
   getHostFromDomainInput,
   nDaysFromNow,
-  oneDayFromNow,
   oneDay,
+  oneDayFromNow,
   oneHour,
   oneMinute,
   oneSecond,

diff --git a/src/js/webrequest.js b/src/js/webrequest.js
@@ -80,7 +80,7 @@ function onBeforeRequest(details) {
   // if we are no longer on the page the request is coming for,
   // don't log in popup or attempt to replace social widgets
   // but do block request/modify headers
-  const request_doc_host = getDocumentHostForRequest(details),
+  const request_doc_host = utils.getDocumentHostForRequest(details),
     misattribution = request_doc_host && request_doc_host != tab_host;
   if (misattribution) {
     tab_host = request_doc_host;
@@ -178,7 +178,7 @@ function onBeforeSendHeaders(details) {
   // if we are no longer on the page the request is coming for,
   // don't log in popup or attempt to replace social widgets
   // but do block request/modify headers
-  const request_doc_host = getDocumentHostForRequest(details),
+  const request_doc_host = utils.getDocumentHostForRequest(details),
     misattribution = request_doc_host && request_doc_host != tab_host;
   if (misattribution) {
     tab_host = request_doc_host;
@@ -308,7 +308,7 @@ function onHeadersReceived(details) {
   // if we are no longer on the page the request is coming for,
   // don't log in popup or attempt to replace social widgets
   // but do block request/modify headers
-  const request_doc_host = getDocumentHostForRequest(details),
+  const request_doc_host = utils.getDocumentHostForRequest(details),
     misattribution = request_doc_host && request_doc_host != tab_host;
   if (misattribution) {
     tab_host = request_doc_host;
@@ -381,65 +381,6 @@ function isThirdPartyDomain(domain1, domain2) {
   return false;
 }
 
-/**
- * Gets the hostname for a given request's top-level document.
- *
- * The request's document may be different from the current top-level document
- * loaded in tab as requests can come out of order:
- *
- * - "main_frame" requests usually but not always mark a boundary
- *   (navigating to another site while the current page is still loading)
- * - sometimes there is no "main_frame" request
- *   (service worker pages in Firefox)
- *
- * @param {Object} details chrome.webRequest request details object
- *
- * @return {String} the hostname for the request's top-level document
- */
-function getDocumentHostForRequest(details) {
-  let host, url;
-
-  // Firefox 54+
-  if (details.hasOwnProperty("documentUrl")) {
-    if (details.type == "main_frame") {
-      // the top-level document itself
-      url = details.url;
-    } else if (details.hasOwnProperty("frameAncestors")) {
-      // Firefox 58+
-      if (details.frameAncestors.length) {
-        // inside a frame
-        url = details.frameAncestors[details.frameAncestors.length - 1].url;
-      } else {
-        // inside the top-level document
-        url = details.documentUrl;
-      }
-    } else {
-      // TODO Firefox 54-57 or a service worker request
-      if (details.documentUrl.endsWith("/sw.js")) {
-        url = details.documentUrl;
-      }
-    }
-
-  // Chrome 63+
-  } else if (details.hasOwnProperty("initiator")) {
-    if (details.initiator && details.initiator != "null") {
-      if (details.type == "main_frame") {
-        url = details.url;
-      } else if (details.parentFrameId == -1 || details.type == "sub_frame" && details.parentFrameId === 0) {
-        // TODO can only rely on initiator for main frame resources:
-        // https://crbug.com/838242#c17
-        url = details.initiator + '/';
-      }
-    }
-  }
-
-  if (url) {
-    host = window.extractHostFromURL(url);
-  }
-
-  return host;
-}
-
 /**
  * Gets the host name for a given tab id
  * @param {Integer} tabId chrome tab id