Bug 1877419 - Part 1: Added nested stripping of urls to copy without site tracking. r=pbz,necko-reviewers

Depends on D204914

Differential Revision: https://phabricator.services.mozilla.com/D202943

netwerk/base/nsURLHelper.cpp

@@ -1302,9 +1302,7 @@ void URLParams::ParseInput(const nsACString& aInput) {
   });
 }
 
-namespace {
-
-void SerializeString(const nsCString& aInput, nsAString& aValue) {
+void URLParams::SerializeString(const nsCString& aInput, nsAString& aValue) {
   const unsigned char* p = (const unsigned char*)aInput.get();
   const unsigned char* end = p + aInput.Length();
 
@@ -1325,8 +1323,6 @@ void SerializeString(const nsCString& aInput, nsAString& aValue) {
   }
 }
 
-}  // namespace
-
 void URLParams::Serialize(nsAString& aValue, bool aEncode) const {
   aValue.Truncate();
   bool first = true;

netwerk/base/nsURLHelper.h

@@ -308,6 +308,8 @@ class URLParams final {
    */
   void Serialize(nsAString& aValue, bool aEncode) const;
 
+  static void SerializeString(const nsCString& aInput, nsAString& aValue);
+
   void Get(const nsAString& aName, nsString& aRetval);
 
   void GetAll(const nsAString& aName, nsTArray<nsString>& aRetval);
@@ -338,6 +340,8 @@ class URLParams final {
 
   uint32_t Length() const { return mParams.Length(); }
 
+  static void DecodeString(const nsACString& aInput, nsAString& aOutput);
+
   const nsAString& GetKeyAtIndex(uint32_t aIndex) const {
     MOZ_ASSERT(aIndex < mParams.Length());
     return mParams[aIndex].mKey;
@@ -355,7 +359,6 @@ class URLParams final {
   void Sort();
 
  private:
-  static void DecodeString(const nsACString& aInput, nsAString& aOutput);
   static void ConvertString(const nsACString& aInput, nsAString& aOutput);
   static bool ParseNextInternal(const char*& aStart, const char* aEnd,
                                 bool aShouldDecode, nsAString* aOutputName,

toolkit/components/antitracking/URLQueryStringStripper.cpp

@@ -17,6 +17,7 @@
 #include "nsIURIMutator.h"
 #include "nsUnicharUtils.h"
 #include "nsURLHelper.h"
+#include "nsNetUtil.h"
 #include "mozilla/dom/StripOnShareRuleBinding.h"
 
 namespace {
@@ -84,50 +85,9 @@ URLQueryStringStripper::StripForCopyOrShare(nsIURI* aURI,
   NS_ENSURE_ARG_POINTER(strippedURI);
   int aStripCount = 0;
 
-  nsAutoCString query;
-  nsresult rv = aURI->GetQuery(query);
+  nsresult rv =
+      StripForCopyOrShareInternal(aURI, strippedURI, aStripCount, false);
   NS_ENSURE_SUCCESS(rv, rv);
-  // We don't need to do anything if there is no query string.
-  if (query.IsEmpty()) {
-    Telemetry::Accumulate(Telemetry::STRIP_ON_SHARE_PARAMS_REMOVED, 0);
-    return NS_OK;
-  }
-  nsAutoCString host;
-  rv = aURI->GetHost(host);
-  NS_ENSURE_SUCCESS(rv, rv);
-
-  URLParams params;
-
-  URLParams::Parse(query, true, [&](nsString&& name, nsString&& value) {
-    nsAutoString lowerCaseName;
-    ToLowerCase(name, lowerCaseName);
-    // Look through the global rules.
-    dom::StripRule globalRule;
-    bool keyExists = mStripOnShareMap.Get("*"_ns, &globalRule);
-    // There should always be a global rule.
-    MOZ_ASSERT(keyExists);
-    for (const auto& param : globalRule.mQueryParams) {
-      if (param == lowerCaseName) {
-        aStripCount++;
-        return true;
-      }
-    }
-
-    // Check for site specific rules.
-    dom::StripRule siteSpecificRule;
-    keyExists = mStripOnShareMap.Get(host, &siteSpecificRule);
-    if (keyExists) {
-      for (const auto& param : siteSpecificRule.mQueryParams) {
-        if (param == lowerCaseName) {
-          aStripCount++;
-          return true;
-        }
-      }
-    }
-
-    params.Append(name, value);
-    return true;
-  });
 
   Telemetry::Accumulate(Telemetry::STRIP_ON_SHARE_PARAMS_REMOVED, aStripCount);
 
@@ -135,13 +95,6 @@ URLQueryStringStripper::StripForCopyOrShare(nsIURI* aURI,
     return NS_OK;
   }
 
-  nsAutoString newQuery;
-  params.Serialize(newQuery, false);
-
-  Unused << NS_MutateURI(aURI)
-                .SetQuery(NS_ConvertUTF16toUTF8(newQuery))
-                .Finalize(strippedURI);
-
   // To calculate difference in length of the URL
   // after stripping occurs for Telemetry
   nsAutoCString specOriginalURI;
@@ -426,4 +379,103 @@ URLQueryStringStripper::Observe(nsISupports*, const char* aTopic,
   return NS_OK;
 }
 
+nsresult URLQueryStringStripper::StripForCopyOrShareInternal(
+    nsIURI* aURI, nsIURI** strippedURI, int& aStripCount,
+    bool aStripNestedURIs) {
+  nsAutoCString query;
+  nsresult rv = aURI->GetQuery(query);
+  NS_ENSURE_SUCCESS(rv, rv);
+
+  // We don't need to do anything if there is no query string.
+  if (query.IsEmpty()) {
+    Telemetry::Accumulate(Telemetry::STRIP_ON_SHARE_PARAMS_REMOVED, 0);
+    return NS_OK;
+  }
+
+  nsAutoCString host;
+  rv = aURI->GetHost(host);
+  NS_ENSURE_SUCCESS(rv, rv);
+
+  URLParams params;
+
+  URLParams::Parse(query, false, [&](nsString&& name, nsString&& value) {
+    nsAutoString lowerCaseName;
+    ToLowerCase(name, lowerCaseName);
+
+    // Look through the global rules.
+    dom::StripRule globalRule;
+    bool keyExists = mStripOnShareMap.Get("*"_ns, &globalRule);
+    // There should always be a global rule.
+    MOZ_ASSERT(keyExists);
+
+    // Look through the global rules.
+    for (const auto& param : globalRule.mQueryParams) {
+      if (param == lowerCaseName) {
+        aStripCount++;
+        return true;
+      }
+    }
+
+    // Check for site specific rules.
+    dom::StripRule siteSpecificRule;
+    keyExists = mStripOnShareMap.Get(host, &siteSpecificRule);
+    if (keyExists) {
+      for (const auto& param : siteSpecificRule.mQueryParams) {
+        if (param == lowerCaseName) {
+          aStripCount++;
+          return true;
+        }
+      }
+    }
+
+    // Only if it is top layer of the recursion then it
+    // checks if the value of the query parameter is a valid URI
+    // if not then it gets added back to the query, if it is then
+    // it gets passed back into this method but with the recursive
+    // stripping flag set to true
+    if (!aStripNestedURIs) {
+      nsAutoString decodeValue;
+      URLParams::DecodeString(NS_ConvertUTF16toUTF8(value), decodeValue);
+
+      nsCOMPtr<nsIURI> nestedURI;
+      rv = NS_NewURI(getter_AddRefs(nestedURI), decodeValue);
+
+      if (NS_WARN_IF(NS_FAILED(rv))) {
+        params.Append(name, value);
+        return true;
+      }
+
+      nsCOMPtr<nsIURI> strippedNestedURI;
+      rv = StripForCopyOrShareInternal(
+          nestedURI, getter_AddRefs(strippedNestedURI), aStripCount, true);
+      if (NS_WARN_IF(NS_FAILED(rv))) {
+        return false;
+      }
+
+      nsAutoCString nestedURIString;
+      rv = strippedNestedURI->GetSpec(nestedURIString);
+      if (NS_WARN_IF(NS_FAILED(rv))) {
+        return false;
+      }
+
+      // Encodes URI
+      nsAutoString encodedURI;
+      URLParams::SerializeString(nestedURIString, encodedURI);
+
+      params.Append(name, encodedURI);
+      return true;
+    }
+
+    params.Append(name, value);
+    return true;
+  });
+
+  nsAutoString newQuery;
+  params.Serialize(newQuery, false);
+
+  return NS_MutateURI(aURI)
+      .SetQuery(NS_ConvertUTF16toUTF8(newQuery))
+      .Finalize(strippedURI);
+}
+
 }  // namespace mozilla

toolkit/components/antitracking/URLQueryStringStripper.h

@@ -49,6 +49,12 @@ class URLQueryStringStripper final : public nsIObserver,
   void PopulateStripList(const nsAString& aList);
   void PopulateAllowList(const nsACString& aList);
 
+  // Recursive helper function that helps strip URIs of tracking parameters
+  // and enables the stripping of tracking paramerters that are in a URI which
+  // is nested in a query parameter
+  nsresult StripForCopyOrShareInternal(nsIURI* aURI, nsIURI** strippedURI,
+                                       int& aStripCount, bool aStripNestedURIs);
+
   nsTHashSet<nsString> mList;
   nsTHashSet<nsCString> mAllowList;
   nsCOMPtr<nsIURLQueryStrippingListService> mListService;