[Extensions] Add support for matching data URLs in manifest parsing

Add support for a "match_data_urls" field in content scripts specified
in an extension's manifest. This CL adds the parsing logic and related
unittests, behind a base::Feature. Future CLs will add support for
the actual injection of scripts into these data URLs, as well as
expanding the support to include dynamic script injections.

Bug: 55084

Change-Id: I288c66dfb901e9cb77d09562cf943a223adcb8d7
Reviewed-on: https://chromium-review.googlesource.com/c/chromium/src/+/2292697Reviewed-by: Karan Bhatia <karandeepb@chromium.org>
Commit-Queue: Devlin <rdevlin.cronin@chromium.org>
Cr-Commit-Position: refs/heads/master@{#791068}

chrome/common/extensions/manifest_handlers/content_scripts_manifest_unittest.cc

@@ -7,11 +7,13 @@
 #include "base/path_service.h"
 #include "base/stl_util.h"
 #include "base/strings/string_number_conversions.h"
+#include "base/test/scoped_feature_list.h"
 #include "chrome/common/chrome_paths.h"
 #include "chrome/common/extensions/manifest_tests/chrome_manifest_test.h"
 #include "chrome/common/webui_url_constants.h"
 #include "extensions/common/error_utils.h"
 #include "extensions/common/extension.h"
+#include "extensions/common/extension_features.h"
 #include "extensions/common/file_util.h"
 #include "extensions/common/manifest_constants.h"
 #include "extensions/common/manifest_handlers/content_scripts_handler.h"
@@ -102,4 +104,43 @@ TEST_F(ContentScriptsManifestTest, FailLoadingNonUTF8Scripts) {
       error.c_str());
 }
 
+TEST_F(ContentScriptsManifestTest, MatchDataURLs_FeatureEnabled) {
+  base::test::ScopedFeatureList scoped_feature_list;
+  scoped_feature_list.InitAndEnableFeature(
+      extensions_features::kContentScriptsOnDataUrls);
+
+  scoped_refptr<const Extension> extension =
+      LoadAndExpectSuccess("content_script_match_data_urls.json");
+  ASSERT_TRUE(extension);
+  const UserScriptList& user_scripts =
+      ContentScriptsInfo::GetContentScripts(extension.get());
+  ASSERT_EQ(3u, user_scripts.size());
+
+  // The first script specifies `"match_data_urls": true`.
+  EXPECT_TRUE(user_scripts[0]->match_data_urls());
+  // The second specifies `"match_data_urls": false`.
+  EXPECT_FALSE(user_scripts[1]->match_data_urls());
+  // The third doesn't specify a value for "match_data_urls"; it should
+  // default to false.
+  EXPECT_FALSE(user_scripts[2]->match_data_urls());
+}
+
+TEST_F(ContentScriptsManifestTest, MatchDataURLs_FeatureDisabled) {
+  base::test::ScopedFeatureList scoped_feature_list;
+  scoped_feature_list.InitAndDisableFeature(
+      extensions_features::kContentScriptsOnDataUrls);
+
+  scoped_refptr<const Extension> extension =
+      LoadAndExpectSuccess("content_script_match_data_urls.json");
+  ASSERT_TRUE(extension);
+  const UserScriptList& user_scripts =
+      ContentScriptsInfo::GetContentScripts(extension.get());
+  ASSERT_EQ(3u, user_scripts.size());
+
+  // Without the feature enabled, match_data_urls should always be false.
+  EXPECT_FALSE(user_scripts[0]->match_data_urls());
+  EXPECT_FALSE(user_scripts[1]->match_data_urls());
+  EXPECT_FALSE(user_scripts[2]->match_data_urls());
+}
+
 }  // namespace extensions

chrome/test/data/extensions/manifest_tests/content_script_match_data_urls.json

+{
+  "name": "Content scripts matching data urls",
+  "description": "ContentScriptsManifestTest.MatchDataURLs*",
+  "version": "0.1",
+  "manifest_version": 2,
+  "content_scripts": [{
+    "matches": ["https://example.com/*"],
+    "match_data_urls": true,
+    "js": ["file.js"]
+  }, {
+    "matches": ["https://foo.example/*"],
+    "match_data_urls": false,
+    "js": ["file.js"]
+  }, {
+    "matches": ["https://bar.example/*"],
+    "js": ["file.js"]
+  }]
+}

extensions/common/extension_features.cc

@@ -43,6 +43,10 @@ const base::Feature kAllowWithholdingExtensionPermissionsOnInstall{
     "AllowWithholdingExtensionPermissionsOnInstall",
     base::FEATURE_DISABLED_BY_DEFAULT};
 
+// Enables content scripts on data URLs.
+const base::Feature kContentScriptsOnDataUrls{
+    "ContentScriptsOnDataUrls", base::FEATURE_DISABLED_BY_DEFAULT};
+
 // Reports Extensions.WebRequest.KeepaliveRequestFinished when enabled.
 const base::Feature kReportKeepaliveUkm{"ReportKeepaliveUkm",
                                         base::FEATURE_ENABLED_BY_DEFAULT};

extensions/common/extension_features.h

@@ -25,6 +25,8 @@ extern const base::Feature kForceWebRequestProxyForTest;
 
 extern const base::Feature kAllowWithholdingExtensionPermissionsOnInstall;
 
+extern const base::Feature kContentScriptsOnDataUrls;
+
 extern const base::Feature kReportKeepaliveUkm;
 
 }  // namespace extensions_features

extensions/common/manifest_constants.cc

@@ -105,6 +105,7 @@ const char kLinkedAppIconURL[] = "url";
 const char kLinkedAppIconSize[] = "size";
 const char kManifestVersion[] = "manifest_version";
 const char kMatchAboutBlank[] = "match_about_blank";
+const char kMatchDataUrls[] = "match_data_urls";
 const char kMatches[] = "matches";
 const char kMinimumChromeVersion[] = "minimum_chrome_version";
 const char kMinimumVersion[] = "minimum_version";
@@ -568,6 +569,8 @@ const char kInvalidMatch[] =
     "Invalid value for 'content_scripts[*].matches[*]': *";
 const char kInvalidMatchAboutBlank[] =
     "Invalid value for 'content_scripts[*].match_about_blank'.";
+const char kInvalidMatchDataUrls[] =
+    "Invalid value for 'content_scripts[*].match_data_urls'.";
 const char kInvalidMatchCount[] =
     "Invalid value for 'content_scripts[*].matches'. There must be at least "
     "one match specified.";

extensions/common/manifest_constants.h

@@ -108,6 +108,7 @@ extern const char kLinkedAppIconURL[];
 extern const char kLinkedAppIconSize[];
 extern const char kManifestVersion[];
 extern const char kMatchAboutBlank[];
+extern const char kMatchDataUrls[];
 extern const char kMatches[];
 extern const char kMIMETypes[];
 extern const char kMimeTypesHandler[];
@@ -418,6 +419,7 @@ extern const char kInvalidManifestVersion[];
 extern const char kInvalidManifestVersionOld[];
 extern const char kInvalidMatch[];
 extern const char kInvalidMatchAboutBlank[];
+extern const char kInvalidMatchDataUrls[];
 extern const char kInvalidMatchCount[];
 extern const char kInvalidMatches[];
 extern const char kInvalidMIMETypes[];

extensions/common/manifest_handlers/content_scripts_handler.cc

@@ -18,6 +18,7 @@
 #include "content/public/common/url_constants.h"
 #include "extensions/common/error_utils.h"
 #include "extensions/common/extension.h"
+#include "extensions/common/extension_features.h"
 #include "extensions/common/extension_resource.h"
 #include "extensions/common/host_id.h"
 #include "extensions/common/manifest_constants.h"
@@ -126,6 +127,22 @@ std::unique_ptr<UserScript> LoadUserScriptFromDictionary(
     result->set_match_about_blank(match_about_blank->GetBool());
   }
 
+  // match data urls
+  if (base::FeatureList::IsEnabled(
+          extensions_features::kContentScriptsOnDataUrls)) {
+    const base::Value* match_data_urls =
+        content_script.FindKey(keys::kMatchDataUrls);
+    if (match_data_urls) {
+      if (!match_data_urls->is_bool()) {
+        *error = ErrorUtils::FormatErrorMessageUTF16(
+            errors::kInvalidMatchDataUrls,
+            base::NumberToString(definition_index));
+        return nullptr;
+      }
+      result->set_match_data_urls(match_data_urls->GetBool());
+    }
+  }
+
   // matches (required)
   const base::Value* matches =
       content_script.FindKeyOfType(keys::kMatches, base::Value::Type::LIST);

extensions/common/user_script.cc

@@ -98,6 +98,7 @@ UserScript::UserScript()
       emulate_greasemonkey_(false),
       match_all_frames_(false),
       match_about_blank_(false),
+      match_data_urls_(false),
       incognito_enabled_(false) {}
 
 UserScript::~UserScript() {
@@ -132,6 +133,7 @@ std::unique_ptr<UserScript> UserScript::CopyMetadataFrom(
   script->emulate_greasemonkey_ = other.emulate_greasemonkey_;
   script->match_all_frames_ = other.match_all_frames_;
   script->match_about_blank_ = other.match_about_blank_;
+  script->match_data_urls_ = other.match_data_urls_;
   script->incognito_enabled_ = other.incognito_enabled_;
 
   return script;
@@ -198,6 +200,7 @@ void UserScript::Pickle(base::Pickle* pickle) const {
   pickle->WriteBool(emulate_greasemonkey());
   pickle->WriteBool(match_all_frames());
   pickle->WriteBool(match_about_blank());
+  pickle->WriteBool(match_data_urls());
   pickle->WriteBool(is_incognito_enabled());
 
   PickleHostID(pickle, host_id_);
@@ -253,6 +256,7 @@ void UserScript::Unpickle(const base::Pickle& pickle,
   CHECK(iter->ReadBool(&emulate_greasemonkey_));
   CHECK(iter->ReadBool(&match_all_frames_));
   CHECK(iter->ReadBool(&match_about_blank_));
+  CHECK(iter->ReadBool(&match_data_urls_));
   CHECK(iter->ReadBool(&incognito_enabled_));
 
   UnpickleHostID(pickle, iter, &host_id_);

extensions/common/user_script.h

@@ -174,6 +174,10 @@ class UserScript {
   bool match_about_blank() const { return match_about_blank_; }
   void set_match_about_blank(bool val) { match_about_blank_ = val; }
 
+  // Whether to match data:-scheme URLs.
+  bool match_data_urls() const { return match_data_urls_; }
+  void set_match_data_urls(bool val) { match_data_urls_ = val; }
+
   // The globs, if any, that determine which pages this script runs against.
   // These are only used with "standalone" Greasemonkey-like user scripts.
   const std::vector<std::string>& globs() const { return globs_; }
@@ -231,9 +235,9 @@ class UserScript {
   bool MatchesURL(const GURL& url) const;
 
   // Returns true if the script should be applied to the given
-  // |effective_document_url| (calculated by the caller based on
-  // match_about_blank()| while also taking into account whether the document's
-  // frame |is_subframe| and what the |top_level_origin| is.
+  // |effective_document_url|. It is the caller's responsibility to calculate
+  // |effective_document_url| based on match_about_blank() and
+  // match_data_urls().
   bool MatchesDocument(const GURL& effective_document_url,
                        bool is_subframe) const;
 
@@ -326,6 +330,10 @@ class UserScript {
   // Defaults to false.
   bool match_about_blank_;
 
+  // Whether the user script should run in data:-scheme frames.
+  // Defaults to false.
+  bool match_data_urls_;
+
   // True if the script should be injected into an incognito tab.
   bool incognito_enabled_;