Parse metadata block in user scripts

This allows WebEngine to build up user script objects from the
metadata section of their source code. It also implements @include,
@exclude and @match rules for restricting script injection to given
URL patterns.

Change-Id: Ic7b3023c0143643bfbf71adbfa25a8022b223fcf
Reviewed-by: Allan Sandfeld Jensen <allan.jensen@qt.io>

6 files changed, 163 insertions, 4 deletions

diff --git a/src/core/chrome_qt.gyp b/src/core/chrome_qt.gyp
index 394e72733..ff7988f09 100644
--- a/src/core/chrome_qt.gyp
+++ b/src/core/chrome_qt.gyp
@@ -78,8 +78,16 @@
         '<(DEPTH)/chrome/browser/media/desktop_streams_registry.h',
         '<(DEPTH)/chrome/common/chrome_switches.cc',
         '<(DEPTH)/chrome/common/chrome_switches.h',
+        '<(DEPTH)/extensions/common/constants.cc',
+        '<(DEPTH)/extensions/common/constants.h',
+        '<(DEPTH)/extensions/common/url_pattern.cc',
+        '<(DEPTH)/extensions/common/url_pattern.h',
       ],
       'conditions': [
+        ['OS == "win"', {
+          # crbug.com/167187 fix size_t to int truncations
+          'msvs_disabled_warnings': [4267, ],
+        }],
         ['enable_spellcheck==1', {
           'sources': [ '<@(chrome_spellchecker_sources)' ],
           'include_dirs': [
@@ -99,10 +107,6 @@
               '__STDC_FORMAT_MACROS',
           ],
           'conditions': [
-            ['OS == "win"', {
-              # crbug.com/167187 fix size_t to int truncations
-              'msvs_disabled_warnings': [4267, ],
-            }],
             [ 'OS != "mac"', {
               'sources/': [
                 ['exclude', '_mac\\.(cc|cpp|mm?)$'],
diff --git a/src/core/common/qt_messages.h b/src/core/common/qt_messages.h
index b8991a2b3..2c971aab2 100644
--- a/src/core/common/qt_messages.h
+++ b/src/core/common/qt_messages.h
@@ -16,6 +16,9 @@ IPC_STRUCT_TRAITS_BEGIN(UserScriptData)
     IPC_STRUCT_TRAITS_MEMBER(injectForSubframes)
     IPC_STRUCT_TRAITS_MEMBER(worldId)
     IPC_STRUCT_TRAITS_MEMBER(scriptId)
+    IPC_STRUCT_TRAITS_MEMBER(globs)
+    IPC_STRUCT_TRAITS_MEMBER(excludeGlobs)
+    IPC_STRUCT_TRAITS_MEMBER(urlPatterns)
 IPC_STRUCT_TRAITS_END()
 
 
diff --git a/src/core/common/user_script_data.h b/src/core/common/user_script_data.h
index 943d61798..8d98890e3 100644
--- a/src/core/common/user_script_data.h
+++ b/src/core/common/user_script_data.h
@@ -60,6 +60,9 @@ struct UserScriptData {
     bool injectForSubframes;
     uint worldId;
     uint64_t scriptId;
+    std::vector<std::string> globs;
+    std::vector<std::string> excludeGlobs;
+    std::vector<std::string> urlPatterns;
 };
 
 QT_BEGIN_NAMESPACE
diff --git a/src/core/renderer/user_resource_controller.cpp b/src/core/renderer/user_resource_controller.cpp
index 68b5f3d40..8c603b805 100644
--- a/src/core/renderer/user_resource_controller.cpp
+++ b/src/core/renderer/user_resource_controller.cpp
@@ -39,9 +39,12 @@
 
 #include "user_resource_controller.h"
 
+#include "base/strings/pattern.h"
 #include "content/public/renderer/render_frame.h"
 #include "content/public/renderer/render_view.h"
 #include "content/public/renderer/render_view_observer.h"
+#include "extensions/common/url_pattern.h"
+#include "third_party/WebKit/public/web/WebDocument.h"
 #include "third_party/WebKit/public/web/WebLocalFrame.h"
 #include "third_party/WebKit/public/web/WebScriptSource.h"
 #include "third_party/WebKit/public/web/WebView.h"
@@ -49,6 +52,8 @@
 
 #include "common/qt_messages.h"
 #include "common/user_script_data.h"
+#include "type_conversion.h"
+#include "user_script.h"
 
 Q_GLOBAL_STATIC(UserResourceController, qt_webengine_userResourceController)
 
@@ -57,6 +62,40 @@ static content::RenderView * const globalScriptsIndex = 0;
 // Scripts meant to run after the load event will be run 500ms after DOMContentLoaded if the load event doesn't come within that delay.
 static const int afterLoadTimeout = 500;
 
+static bool scriptMatchesURL(const UserScriptData &scriptData, const GURL &url) {
+    // Logic taken from Chromium (extensions/common/user_script.cc)
+    bool matchFound;
+    if (!scriptData.urlPatterns.empty()) {
+        matchFound = false;
+        for (auto it = scriptData.urlPatterns.begin(), end = scriptData.urlPatterns.end(); it != end; ++it) {
+            URLPattern urlPattern(QtWebEngineCore::UserScript::validUserScriptSchemes(), *it);
+            if (urlPattern.MatchesURL(url))
+                matchFound = true;
+        }
+        if (!matchFound)
+            return false;
+    }
+
+    if (!scriptData.globs.empty()) {
+        matchFound = false;
+        for (auto it = scriptData.globs.begin(), end = scriptData.globs.end(); it != end; ++it) {
+            if (base::MatchPattern(url.spec(), *it))
+                matchFound = true;
+        }
+        if (!matchFound)
+            return false;
+    }
+
+    if (!scriptData.excludeGlobs.empty()) {
+        for (auto it = scriptData.excludeGlobs.begin(), end = scriptData.excludeGlobs.end(); it != end; ++it) {
+            if (base::MatchPattern(url.spec(), *it))
+                return false;
+        }
+    }
+
+    return true;
+}
+
 class UserResourceController::RenderViewObserverHelper : public content::RenderViewObserver
 {
 public:
@@ -99,6 +138,8 @@ void UserResourceController::runScripts(UserScriptData::InjectionPoint p, blink:
         if (script.injectionPoint != p
                 || (!script.injectForSubframes && !isMainFrame))
             continue;
+        if (!scriptMatchesURL(script, frame->document().url()))
+            continue;
         blink::WebScriptSource source(blink::WebString::fromUTF8(script.source), script.url);
         if (script.worldId)
             frame->executeScriptInIsolatedWorld(script.worldId, &source, /*numSources = */1, /*contentScriptExtentsionGroup = */ 0);
diff --git a/src/core/user_script.cpp b/src/core/user_script.cpp
index 839eff366..b33dd6a7d 100644
--- a/src/core/user_script.cpp
+++ b/src/core/user_script.cpp
@@ -38,11 +38,39 @@
 ****************************************************************************/
 
 #include "common/user_script_data.h"
+#include "extensions/common/url_pattern.h"
 #include "user_script.h"
 #include "type_conversion.h"
 
+namespace {
+
+// Helper function to parse Greasemonkey headers
+bool GetDeclarationValue(const base::StringPiece& line,
+                         const base::StringPiece& prefix,
+                         std::string* value) {
+    base::StringPiece::size_type index = line.find(prefix);
+    if (index == base::StringPiece::npos)
+        return false;
+
+    std::string temp(line.data() + index + prefix.length(),
+                     line.length() - index - prefix.length());
+
+    if (temp.empty() || !base::IsUnicodeWhitespace(temp[0]))
+        return false;
+
+    base::TrimWhitespaceASCII(temp, base::TRIM_ALL, value);
+        return true;
+}
+
+}  // namespace
+
 namespace QtWebEngineCore {
 
+int UserScript::validUserScriptSchemes()
+{
+    return URLPattern::SCHEME_HTTP | URLPattern::SCHEME_HTTPS | URLPattern::SCHEME_FILE;
+}
+
 ASSERT_ENUMS_MATCH(UserScript::AfterLoad, UserScriptData::AfterLoad)
 ASSERT_ENUMS_MATCH(UserScript::DocumentLoadFinished, UserScriptData::DocumentLoadFinished)
 ASSERT_ENUMS_MATCH(UserScript::DocumentElementCreation, UserScriptData::DocumentElementCreation)
@@ -100,6 +128,7 @@ void UserScript::setSourceCode(const QString &source)
 {
     initData();
     scriptData->source = source.toStdString();
+    parseMetadataHeader();
 }
 
 UserScript::InjectionPoint UserScript::injectionPoint() const
@@ -169,4 +198,80 @@ UserScriptData &UserScript::data() const
     return *(scriptData.data());
 }
 
+void UserScript::parseMetadataHeader()
+{
+    // Logic taken from Chromium (extensions/browser/user_script_loader.cc)
+    // http://wiki.greasespot.net/Metadata_block
+    const std::string &script_text = scriptData->source;
+    base::StringPiece line;
+    size_t line_start = 0;
+    size_t line_end = line_start;
+    bool in_metadata = false;
+
+    static const base::StringPiece kUserScriptBegin("// ==UserScript==");
+    static const base::StringPiece kUserScriptEnd("// ==/UserScript==");
+    static const base::StringPiece kNameDeclaration("// @name");
+    static const base::StringPiece kIncludeDeclaration("// @include");
+    static const base::StringPiece kExcludeDeclaration("// @exclude");
+    static const base::StringPiece kMatchDeclaration("// @match");
+    static const base::StringPiece kRunAtDeclaration("// @run-at");
+    static const base::StringPiece kRunAtDocumentStartValue("document-start");
+    static const base::StringPiece kRunAtDocumentEndValue("document-end");
+    static const base::StringPiece kRunAtDocumentIdleValue("document-idle");
+    // FIXME: Scripts don't run in subframes by default. If we would like to
+    // support @noframes rule, we have to change the current default behavior.
+    // static const base::StringPiece kNoFramesDeclaration("// @noframes");
+
+    static URLPattern urlPatternParser(validUserScriptSchemes());
+
+    while (line_start < script_text.length()) {
+        line_end = script_text.find('\n', line_start);
+
+        // Handle the case where there is no trailing newline in the file.
+        if (line_end == std::string::npos)
+            line_end = script_text.length() - 1;
+
+        line.set(script_text.data() + line_start, line_end - line_start);
+
+        if (!in_metadata) {
+            if (line.starts_with(kUserScriptBegin))
+                in_metadata = true;
+        } else {
+            if (line.starts_with(kUserScriptEnd))
+                break;
+
+            std::string value;
+            if (GetDeclarationValue(line, kNameDeclaration, &value)) {
+                setName(toQt(value));
+            } else if (GetDeclarationValue(line, kIncludeDeclaration, &value)) {
+                // We escape some characters that MatchPattern() considers special.
+                base::ReplaceSubstringsAfterOffset(&value, 0, "\\", "\\\\");
+                base::ReplaceSubstringsAfterOffset(&value, 0, "?", "\\?");
+                scriptData->globs.push_back(value);
+            } else if (GetDeclarationValue(line, kExcludeDeclaration, &value)) {
+                base::ReplaceSubstringsAfterOffset(&value, 0, "\\", "\\\\");
+                base::ReplaceSubstringsAfterOffset(&value, 0, "?", "\\?");
+                scriptData->excludeGlobs.push_back(value);
+            } else if (GetDeclarationValue(line, kMatchDeclaration, &value)) {
+                if (URLPattern::PARSE_SUCCESS == urlPatternParser.Parse(value))
+                    scriptData->urlPatterns.push_back(value);
+            } else if (GetDeclarationValue(line, kRunAtDeclaration, &value)) {
+                if (value == kRunAtDocumentStartValue)
+                    scriptData->injectionPoint = DocumentElementCreation;
+                else if (value == kRunAtDocumentEndValue)
+                    scriptData->injectionPoint = DocumentLoadFinished;
+                else if (value == kRunAtDocumentIdleValue)
+                    scriptData->injectionPoint = AfterLoad;
+            }
+        }
+
+        line_start = line_end + 1;
+    }
+
+    // If no patterns were specified, default to @include *. This is what
+    // Greasemonkey does.
+    if (scriptData->globs.empty() && scriptData->urlPatterns.empty())
+        scriptData->globs.push_back("*");
+}
+
 } // namespace QtWebEngineCore
diff --git a/src/core/user_script.h b/src/core/user_script.h
index 9d7d66a58..e44efd3e9 100644
--- a/src/core/user_script.h
+++ b/src/core/user_script.h
@@ -85,9 +85,12 @@ public:
 
     bool operator==(const UserScript &) const;
 
+    static int validUserScriptSchemes();
+
 private:
     void initData();
     UserScriptData &data() const;
+    void parseMetadataHeader();
     friend class UserResourceControllerHost;
 
     QScopedPointer<UserScriptData> scriptData;