Move parsing of metadata header into browser process. This is a prerequisite

to getting user scripts working in extensions because extensions won't express
their metadata using the UserScript header, so parsing can't be done in the
renderer.

Review URL: http://codereview.chromium.org/18308

git-svn-id: svn://svn.chromium.org/chrome/trunk/src@8249 0039d316-1c4b-4281-b951-d872f2087c98

3 files changed, 40 insertions, 129 deletions

diff --git a/chrome/renderer/user_script_slave.cc b/chrome/renderer/user_script_slave.cc
index 52d9529..d227d5e 100644
--- a/chrome/renderer/user_script_slave.cc
+++ b/chrome/renderer/user_script_slave.cc
@@ -18,16 +18,6 @@ static const char kUserScriptTail[] = "\n})(window);";
 
 // UserScript
 
-void UserScript::Parse(const StringPiece& script_text) {
-  ParseMetadata(script_text);
-
-  // TODO(aa): Set body to just the part after the metadata block? This would
-  // significantly cut down on the size of the injected script in some cases.
-  // Would require remembering the line number the body begins at, for correct
-  // error line number reporting.
-  body_ = script_text;
-}
-
 bool UserScript::MatchesUrl(const GURL& url) {
   for (std::vector<std::string>::iterator pattern = include_patterns_.begin();
        pattern != include_patterns_.end(); ++pattern) {
@@ -39,57 +29,6 @@ bool UserScript::MatchesUrl(const GURL& url) {
   return false;
 }
 
-void UserScript::ParseMetadata(const StringPiece& script_text) {
-  // http://wiki.greasespot.net/Metadata_block
-  StringPiece line;
-  size_t line_start = 0;
-  size_t line_end = 0;
-  bool in_metadata = false;
-
-  static const StringPiece kUserScriptBegin("// ==UserScript==");
-  static const StringPiece kUserScriptEng("// ==/UserScript==");
-  static const StringPiece kIncludeDeclaration("// @include ");
-
-  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(kUserScriptEng)) {
-        break;
-      }
-
-      if (line.starts_with(kIncludeDeclaration)) {
-        std::string pattern(line.data() + kIncludeDeclaration.length(),
-                            line.length() - kIncludeDeclaration.length());
-        std::string pattern_trimmed;
-        TrimWhitespace(pattern, TRIM_ALL, &pattern_trimmed);
-        AddInclude(pattern_trimmed);
-      }
-
-      // TODO(aa): Handle more types of metadata.
-    }
-
-    line_start = line_end + 1;
-  }
-
-  // If no @include patterns were specified, default to @include *.
-  // This is what Greasemonkey does.
-  if (include_patterns_.size() == 0) {
-    AddInclude("*");
-  }
-}
-
 void UserScript::AddInclude(const std::string &glob_pattern) {
   include_patterns_.push_back(EscapeGlob(glob_pattern));
 }
@@ -160,12 +99,12 @@ bool UserScriptSlave::UpdateScripts(base::SharedMemoryHandle shared_memory) {
 
   // Unpickle scripts.
   void* iter = NULL;
-  int num_scripts = 0;
+  size_t num_scripts = 0;
   Pickle pickle(reinterpret_cast<char*>(shared_memory_->memory()),
                 pickle_size);
-  pickle.ReadInt(&iter, &num_scripts);
+  pickle.ReadSize(&iter, &num_scripts);
 
-  for (int i = 0; i < num_scripts; ++i) {
+  for (size_t i = 0; i < num_scripts; ++i) {
     const char* url = NULL;
     int url_length = 0;
     const char* body = NULL;
@@ -174,9 +113,17 @@ bool UserScriptSlave::UpdateScripts(base::SharedMemoryHandle shared_memory) {
     pickle.ReadData(&iter, &url, &url_length);
     pickle.ReadData(&iter, &body, &body_length);
 
-    scripts_.push_back(UserScript(StringPiece(url, url_length)));
+    scripts_.push_back(UserScript(StringPiece(url, url_length),
+                                  StringPiece(body, body_length)));
     UserScript& script = scripts_.back();
-    script.Parse(StringPiece(body, body_length));
+
+    size_t num_includes;
+    pickle.ReadSize(&iter, &num_includes);
+    for (size_t j = 0; j < num_includes; ++j) {
+      std::string include;
+      pickle.ReadString(&iter, &include);
+      script.AddInclude(include);
+    }
   }
 
   return true;
diff --git a/chrome/renderer/user_script_slave.h b/chrome/renderer/user_script_slave.h
index 21b67dc..b4558a6 100644
--- a/chrome/renderer/user_script_slave.h
+++ b/chrome/renderer/user_script_slave.h
@@ -16,21 +16,23 @@
 // Parsed representation of a user script.
 class UserScript {
  public:
-  UserScript(const StringPiece& script_url)
-    : url_(script_url) {}
-
-  // Gets the script body that should be injected into matching content.
-  const StringPiece& GetBody() const {
-    return body_;
-  }
+  UserScript(const StringPiece& script_url,
+             const StringPiece& body)
+    : url_(script_url), body_(body) {}
 
   // Gets a URL where this script can be found.
   const StringPiece& GetURL() const {
     return url_;
   }
 
-  // Parses the text content of a user script file.
-  void Parse(const StringPiece& script_text);
+  // Gets the script body that should be injected into matching content.
+  const StringPiece& GetBody() const {
+    return body_;
+  }
+
+  // Adds an include pattern that will be checked to determine whether to
+  // include a script on a given page.
+  void AddInclude(const std::string &glob_pattern);
 
   // Returns true if the script should be applied to the specified URL, false
   // otherwise.
@@ -38,30 +40,20 @@ class UserScript {
 
  private:
   FRIEND_TEST(UserScriptSlaveTest, EscapeGlob);
-  FRIEND_TEST(UserScriptSlaveTest, Parse1);
-  FRIEND_TEST(UserScriptSlaveTest, Parse2);
-  FRIEND_TEST(UserScriptSlaveTest, Parse3);
 
   // Helper function to convert the user script glob format to the patterns
   // used internally to test URLs.
   static std::string EscapeGlob(const std::string& glob);
 
-  // Parses the metadata block from the script.
-  void ParseMetadata(const StringPiece& script_text);
-
-  // Adds an include pattern that will be checked to determine whether to
-  // include a script on a given page.
-  void AddInclude(const std::string &glob_pattern);
+  // The url of the file the script came from. This references shared_memory_,
+  // and is valid until that memory is either deleted or Unmap()'d.
+  StringPiece url_;
 
   // The body of the script, which will be injected into content pages. This
   // references shared_memory_, and is valid until that memory is either
   // deleted or Unmap()'d.
   StringPiece body_;
 
-  // The url of the file the script came from. This references shared_memory_,
-  // and is valid until that memory is either deleted or Unmap()'d.
-  StringPiece url_;
-
   // List of patterns to test URLs against for this script. These patterns have
   // been escaped for use with MatchPattern() in string_utils ('?' and '\' are
   // escaped).
diff --git a/chrome/renderer/user_script_slave_unittest.cc b/chrome/renderer/user_script_slave_unittest.cc
index 57d34a9..eeea749 100644
--- a/chrome/renderer/user_script_slave_unittest.cc
+++ b/chrome/renderer/user_script_slave_unittest.cc
@@ -16,28 +16,11 @@ TEST(UserScriptSlaveTest, EscapeGlob) {
             UserScript::EscapeGlob("foo\\bar?hot=dog"));

 }

 

-TEST(UserScriptSlaveTest, Parse1) {

-  const std::string text(

-    "// This is my awesome script\n"

-    "// It does stuff.\n"

-    "// ==UserScript==   trailing garbage\n"

-    "// @name foobar script\n"

-    "// @namespace http://www.google.com/\n"

-    "// @include *mail.google.com*\n"

-    "// \n"

-    "// @othergarbage\n"

-    "// @include *mail.yahoo.com*\r\n"

-    "// @include  \t *mail.msn.com*\n" // extra spaces after "@include" OK

-    "//@include not-recognized\n" // must have one space after "//"

-    "// ==/UserScript==  trailing garbage\n"

-    "\n"

-    "\n"

-    "alert('hoo!');\n");

-

-  UserScript script("foo");

-  script.Parse(text);

-  EXPECT_EQ(3U, script.include_patterns_.size());

-  EXPECT_EQ(text, script.GetBody());

+TEST(UserScriptSlaveTest, Match1) {

+  UserScript script("foo", "bar");

+  script.AddInclude("*mail.google.com*");

+  script.AddInclude("*mail.yahoo.com*");

+  script.AddInclude("*mail.msn.com*");

   EXPECT_TRUE(script.MatchesUrl(GURL("http://mail.google.com")));

   EXPECT_TRUE(script.MatchesUrl(GURL("http://mail.google.com/foo")));

   EXPECT_TRUE(script.MatchesUrl(GURL("http://mail.yahoo.com/bar")));

@@ -45,27 +28,16 @@ TEST(UserScriptSlaveTest, Parse1) {
   EXPECT_FALSE(script.MatchesUrl(GURL("http://www.hotmail.com")));

 }

 

-TEST(UserScriptSlaveTest, Parse2) {

-  const std::string text("default to @include *");

-

-  UserScript script("foo");

-  script.Parse(text);

-  EXPECT_EQ(1U, script.include_patterns_.size());

-  EXPECT_EQ(text, script.GetBody());

-  EXPECT_TRUE(script.MatchesUrl(GURL("foo")));

-  EXPECT_TRUE(script.MatchesUrl(GURL("bar")));

+TEST(UserScriptSlaveTest, Match2) {

+  UserScript script("foo", "bar");

+  script.AddInclude("*");

+  EXPECT_TRUE(script.MatchesUrl(GURL("http://foo.com/bar")));

+  EXPECT_TRUE(script.MatchesUrl(GURL("http://hot.com/dog")));

 }

 

-TEST(UserScriptSlaveTest, Parse3) {

-  const std::string text(

-    "// ==UserScript==\n"

-    "// @include *foo*\n"

-    "// ==/UserScript=="); // no trailing newline

-

-  UserScript script("foo");

-  script.Parse(text);

-  EXPECT_EQ(1U, script.include_patterns_.size());

-  EXPECT_EQ(text, script.GetBody());

+TEST(UserScriptSlaveTest, Match3) {

+  UserScript script("foo", "bar");

+  script.AddInclude("*foo*");

   EXPECT_TRUE(script.MatchesUrl(GURL("http://foo.com/bar")));

   EXPECT_FALSE(script.MatchesUrl(GURL("http://baz.org")));

 }