Migrate WebRequestRedirectByRegExAction to use RE2

BUG=112155
TBR=ben@chromium.org

Review URL: https://chromiumcodereview.appspot.com/10826120

git-svn-id: svn://svn.chromium.org/chrome/trunk/src@149852 0039d316-1c4b-4281-b951-d872f2087c98

chrome/browser/DEPS

@@ -65,6 +65,7 @@ include_rules = [
   "+third_party/libevent",  # For the remote V8 debugging server
   "+third_party/libjingle",
   "+third_party/protobuf/src/google/protobuf",
+  "+third_party/re2",
   "+third_party/sqlite",
   "+third_party/undoview",
 

chrome/browser/extensions/api/declarative_webrequest/webrequest_action.cc

@@ -10,7 +10,6 @@
 #include "base/logging.h"
 #include "base/stringprintf.h"
 #include "base/string_util.h"
-#include "base/utf_string_conversions.h"
 #include "base/values.h"
 #include "chrome/browser/extensions/api/declarative_webrequest/request_stages.h"
 #include "chrome/browser/extensions/api/declarative_webrequest/webrequest_constants.h"
@@ -19,6 +18,7 @@
 #include "chrome/browser/extensions/extension_info_map.h"
 #include "chrome/common/extensions/extension.h"
 #include "net/url_request/url_request.h"
+#include "third_party/re2/re2/re2.h"
 
 namespace extensions {
 
@@ -108,20 +108,18 @@ scoped_ptr<WebRequestAction> CreateRedirectRequestByRegExAction(
   INPUT_FORMAT_VALIDATE(dict->GetString(keys::kFromKey, &from));
   INPUT_FORMAT_VALIDATE(dict->GetString(keys::kToKey, &to));
 
-  // TODO(battre): Add this line once we migrate from ICU RegEx to RE2 RegEx.s
-  // to = WebRequestRedirectByRegExAction::PerlToRe2Style(to);
+  to = WebRequestRedirectByRegExAction::PerlToRe2Style(to);
 
-  UParseError parse_error;
-  UErrorCode status = U_ZERO_ERROR;
-  scoped_ptr<icu::RegexPattern> pattern(
-      icu::RegexPattern::compile(icu::UnicodeString(from.data(), from.size()),
-                                 0, parse_error, status));
-  if (U_FAILURE(status) || !pattern.get()) {
+  RE2::Options options;
+  options.set_case_sensitive(false);
+  scoped_ptr<RE2> from_pattern(new RE2(from, options));
+
+  if (!from_pattern->ok()) {
     *error = "Invalid pattern '" + from + "' -> '" + to + "'";
     return scoped_ptr<WebRequestAction>(NULL);
   }
   return scoped_ptr<WebRequestAction>(
-      new WebRequestRedirectByRegExAction(pattern.Pass(), to));
+      new WebRequestRedirectByRegExAction(from_pattern.Pass(), to));
 }
 
 scoped_ptr<WebRequestAction> CreateSetRequestHeaderAction(
@@ -590,7 +588,7 @@ WebRequestRedirectToEmptyDocumentAction::CreateDelta(
 //
 
 WebRequestRedirectByRegExAction::WebRequestRedirectByRegExAction(
-    scoped_ptr<icu::RegexPattern> from_pattern,
+    scoped_ptr<RE2> from_pattern,
     const std::string& to_pattern)
     : from_pattern_(from_pattern.Pass()),
       to_pattern_(to_pattern.data(), to_pattern.size()) {}
@@ -667,29 +665,17 @@ LinkedPtrEventResponseDelta WebRequestRedirectByRegExAction::CreateDelta(
   CHECK(request_stage & GetStages());
   CHECK(from_pattern_.get());
 
-  UErrorCode status = U_ZERO_ERROR;
   const std::string& old_url = request->url().spec();
-  icu::UnicodeString old_url_unicode(old_url.data(), old_url.size());
-
-  scoped_ptr<icu::RegexMatcher> matcher(
-      from_pattern_->matcher(old_url_unicode, status));
-  if (U_FAILURE(status) || !matcher.get())
-    return LinkedPtrEventResponseDelta(NULL);
-
-  icu::UnicodeString new_url = matcher->replaceAll(to_pattern_, status);
-  if (U_FAILURE(status))
-    return LinkedPtrEventResponseDelta(NULL);
-
-  std::string new_url_utf8;
-  UTF16ToUTF8(new_url.getBuffer(), new_url.length(), &new_url_utf8);
-
-  if (new_url_utf8 == request->url().spec())
+  std::string new_url = old_url;
+  if (!RE2::Replace(&new_url, *from_pattern_, to_pattern_) ||
+      new_url == old_url) {
     return LinkedPtrEventResponseDelta(NULL);
+  }
 
   LinkedPtrEventResponseDelta result(
       new extension_web_request_api_helpers::EventResponseDelta(
           extension_id, extension_install_time));
-  result->new_url = GURL(new_url_utf8);
+  result->new_url = GURL(new_url);
   return result;
 }
 

chrome/browser/extensions/api/declarative_webrequest/webrequest_action.h

@@ -16,7 +16,6 @@
 #include "chrome/browser/extensions/api/web_request/web_request_api_helpers.h"
 #include "chrome/common/extensions/api/events.h"
 #include "googleurl/src/gurl.h"
-#include "unicode/regex.h"
 
 class WebRequestPermission;
 
@@ -38,6 +37,10 @@ namespace net {
 class URLRequest;
 }
 
+namespace re2 {
+class RE2;
+}
+
 namespace extensions {
 
 typedef linked_ptr<extension_web_request_api_helpers::EventResponseDelta>
@@ -248,11 +251,10 @@ class WebRequestRedirectToEmptyDocumentAction : public WebRequestAction {
 // Action that instructs to redirect a network request.
 class WebRequestRedirectByRegExAction : public WebRequestAction {
  public:
-  // The |to_pattern| has to be passed in ICU syntax.
-  // TODO(battre): Change this to Perl style when migrated to RE2.
-  explicit WebRequestRedirectByRegExAction(
-      scoped_ptr<icu::RegexPattern> from_pattern,
-      const std::string& to_pattern);
+  // The |to_pattern| has to be passed in RE2 syntax with the exception that
+  // capture groups are referenced in Perl style ($1, $2, ...).
+  WebRequestRedirectByRegExAction(scoped_ptr<re2::RE2> from_pattern,
+                                  const std::string& to_pattern);
   virtual ~WebRequestRedirectByRegExAction();
 
   // Conversion of capture group styles between Perl style ($1, $2, ...) and
@@ -270,8 +272,8 @@ class WebRequestRedirectByRegExAction : public WebRequestAction {
       const base::Time& extension_install_time) const OVERRIDE;
 
  private:
-  scoped_ptr<icu::RegexPattern> from_pattern_;
-  icu::UnicodeString to_pattern_;
+  scoped_ptr<re2::RE2> from_pattern_;
+  std::string to_pattern_;
 
   DISALLOW_COPY_AND_ASSIGN(WebRequestRedirectByRegExAction);
 };

chrome/chrome_browser_extensions.gypi

@@ -38,6 +38,7 @@
         '../third_party/icu/icu.gyp:icuuc',
         '../third_party/leveldatabase/leveldatabase.gyp:leveldatabase',
         '../third_party/libusb/libusb.gyp:libusb',
+        '../third_party/re2/re2.gyp:re2',
         '../ui/base/strings/ui_strings.gyp:ui_strings',
         '../ui/ui.gyp:ui',
         '../ui/ui.gyp:ui_resources',