[Android] Introduce email address detection.

For more context see: https://chromiumcodereview.appspot.com/10187020/

BUG=125390
TEST=email_detector_unittest.cc

Review URL: https://chromiumcodereview.appspot.com/10443015

git-svn-id: svn://svn.chromium.org/chrome/trunk/src@138887 0039d316-1c4b-4281-b951-d872f2087c98

5 files changed, 166 insertions, 0 deletions

diff --git a/content/content_renderer.gypi b/content/content_renderer.gypi
index 0d67650..fd8eca1 100644
--- a/content/content_renderer.gypi
+++ b/content/content_renderer.gypi
@@ -44,6 +44,8 @@
     'renderer/android/address_detector.h',
     'renderer/android/content_detector.cc',
     'renderer/android/content_detector.h',
+    'renderer/android/email_detector.cc',
+    'renderer/android/email_detector.h',
     'renderer/active_notification_tracker.cc',
     'renderer/active_notification_tracker.h',
     'renderer/device_orientation_dispatcher.cc',
diff --git a/content/content_tests.gypi b/content/content_tests.gypi
index 4758bc8..e449b26 100644
--- a/content/content_tests.gypi
+++ b/content/content_tests.gypi
@@ -285,6 +285,7 @@
         'gpu/gpu_info_collector_unittest_win.cc',
         'renderer/active_notification_tracker_unittest.cc',
         'renderer/android/address_detector_unittest.cc',
+        'renderer/android/email_detector_unittest.cc',
         'renderer/gpu/input_event_filter_unittest.cc',
         'renderer/media/audio_message_filter_unittest.cc',
         'renderer/media/capture_video_decoder_unittest.cc',
diff --git a/content/renderer/android/email_detector.cc b/content/renderer/android/email_detector.cc
new file mode 100644
index 0000000..1178492
--- /dev/null
+++ b/content/renderer/android/email_detector.cc
@@ -0,0 +1,74 @@
+// Copyright (c) 2012 The Chromium Authors. All rights reserved.
+// Use of this source code is governed by a BSD-style license that can be
+// found in the LICENSE file.
+
+#include "content/renderer/android/email_detector.h"
+
+#include "base/memory/scoped_ptr.h"
+#include "base/utf_string_conversions.h"
+#include "net/base/escape.h"
+#include "unicode/regex.h"
+
+namespace {
+
+// Maximum length of an email address.
+const size_t kMaximumEmailLength = 254;
+
+// Prefix used for email intent URIs.
+const char kEmailSchemaPrefix[] = "mailto:";
+
+// Regex to match email addresses.
+// This is more specific than RFC 2822 (uncommon special characters are
+// disallowed) in order to avoid false positives.
+// Delimiters are word boundaries to allow punctuation, quote marks etc. around
+// the address.
+const char kEmailRegex[] = "\\b[A-Z0-9._%+-]+@[A-Z0-9.-]+\\.[A-Z]{2,6}\\b";
+
+}  // anonymous namespace
+
+namespace content {
+
+EmailDetector::EmailDetector() {
+}
+
+size_t EmailDetector::GetMaximumContentLength() {
+  return kMaximumEmailLength;
+}
+
+GURL EmailDetector::GetIntentURL(const std::string& content_text) {
+  if (content_text.empty())
+    return GURL();
+
+  return GURL(kEmailSchemaPrefix +
+      net::EscapeQueryParamValue(content_text, true));
+}
+
+bool EmailDetector::FindContent(const string16::const_iterator& begin,
+                                const string16::const_iterator& end,
+                                size_t* start_pos,
+                                size_t* end_pos,
+                                std::string* content_text) {
+  string16 utf16_input = string16(begin, end);
+  icu::UnicodeString pattern(kEmailRegex);
+  icu::UnicodeString input(utf16_input.data(), utf16_input.length());
+  UErrorCode status = U_ZERO_ERROR;
+  scoped_ptr<icu::RegexMatcher> matcher(
+      new icu::RegexMatcher(pattern,
+                            input,
+                            UREGEX_CASE_INSENSITIVE,
+                            status));
+  if (matcher->find()) {
+    *start_pos = matcher->start(status);
+    DCHECK(U_SUCCESS(status));
+    *end_pos = matcher->end(status);
+    DCHECK(U_SUCCESS(status));
+    icu::UnicodeString content_ustr(matcher->group(status));
+    DCHECK(U_SUCCESS(status));
+    UTF16ToUTF8(content_ustr.getBuffer(), content_ustr.length(), content_text);
+    return true;
+  }
+
+  return false;
+}
+
+}  // namespace content
diff --git a/content/renderer/android/email_detector.h b/content/renderer/android/email_detector.h
new file mode 100644
index 0000000..cda43c6
--- /dev/null
+++ b/content/renderer/android/email_detector.h
@@ -0,0 +1,38 @@
+// Copyright (c) 2012 The Chromium Authors. All rights reserved.
+// Use of this source code is governed by a BSD-style license that can be
+// found in the LICENSE file.
+
+#ifndef CONTENT_RENDERER_ANDROID_EMAIL_DETECTOR_H_
+#define CONTENT_RENDERER_ANDROID_EMAIL_DETECTOR_H_
+#pragma once
+
+#include "content/renderer/android/content_detector.h"
+
+class EmailDetectorTest;
+
+namespace content {
+
+// Finds email addresses (in most common formats, but not including special
+// characters) in the given text string.
+class EmailDetector : public ContentDetector {
+ public:
+  EmailDetector();
+
+ private:
+  friend class ::EmailDetectorTest;
+
+  // Implementation of ContentDetector.
+  virtual bool FindContent(const string16::const_iterator& begin,
+                           const string16::const_iterator& end,
+                           size_t* start_pos,
+                           size_t* end_pos,
+                           std::string* content_text) OVERRIDE;
+  virtual GURL GetIntentURL(const std::string& content_text) OVERRIDE;
+  virtual size_t GetMaximumContentLength() OVERRIDE;
+
+  DISALLOW_COPY_AND_ASSIGN(EmailDetector);
+};
+
+}  // namespace content
+
+#endif  // CONTENT_RENDERER_ANDROID_EMAIL_DETECTOR_H_
diff --git a/content/renderer/android/email_detector_unittest.cc b/content/renderer/android/email_detector_unittest.cc
new file mode 100644
index 0000000..a68184e
--- /dev/null
+++ b/content/renderer/android/email_detector_unittest.cc
@@ -0,0 +1,51 @@
+// Copyright (c) 2012 The Chromium Authors. All rights reserved.
+// Use of this source code is governed by a BSD-style license that can be
+// found in the LICENSE file.
+
+#include "content/renderer/android/email_detector.h"
+
+#include "base/utf_string_conversions.h"
+#include "testing/gtest/include/gtest/gtest.h"
+
+using content::EmailDetector;
+
+class EmailDetectorTest : public testing::Test {
+ public:
+  static void FindAndCheckEmail(const std::string& content,
+                                const std::string& expected) {
+    string16 content_16 = UTF8ToUTF16(content);
+    string16 result_16;
+    size_t start, end;
+    EmailDetector detector;
+    std::string content_text;
+    if (detector.FindContent(content_16.begin(), content_16.end(),
+                             &start, &end, &content_text)) {
+      result_16 = content_16.substr(start, end - start);
+    }
+    EXPECT_EQ(expected, UTF16ToUTF8(result_16));
+    EXPECT_EQ(expected, content_text);
+  }
+};
+
+TEST_F(EmailDetectorTest, FindEmail) {
+  FindAndCheckEmail("please email test@testing.com", "test@testing.com");
+  FindAndCheckEmail("please email test@123.456.co.uk.", "test@123.456.co.uk");
+  FindAndCheckEmail("My email is 'a@b.org'.", "a@b.org");
+  FindAndCheckEmail("123@bcd.org", "123@bcd.org");
+  FindAndCheckEmail("[quitelongwelllongemailaddress@somequitelongdomain.org]",
+                    "quitelongwelllongemailaddress@somequitelongdomain.org");
+  FindAndCheckEmail("Should find the first@email.org not the second@email.org",
+                    "first@email.org");
+  FindAndCheckEmail("Email:HELLO@SOMETHING.COM", "HELLO@SOMETHING.COM");
+  FindAndCheckEmail("Email SOMEONE@GOOGLE.COM for details.",
+                    "SOMEONE@GOOGLE.COM");
+  FindAndCheckEmail("It's \"testadd@company.fr\"", "testadd@company.fr");
+  FindAndCheckEmail("This is not an @emailaddress.com", "");
+  FindAndCheckEmail("Apples @2.50 each", "");
+  FindAndCheckEmail("Log on to google.com", "");
+  FindAndCheckEmail("Try someone@, they might know.", "");
+  FindAndCheckEmail("No, bob@com is not an email address.", "");
+  FindAndCheckEmail("@", "");
+  FindAndCheckEmail("Just bob @google.com", "");
+  FindAndCheckEmail("Why not call larry@google and ask him.", "");
+}