[Android] Introduce phone number detection.

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

BUG=125390
TEST=phone_number_detector_unittest.cc

Committed: https://src.chromium.org/viewvc/chrome?view=rev&revision=139019

Review URL: https://chromiumcodereview.appspot.com/10440021

git-svn-id: svn://svn.chromium.org/chrome/trunk/src@139026 0039d316-1c4b-4281-b951-d872f2087c98

content/content_renderer.gypi

@@ -46,6 +46,8 @@
     'renderer/android/content_detector.h',
     'renderer/android/email_detector.cc',
     'renderer/android/email_detector.h',
+    'renderer/android/phone_number_detector.cc',
+    'renderer/android/phone_number_detector.h',
     'renderer/active_notification_tracker.cc',
     'renderer/active_notification_tracker.h',
     'renderer/device_orientation_dispatcher.cc',
@@ -268,6 +270,11 @@
           '../base/allocator/allocator.gyp:allocator',
       ],
     }],
+    ['OS=="android"', {
+      'dependencies': [
+        '../third_party/libphonenumber/libphonenumber.gyp:libphonenumber',
+      ],
+    }],
     # TODO(jrg): remove the OS=="android" section?
     # http://crbug.com/113172
     # Understand better how media_stream_ is tied into Chromium.

content/content_tests.gypi

@@ -286,6 +286,7 @@
         'renderer/active_notification_tracker_unittest.cc',
         'renderer/android/address_detector_unittest.cc',
         'renderer/android/email_detector_unittest.cc',
+        'renderer/android/phone_number_detector_unittest.cc',
         'renderer/gpu/input_event_filter_unittest.cc',
         'renderer/media/audio_message_filter_unittest.cc',
         'renderer/media/capture_video_decoder_unittest.cc',

content/renderer/android/DEPS

@@ -0,0 +1,3 @@
+include_rules = [
+  "+third_party/libphonenumber",  # For phone number detection.
+]

content/renderer/android/phone_number_detector.cc

@@ -0,0 +1,90 @@
+// 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/phone_number_detector.h"
+
+#include <algorithm>
+
+#include "base/utf_string_conversions.h"
+#include "net/base/escape.h"
+#include "third_party/libphonenumber/src/phonenumber_api.h"
+#include "third_party/libphonenumber/src/phonenumbers/phonenumbermatch.h"
+#include "third_party/libphonenumber/src/phonenumbers/phonenumbermatcher.h"
+#include "third_party/libphonenumber/src/phonenumbers/region_code.h"
+
+using i18n::phonenumbers::PhoneNumberMatch;
+using i18n::phonenumbers::PhoneNumberMatcher;
+using i18n::phonenumbers::PhoneNumberUtil;
+using i18n::phonenumbers::RegionCode;
+
+namespace {
+
+// Maximum number of characters to look around for phone number detection.
+const size_t kMaximumTelephoneLength = 20;
+
+// Prefix used for telephone number intent URIs.
+const char kPhoneNumberSchemaPrefix[] = "tel:";
+
+}  // anonymous namespace
+
+namespace content {
+
+PhoneNumberDetector::PhoneNumberDetector()
+    : region_code_(RegionCode::GetUnknown()) {
+}
+
+// Region should be empty or an ISO 3166-1 alpha-2 country code.
+PhoneNumberDetector::PhoneNumberDetector(const std::string& region)
+    : region_code_(region.empty() ? RegionCode::GetUnknown()
+                                  : StringToUpperASCII(region)) {
+}
+
+PhoneNumberDetector::~PhoneNumberDetector() {
+}
+
+size_t PhoneNumberDetector::GetMaximumContentLength() {
+  return kMaximumTelephoneLength;
+}
+
+GURL PhoneNumberDetector::GetIntentURL(const std::string& content_text) {
+  if (content_text.empty())
+    return GURL();
+
+  return GURL(kPhoneNumberSchemaPrefix +
+      net::EscapeQueryParamValue(content_text, true));
+}
+
+bool PhoneNumberDetector::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);
+  std::string utf8_input = UTF16ToUTF8(utf16_input);
+
+  PhoneNumberMatcher matcher(utf8_input, region_code_);
+  if (matcher.HasNext()) {
+    PhoneNumberMatch match;
+    matcher.Next(&match);
+
+    PhoneNumberUtil* phone_util = PhoneNumberUtil::GetInstance();
+    phone_util->FormatNumberForMobileDialing(match.number(), region_code_,
+                                             false, /* with_formatting */
+                                             content_text);
+    // If the number can't be dialed from the current region, the formatted
+    // string will be empty.
+    if (content_text->empty())
+      return false;
+
+    // Need to return start_pos and end_pos relative to a UTF16 encoding.
+    *start_pos = UTF8ToUTF16(utf8_input.substr(0, match.start())).length();
+    *end_pos = *start_pos + UTF8ToUTF16(match.raw_string()).length();
+
+    return true;
+  }
+
+  return false;
+}
+
+}  // namespace content

content/renderer/android/phone_number_detector.h

@@ -0,0 +1,41 @@
+// 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_PHONE_NUMBER_DETECTOR_H_
+#define CONTENT_RENDERER_ANDROID_PHONE_NUMBER_DETECTOR_H_
+#pragma once
+
+#include "content/renderer/android/content_detector.h"
+
+class PhoneNumberDetectorTest;
+
+namespace content {
+
+// Finds a telephone number in the given content text string.
+class PhoneNumberDetector : public ContentDetector {
+ public:
+  PhoneNumberDetector();
+  explicit PhoneNumberDetector(const std::string& region);
+  virtual ~PhoneNumberDetector();
+
+ private:
+  friend class ::PhoneNumberDetectorTest;
+
+  // 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;
+
+  const std::string region_code_;
+
+  DISALLOW_COPY_AND_ASSIGN(PhoneNumberDetector);
+};
+
+}  // namespace content
+
+#endif  // CONTENT_RENDERER_ANDROID_PHONE_NUMBER_DETECTOR_H_

content/renderer/android/phone_number_detector_unittest.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/phone_number_detector.h"
+
+#include "base/utf_string_conversions.h"
+#include "testing/gtest/include/gtest/gtest.h"
+
+using content::PhoneNumberDetector;
+
+class PhoneNumberDetectorTest : public testing::Test {
+ public:
+  static std::string FindNumber(const std::string& content,
+                                const std::string& region) {
+    string16 content_16 = UTF8ToUTF16(content);
+    string16 result_16;
+    size_t start, end;
+    PhoneNumberDetector detector(region);
+    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);
+    return UTF16ToUTF8(result_16);
+  }
+
+  static std::string FindAndFormatNumber(const std::string& content,
+                                         const std::string& region) {
+    string16 content_16 = UTF8ToUTF16(content);
+    string16 result_16;
+    size_t start, end;
+    PhoneNumberDetector detector(region);
+    std::string content_text;
+    detector.FindContent(content_16.begin(), content_16.end(),
+                         &start, &end, &content_text);
+    return content_text;
+  }
+};
+
+TEST_F(PhoneNumberDetectorTest, FindNumber) {
+  // Tests cases with valid home numbers.
+  EXPECT_EQ("617-426-3000", FindNumber("hello 617-426-3000 blah", "us"));
+  EXPECT_EQ("", FindNumber("hello 617-426-3000 blah", "gb"));
+  EXPECT_EQ("020-7617-4426", FindNumber("<div>020-7617-4426</div>", "gb"));
+  EXPECT_EQ("", FindNumber("<div>020-7617-4426</div>", "fr"));
+  EXPECT_EQ("02.38.96.68.88", FindNumber("Tel:02.38.96.68.88", "fr"));
+  EXPECT_EQ("", FindNumber("Tel:02.38.96.68.88", "gb"));
+  EXPECT_EQ("1-800-866-2453",
+            FindNumber("You can call this number:1-800-866-2453 for more "
+                       "information", "us"));
+  EXPECT_EQ("+1 203-925-4602", FindNumber("+1 203-925-4602", "us"));
+}
+
+TEST_F(PhoneNumberDetectorTest, FindAndFormatNumber) {
+  EXPECT_EQ("+16174263000",
+            FindAndFormatNumber("hello 617-426-3000 blah", "us"));
+  EXPECT_EQ("", FindAndFormatNumber("hello 617-426-3000 blah", "gb"));
+  EXPECT_EQ("+442076174426",
+            FindAndFormatNumber("<div>020-7617-4426</div>", "gb"));
+  EXPECT_EQ("", FindAndFormatNumber("<div>020-7617-4426</div>", "fr"));
+  EXPECT_EQ("+33238966888", FindAndFormatNumber("Tel:02.38.96.68.88", "fr"));
+  EXPECT_EQ("+18008662453",
+            FindAndFormatNumber("You can call this number:1-800-866-2453 for"
+                                "more information", "us"));
+  EXPECT_EQ("+12039254602", FindAndFormatNumber("+1 203-925-4602", "us"));
+
+  // "+1 (650) 333-6000" using fullwidth UTF-8 characters.
+  EXPECT_EQ("+16503336000", FindAndFormatNumber(
+            "\xEF\xBC\x8B\xEF\xBC\x91\xE3\x80\x80\xEF\xBC\x88"
+            "\xEF\xBC\x96\xEF\xBC\x95\xEF\xBC\x90\xEF\xBC\x89"
+            "\xE3\x80\x80\xEF\xBC\x93\xEF\xBC\x93\xEF\xBC\x93"
+            "\xE3\x83\xBC\xEF\xBC\x96\xEF\xBC\x90\xEF\xBC\x90"
+            "\xEF\xBC\x90", "us"));
+}

third_party/libphonenumber/libphonenumber.gyp

@@ -58,6 +58,7 @@
     'direct_dependent_settings': {
       'include_dirs': [
         '<(SHARED_INTERMEDIATE_DIR)/protoc_out/third_party/libphonenumber',
+        'src',
       ],
     },
     'variables': {