Make it possible to build url/ without ICU on android.

This is needed to reduce binary size of net/ when build as a
library.

BUG=362608

Review URL: https://codereview.chromium.org/257673002

git-svn-id: svn://svn.chromium.org/chrome/trunk/src@268740 0039d316-1c4b-4281-b951-d872f2087c98

11 files changed, 435 insertions, 108 deletions

diff --git a/url/BUILD.gn b/url/BUILD.gn
index 6445bfb..faf8dd8 100644
--- a/url/BUILD.gn
+++ b/url/BUILD.gn
@@ -2,12 +2,26 @@
 # Use of this source code is governed by a BSD-style license that can be
 # found in the LICENSE file.
 
+declare_args() {
+  # Switches to using platform functions instead of ICU on Android.
+  use_icu_alternatives_on_android = false
+}
+
+# Sets the USE_ICU_ALTERNATIVES_ON_ANDROID define based on the build flag.
+config("url_icu_config") {
+  if (use_icu_alternatives_on_android) {
+    defines = [ "USE_ICU_ALTERNATIVES_ON_ANDROID=1" ]
+  }
+}
+
 component("url") {
   if (is_win) {
     # Don't conflict with Windows' "url.dll".
     output_name = "url_lib"
   }
   sources = [
+    "android/url_jni_registrar.cc",
+    "android/url_jni_registrar.h",
     "gurl.cc",
     "gurl.h",
     "third_party/mozilla/url_parse.cc",
@@ -42,6 +56,8 @@ component("url") {
 
   defines = [ "URL_IMPLEMENTATION" ]
 
+  configs += [ ":url_icu_config" ]
+
   # if (is_win) {
   #   TODO(jschuh): crbug.com/167187 fix size_t to int truncations.
   #   'msvs_disabled_warnings': [ 4267, ]
@@ -52,6 +68,22 @@ component("url") {
     "//third_party/icu:icudata",
     "//third_party/icu",
   ]
+
+  if (use_icu_alternatives_on_android) {
+    sources -= [
+      "url_canon_icu.cc",
+      "url_canon_icu.h",
+    ]
+    deps -= [
+      "//third_party/icu:icudata",
+      "//third_party/icu",
+    ]
+
+    sources += [
+      "url_canon_icu_alternatives_android.cc",
+      "url_canon_icu_alternatives_android.h",
+    ]
+  }
 }
 
 # TODO(dpranke): crbug.com/360936. Get this to build and run on Android.
@@ -59,6 +91,7 @@ if (!is_android) {
   test("url_unittests") {
     sources = [
       "gurl_unittest.cc",
+      "url_canon_icu_unittest.cc",
       "url_canon_unittest.cc",
       "url_parse_unittest.cc",
       "url_test_utils.h",
@@ -79,10 +112,18 @@ if (!is_android) {
 
     deps = [
       ":url",
-      "//base:i18n",
       "//base/test:run_all_unittests",
       "//testing/gtest",
       "//third_party/icu:icuuc",
     ]
+
+    if (use_icu_alternatives_on_android) {
+      sources -= [
+        "url_canon_icu_unittest.cc",
+      ]
+      deps -= [
+        "//third_party/icu:icuuc",
+      ]
+    }
   }
 }
diff --git a/url/DEPS b/url/DEPS
new file mode 100644
index 0000000..9643291
--- /dev/null
+++ b/url/DEPS
@@ -0,0 +1,13 @@
+include_rules = [
+  "+jni",
+
+  # Limit files that can depend on icu.
+  "-base/i18n",
+  "-third_party/icu",
+]
+
+specific_include_rules = {
+  "url_canon_icu(\.cc|_unittest\.cc)": [
+    "+third_party/icu",
+  ],
+}
diff --git a/url/android/java/src/org/chromium/url/IDNStringUtil.java b/url/android/java/src/org/chromium/url/IDNStringUtil.java
new file mode 100644
index 0000000..32000fd
--- /dev/null
+++ b/url/android/java/src/org/chromium/url/IDNStringUtil.java
@@ -0,0 +1,33 @@
+// Copyright 2014 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.
+
+package org.chromium.url;
+
+import org.chromium.base.CalledByNative;
+import org.chromium.base.JNINamespace;
+
+import java.net.IDN;
+
+/**
+ * This class is used to convert unicode IDN domain names to ASCII, when not
+ * building with ICU.
+ */
+@JNINamespace("url::android")
+public class IDNStringUtil {
+    /**
+     * Attempts to convert a Unicode string to an ASCII string using IDN rules.
+     * As of May 2014, the underlying Java function IDNA2003.
+     * @param src String to convert.
+     * @return: String containing only ASCII characters on success, null on
+     *                 failure.
+     */
+    @CalledByNative
+    private static String idnToASCII(String src) {
+        try {
+            return IDN.toASCII(src, IDN.USE_STD3_ASCII_RULES);
+        } catch (Exception e) {
+            return null;
+        }
+    }
+}
\ No newline at end of file
diff --git a/url/android/url_jni_registrar.cc b/url/android/url_jni_registrar.cc
new file mode 100644
index 0000000..82a6e1b
--- /dev/null
+++ b/url/android/url_jni_registrar.cc
@@ -0,0 +1,24 @@
+// Copyright 2014 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 "url/android/url_jni_registrar.h"
+
+#ifdef USE_ICU_ALTERNATIVES_ON_ANDROID
+#include "url/url_canon_icu_alternatives_android.h"
+#endif
+
+namespace url {
+namespace android {
+
+bool RegisterJni(JNIEnv* env) {
+#ifdef USE_ICU_ALTERNATIVES_ON_ANDROID
+  return RegisterIcuAlternativesJni(env);
+#endif
+
+  // Do nothing if USE_ICU_ALTERNATIVES_ON_ANDROID is not defined.
+  return true;
+}
+
+}  // namespace android
+}  // namespace url
diff --git a/url/android/url_jni_registrar.h b/url/android/url_jni_registrar.h
new file mode 100644
index 0000000..8fe7ded
--- /dev/null
+++ b/url/android/url_jni_registrar.h
@@ -0,0 +1,21 @@
+// Copyright 2014 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 URL_ANDROID_URL_JNI_REGISTRAR_H_
+#define URL_ANDROID_URL_JNI_REGISTRAR_H_
+
+#include <jni.h>
+
+#include "url/url_export.h"
+
+namespace url {
+namespace android {
+
+// Register all JNI bindings necessary for url.
+URL_EXPORT bool RegisterJni(JNIEnv* env);
+
+}  // namespace android
+}  // namespace url
+
+#endif  // URL_ANDROID_URL_JNI_REGISTRAR_H_
diff --git a/url/url.gyp b/url/url.gyp
index 93bbee2..d1c3bf1 100644
--- a/url/url.gyp
+++ b/url/url.gyp
@@ -33,6 +33,28 @@
       'defines': [
         'URL_IMPLEMENTATION',
       ],
+      'conditions': [
+        ['use_icu_alternatives_on_android==1', {
+          'sources!': [
+            'url_canon_icu.cc',
+            'url_canon_icu.h',
+          ],
+          'dependencies!': [
+            '../third_party/icu/icu.gyp:icui18n',
+            '../third_party/icu/icu.gyp:icuuc',
+          ],
+        }],
+        ['use_icu_alternatives_on_android==1 and OS=="android"', {
+          'dependencies': [
+            'url_java',
+            'url_jni_headers',
+          ],
+          'sources': [
+            'url_canon_icu_alternatives_android.cc',
+            'url_canon_icu_alternatives_android.h',
+          ],
+        }],
+      ],
       # TODO(jschuh): crbug.com/167187 fix size_t to int truncations.
       'msvs_disabled_warnings': [4267, ],
     },
@@ -40,7 +62,6 @@
       'target_name': 'url_unittests',
       'type': 'executable',
       'dependencies': [
-        '../base/base.gyp:base_i18n',
         '../base/base.gyp:run_all_unittests',
         '../testing/gtest.gyp:gtest',
         '../third_party/icu/icu.gyp:icuuc',
@@ -49,6 +70,7 @@
       'sources': [
         'gurl_unittest.cc',
         'origin_unittest.cc',
+        'url_canon_icu_unittest.cc',
         'url_canon_unittest.cc',
         'url_parse_unittest.cc',
         'url_test_utils.h',
@@ -63,9 +85,47 @@
             ],
           }
         ],
+        ['use_icu_alternatives_on_android==1',
+          {
+            'sources!': [
+              'url_canon_icu_unittest.cc',
+            ],
+            'dependencies!': [
+              '../third_party/icu/icu.gyp:icuuc',
+            ],
+          }
+        ],
       ],
       # TODO(jschuh): crbug.com/167187 fix size_t to int truncations.
       'msvs_disabled_warnings': [4267, ],
     },
   ],
+  'conditions': [
+    ['use_icu_alternatives_on_android==1 and OS=="android"', {
+      'targets': [
+        {
+          'target_name': 'url_jni_headers',
+          'type': 'none',
+          'sources': [
+            'android/java/src/org/chromium/url/IDNStringUtil.java'
+          ],
+          'variables': {
+            'jni_gen_package': 'url',
+          },
+          'includes': [ '../build/jni_generator.gypi' ],
+        },
+        {
+          'target_name': 'url_java',
+          'type': 'none',
+          'variables': {
+            'java_in_dir': '../url/android/java',
+          },
+          'dependencies': [
+            '../base/base.gyp:base',
+          ],
+          'includes': [ '../build/java.gypi' ],
+        },
+      ],
+    }],
+  ],
 }
diff --git a/url/url_canon_icu_alternatives_android.cc b/url/url_canon_icu_alternatives_android.cc
new file mode 100644
index 0000000..f43efce
--- /dev/null
+++ b/url/url_canon_icu_alternatives_android.cc
@@ -0,0 +1,41 @@
+// Copyright 2014 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 <string.h>
+
+#include "base/android/jni_android.h"
+#include "base/android/jni_string.h"
+#include "base/strings/string16.h"
+#include "base/strings/string_piece.h"
+#include "jni/IDNStringUtil_jni.h"
+#include "url/url_canon_internal.h"
+
+namespace url {
+
+// This uses the JDK's conversion function, which uses IDNA 2003, unlike the
+// ICU implementation.
+bool IDNToASCII(const base::char16* src, int src_len, CanonOutputW* output) {
+  DCHECK_EQ(0, output->length());  // Output buffer is assumed empty.
+
+  JNIEnv* env = base::android::AttachCurrentThread();
+  base::android::ScopedJavaLocalRef<jstring> java_src =
+      base::android::ConvertUTF16ToJavaString(
+          env, base::StringPiece16(src, src_len));
+  ScopedJavaLocalRef<jstring> java_result =
+      android::Java_IDNStringUtil_idnToASCII(env, java_src.obj());
+  // NULL indicates failure.
+  if (java_result.is_null())
+    return false;
+
+  base::string16 utf16_result =
+      base::android::ConvertJavaStringToUTF16(java_result);
+  output->Append(utf16_result.data(), static_cast<int>(utf16_result.size()));
+  return true;
+}
+
+bool RegisterIcuAlternativesJni(JNIEnv* env) {
+  return android::RegisterNativesImpl(env);
+}
+
+}  // namespace url
diff --git a/url/url_canon_icu_alternatives_android.h b/url/url_canon_icu_alternatives_android.h
new file mode 100644
index 0000000..67306db
--- /dev/null
+++ b/url/url_canon_icu_alternatives_android.h
@@ -0,0 +1,18 @@
+// Copyright 2014 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 URL_URL_CANON_ICU_ALTERNATIVES_ANDROID_H_
+#define URL_URL_CANON_ICU_ALTERNATIVES_ANDROID_H_
+
+#include <jni.h>
+
+namespace url {
+
+// Explicitly register static JNI functions needed when not using ICU.
+bool RegisterIcuAlternativesJni(JNIEnv* env);
+
+}  // namespace url
+
+#endif  // URL_URL_CANON_ICU_ALTERNATIVES_ANDROID_H_
+
diff --git a/url/url_canon_icu_unittest.cc b/url/url_canon_icu_unittest.cc
new file mode 100644
index 0000000..b28c30a
--- /dev/null
+++ b/url/url_canon_icu_unittest.cc
@@ -0,0 +1,167 @@
+// Copyright 2014 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 "base/macros.h"
+#include "testing/gtest/include/gtest/gtest.h"
+#include "third_party/icu/source/common/unicode/ucnv.h"
+#include "url/url_canon.h"
+#include "url/url_canon_icu.h"
+#include "url/url_canon_stdstring.h"
+#include "url/url_test_utils.h"
+
+// Some implementations of base/basictypes.h may define ARRAYSIZE.
+// If it's not defined, we define it to the ARRAYSIZE_UNSAFE macro
+// which is in our version of basictypes.h.
+#ifndef ARRAYSIZE
+#define ARRAYSIZE ARRAYSIZE_UNSAFE
+#endif
+
+namespace url {
+
+using test_utils::WStringToUTF16;
+
+namespace {
+
+// Wrapper around a UConverter object that managers creation and destruction.
+class UConvScoper {
+ public:
+  explicit UConvScoper(const char* charset_name) {
+    UErrorCode err = U_ZERO_ERROR;
+    converter_ = ucnv_open(charset_name, &err);
+  }
+
+  ~UConvScoper() {
+    if (converter_)
+      ucnv_close(converter_);
+  }
+
+  // Returns the converter object, may be NULL.
+  UConverter* converter() const { return converter_; }
+
+ private:
+  UConverter* converter_;
+};
+
+TEST(URLCanonIcuTest, ICUCharsetConverter) {
+  struct ICUCase {
+    const wchar_t* input;
+    const char* encoding;
+    const char* expected;
+  } icu_cases[] = {
+      // UTF-8.
+    {L"Hello, world", "utf-8", "Hello, world"},
+    {L"\x4f60\x597d", "utf-8", "\xe4\xbd\xa0\xe5\xa5\xbd"},
+      // Non-BMP UTF-8.
+    {L"!\xd800\xdf00!", "utf-8", "!\xf0\x90\x8c\x80!"},
+      // Big5
+    {L"\x4f60\x597d", "big5", "\xa7\x41\xa6\x6e"},
+      // Unrepresentable character in the destination set.
+    {L"hello\x4f60\x06de\x597dworld", "big5",
+      "hello\xa7\x41%26%231758%3B\xa6\x6eworld"},
+  };
+
+  for (size_t i = 0; i < ARRAYSIZE(icu_cases); i++) {
+    UConvScoper conv(icu_cases[i].encoding);
+    ASSERT_TRUE(conv.converter() != NULL);
+    ICUCharsetConverter converter(conv.converter());
+
+    std::string str;
+    StdStringCanonOutput output(&str);
+
+    base::string16 input_str(WStringToUTF16(icu_cases[i].input));
+    int input_len = static_cast<int>(input_str.length());
+    converter.ConvertFromUTF16(input_str.c_str(), input_len, &output);
+    output.Complete();
+
+    EXPECT_STREQ(icu_cases[i].expected, str.c_str());
+  }
+
+  // Test string sizes around the resize boundary for the output to make sure
+  // the converter resizes as needed.
+  const int static_size = 16;
+  UConvScoper conv("utf-8");
+  ASSERT_TRUE(conv.converter());
+  ICUCharsetConverter converter(conv.converter());
+  for (int i = static_size - 2; i <= static_size + 2; i++) {
+    // Make a string with the appropriate length.
+    base::string16 input;
+    for (int ch = 0; ch < i; ch++)
+      input.push_back('a');
+
+    RawCanonOutput<static_size> output;
+    converter.ConvertFromUTF16(input.c_str(), static_cast<int>(input.length()),
+                               &output);
+    EXPECT_EQ(input.length(), static_cast<size_t>(output.length()));
+  }
+}
+
+TEST(URLCanonIcuTest, QueryWithConverter) {
+  struct QueryCase {
+    const char* input8;
+    const wchar_t* input16;
+    const char* encoding;
+    const char* expected;
+  } query_cases[] = {
+      // Regular ASCII case in some different encodings.
+    {"foo=bar", L"foo=bar", "utf-8", "?foo=bar"},
+    {"foo=bar", L"foo=bar", "shift_jis", "?foo=bar"},
+    {"foo=bar", L"foo=bar", "gb2312", "?foo=bar"},
+      // Chinese input/output
+    {"q=\xe4\xbd\xa0\xe5\xa5\xbd", L"q=\x4f60\x597d", "gb2312",
+      "?q=%C4%E3%BA%C3"},
+    {"q=\xe4\xbd\xa0\xe5\xa5\xbd", L"q=\x4f60\x597d", "big5", "?q=%A7A%A6n"},
+      // Unencodable character in the destination character set should be
+      // escaped. The escape sequence unescapes to be the entity name:
+      // "?q=&#20320;"
+    {"q=Chinese\xef\xbc\xa7", L"q=Chinese\xff27", "iso-8859-1",
+      "?q=Chinese%26%2365319%3B"},
+  };
+
+  for (size_t i = 0; i < ARRAYSIZE(query_cases); i++) {
+    Component out_comp;
+
+    UConvScoper conv(query_cases[i].encoding);
+    ASSERT_TRUE(!query_cases[i].encoding || conv.converter());
+    ICUCharsetConverter converter(conv.converter());
+
+    if (query_cases[i].input8) {
+      int len = static_cast<int>(strlen(query_cases[i].input8));
+      Component in_comp(0, len);
+      std::string out_str;
+
+      StdStringCanonOutput output(&out_str);
+      CanonicalizeQuery(query_cases[i].input8, in_comp, &converter, &output,
+                        &out_comp);
+      output.Complete();
+
+      EXPECT_EQ(query_cases[i].expected, out_str);
+    }
+
+    if (query_cases[i].input16) {
+      base::string16 input16(WStringToUTF16(query_cases[i].input16));
+      int len = static_cast<int>(input16.length());
+      Component in_comp(0, len);
+      std::string out_str;
+
+      StdStringCanonOutput output(&out_str);
+      CanonicalizeQuery(input16.c_str(), in_comp, &converter, &output,
+                        &out_comp);
+      output.Complete();
+
+      EXPECT_EQ(query_cases[i].expected, out_str);
+    }
+  }
+
+  // Extra test for input with embedded NULL;
+  std::string out_str;
+  StdStringCanonOutput output(&out_str);
+  Component out_comp;
+  CanonicalizeQuery("a \x00z\x01", Component(0, 5), NULL, &output, &out_comp);
+  output.Complete();
+  EXPECT_EQ("?a%20%00z%01", out_str);
+}
+
+}  // namespace
+
+}  // namespace url
diff --git a/url/url_canon_unittest.cc b/url/url_canon_unittest.cc
index 55f9608..9a766e3 100644
--- a/url/url_canon_unittest.cc
+++ b/url/url_canon_unittest.cc
@@ -4,10 +4,9 @@
 
 #include <errno.h>
 
+#include "base/macros.h"
 #include "testing/gtest/include/gtest/gtest.h"
-#include "third_party/icu/source/common/unicode/ucnv.h"
 #include "url/url_canon.h"
-#include "url/url_canon_icu.h"
 #include "url/url_canon_internal.h"
 #include "url/url_canon_stdstring.h"
 #include "url/url_parse.h"
@@ -84,26 +83,6 @@ struct ReplaceCase {
   const char* expected;
 };
 
-// Wrapper around a UConverter object that managers creation and destruction.
-class UConvScoper {
- public:
-  explicit UConvScoper(const char* charset_name) {
-    UErrorCode err = U_ZERO_ERROR;
-    converter_ = ucnv_open(charset_name, &err);
-  }
-
-  ~UConvScoper() {
-    if (converter_)
-      ucnv_close(converter_);
-  }
-
-  // Returns the converter object, may be NULL.
-  UConverter* converter() const { return converter_; }
-
- private:
-  UConverter* converter_;
-};
-
 // Magic string used in the replacements code that tells SetupReplComp to
 // call the clear function.
 const char kDeleteComp[] = "|";
@@ -244,58 +223,6 @@ TEST(URLCanonTest, UTF) {
   }
 }
 
-TEST(URLCanonTest, ICUCharsetConverter) {
-  struct ICUCase {
-    const wchar_t* input;
-    const char* encoding;
-    const char* expected;
-  } icu_cases[] = {
-      // UTF-8.
-    {L"Hello, world", "utf-8", "Hello, world"},
-    {L"\x4f60\x597d", "utf-8", "\xe4\xbd\xa0\xe5\xa5\xbd"},
-      // Non-BMP UTF-8.
-    {L"!\xd800\xdf00!", "utf-8", "!\xf0\x90\x8c\x80!"},
-      // Big5
-    {L"\x4f60\x597d", "big5", "\xa7\x41\xa6\x6e"},
-      // Unrepresentable character in the destination set.
-    {L"hello\x4f60\x06de\x597dworld", "big5", "hello\xa7\x41%26%231758%3B\xa6\x6eworld"},
-  };
-
-  for (size_t i = 0; i < ARRAYSIZE(icu_cases); i++) {
-    UConvScoper conv(icu_cases[i].encoding);
-    ASSERT_TRUE(conv.converter() != NULL);
-    ICUCharsetConverter converter(conv.converter());
-
-    std::string str;
-    StdStringCanonOutput output(&str);
-
-    base::string16 input_str(WStringToUTF16(icu_cases[i].input));
-    int input_len = static_cast<int>(input_str.length());
-    converter.ConvertFromUTF16(input_str.c_str(), input_len, &output);
-    output.Complete();
-
-    EXPECT_STREQ(icu_cases[i].expected, str.c_str());
-  }
-
-  // Test string sizes around the resize boundary for the output to make sure
-  // the converter resizes as needed.
-  const int static_size = 16;
-  UConvScoper conv("utf-8");
-  ASSERT_TRUE(conv.converter());
-  ICUCharsetConverter converter(conv.converter());
-  for (int i = static_size - 2; i <= static_size + 2; i++) {
-    // Make a string with the appropriate length.
-    base::string16 input;
-    for (int ch = 0; ch < i; ch++)
-      input.push_back('a');
-
-    RawCanonOutput<static_size> output;
-    converter.ConvertFromUTF16(input.c_str(), static_cast<int>(input.length()),
-                               &output);
-    EXPECT_EQ(input.length(), static_cast<size_t>(output.length()));
-  }
-}
-
 TEST(URLCanonTest, Scheme) {
   // Here, we're mostly testing that unusual characters are handled properly.
   // The canonicalizer doesn't do any parsing or whitespace detection. It will
@@ -1198,57 +1125,38 @@ TEST(URLCanonTest, Query) {
   struct QueryCase {
     const char* input8;
     const wchar_t* input16;
-    const char* encoding;
     const char* expected;
   } query_cases[] = {
-      // Regular ASCII case in some different encodings.
-    {"foo=bar", L"foo=bar", NULL, "?foo=bar"},
-    {"foo=bar", L"foo=bar", "utf-8", "?foo=bar"},
-    {"foo=bar", L"foo=bar", "shift_jis", "?foo=bar"},
-    {"foo=bar", L"foo=bar", "gb2312", "?foo=bar"},
+      // Regular ASCII case.
+    {"foo=bar", L"foo=bar", "?foo=bar"},
       // Allow question marks in the query without escaping
-    {"as?df", L"as?df", NULL, "?as?df"},
+    {"as?df", L"as?df", "?as?df"},
       // Always escape '#' since it would mark the ref.
-    {"as#df", L"as#df", NULL, "?as%23df"},
+    {"as#df", L"as#df", "?as%23df"},
       // Escape some questionable 8-bit characters, but never unescape.
-    {"\x02hello\x7f bye", L"\x02hello\x7f bye", NULL, "?%02hello%7F%20bye"},
-    {"%40%41123", L"%40%41123", NULL, "?%40%41123"},
+    {"\x02hello\x7f bye", L"\x02hello\x7f bye", "?%02hello%7F%20bye"},
+    {"%40%41123", L"%40%41123", "?%40%41123"},
       // Chinese input/output
-    {"q=\xe4\xbd\xa0\xe5\xa5\xbd", L"q=\x4f60\x597d", NULL, "?q=%E4%BD%A0%E5%A5%BD"},
-    {"q=\xe4\xbd\xa0\xe5\xa5\xbd", L"q=\x4f60\x597d", "gb2312", "?q=%C4%E3%BA%C3"},
-    {"q=\xe4\xbd\xa0\xe5\xa5\xbd", L"q=\x4f60\x597d", "big5", "?q=%A7A%A6n"},
-      // Unencodable character in the destination character set should be
-      // escaped. The escape sequence unescapes to be the entity name:
-      // "?q=&#20320;"
-    {"q=Chinese\xef\xbc\xa7", L"q=Chinese\xff27", "iso-8859-1", "?q=Chinese%26%2365319%3B"},
+    {"q=\xe4\xbd\xa0\xe5\xa5\xbd", L"q=\x4f60\x597d", "?q=%E4%BD%A0%E5%A5%BD"},
       // Invalid UTF-8/16 input should be replaced with invalid characters.
-    {"q=\xed\xed", L"q=\xd800\xd800", NULL, "?q=%EF%BF%BD%EF%BF%BD"},
+    {"q=\xed\xed", L"q=\xd800\xd800", "?q=%EF%BF%BD%EF%BF%BD"},
       // Don't allow < or > because sometimes they are used for XSS if the
       // URL is echoed in content. Firefox does this, IE doesn't.
-    {"q=<asdf>", L"q=<asdf>", NULL, "?q=%3Casdf%3E"},
+    {"q=<asdf>", L"q=<asdf>", "?q=%3Casdf%3E"},
       // Escape double quotemarks in the query.
-    {"q=\"asdf\"", L"q=\"asdf\"", NULL, "?q=%22asdf%22"},
+    {"q=\"asdf\"", L"q=\"asdf\"", "?q=%22asdf%22"},
   };
 
   for (size_t i = 0; i < ARRAYSIZE(query_cases); i++) {
     Component out_comp;
 
-    UConvScoper conv(query_cases[i].encoding);
-    ASSERT_TRUE(!query_cases[i].encoding || conv.converter());
-    ICUCharsetConverter converter(conv.converter());
-
-    // Map NULL to a NULL converter pointer.
-    ICUCharsetConverter* conv_pointer = &converter;
-    if (!query_cases[i].encoding)
-      conv_pointer = NULL;
-
     if (query_cases[i].input8) {
       int len = static_cast<int>(strlen(query_cases[i].input8));
       Component in_comp(0, len);
       std::string out_str;
 
       StdStringCanonOutput output(&out_str);
-      CanonicalizeQuery(query_cases[i].input8, in_comp, conv_pointer, &output,
+      CanonicalizeQuery(query_cases[i].input8, in_comp, NULL, &output,
                         &out_comp);
       output.Complete();
 
@@ -1262,8 +1170,7 @@ TEST(URLCanonTest, Query) {
       std::string out_str;
 
       StdStringCanonOutput output(&out_str);
-      CanonicalizeQuery(input16.c_str(), in_comp, conv_pointer, &output,
-                        &out_comp);
+      CanonicalizeQuery(input16.c_str(), in_comp, NULL, &output, &out_comp);
       output.Complete();
 
       EXPECT_EQ(query_cases[i].expected, out_str);
diff --git a/url/url_srcs.gypi b/url/url_srcs.gypi
index 3114620..e8fa82e 100644
--- a/url/url_srcs.gypi
+++ b/url/url_srcs.gypi
@@ -5,6 +5,8 @@
 {
   'variables': {
     'gurl_sources': [
+      'android/url_jni_registrar.cc',
+      'android/url_jni_registrar.h',
       'gurl.cc',
       'gurl.h',
       'origin.cc',