chrome/browser/browser_encoding_browsertest.cc

   1 // Copyright (c) 2012 The Chromium Authors. All rights reserved.
   2 // Use of this source code is governed by a BSD-style license that can be
   3 // found in the LICENSE file.
   4
   5 #include <stddef.h>
   6
   7 #include "base/files/file_util.h"
   8 #include "base/files/scoped_temp_dir.h"
   9 #include "base/path_service.h"
  10 #include "base/stl_util.h"
  11 #include "base/strings/stringprintf.h"
  12 #include "base/threading/thread_restrictions.h"
  13 #include "build/build_config.h"
  14 #include "chrome/browser/profiles/profile.h"
  15 #include "chrome/browser/ui/browser.h"
  16 #include "chrome/browser/ui/browser_commands.h"
  17 #include "chrome/browser/ui/tabs/tab_strip_model.h"
  18 #include "chrome/common/chrome_paths.h"
  19 #include "chrome/common/pref_names.h"
  20 #include "chrome/test/base/in_process_browser_test.h"
  21 #include "chrome/test/base/ui_test_utils.h"
  22 #include "components/prefs/pref_service.h"
  23 #include "content/public/browser/download_manager.h"
  24 #include "content/public/browser/navigation_controller.h"
  25 #include "content/public/browser/notification_service.h"
  26 #include "content/public/browser/notification_source.h"
  27 #include "content/public/browser/notification_types.h"
  28 #include "content/public/browser/web_contents.h"
  29 #include "content/public/test/browser_test.h"
  30 #include "content/public/test/download_test_observer.h"
  31 #include "content/public/test/test_navigation_observer.h"
  32 #include "net/test/embedded_test_server/embedded_test_server.h"
  33
  34 namespace {
  35
  36 struct EncodingTestData {
  37   const char* file_name;
  38   const char* encoding_name;
  39 };
  40
  41 const EncodingTestData kEncodingTestDatas[] = {
  42   { "Big5.html", "Big5" },
  43   { "EUC-JP.html", "EUC-JP" },
  44   { "gb18030.html", "gb18030" },
  45   { "iso-8859-1.html", "windows-1252" },
  46   { "ISO-8859-2.html", "ISO-8859-2" },
  47   { "ISO-8859-4.html", "ISO-8859-4" },
  48   { "ISO-8859-5.html", "ISO-8859-5" },
  49   { "ISO-8859-6.html", "ISO-8859-6" },
  50   { "ISO-8859-7.html", "ISO-8859-7" },
  51   { "ISO-8859-8.html", "ISO-8859-8" },
  52   { "ISO-8859-13.html", "ISO-8859-13" },
  53   { "ISO-8859-15.html", "ISO-8859-15" },
  54   { "KOI8-R.html", "KOI8-R" },
  55   { "KOI8-U.html", "KOI8-U" },
  56   { "macintosh.html", "macintosh" },
  57   { "Shift-JIS.html", "Shift_JIS" },
  58   { "US-ASCII.html", "windows-1252" },  // http://crbug.com/15801
  59   { "UTF-8.html", "UTF-8" },
  60   { "UTF-16LE.html", "UTF-16LE" },
  61   { "windows-874.html", "windows-874" },
  62   { "EUC-KR.html", "EUC-KR" },
  63   { "windows-1250.html", "windows-1250" },
  64   { "windows-1251.html", "windows-1251" },
  65   { "windows-1252.html", "windows-1252" },
  66   { "windows-1253.html", "windows-1253" },
  67   { "windows-1254.html", "windows-1254" },
  68   { "windows-1255.html", "windows-1255" },
  69   { "windows-1256.html", "windows-1256" },
  70   { "windows-1257.html", "windows-1257" },
  71   { "windows-1258.html", "windows-1258" }
  72 };
  73
  74 }  // namespace
  75
  76 static const base::FilePath::CharType* kTestDir =
  77     FILE_PATH_LITERAL("encoding_tests");
  78
  79 class BrowserEncodingTest
  80     : public InProcessBrowserTest,
  81       public testing::WithParamInterface<EncodingTestData> {
  82  protected:
  83   BrowserEncodingTest() {}
  84
  85   // Saves the current page and verifies that the output matches the expected
  86   // result.
  87   void SaveAndCompare(const char* filename_to_write,
  88                       const base::FilePath& expected,
  89                       const GURL& url) {
  90     // Dump the page, the content of dump page should be identical to the
  91     // expected result file.
  92     base::FilePath full_file_name = save_dir_.AppendASCII(filename_to_write);
  93     // We save the page as way of complete HTML file, which requires a directory
  94     // name to save sub resources in it. Although this test file does not have
  95     // sub resources, but the directory name is still required.
  96     scoped_refptr<content::MessageLoopRunner> loop_runner(
  97         new content::MessageLoopRunner);
  98     content::SavePackageFinishedObserver observer(
  99         content::BrowserContext::GetDownloadManager(browser()->profile()),
 100         loop_runner->QuitClosure());
 101     browser()->tab_strip_model()->GetActiveWebContents()->SavePage(
 102         full_file_name, temp_sub_resource_dir_,
 103         content::SAVE_PAGE_TYPE_AS_COMPLETE_HTML);
 104     loop_runner->Run();
 105
 106     base::FilePath expected_file_name = ui_test_utils::GetTestFilePath(
 107         base::FilePath(kTestDir), expected);
 108
 109     std::string actual_contents;
 110     std::string expected_contents;
 111
 112     {
 113       base::ScopedAllowBlockingForTesting allow_blocking;
 114       ASSERT_TRUE(base::ReadFileToString(full_file_name, &actual_contents));
 115       ASSERT_TRUE(
 116           base::ReadFileToString(expected_file_name, &expected_contents));
 117     }
 118
 119     // Add "Mark of the Web" path with source URL.
 120     expected_contents = base::StringPrintf(
 121         expected_contents.c_str(), url.spec().length(), url.spec().c_str());
 122
 123     EXPECT_EQ(expected_contents, actual_contents);
 124   }
 125
 126   void SetUpOnMainThread() override {
 127     base::FilePath test_data_dir;
 128     ASSERT_TRUE(base::PathService::Get(chrome::DIR_TEST_DATA, &test_data_dir));
 129     embedded_test_server()->ServeFilesFromDirectory(test_data_dir);
 130     ASSERT_TRUE(embedded_test_server()->Start());
 131
 132     ASSERT_TRUE(temp_dir_.CreateUniqueTempDir());
 133     save_dir_ = temp_dir_.GetPath();
 134     temp_sub_resource_dir_ = save_dir_.AppendASCII("sub_resource_files");
 135   }
 136
 137   base::ScopedTempDir temp_dir_;
 138   base::FilePath save_dir_;
 139   base::FilePath temp_sub_resource_dir_;
 140 };
 141
 142 // TODO(jnd): 1. Some encodings are missing here. It'll be added later. See
 143 // http://crbug.com/13306.
 144 // 2. Add more files with multiple encoding name variants for each canonical
 145 // encoding name). Webkit layout tests cover some, but testing in the UI test is
 146 // also necessary.
 147 IN_PROC_BROWSER_TEST_P(BrowserEncodingTest, TestEncodingAliasMapping) {
 148   const char* const kAliasTestDir = "alias_mapping";
 149
 150   base::FilePath test_dir_path = base::FilePath(kTestDir).AppendASCII(
 151       kAliasTestDir);
 152   base::FilePath test_file_path(test_dir_path);
 153   test_file_path = test_file_path.AppendASCII(
 154       GetParam().file_name);
 155
 156   GURL url =
 157       embedded_test_server()->GetURL("/" + test_file_path.MaybeAsASCII());
 158   ui_test_utils::NavigateToURL(browser(), url);
 159   EXPECT_EQ(GetParam().encoding_name,
 160             browser()->tab_strip_model()->GetActiveWebContents()->
 161                 GetEncoding());
 162 }
 163
 164 INSTANTIATE_TEST_SUITE_P(EncodingAliases,
 165                          BrowserEncodingTest,
 166                          testing::ValuesIn(kEncodingTestDatas));
 167
 168 // The following encodings are excluded from the auto-detection test because
 169 // it's a known issue that the current encoding detector does not detect them:
 170 // ISO-8859-4
 171 // ISO-8859-13
 172 // KOI8-U
 173 // macintosh
 174 // windows-874
 175 // windows-1252
 176 // windows-1253
 177 // windows-1257
 178 // windows-1258
 179
 180 IN_PROC_BROWSER_TEST_F(BrowserEncodingTest, TestEncodingAutoDetect) {
 181   struct EncodingAutoDetectTestData {
 182     const char* test_file_name;   // File name of test data.
 183     const char* expected_result;  // File name of expected results.
 184     const char* expected_encoding;   // expected encoding.
 185   };
 186   const EncodingAutoDetectTestData kTestDatas[] = {
 187       { "Big5_with_no_encoding_specified.html",
 188         "expected_Big5_saved_from_no_encoding_specified.html",
 189         "Big5" },
 190       { "GBK_with_no_encoding_specified.html",
 191         "expected_GBK_saved_from_no_encoding_specified.html",
 192         "GBK" },
 193       { "iso-8859-1_with_no_encoding_specified.html",
 194         "expected_iso-8859-1_saved_from_no_encoding_specified.html",
 195         "windows-1252" },
 196       { "ISO-8859-5_with_no_encoding_specified.html",
 197         "expected_ISO-8859-5_saved_from_no_encoding_specified.html",
 198         "ISO-8859-5" },
 199       { "ISO-8859-6_with_no_encoding_specified.html",
 200         "expected_ISO-8859-6_saved_from_no_encoding_specified.html",
 201         "ISO-8859-6" },
 202       { "ISO-8859-7_with_no_encoding_specified.html",
 203         "expected_ISO-8859-7_saved_from_no_encoding_specified.html",
 204         "ISO-8859-7" },
 205       { "ISO-8859-8-I_with_no_encoding_specified.html",
 206         "expected_ISO-8859-8-I_saved_from_no_encoding_specified.html",
 207         "windows-1255" },
 208       { "KOI8-R_with_no_encoding_specified.html",
 209         "expected_KOI8-R_saved_from_no_encoding_specified.html",
 210         "KOI8-R" },
 211       { "Shift-JIS_with_no_encoding_specified.html",
 212         "expected_Shift-JIS_saved_from_no_encoding_specified.html",
 213         "Shift_JIS" },
 214       { "EUC-KR_with_no_encoding_specified.html",
 215         "expected_EUC-KR_saved_from_no_encoding_specified.html",
 216         "EUC-KR" },
 217       { "windows-1251_with_no_encoding_specified.html",
 218         "expected_windows-1251_saved_from_no_encoding_specified.html",
 219         "windows-1251" },
 220       { "windows-1254_with_no_encoding_specified.html",
 221         "expected_windows-1254_saved_from_no_encoding_specified.html",
 222         "windows-1254" },
 223       { "windows-1255_with_no_encoding_specified.html",
 224         "expected_windows-1255_saved_from_no_encoding_specified.html",
 225         "windows-1255" },
 226       { "windows-1256_with_no_encoding_specified.html",
 227         "expected_windows-1256_saved_from_no_encoding_specified.html",
 228         "windows-1256" }
 229     };
 230   const char* const kAutoDetectDir = "auto_detect";
 231   // Directory of the files of expected results.
 232   const char* const kExpectedResultDir = "expected_results";
 233
 234   base::FilePath test_dir_path =
 235       base::FilePath(kTestDir).AppendASCII(kAutoDetectDir);
 236
 237   // Set the default charset to one of encodings not supported by the current
 238   // auto-detector (Please refer to the above comments) to make sure we
 239   // incorrectly decode the page. Now we use ISO-8859-4.
 240   browser()->profile()->GetPrefs()->SetString(prefs::kDefaultCharset,
 241                                               "ISO-8859-4");
 242
 243   content::WebContents* web_contents =
 244       browser()->tab_strip_model()->GetActiveWebContents();
 245   for (size_t i = 0; i < base::size(kTestDatas); ++i) {
 246     base::FilePath test_file_path(test_dir_path);
 247     test_file_path = test_file_path.AppendASCII(kTestDatas[i].test_file_name);
 248     GURL url =
 249         embedded_test_server()->GetURL("/" + test_file_path.MaybeAsASCII());
 250     ui_test_utils::NavigateToURL(browser(), url);
 251
 252     // Get the encoding of page. It should return the real encoding now.
 253     EXPECT_EQ(kTestDatas[i].expected_encoding, web_contents->GetEncoding());
 254
 255     // Dump the page, the content of dump page should be equal with our expect
 256     // result file.
 257     base::FilePath expected_result_file_name =
 258         base::FilePath().AppendASCII(kAutoDetectDir).
 259         AppendASCII(kExpectedResultDir).
 260         AppendASCII(kTestDatas[i].expected_result);
 261     SaveAndCompare(kTestDatas[i].test_file_name, expected_result_file_name,
 262                    url);
 263   }
 264 }