Add a UI test for "Encoding" menu

chrome/browser/browser_encoding_uitest.cc

Index: chrome/browser/browser_encoding_uitest.cc
===================================================================
--- chrome/browser/browser_encoding_uitest.cc	(revision 0)
+++ chrome/browser/browser_encoding_uitest.cc	(revision 0)
@@ -0,0 +1,306 @@
+// Copyright (c) 2006-2008 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/file_util.h"
+#include "base/path_service.h"
+#include "base/string_util.h"
+#include "chrome/browser/automation/url_request_mock_http_job.h"
+#include "chrome/browser/download/save_package.h"
+#include "chrome/common/chrome_paths.h"
+#include "chrome/test/automation/automation_messages.h"
+#include "chrome/test/automation/automation_proxy.h"
+#include "chrome/test/automation/browser_proxy.h"
+#include "chrome/test/automation/tab_proxy.h"
+#include "chrome/test/ui/ui_test.h"
+#include "net/url_request/url_request_unittest.h"
+#include "chrome/common/pref_names.h"
+
+const std::wstring kTestDir = L"encoding_tests";
+
+class BrowserEncodingTest : public UITest {
+ protected:
+  BrowserEncodingTest() : UITest() {}
+
+  // Make sure the content of the page are expected
+  // after override or auto-detect
+  void CheckFile(const std::wstring& generated_file,

[Johnny(Jianning) Ding, 2009/03/09 12:31:46]

would please move this function to UITest with adding one additional parameter
"test_dir_path". Then we can remove the  duplicated duplicate CheckFile both
encoding_uitest and savepage_uitest

+                 const std::wstring& expect_result_file,
+                 bool check_equal) {
+    std::wstring expect_result_filepath =
+        UITest::GetTestFilePath(kTestDir, expect_result_file);
+
+    ASSERT_TRUE(file_util::PathExists(expect_result_filepath));
+    WaitForGeneratedFileAndCheck(generated_file,
+                                 expect_result_filepath,
+                                 check_equal, true);
+  }
+
+  virtual void SetUp() {
+    UITest::SetUp();
+    EXPECT_TRUE(file_util::CreateNewTempDirectory(L"", &save_dir_));
+    save_dir_ += FilePath::kSeparators[0];
+  }
+
+  std::wstring save_dir_;
+  std::wstring encoding_;
+  GURL url;
+};
+
+// Below encodings are ignored because I don't have an editor to
+// create corresponding encoded test file for now :
+// GBK
+// Big5-HKSCS
+// ISO-2022-JP
+// ISO-8859-2
+// ISO-8859-3
+// ISO-8859-10
+// ISO-8859-14
+// ISO-8859-16

[Johnny(Jianning) Ding, 2009/03/09 12:31:46]

would you please file a bug for those encoding, the once you  have practicable
editor, you can add those tests.

+TEST_F(BrowserEncodingTest, TestEncodingAliasMapping) {
+  struct EncodingTestData {

[Johnny(Jianning) Ding, 2009/03/09 12:31:46]

you can move the definition to class BrowserEncodingTest, then we use can it in
those test functions.

+    const wchar_t* file_name;
+    const wchar_t* encoding_name;
+  };
+  
+  EncodingTestData encoding_test_data[] = {
+  { L"encoding-UTF-8.htm", L"UTF-8" },
+  { L"encoding-UTF-16LE.htm", L"UTF-16LE" },
+  { L"encoding-iso-8859-1.htm", L"ISO-8859-1" },
+  { L"encoding-windows-1252.htm", L"windows-1252" },
+  { L"encoding-gb18030.htm", L"gb18030" },
+  { L"encoding-Big5.htm", L"Big5" },
+  { L"encoding-windows-949.htm", L"windows-949" },
+  { L"encoding-Shift-JIS.htm", L"Shift_JIS" },
+  { L"encoding-EUC-JP.htm", L"EUC-JP" },
+  { L"encoding-windows-874.htm", L"windows-874" },
+  { L"encoding-ISO-8859-15.htm", L"ISO-8859-15" },
+  { L"encoding-macintosh.htm", L"macintosh" },
+  { L"encoding-ISO-8859-2.htm", L"ISO-8859-2" },
+  { L"encoding-windows-1250.htm", L"windows-1250" },
+  { L"encoding-ISO-8859-5.htm", L"ISO-8859-5" },
+  { L"encoding-windows-1251.htm", L"windows-1251" },
+  { L"encoding-KOI8-R.htm", L"KOI8-R" },
+  { L"encoding-KOI8-U.htm", L"KOI8-U" },
+  { L"encoding-ISO-8859-7.htm", L"ISO-8859-7" },
+  { L"encoding-windows-1253.htm", L"windows-1253" },
+  { L"encoding-windows-1254.htm", L"windows-1254" },
+  { L"encoding-ISO-8859-6.htm", L"ISO-8859-6" },
+  { L"encoding-windows-1256.htm", L"windows-1256" },
+  { L"encoding-ISO-8859-8.htm", L"ISO-8859-8" },
+  { L"encoding-windows-1255.htm", L"windows-1255" },
+  { L"encoding-windows-1258.htm", L"windows-1258" },
+  { L"encoding-ISO-8859-4.htm", L"ISO-8859-4" },
+  { L"encoding-ISO-8859-13.htm", L"ISO-8859-13" },
+  { L"encoding-windows-1257.htm", L"windows-1257" },
+  };
+
+  for (int i = 0;i < arraysize (encoding_test_data); i++) {
+    url = URLRequestMockHTTPJob::GetMockUrl(
+    kTestDir + L"/encoding_alias_mapping/" + encoding_test_data[i].file_name);

[Johnny(Jianning) Ding, 2009/03/09 12:31:46]

needs 4 spaces indent, you can write it like
url = URLRequestMockHTTPJob::GetMockUrl(kTestDir + 
    L"/encoding_alias_mapping/" + encoding_test_data[i].file_name);
or
url = URLRequestMockHTTPJob::GetMockUrl(kTestDir + 
                                        L"/encoding_alias_mapping/" +
                                        encoding_test_data[i].file_name);

+
+    scoped_ptr<TabProxy> tab(GetActiveTab());
+    ASSERT_TRUE(tab->NavigateToURL(url));
+    WaitUntilTabCount(1);
+
+    EXPECT_TRUE(tab->GetPageCurrentEncoding(&encoding_));
+    EXPECT_EQ(encoding_, encoding_test_data[i].encoding_name);
+  }
+}
+
+TEST_F(BrowserEncodingTest, TestOverrideEncoding) {
+  std::wstring file_name = L"gb18030_content_with_encoding_iso88591.htm";
+  std::wstring expect_file_name =
+    L"encoding_user_override\\expect_gb18030_content_save_from_iso88591.htm";
+
+  url = URLRequestMockHTTPJob::GetMockUrl(
+            kTestDir + L"/encoding_user_override/" + file_name);

[Johnny(Jianning) Ding, 2009/03/09 12:31:46]

4 spaces indent

+  scoped_ptr<TabProxy> tab(GetActiveTab());
+  ASSERT_TRUE(tab->NavigateToURL(url));
+  WaitUntilTabCount(1);
+
+  // Get the original defined encoding in the page
+  // std::wstring encoding;
+  EXPECT_TRUE(tab->GetPageCurrentEncoding(&encoding_));
+  EXPECT_EQ(encoding_, L"ISO-8859-1");
+
+  // Override the encoding to "gb18030".
+  int64 last_nav_time = 0;
+  EXPECT_TRUE(tab->GetLastNavigationTime(&last_nav_time));
+  EXPECT_TRUE(tab->OverrideEncoding(L"gb18030"));
+  EXPECT_TRUE(tab->WaitForNavigation(last_nav_time));
+
+  // Re-get the encoding of page. It should be gb18030.
+  EXPECT_TRUE(tab->GetPageCurrentEncoding(&encoding_));
+  EXPECT_EQ(encoding_, L"gb18030");
+
+  // Dump the page, the content of dump page should be equal with our expect
+  // result file gb18030_content_with_encoding_gb18030.htm.
+  std::wstring full_file_name = save_dir_ + file_name;
+  std::wstring dir = save_dir_ +
+      L"gb18030_content_with_encoding_iso88591_files";
+  EXPECT_TRUE(tab->SavePage(full_file_name, dir,
+                            SavePackage::SAVE_AS_COMPLETE_HTML));
+  EXPECT_TRUE(WaitForDownloadShelfVisible(tab.get()));
+  CheckFile(full_file_name, expect_file_name, true);
+}
+
+// Below encoding are disabled because it is known issue that
+// auto detect doesn't return right value for them
+// ISO-8859-4
+// ISO-8859-13
+// windows-1257
+// KOI8-U
+// macintosh
+// windows-1252
+

[Johnny(Jianning) Ding, 2009/03/09 12:31:46]

please remove this empty line.

+// Also disabled Thai and Vietnamese because auto detect doesn't work for them
+// windows-874
+// windows-1258
+// windows-1253
+

[Johnny(Jianning) Ding, 2009/03/09 12:31:46]

please remove this empty line.

+// For Hebrew, I made the expecting encoding value as ISO-8859-8-I because of
+// bug crbug.com/2927
+
+TEST_F(BrowserEncodingTest, TestEncodingAutoDetect) {
+  // Test file name
+  const wchar_t* file_name[] = {

[Johnny(Jianning) Ding, 2009/03/09 12:31:46]

please use the previous defined "EncodingTestData" structure instead of using
two separated string arrays.

+      L"Big5_content_without_encoding_declaration.htm",
+      L"gb18030_content_without_encoding_declaration.htm",
+      L"iso-8859-1_content_without_encoding_declaration.htm",
+      L"ISO-8859-5_content_without_encoding_declaration.htm",
+      L"ISO-8859-6_content_without_encoding_declaration.htm",
+      L"ISO-8859-7_content_without_encoding_declaration.htm",
+      L"ISO-8859-8_content_without_encoding_declaration.htm",
+      L"KOI8-R_content_without_encoding_declaration.htm",
+      L"Shift-JIS_content_without_encoding_declaration.htm",
+      L"UTF-8_content_without_encoding_declaration.htm",
+      L"windows-949_content_without_encoding_declaration.htm",
+      L"windows-1251_content_without_encoding_declaration.htm",
+      L"windows-1254_content_without_encoding_declaration.htm",
+      L"windows-1255_content_without_encoding_declaration.htm",
+      L"windows-1256_content_without_encoding_declaration.htm",
+  };
+
+  // Expect file directory and file name
+  std::wstring expect_folder_name = L"encoding_auto_detect\\expect_results\\";
+  const wchar_t* expect_file_name[] = {
+    L"expect_Big5_content_save_from_without_encoding_declaration.htm",
+    L"expect_gb18030_content_save_from_without_encoding_declaration.htm",
+    L"expect_iso-8859-1_content_save_from_without_encoding_declaration.htm",
+    L"expect_ISO-8859-5_content_save_from_without_encoding_declaration.htm",
+    L"expect_ISO-8859-6_content_save_from_without_encoding_declaration.htm",
+    L"expect_ISO-8859-7_content_save_from_without_encoding_declaration.htm",
+    L"expect_ISO-8859-8_content_save_from_without_encoding_declaration.htm",
+    L"expect_KOI8-R_content_save_from_without_encoding_declaration.htm",
+    L"expect_Shift-JIS_content_save_from_without_encoding_declaration.htm",
+    L"expect_UTF-8_content_save_from_without_encoding_declaration.htm",
+    L"expect_windows-949_content_save_from_without_encoding_declaration.htm",
+    L"expect_windows-1251_content_save_from_without_encoding_declaration.htm",
+    L"expect_windows-1254_content_save_from_without_encoding_declaration.htm",
+    L"expect_windows-1255_content_save_from_without_encoding_declaration.htm",
+    L"expect_windows-1256_content_save_from_without_encoding_declaration.htm",
+  };
+
+  // Real expect file name
+  std::wstring full_expect_file_name;
+
+  // Expected encoding
+  const wchar_t* expect_encoding[] = {L"Big5",
+                                      L"gb18030",
+                                      L"ISO-8859-1",
+                                      L"ISO-8859-5",
+                                      L"ISO-8859-6",
+                                      L"ISO-8859-7",
+                                      L"ISO-8859-8-I",
+                                      L"KOI8-R",
+                                      L"Shift_JIS",
+                                      L"UTF-8",
+                                      L"windows-949",
+                                      L"windows-1251",
+                                      L"windows-1254",
+                                      L"windows-1255",
+                                      L"windows-1256",
+  };
+
+  // Save as file directory
+  const wchar_t* save_dir_sub[] = {
+      L"Big5_content_without_encoding_declaration_files",

[Johnny(Jianning) Ding, 2009/03/09 12:31:46]

you can define the prefix name like EncodingTestData EncodingTestData
encoding_test_data[] = {
{ L"Big5_content_without_encoding_declaration", L"Big5" },
...
};
so 
test_file_name = encoding_test_data[i] + L".htm";
expact_file_name = L"expect_" + encoding_test_data[i] + L".htm";
sub_dir_name =  encoding_test_data[i] + L"_files";

it will make code more readable.

+      L"gb18030_content_without_encoding_declaration_files",
+      L"iso-8859-1_content_without_encoding_declaration_files",
+      L"ISO-8859-5_content_without_encoding_declaration_files",
+      L"ISO-8859-6_content_without_encoding_declaration_files",
+      L"ISO-8859-7_content_without_encoding_declaration_files",
+      L"ISO-8859-8_content_without_encoding_declaration_files",
+      L"KOI8-R_content_without_encoding_declaration_files",
+      L"Shift-JIS_content_without_encoding_declaration_files",
+      L"UTF-8_content_without_encoding_declaration_files",
+      L"windows-949_content_without_encoding_declaration_files",
+      L"windows-1251_content_without_encoding_declaration_files",
+      L"windows-1254_content_without_encoding_declaration_files",
+      L"windows-1255_content_without_encoding_declaration_files",
+      L"windows-1256_content_without_encoding_declaration_files",
+  };
+
+  // Real save as location: dir = save_dir_ + save_dir_sub[i];
+  std::wstring dir;
+
+  // Real save as file name: full_file_name = save_dir_ + file_name[i];
+  std::wstring full_file_name;
+
+  int64 last_nav_time;
+  bool encoding_auto_detect = false;
+  bool new_encoding_auto_detect = false;
+
+  for(int i = 0;i < arraysize (file_name);i++) {
+    scoped_ptr<BrowserProxy> browser(automation()->GetBrowserWindow(0));
+    ASSERT_TRUE(browser.get());
+
+    // Set the default charset to non-GB related encoding to make sure we
+    // incorrectly decode the page.
+    browser->SetStringPreference(prefs::kDefaultCharset, L"ISO-8859-1");
+
+    url = URLRequestMockHTTPJob::GetMockUrl(
+              kTestDir + L"/encoding_auto_detect/" + file_name[i]);

[Johnny(Jianning) Ding, 2009/03/09 12:31:46]

4 spaces indent

+    scoped_ptr<TabProxy> tab(GetActiveTab());
+    ASSERT_TRUE(tab->NavigateToURL(url));
+    WaitUntilTabCount(1);
+
+    // Disable auto detect if it is on
+    last_nav_time = 0;
+    EXPECT_TRUE(browser->SetBooleanPreference(
+                  prefs::kWebKitUsesUniversalDetector, encoding_auto_detect));
+    EXPECT_TRUE(tab->Reload());
+
+    // Get the encoding used for the page, it must be the default charset we
+    // just set
+    // std::wstring encoding;
+    EXPECT_TRUE(tab->GetPageCurrentEncoding(&encoding_));
+    EXPECT_EQ(encoding_, L"ISO-8859-1");
+
+    // Enable the encoding auto detection.
+    last_nav_time = 0;
+    EXPECT_TRUE(browser->SetBooleanPreference(
+                  prefs::kWebKitUsesUniversalDetector, !encoding_auto_detect));
+    EXPECT_TRUE(tab->Reload());
+
+    // Re-get the encoding of page. It should return the real encoding now.
+    EXPECT_TRUE(browser->GetBooleanPreference(
+              prefs::kWebKitUsesUniversalDetector, &new_encoding_auto_detect));
+    EXPECT_EQ(new_encoding_auto_detect, !encoding_auto_detect);
+    EXPECT_TRUE(tab->GetPageCurrentEncoding(&encoding_));
+    EXPECT_EQ(encoding_, expect_encoding[i]);
+
+    // Dump the page, the content of dump page should be equal with our expect
+    // result file gb18030_content_with_encoding_gb18030.htm.
+    full_file_name = save_dir_ + file_name[i];
+    dir = save_dir_ + save_dir_sub[i];
+    full_expect_file_name = expect_folder_name + expect_file_name[i];
+
+    EXPECT_TRUE(tab->SavePage(full_file_name, dir,
+                              SavePackage::SAVE_AS_COMPLETE_HTML));
+    EXPECT_TRUE(WaitForDownloadShelfVisible(tab.get()));
+
+    CheckFile(full_file_name, full_expect_file_name, true);
+  }
+}