Fixed ExtractSearchTermsFromURL for search engines with encoding != "UTF-8".

components/search_engines/template_url_unittest.cc

Index: components/search_engines/template_url_unittest.cc
diff --git a/components/search_engines/template_url_unittest.cc b/components/search_engines/template_url_unittest.cc
index 6d09bf89c38f3db31adc90576fe14af1150ab690..7194a05894c47066873cf56a05f30f0ccccf52b3 100644
--- a/components/search_engines/template_url_unittest.cc
+++ b/components/search_engines/template_url_unittest.cc
@@ -1142,6 +1142,56 @@ TEST_F(TemplateURLTest, ExtractSearchTermsFromURLPath) {
   EXPECT_EQ(base::string16(), result);
 }
 
+TEST_F(TemplateURLTest, ExtractSearchTermsFromURL_CyrillicInWindows1251) {

[Peter Kasting, 2015/04/13 18:59:48]

Nit: I would name this "ExtraceSearchTermsFromNonUTF8URL" and the next one
"ExtraceSearchTermsFromUTF8URL".

Add comments on these describing what they're trying to test.

[Vitaly Baranov, 2015/04/14 13:26:15]

Done.

+  TemplateURLData data;
+  data.SetURL("http://windows-1251.ru/?q={searchTerms}");
+  data.alternate_urls.push_back("http://windows-1251.ru/#q={searchTerms}");
+  data.alternate_urls.push_back("http://windows-1251.ru/path/{searchTerms}");
+  data.input_encodings.push_back("windows-1251");
+  TemplateURL url(data);
+  base::string16 result;
+
+  EXPECT_TRUE(url.ExtractSearchTermsFromURL(
+      GURL("http://windows-1251.ru/?q="
+           "%C7%E4%F0%E0%E2%F1%F2%E2%F3%E9%2C+%EC%E8%F0!"),
+      search_terms_data_, &result));
+  EXPECT_EQ(base::UTF8ToUTF16("Здравствуй, мир!"), result);

[Peter Kasting, 2015/04/13 18:59:48]

Don't put extended characters directly into source code -- use C++ escaped
values.

That unfortunately makes this less readable, but it's safer and less likely to
break on systems with different default encodings.

[Vitaly Baranov, 2015/04/14 13:26:15]

Done.

+
+  EXPECT_TRUE(url.ExtractSearchTermsFromURL(
+      GURL("http://windows-1251.ru/#q=%E4%E2%E0+%F1%EB%EE%E2%E0"),
+      search_terms_data_, &result));
+  EXPECT_EQ(base::UTF8ToUTF16("два слова"), result);
+
+  EXPECT_TRUE(url.ExtractSearchTermsFromURL(
+      GURL("http://windows-1251.ru/path/%E1%F3%EA%E2%FB%20%C0%20%E8%20A"),
+      search_terms_data_, &result));
+  EXPECT_EQ(base::UTF8ToUTF16("буквы А и A"), result);
+}
+
+TEST_F(TemplateURLTest, ExtractSearchTermsFromURL_CyrillicInUTF8) {
+  TemplateURLData data;
+  data.SetURL("http://utf-8.ru/?q={searchTerms}");
+  data.alternate_urls.push_back("http://utf-8.ru/#q={searchTerms}");
+  data.alternate_urls.push_back("http://utf-8.ru/path/{searchTerms}");
+  TemplateURL url(data);
+  base::string16 result;
+
+  EXPECT_TRUE(url.ExtractSearchTermsFromURL(
+      GURL("http://utf-8.ru/?q=Здравствуй%2C+мир!"),
+      search_terms_data_, &result));
+  EXPECT_EQ(base::UTF8ToUTF16("Здравствуй, мир!"), result);
+
+  EXPECT_TRUE(url.ExtractSearchTermsFromURL(
+      GURL("http://utf-8.ru/#q=два+слова"),
+      search_terms_data_, &result));
+  EXPECT_EQ(base::UTF8ToUTF16("два слова"), result);
+
+  EXPECT_TRUE(url.ExtractSearchTermsFromURL(
+      GURL("http://utf-8.ru/path/буквы%20А%20и%20A"),
+      search_terms_data_, &result));
+  EXPECT_EQ(base::UTF8ToUTF16("буквы А и A"), result);
+}
+
 TEST_F(TemplateURLTest, HasSearchTermsReplacementKey) {
   TemplateURLData data;
   data.SetURL("http://google.com/?q={searchTerms}");