Merge "Pass more hints to encoding detector." to M57 branch

third_party/WebKit/Source/platform/text/TextEncodingDetectorTest.cpp

Index: third_party/WebKit/Source/platform/text/TextEncodingDetectorTest.cpp
diff --git a/third_party/WebKit/Source/platform/text/TextEncodingDetectorTest.cpp b/third_party/WebKit/Source/platform/text/TextEncodingDetectorTest.cpp
index f768df9b1d6bc5c85f1e5e1891fac1b7cec7c728..1900b172b876617ce573f891f67f90b49bb727b3 100644
--- a/third_party/WebKit/Source/platform/text/TextEncodingDetectorTest.cpp
+++ b/third_party/WebKit/Source/platform/text/TextEncodingDetectorTest.cpp
@@ -16,7 +16,7 @@ TEST(TextResourceDecoderTest, RespectIso2022Jp) {
       "$BKL3$F;F|K\\%O%`%U%!%$%?!<%:$,%=%U%H%P%s%/$H$N%W%l!<%*%U$r@)$7!\"";
   WTF::TextEncoding encoding;
   bool result = detectTextEncoding(iso2022jp.c_str(), iso2022jp.length(),
-                                   nullptr, &encoding);
+                                   nullptr, nullptr, nullptr, &encoding);
   EXPECT_TRUE(result);
   EXPECT_EQ(WTF::TextEncoding("ISO-2022-JP"), encoding);
 }
@@ -28,9 +28,49 @@ TEST(TextResourceDecoderTest, Ignore7BitEncoding) {
       " ~{\x54\x42\x31\x7D\x37\x22\x55\x39\x35\x3D\x3D\x71~} abc";
   WTF::TextEncoding encoding;
   bool result = detectTextEncoding(hzGb2312.c_str(), hzGb2312.length(), nullptr,
-                                   &encoding);
+                                   nullptr, nullptr, &encoding);
   EXPECT_TRUE(result);
   EXPECT_EQ(WTF::TextEncoding("US-ASCII"), encoding);
 }
 
+TEST(TextEncodingDetectorTest, UrlHintHelpsEUCJP) {
+  std::string eucjpBytes =
+      "<TITLE>"
+      "\xA5\xD1\xA5\xEF\xA1\xBC\xA5\xC1\xA5\xE3\xA1\xBC\xA5\xC8\xA1\xC3\xC5\xEA"
+      "\xBB\xF1\xBE\xF0\xCA\xF3\xA4\xCE\xA5\xD5\xA5\xA3\xA5\xB9\xA5\xB3</"
+      "TITLE>";
+  WTF::TextEncoding encoding;
+  bool result = detectTextEncoding(eucjpBytes.c_str(), eucjpBytes.length(),
+                                   nullptr, nullptr, nullptr, &encoding);
+  EXPECT_TRUE(result);
+  EXPECT_EQ(WTF::TextEncoding("GBK"), encoding)
+      << "Without language hint, it's detected as GBK";
+
+  result = detectTextEncoding(eucjpBytes.c_str(), eucjpBytes.length(), nullptr,
+                              "http://example.co.jp/", nullptr, &encoding);
+  EXPECT_TRUE(result);
+  EXPECT_EQ(WTF::TextEncoding("EUC-JP"), encoding)
+      << "With URL hint including '.jp', it's detected as EUC-JP";
+}
+
+TEST(TextEncodingDetectorTest, LanguageHintHelpsEUCJP) {
+  std::string eucjpBytes =
+      "<TITLE>"
+      "\xA5\xD1\xA5\xEF\xA1\xBC\xA5\xC1\xA5\xE3\xA1\xBC\xA5\xC8\xA1\xC3\xC5\xEA"
+      "\xBB\xF1\xBE\xF0\xCA\xF3\xA4\xCE\xA5\xD5\xA5\xA3\xA5\xB9\xA5\xB3</"
+      "TITLE>";
+  WTF::TextEncoding encoding;
+  bool result = detectTextEncoding(eucjpBytes.c_str(), eucjpBytes.length(),
+                                   nullptr, nullptr, nullptr, &encoding);
+  EXPECT_TRUE(result);
+  EXPECT_EQ(WTF::TextEncoding("GBK"), encoding)
+      << "Without language hint, it's detected as GBK";
+
+  result = detectTextEncoding(eucjpBytes.c_str(), eucjpBytes.length(), nullptr,
+                              nullptr, "ja", &encoding);
+  EXPECT_TRUE(result);
+  EXPECT_EQ(WTF::TextEncoding("EUC-JP"), encoding)
+      << "With language hint 'ja', it's detected as EUC-JP";
+}
+
 }  // namespace blink