Encode Subject header correctly per RFC 2047

Thee're some differences between RFC 2047 which we should be
used to encode header value and RFC 2045 for body:
1) Use CRLF+SPACE for soft line break.
2) SPACE and TAB should always be encoded.
3) Multiple encoded text should be used

Did manual test with FAR file manager w/ Observer plugin and
7-Zip w/ eDecoder plugin.

Bug: 794835
Change-Id: I5b87b7392d2208dd58bf512c7ee59c87bc32a85a
Reviewed-on: https://chromium-review.googlesource.com/835009
Reviewed-by: Xianzhu Wang <wangxianzhu@chromium.org>
Reviewed-by: Łukasz Anforowicz <lukasza@chromium.org>
Reviewed-by: Daniel Cheng <dcheng@chromium.org>
Commit-Queue: Jian Li <jianli@chromium.org>
Cr-Commit-Position: refs/heads/master@{#530371}

third_party/WebKit/Source/core/frame/MHTMLTest.cpp

@@ -157,13 +157,29 @@ class MHTMLTest : public ::testing::Test {
     LineReader line_reader(
         std::string(mhtml_data->data(), mhtml_data->length()));
     std::string line;
-    while (line_reader.GetNextLine(&line) && line.length()) {
+    line_reader.GetNextLine(&line);
+    while (line.length()) {
+      // Peek next line to see if it starts with soft line break. If yes, append
+      // to current line.
+      std::string next_line;
+      while (true) {
+        line_reader.GetNextLine(&next_line);
+        if (next_line.length() > 1 &&
+            (next_line[0] == ' ' || next_line[0] == '\t')) {
+          line += &(next_line.at(1));
+          continue;
+        }
+        break;
+      }
+
       std::string::size_type pos = line.find(':');
       if (pos == std::string::npos)
         continue;
       std::string key = line.substr(0, pos);
       std::string value = line.substr(pos + 2);
       mhtml_headers.emplace(key, value);
+
+      line = next_line;
     }
     return mhtml_headers;
   }
@@ -251,14 +267,16 @@ TEST_F(MHTMLTest, TestMHTMLHeadersWithTitleContainingAllPrintableCharacters) {
 
   EXPECT_EQ("<Saved by Blink>", mhtml_headers["From"]);
   EXPECT_FALSE(mhtml_headers["Date"].empty());
-  EXPECT_EQ("multipart/related;", mhtml_headers["Content-Type"]);
+  EXPECT_EQ(
+      "multipart/related;type=\"text/html\";boundary=\"boundary-example\"",
+      mhtml_headers["Content-Type"]);
   EXPECT_EQ("abc", mhtml_headers["Subject"]);
   EXPECT_EQ(kURL, mhtml_headers["Snapshot-Content-Location"]);
 }
 
 TEST_F(MHTMLTest, TestMHTMLHeadersWithTitleContainingNonPrintableCharacters) {
   const char kURL[] = "http://www.example.com/";
-  const char kTitle[] = u8"abc=\u261D\U0001F3FB";
+  const char kTitle[] = "abc \t=\xe2\x98\x9d\xf0\x9f\x8f\xbb";
   AddTestResources();
   scoped_refptr<RawData> data =
       Serialize(ToKURL(kURL), String::FromUTF8(kTitle), "text/html",
@@ -268,12 +286,43 @@ TEST_F(MHTMLTest, TestMHTMLHeadersWithTitleContainingNonPrintableCharacters) {
 
   EXPECT_EQ("<Saved by Blink>", mhtml_headers["From"]);
   EXPECT_FALSE(mhtml_headers["Date"].empty());
-  EXPECT_EQ("multipart/related;", mhtml_headers["Content-Type"]);
-  EXPECT_EQ("=?utf-8?Q?abc=3D=E2=98=9D=F0=9F=8F=BB?=",
+  EXPECT_EQ(
+      "multipart/related;type=\"text/html\";boundary=\"boundary-example\"",
+      mhtml_headers["Content-Type"]);
+  EXPECT_EQ("=?utf-8?Q?abc=20=09=3D=E2=98=9D=F0=9F=8F=BB?=",
             mhtml_headers["Subject"]);
   EXPECT_EQ(kURL, mhtml_headers["Snapshot-Content-Location"]);
 }
 
+TEST_F(MHTMLTest,
+       TestMHTMLHeadersWithLongTitleContainingNonPrintableCharacters) {
+  const char kURL[] = "http://www.example.com/";
+  const char kTitle[] =
+      "01234567890123456789012345678901234567890123456789"
+      "01234567890123456789012345678901234567890123456789"
+      " \t=\xe2\x98\x9d\xf0\x9f\x8f\xbb";
+  AddTestResources();
+  scoped_refptr<RawData> data =
+      Serialize(ToKURL(kURL), String::FromUTF8(kTitle), "text/html",
+                MHTMLArchive::kUseDefaultEncoding);
+
+  std::map<std::string, std::string> mhtml_headers = ExtractMHTMLHeaders(data);
+
+  EXPECT_EQ("<Saved by Blink>", mhtml_headers["From"]);
+  EXPECT_FALSE(mhtml_headers["Date"].empty());
+  EXPECT_EQ(
+      "multipart/related;type=\"text/html\";boundary=\"boundary-example\"",
+      mhtml_headers["Content-Type"]);
+  EXPECT_EQ(
+      "=?utf-8?Q?012345678901234567890123456789"
+      "012345678901234567890123456789012?="
+      "=?utf-8?Q?345678901234567890123456789"
+      "0123456789=20=09=3D=E2=98=9D=F0=9F?="
+      "=?utf-8?Q?=8F=BB?=",
+      mhtml_headers["Subject"]);
+  EXPECT_EQ(kURL, mhtml_headers["Snapshot-Content-Location"]);
+}
+
 TEST_F(MHTMLTest, TestMHTMLEncoding) {
   const char kURL[] = "http://www.example.com";
   AddTestResources();
@@ -466,4 +515,24 @@ TEST_F(MHTMLTest, FormControlElements) {
   EXPECT_FALSE(document->getElementById("fm")->IsDisabledFormControl());
 }
 
+TEST_F(MHTMLTest, LoadMHTMLContainingSoftLineBreaks) {
+  const char kURL[] = "http://www.example.com";
+
+  // Register the mocked frame and load it.
+  RegisterMockedURLLoad(kURL, "soft_line_break.mht");
+  LoadURLInTopFrame(ToKURL(kURL));
+  ASSERT_TRUE(GetPage());
+  LocalFrame* frame = ToLocalFrame(GetPage()->MainFrame());
+  ASSERT_TRUE(frame);
+  // We should not have problem to concatenate header lines separated by soft
+  // line breaks.
+  Document* document = frame->GetDocument();
+  ASSERT_TRUE(document);
+
+  // We should not have problem to concatenate body lines separated by soft
+  // line breaks.
+  EXPECT_TRUE(document->getElementById(
+      "AVeryLongID012345678901234567890123456789012345678901234567890End"));
+}
+
 }  // namespace blink

third_party/WebKit/Source/core/testing/data/mhtml/soft_line_break.mht

@@ -0,0 +1,27 @@
+From: <Saved by Blink>
+Subject: =?utf-8?Q?012345678901234567890123456789012345678901234567890
+	1234567890123456789012345678901234567890123456789
+	=3D=E2=98=9D=F0=9F=8F=BB?=
+Date: Thu, 4 Oct 2017 21:18:18 -0000
+MIME-Version: 1.0
+Content-Type: multipart/related;
+	type="text/html";
+	boundary="----MultipartBoundary--e77OylKXx1PBMEF67x53AwnQLf4DUmwdt037X9MjPK----"
+
+------MultipartBoundary--e77OylKXx1PBMEF67x53AwnQLf4DUmwdt037X9MjPK----
+Content-Type: text/html
+Content-ID: <frame-31894-fca076b5-329b-490d-a9ee-6974bf0c4bcd@mhtml.blink>
+Content-Transfer-Encoding: quoted-printable
+Content-Location: http://localhost/soft_line_break.html
+
+<html>
+<head><meta http-equiv=3D"Content-Type" content=3D"text/html; charset=
+=3DUTF-8">
+</head>
+<body>
+<div id=3D"AVeryLongID01234567890123456789012345678901234567890123456=
+7890End">
+</div>
+</body>
+</html>
+------MultipartBoundary--e77OylKXx1PBMEF67x53AwnQLf4DUmwdt037X9MjPK------

third_party/WebKit/Source/platform/mhtml/MHTMLArchive.cpp

@@ -48,9 +48,78 @@
 
 namespace blink {
 
-const char* const kQuotedPrintable = "quoted-printable";
-const char* const kBase64 = "base64";
-const char* const kBinary = "binary";
+namespace {
+
+const size_t kMaximumLineLength = 76;
+const char kCrlfLineEnding[] = "\r\n";
+
+const char kRFC2047EncodingPrefix[] = "=?utf-8?Q?";
+const size_t kRFC2047EncodingPrefixLength = 10;
+const char kRFC2047EncodingSuffix[] = "?=";
+const size_t kRFC2047EncodingSuffixLength = 2;
+
+const char kQuotedPrintable[] = "quoted-printable";
+const char kBase64[] = "base64";
+const char kBinary[] = "binary";
+
+}  // namespace
+
+// Controls quoted-printable encoding characters in body, per RFC 2045.
+class QuotedPrintableEncodeBodyDelegate : public QuotedPrintableEncodeDelegate {
+ public:
+  QuotedPrintableEncodeBodyDelegate() = default;
+  ~QuotedPrintableEncodeBodyDelegate() override = default;
+
+  size_t GetMaxLineLengthForEncodedContent() const override {
+    return kMaximumLineLength;
+  }
+
+  bool ShouldEncodeWhiteSpaceCharacters(bool end_of_line) const override {
+    // They should be encoded only if they appear at the end of a body line.
+    return end_of_line;
+  }
+
+  void DidStartLine(Vector<char>& out) override {
+    // Nothing to add.
+  }
+
+  void DidFinishLine(bool last_line, Vector<char>& out) override {
+    if (!last_line) {
+      out.push_back('=');
+      out.Append(kCrlfLineEnding, strlen(kCrlfLineEnding));
+    }
+  }
+};
+
+// Controls quoted-printable encoding characters in headers, per RFC 2047.
+class QuotedPrintableEncodeHeaderDelegate
+    : public QuotedPrintableEncodeDelegate {
+ public:
+  QuotedPrintableEncodeHeaderDelegate() = default;
+  ~QuotedPrintableEncodeHeaderDelegate() override = default;
+
+  size_t GetMaxLineLengthForEncodedContent() const override {
+    return kMaximumLineLength - kRFC2047EncodingPrefixLength -
+           kRFC2047EncodingSuffixLength;
+  }
+
+  bool ShouldEncodeWhiteSpaceCharacters(bool end_of_line) const override {
+    // They should always be encoded if they appear anywhere in the header.
+    return true;
+  }
+
+  void DidStartLine(Vector<char>& out) override {
+    out.Append(kRFC2047EncodingPrefix, kRFC2047EncodingPrefixLength);
+  }
+
+  void DidFinishLine(bool last_line, Vector<char>& out) override {
+    out.Append(kRFC2047EncodingSuffix, kRFC2047EncodingSuffixLength);
+    if (!last_line) {
+      out.Append(kCrlfLineEnding, strlen(kCrlfLineEnding));
+      out.push_back(' ');
+    }
+  }
+};
 
 static String ConvertToPrintableCharacters(const String& text) {
   // If the text contains all printable ASCII characters, no need for encoding.
@@ -70,9 +139,11 @@ static String ConvertToPrintableCharacters(const String& text) {
   // where, "utf-8" is the chosen charset to represent the text and "Q" is the
   // Quoted-Printable format to convert to 7-bit printable ASCII characters.
   CString utf8_text = text.Utf8();
+  QuotedPrintableEncodeHeaderDelegate header_delegate;
   Vector<char> encoded_text;
-  QuotedPrintableEncode(utf8_text.data(), utf8_text.length(), encoded_text);
-  return "=?utf-8?Q?" + String(encoded_text.data(), encoded_text.size()) + "?=";
+  QuotedPrintableEncode(utf8_text.data(), utf8_text.length(), &header_delegate,
+                        encoded_text);
+  return String(encoded_text.data(), encoded_text.size());
 }
 
 MHTMLArchive::MHTMLArchive() = default;
@@ -245,15 +316,15 @@ void MHTMLArchive::GenerateMHTMLPart(const String& boundary,
     size_t data_length = flat_data.size();
     Vector<char> encoded_data;
     if (!strcmp(content_encoding, kQuotedPrintable)) {
-      QuotedPrintableEncode(data, data_length, encoded_data);
+      QuotedPrintableEncodeBodyDelegate body_delegate;
+      QuotedPrintableEncode(data, data_length, &body_delegate, encoded_data);
       output_buffer.Append(encoded_data.data(), encoded_data.size());
       output_buffer.Append("\r\n", 2u);
     } else {
       DCHECK(!strcmp(content_encoding, kBase64));
       // We are not specifying insertLFs = true below as it would cut the lines
       // with LFs and MHTML requires CRLFs.
       Base64Encode(data, data_length, encoded_data);
-      const size_t kMaximumLineLength = 76;
       size_t index = 0;
       size_t encoded_data_length = encoded_data.size();
       do {

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

@@ -34,8 +34,6 @@
 
 namespace blink {
 
-static const size_t kMaximumLineLength = 76;
-
 static const char kCrlfLineEnding[] = "\r\n";
 
 static size_t LengthOfLineEndingAtIndex(const char* input,
@@ -54,15 +52,13 @@ static size_t LengthOfLineEndingAtIndex(const char* input,
   return 0;
 }
 
-void QuotedPrintableEncode(const Vector<char>& in, Vector<char>& out) {
-  QuotedPrintableEncode(in.data(), in.size(), out);
-}
-
 void QuotedPrintableEncode(const char* input,
                            size_t input_length,
+                           QuotedPrintableEncodeDelegate* delegate,
                            Vector<char>& out) {
   out.clear();
   out.ReserveCapacity(input_length);
+  delegate->DidStartLine(out);
   size_t current_line_length = 0;
   for (size_t i = 0; i < input_length; ++i) {
     bool is_last_character = (i == input_length - 1);
@@ -74,13 +70,14 @@ void QuotedPrintableEncode(const char* input,
         current_character != '\t')
       requires_encoding = true;
 
-    // Space and tab characters have to be encoded if they appear at the end of
-    // a line.
+    // Decide if space and tab characters need to be encoded.
     if (!requires_encoding &&
-        (current_character == '\t' || current_character == ' ') &&
-        (is_last_character ||
-         LengthOfLineEndingAtIndex(input, input_length, i + 1)))
-      requires_encoding = true;
+        (current_character == '\t' || current_character == ' ')) {
+      bool end_of_line = is_last_character ||
+                         LengthOfLineEndingAtIndex(input, input_length, i + 1);
+      requires_encoding =
+          delegate->ShouldEncodeWhiteSpaceCharacters(end_of_line);
+    }
 
     // End of line should be converted to CR-LF sequences.
     if (!is_last_character) {
@@ -103,10 +100,10 @@ void QuotedPrintableEncode(const char* input,
 
     // Insert a soft line break if necessary.
     if (current_line_length + length_of_encoded_character >
-        kMaximumLineLength) {
-      out.push_back('=');
-      out.Append(kCrlfLineEnding, strlen(kCrlfLineEnding));
+        delegate->GetMaxLineLengthForEncodedContent()) {
+      delegate->DidFinishLine(false /*last_line*/, out);
       current_line_length = 0;
+      delegate->DidStartLine(out);
     }
 
     // Finally, insert the actual character(s).
@@ -120,6 +117,7 @@ void QuotedPrintableEncode(const char* input,
       current_line_length++;
     }
   }
+  delegate->DidFinishLine(true /*last_line*/, out);
 }
 
 void QuotedPrintableDecode(const Vector<char>& in, Vector<char>& out) {

third_party/WebKit/Source/platform/text/QuotedPrintable.h

@@ -36,10 +36,38 @@
 
 namespace blink {
 
-PLATFORM_EXPORT void QuotedPrintableEncode(const Vector<char>&, Vector<char>&);
-PLATFORM_EXPORT void QuotedPrintableEncode(const char*, size_t, Vector<char>&);
+// Delegate for controling the behavior of quoted-printable encoding. The
+// original characters may be encoded a bit differently depending on where
+// they live, header or body. For example, "=CRLF" should be used to break
+// long line in body while "CRLF+SPACE" should be used to break long line in
+// header.
+class PLATFORM_EXPORT QuotedPrintableEncodeDelegate {
+ public:
+  QuotedPrintableEncodeDelegate() = default;
+  virtual ~QuotedPrintableEncodeDelegate() = default;
+
+  // Returns maximum number of characters allowed for an encoded line, excluding
+  // prefix and soft line break.
+  virtual size_t GetMaxLineLengthForEncodedContent() const = 0;
+
+  // Returns true if space and tab characters need to be encoded.
+  virtual bool ShouldEncodeWhiteSpaceCharacters(bool end_of_line) const = 0;
+
+  // Called when an encoded line starts. The delegate can take this chance to
+  // add any prefix.
+  virtual void DidStartLine(Vector<char>& out) = 0;
+
+  // Called when an encoded line ends. The delegate can take this chance to add
+  // any suffix. If it is not last line, a soft line break should also
+  // be added after the suffix.
+  virtual void DidFinishLine(bool last_line, Vector<char>& out) = 0;
+};
+
+PLATFORM_EXPORT void QuotedPrintableEncode(const char*,
+                                           size_t,
+                                           QuotedPrintableEncodeDelegate*,
+                                           Vector<char>&);
 
-PLATFORM_EXPORT void QuotedPrintableDecode(const Vector<char>&, Vector<char>&);
 PLATFORM_EXPORT void QuotedPrintableDecode(const char*, size_t, Vector<char>&);
 
 }  // namespace blink