Implement a parser for Content-Disposition header fields

third_party/WebKit/Source/platform/network/ParsedContentType.cpp

Index: third_party/WebKit/Source/platform/network/ParsedContentType.cpp
diff --git a/third_party/WebKit/Source/platform/network/ParsedContentType.cpp b/third_party/WebKit/Source/platform/network/ParsedContentType.cpp
index 6755f4a8e7adf367380744d3112356209625615e..a80114072d062c8a768c380ee167f06a352e2a50 100644
--- a/third_party/WebKit/Source/platform/network/ParsedContentType.cpp
+++ b/third_party/WebKit/Source/platform/network/ParsedContentType.cpp
@@ -31,148 +31,12 @@
 
 #include "platform/network/ParsedContentType.h"
 
+#include "platform/network/ContentHeaderFieldInput.h"
 #include "platform/wtf/text/StringBuilder.h"
 #include "platform/wtf/text/StringView.h"
 
 namespace blink {
 
-using Mode = ParsedContentType::Mode;
-
-namespace {
-
-bool IsTokenCharacter(Mode mode, UChar c) {
-  if (c >= 128)
-    return false;
-  if (c < 0x20)
-    return false;
-
-  switch (c) {
-    case ' ':
-    case ';':
-    case '"':
-      return false;
-    case '(':
-    case ')':
-    case '<':
-    case '>':
-    case '@':
-    case ',':
-    case ':':
-    case '\\':
-    case '/':
-    case '[':
-    case ']':
-    case '?':
-    case '=':
-      return mode == Mode::kRelaxed;
-    default:
-      return true;
-  }
-}
-
-bool Consume(char c, const String& input, unsigned& index) {
-  DCHECK_NE(c, ' ');
-  while (index < input.length() && input[index] == ' ')
-    ++index;
-
-  if (index < input.length() && input[index] == c) {
-    ++index;
-    return true;
-  }
-  return false;
-}
-
-bool ConsumeToken(Mode mode,
-                  const String& input,
-                  unsigned& index,
-                  StringView& output) {
-  DCHECK(output.IsNull());
-
-  while (index < input.length() && input[index] == ' ')
-    ++index;
-
-  auto start = index;
-  while (index < input.length() && IsTokenCharacter(mode, input[index]))
-    ++index;
-
-  if (start == index)
-    return false;
-
-  output = StringView(input, start, index - start);
-  return true;
-}
-
-bool ConsumeQuotedString(const String& input, unsigned& index, String& output) {
-  StringBuilder builder;
-  DCHECK_EQ('"', input[index]);
-  ++index;
-  while (index < input.length()) {
-    if (input[index] == '\\') {
-      ++index;
-      if (index == input.length())
-        return false;
-      builder.Append(input[index]);
-      ++index;
-      continue;
-    }
-    if (input[index] == '"') {
-      output = builder.ToString();
-      ++index;
-      return true;
-    }
-    builder.Append(input[index]);
-    ++index;
-  }
-  return false;
-}
-
-bool ConsumeTokenOrQuotedString(Mode mode,
-                                const String& input,
-                                unsigned& index,
-                                String& output) {
-  while (index < input.length() && input[index] == ' ')
-    ++index;
-  if (input.length() == index)
-    return false;
-  if (input[index] == '"') {
-    return ConsumeQuotedString(input, index, output);
-  }
-  StringView view;
-  auto result = ConsumeToken(mode, input, index, view);
-  output = view.ToString();
-  return result;
-}
-
-bool IsEnd(const String& input, unsigned index) {
-  while (index < input.length()) {
-    if (input[index] != ' ')
-      return false;
-    ++index;
-  }
-  return true;
-}
-
-}  // namespace
-
-ParsedContentType::ParsedContentType(const String& content_type, Mode mode)
-    : mode_(mode) {
-  is_valid_ = Parse(content_type);
-}
-
-String ParsedContentType::Charset() const {
-  return ParameterValueForName("charset");
-}
-
-String ParsedContentType::ParameterValueForName(const String& name) const {
-  if (!name.ContainsOnlyASCII())
-    return String();
-  return parameters_.at(name.DeprecatedLower());
-}
-
-size_t ParsedContentType::ParameterCount() const {
-  return parameters_.size();
-}
-
 // From http://tools.ietf.org/html/rfc2045#section-5.1:
 //
 // content := "Content-Type" ":" type "/" subtype
@@ -218,22 +82,21 @@ size_t ParsedContentType::ParameterCount() const {
 //               "/" / "[" / "]" / "?" / "="
 //               ; Must be in quoted-string,
 //               ; to use within parameter values
-
-bool ParsedContentType::Parse(const String& content_type) {
-  unsigned index = 0;
+ParsedContentType::ParsedContentType(const String& content_type, Mode mode) {
+  ContentHeaderFieldInput input(content_type);
 
   StringView type, subtype;
-  if (!ConsumeToken(Mode::kNormal, content_type, index, type)) {
+  if (!input.ConsumeToken(Mode::kNormal, type)) {
     DVLOG(1) << "Failed to find `type' in '" << content_type << "'";
-    return false;
+    return;
   }
-  if (!Consume('/', content_type, index)) {
+  if (!input.Consume('/')) {
     DVLOG(1) << "Failed to find '/' in '" << content_type << "'";
-    return false;
+    return;
   }
-  if (!ConsumeToken(Mode::kNormal, content_type, index, subtype)) {
-    DVLOG(1) << "Failed to find `type' in '" << content_type << "'";
-    return false;
+  if (!input.ConsumeToken(Mode::kNormal, subtype)) {
+    DVLOG(1) << "Failed to find `subtype' in '" << content_type << "'";
+    return;
   }
 
   StringBuilder builder;
@@ -242,40 +105,11 @@ bool ParsedContentType::Parse(const String& content_type) {
   builder.Append(subtype);
   mime_type_ = builder.ToString();
 
-  KeyValuePairs map;
-  while (!IsEnd(content_type, index)) {
-    if (!Consume(';', content_type, index)) {
-      DVLOG(1) << "Failed to find ';'";
-      return false;
-    }
+  parameters_.ParseParameters(std::move(input), mode);
+}
 
-    StringView key;
-    String value;
-    if (!ConsumeToken(Mode::kNormal, content_type, index, key)) {
-      DVLOG(1) << "Invalid Content-Type parameter name. (at " << index << ")";
-      return false;
-    }
-    if (!Consume('=', content_type, index)) {
-      DVLOG(1) << "Failed to find '='";
-      return false;
-    }
-    if (!ConsumeTokenOrQuotedString(mode_, content_type, index, value)) {
-      DVLOG(1) << "Invalid Content-Type, invalid parameter value (at " << index
-               << ", for '" << key.ToString() << "').";
-      return false;
-    }
-    // As |key| is parsed as a token, it consists of ascii characters
-    // and hence we don't need to care about non-ascii lowercasing.
-    DCHECK(key.ToString().ContainsOnlyASCII());
-    String key_string = key.ToString().DeprecatedLower();
-    if (mode_ == Mode::kStrict && map.find(key_string) != map.end()) {
-      DVLOG(1) << "Parameter " << key_string << " is defined multiple times.";
-      return false;
-    }
-    map.Set(key_string, value);
-  }
-  parameters_ = std::move(map);
-  return true;
+String ParsedContentType::Charset() const {
+  return ParameterValueForName("charset");
 }
 
 }  // namespace blink