net/base/data_url.cc

   1 // Copyright (c) 2012 The Chromium Authors. All rights reserved.
   2 // Use of this source code is governed by a BSD-style license that can be
   3 // found in the LICENSE file.
   4
   5 // NOTE: based loosely on mozilla's nsDataChannel.cpp
   6
   7 #include <algorithm>
   8
   9 #include "net/base/data_url.h"
  10
  11 #include "base/base64.h"
  12 #include "base/basictypes.h"
  13 #include "base/strings/string_split.h"
  14 #include "base/strings/string_util.h"
  15 #include "net/base/escape.h"
  16 #include "url/gurl.h"
  17
  18 namespace net {
  19
  20 // static
  21 bool DataURL::Parse(const GURL& url, std::string* mime_type,
  22                     std::string* charset, std::string* data) {
  23   DCHECK(mime_type->empty());
  24   DCHECK(charset->empty());
  25   std::string::const_iterator begin = url.spec().begin();
  26   std::string::const_iterator end = url.spec().end();
  27
  28   std::string::const_iterator after_colon = std::find(begin, end, ':');
  29   if (after_colon == end)
  30     return false;
  31   ++after_colon;
  32
  33   std::string::const_iterator comma = std::find(after_colon, end, ',');
  34   if (comma == end)
  35     return false;
  36
  37   std::vector<std::string> meta_data;
  38   std::string unparsed_meta_data(after_colon, comma);
  39   base::SplitString(unparsed_meta_data, ';', &meta_data);
  40
  41   std::vector<std::string>::iterator iter = meta_data.begin();
  42   if (iter != meta_data.end()) {
  43     mime_type->swap(*iter);
  44     StringToLowerASCII(mime_type);
  45     ++iter;
  46   }
  47
  48   static const char kBase64Tag[] = "base64";
  49   static const char kCharsetTag[] = "charset=";
  50   const size_t kCharsetTagLength = arraysize(kCharsetTag) - 1;
  51
  52   bool base64_encoded = false;
  53   for (; iter != meta_data.end(); ++iter) {
  54     if (!base64_encoded && *iter == kBase64Tag) {
  55       base64_encoded = true;
  56     } else if (charset->empty() &&
  57                iter->compare(0, kCharsetTagLength, kCharsetTag) == 0) {
  58       charset->assign(iter->substr(kCharsetTagLength));
  59     }
  60   }
  61
  62   // fallback to defaults if nothing specified in the URL:
  63   if (mime_type->empty())
  64     mime_type->assign("text/plain");
  65   if (charset->empty())
  66     charset->assign("US-ASCII");
  67
  68   // The caller may not be interested in receiving the data.
  69   if (!data)
  70     return true;
  71
  72   // Preserve spaces if dealing with text or xml input, same as mozilla:
  73   //   https://bugzilla.mozilla.org/show_bug.cgi?id=138052
  74   // but strip them otherwise:
  75   //   https://bugzilla.mozilla.org/show_bug.cgi?id=37200
  76   // (Spaces in a data URL should be escaped, which is handled below, so any
  77   // spaces now are wrong. People expect to be able to enter them in the URL
  78   // bar for text, and it can't hurt, so we allow it.)
  79   std::string temp_data = std::string(comma + 1, end);
  80
  81   // For base64, we may have url-escaped whitespace which is not part
  82   // of the data, and should be stripped. Otherwise, the escaped whitespace
  83   // could be part of the payload, so don't strip it.
  84   if (base64_encoded) {
  85     temp_data = UnescapeURLComponent(temp_data,
  86         UnescapeRule::SPACES | UnescapeRule::URL_SPECIAL_CHARS |
  87         UnescapeRule::CONTROL_CHARS);
  88   }
  89
  90   // Strip whitespace.
  91   if (base64_encoded || !(mime_type->compare(0, 5, "text/") == 0 ||
  92                           mime_type->find("xml") != std::string::npos)) {
  93     temp_data.erase(std::remove_if(temp_data.begin(), temp_data.end(),
  94                                    IsAsciiWhitespace<wchar_t>),
  95                     temp_data.end());
  96   }
  97
  98   if (!base64_encoded) {
  99     temp_data = UnescapeURLComponent(temp_data,
 100         UnescapeRule::SPACES | UnescapeRule::URL_SPECIAL_CHARS |
 101         UnescapeRule::CONTROL_CHARS);
 102   }
 103
 104   if (base64_encoded) {
 105     size_t length = temp_data.length();
 106     size_t padding_needed = 4 - (length % 4);
 107     // If the input wasn't padded, then we pad it as necessary until we have a
 108     // length that is a multiple of 4 as required by our decoder. We don't
 109     // correct if the input was incorrectly padded. If |padding_needed| == 3,
 110     // then the input isn't well formed and decoding will fail with or without
 111     // padding.
 112     if ((padding_needed == 1 || padding_needed == 2) &&
 113         temp_data[length - 1] != '=') {
 114       temp_data.resize(length + padding_needed, '=');
 115     }
 116     return base::Base64Decode(temp_data, data);
 117   }
 118
 119   temp_data.swap(*data);
 120   return true;
 121 }
 122
 123 }  // namespace net