1// Copyright (c) 2006-2008 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/string_util.h"
13#include "googleurl/src/gurl.h"
14#include "net/base/escape.h"
15
16namespace net {
17
18// static
19bool DataURL::Parse(const GURL& url, std::string* mime_type,
20                    std::string* charset, std::string* data) {
21  std::string::const_iterator begin = url.spec().begin();
22  std::string::const_iterator end = url.spec().end();
23
24  std::string::const_iterator after_colon = std::find(begin, end, ':');
25  if (after_colon == end)
26    return false;
27  ++after_colon;
28
29  // first, find the start of the data
30  std::string::const_iterator comma = std::find(after_colon, end, ',');
31  if (comma == end)
32    return false;
33
34  const char kBase64Tag[] = ";base64";
35  std::string::const_iterator it =
36      std::search(after_colon, comma, kBase64Tag,
37                  kBase64Tag + sizeof(kBase64Tag)-1);
38
39  bool base64_encoded = (it != comma);
40
41  if (comma != after_colon) {
42    // everything else is content type
43    std::string::const_iterator semi_colon = std::find(after_colon, comma, ';');
44    if (semi_colon != after_colon) {
45      mime_type->assign(after_colon, semi_colon);
46      StringToLowerASCII(mime_type);
47    }
48    if (semi_colon != comma) {
49      const char kCharsetTag[] = "charset=";
50      it = std::search(semi_colon + 1, comma, kCharsetTag,
51                       kCharsetTag + sizeof(kCharsetTag)-1);
52      if (it != comma)
53        charset->assign(it + sizeof(kCharsetTag)-1, comma);
54    }
55  }
56
57  // fallback to defaults if nothing specified in the URL:
58  if (mime_type->empty())
59    mime_type->assign("text/plain");
60  if (charset->empty())
61    charset->assign("US-ASCII");
62
63  // The caller may not be interested in receiving the data.
64  if (!data)
65    return true;
66
67  // Preserve spaces if dealing with text or xml input, same as mozilla:
68  //   https://bugzilla.mozilla.org/show_bug.cgi?id=138052
69  // but strip them otherwise:
70  //   https://bugzilla.mozilla.org/show_bug.cgi?id=37200
71  // (Spaces in a data URL should be escaped, which is handled below, so any
72  // spaces now are wrong. People expect to be able to enter them in the URL
73  // bar for text, and it can't hurt, so we allow it.)
74  std::string temp_data = std::string(comma + 1, end);
75
76  // For base64, we may have url-escaped whitespace which is not part
77  // of the data, and should be stripped. Otherwise, the escaped whitespace
78  // could be part of the payload, so don't strip it.
79  if (base64_encoded) {
80    temp_data = UnescapeURLComponent(temp_data,
81        UnescapeRule::SPACES | UnescapeRule::URL_SPECIAL_CHARS |
82        UnescapeRule::CONTROL_CHARS);
83  }
84
85  // Strip whitespace.
86  if (base64_encoded || !(mime_type->compare(0, 5, "text/") == 0 ||
87                          mime_type->find("xml") != std::string::npos)) {
88    temp_data.erase(std::remove_if(temp_data.begin(), temp_data.end(),
89                                   IsAsciiWhitespace<wchar_t>),
90                    temp_data.end());
91  }
92
93  if (!base64_encoded) {
94    temp_data = UnescapeURLComponent(temp_data,
95        UnescapeRule::SPACES | UnescapeRule::URL_SPECIAL_CHARS |
96        UnescapeRule::CONTROL_CHARS);
97  }
98
99  if (base64_encoded)
100    return base::Base64Decode(temp_data, data);
101
102  temp_data.swap(*data);
103  return true;
104}
105
106}  // namespace net
107