url/url_canon_fileurl.cc

   1 // Copyright 2013 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 // Functions for canonicalizing "file:" URLs.
   6
   7 #include "url/url_canon.h"
   8 #include "url/url_canon_internal.h"
   9 #include "url/url_file.h"
  10 #include "url/url_parse_internal.h"
  11
  12 namespace url {
  13
  14 namespace {
  15
  16 #ifdef WIN32
  17
  18 // Given a pointer into the spec, this copies and canonicalizes the drive
  19 // letter and colon to the output, if one is found. If there is not a drive
  20 // spec, it won't do anything. The index of the next character in the input
  21 // spec is returned (after the colon when a drive spec is found, the begin
  22 // offset if one is not).
  23 template<typename CHAR>
  24 int FileDoDriveSpec(const CHAR* spec, int begin, int end,
  25                     CanonOutput* output) {
  26   // The path could be one of several things: /foo/bar, c:/foo/bar, /c:/foo,
  27   // (with backslashes instead of slashes as well).
  28   int num_slashes = CountConsecutiveSlashes(spec, begin, end);
  29   int after_slashes = begin + num_slashes;
  30
  31   if (!DoesBeginWindowsDriveSpec(spec, after_slashes, end))
  32     return begin;  // Haven't consumed any characters
  33
  34   // A drive spec is the start of a path, so we need to add a slash for the
  35   // authority terminator (typically the third slash).
  36   output->push_back('/');
  37
  38   // DoesBeginWindowsDriveSpec will ensure that the drive letter is valid
  39   // and that it is followed by a colon/pipe.
  40
  41   // Normalize Windows drive letters to uppercase
  42   if (spec[after_slashes] >= 'a' && spec[after_slashes] <= 'z')
  43     output->push_back(static_cast<char>(spec[after_slashes] - 'a' + 'A'));
  44   else
  45     output->push_back(static_cast<char>(spec[after_slashes]));
  46
  47   // Normalize the character following it to a colon rather than pipe.
  48   output->push_back(':');
  49   return after_slashes + 2;
  50 }
  51
  52 #endif  // WIN32
  53
  54 template<typename CHAR, typename UCHAR>
  55 bool DoFileCanonicalizePath(const CHAR* spec,
  56                             const Component& path,
  57                             CanonOutput* output,
  58                             Component* out_path) {
  59   // Copies and normalizes the "c:" at the beginning, if present.
  60   out_path->begin = output->length();
  61   int after_drive;
  62 #ifdef WIN32
  63   after_drive = FileDoDriveSpec(spec, path.begin, path.end(), output);
  64 #else
  65   after_drive = path.begin;
  66 #endif
  67
  68   // Copies the rest of the path, starting from the slash following the
  69   // drive colon (if any, Windows only), or the first slash of the path.
  70   bool success = true;
  71   if (after_drive < path.end()) {
  72     // Use the regular path canonicalizer to canonicalize the rest of the
  73     // path. Give it a fake output component to write into. DoCanonicalizeFile
  74     // will compute the full path component.
  75     Component sub_path = MakeRange(after_drive, path.end());
  76     Component fake_output_path;
  77     success = CanonicalizePath(spec, sub_path, output, &fake_output_path);
  78   } else {
  79     // No input path, canonicalize to a slash.
  80     output->push_back('/');
  81   }
  82
  83   out_path->len = output->length() - out_path->begin;
  84   return success;
  85 }
  86
  87 template<typename CHAR, typename UCHAR>
  88 bool DoCanonicalizeFileURL(const URLComponentSource<CHAR>& source,
  89                            const Parsed& parsed,
  90                            CharsetConverter* query_converter,
  91                            CanonOutput* output,
  92                            Parsed* new_parsed) {
  93   // Things we don't set in file: URLs.
  94   new_parsed->username = Component();
  95   new_parsed->password = Component();
  96   new_parsed->port = Component();
  97
  98   // Scheme (known, so we don't bother running it through the more
  99   // complicated scheme canonicalizer).
 100   new_parsed->scheme.begin = output->length();
 101   output->Append("file://", 7);
 102   new_parsed->scheme.len = 4;
 103
 104   // Append the host. For many file URLs, this will be empty. For UNC, this
 105   // will be present.
 106   // TODO(brettw) This doesn't do any checking for host name validity. We
 107   // should probably handle validity checking of UNC hosts differently than
 108   // for regular IP hosts.
 109   bool success = CanonicalizeHost(source.host, parsed.host,
 110                                   output, &new_parsed->host);
 111   success &= DoFileCanonicalizePath<CHAR, UCHAR>(source.path, parsed.path,
 112                                     output, &new_parsed->path);
 113   CanonicalizeQuery(source.query, parsed.query, query_converter,
 114                     output, &new_parsed->query);
 115
 116   // Ignore failure for refs since the URL can probably still be loaded.
 117   CanonicalizeRef(source.ref, parsed.ref, output, &new_parsed->ref);
 118
 119   return success;
 120 }
 121
 122 } // namespace
 123
 124 bool CanonicalizeFileURL(const char* spec,
 125                          int spec_len,
 126                          const Parsed& parsed,
 127                          CharsetConverter* query_converter,
 128                          CanonOutput* output,
 129                          Parsed* new_parsed) {
 130   return DoCanonicalizeFileURL<char, unsigned char>(
 131       URLComponentSource<char>(spec), parsed, query_converter,
 132       output, new_parsed);
 133 }
 134
 135 bool CanonicalizeFileURL(const base::char16* spec,
 136                          int spec_len,
 137                          const Parsed& parsed,
 138                          CharsetConverter* query_converter,
 139                          CanonOutput* output,
 140                          Parsed* new_parsed) {
 141   return DoCanonicalizeFileURL<base::char16, base::char16>(
 142       URLComponentSource<base::char16>(spec), parsed, query_converter,
 143       output, new_parsed);
 144 }
 145
 146 bool FileCanonicalizePath(const char* spec,
 147                           const Component& path,
 148                           CanonOutput* output,
 149                           Component* out_path) {
 150   return DoFileCanonicalizePath<char, unsigned char>(spec, path,
 151                                                      output, out_path);
 152 }
 153
 154 bool FileCanonicalizePath(const base::char16* spec,
 155                           const Component& path,
 156                           CanonOutput* output,
 157                           Component* out_path) {
 158   return DoFileCanonicalizePath<base::char16, base::char16>(spec, path,
 159                                                             output, out_path);
 160 }
 161
 162 bool ReplaceFileURL(const char* base,
 163                     const Parsed& base_parsed,
 164                     const Replacements<char>& replacements,
 165                     CharsetConverter* query_converter,
 166                     CanonOutput* output,
 167                     Parsed* new_parsed) {
 168   URLComponentSource<char> source(base);
 169   Parsed parsed(base_parsed);
 170   SetupOverrideComponents(base, replacements, &source, &parsed);
 171   return DoCanonicalizeFileURL<char, unsigned char>(
 172       source, parsed, query_converter, output, new_parsed);
 173 }
 174
 175 bool ReplaceFileURL(const char* base,
 176                     const Parsed& base_parsed,
 177                     const Replacements<base::char16>& replacements,
 178                     CharsetConverter* query_converter,
 179                     CanonOutput* output,
 180                     Parsed* new_parsed) {
 181   RawCanonOutput<1024> utf8;
 182   URLComponentSource<char> source(base);
 183   Parsed parsed(base_parsed);
 184   SetupUTF16OverrideComponents(base, replacements, &utf8, &source, &parsed);
 185   return DoCanonicalizeFileURL<char, unsigned char>(
 186       source, parsed, query_converter, output, new_parsed);
 187 }
 188
 189 }  // namespace url