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