url/url_util.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 #include "url/url_util.h"
   6
   7 #include <string.h>
   8 #include <vector>
   9
  10 #include "base/logging.h"
  11 #include "url/url_canon_internal.h"
  12 #include "url/url_file.h"
  13 #include "url/url_util_internal.h"
  14
  15 namespace url_util {
  16
  17 const char kFileScheme[] = "file";
  18 const char kFileSystemScheme[] = "filesystem";
  19 const char kMailtoScheme[] = "mailto";
  20
  21 namespace {
  22
  23 // ASCII-specific tolower.  The standard library's tolower is locale sensitive,
  24 // so we don't want to use it here.
  25 template <class Char> inline Char ToLowerASCII(Char c) {
  26   return (c >= 'A' && c <= 'Z') ? (c + ('a' - 'A')) : c;
  27 }
  28
  29 // Backend for LowerCaseEqualsASCII.
  30 template<typename Iter>
  31 inline bool DoLowerCaseEqualsASCII(Iter a_begin, Iter a_end, const char* b) {
  32   for (Iter it = a_begin; it != a_end; ++it, ++b) {
  33     if (!*b || ToLowerASCII(*it) != *b)
  34       return false;
  35   }
  36   return *b == 0;
  37 }
  38
  39 const int kNumStandardURLSchemes = 8;
  40 const char* kStandardURLSchemes[kNumStandardURLSchemes] = {
  41   "http",
  42   "https",
  43   kFileScheme,  // Yes, file urls can have a hostname!
  44   "ftp",
  45   "gopher",
  46   "ws",  // WebSocket.
  47   "wss",  // WebSocket secure.
  48   kFileSystemScheme,
  49 };
  50
  51 // List of the currently installed standard schemes. This list is lazily
  52 // initialized by InitStandardSchemes and is leaked on shutdown to prevent
  53 // any destructors from being called that will slow us down or cause problems.
  54 std::vector<const char*>* standard_schemes = NULL;
  55
  56 // See the LockStandardSchemes declaration in the header.
  57 bool standard_schemes_locked = false;
  58
  59 // Ensures that the standard_schemes list is initialized, does nothing if it
  60 // already has values.
  61 void InitStandardSchemes() {
  62   if (standard_schemes)
  63     return;
  64   standard_schemes = new std::vector<const char*>;
  65   for (int i = 0; i < kNumStandardURLSchemes; i++)
  66     standard_schemes->push_back(kStandardURLSchemes[i]);
  67 }
  68
  69 // Given a string and a range inside the string, compares it to the given
  70 // lower-case |compare_to| buffer.
  71 template<typename CHAR>
  72 inline bool DoCompareSchemeComponent(const CHAR* spec,
  73                                      const url_parse::Component& component,
  74                                      const char* compare_to) {
  75   if (!component.is_nonempty())
  76     return compare_to[0] == 0;  // When component is empty, match empty scheme.
  77   return LowerCaseEqualsASCII(&spec[component.begin],
  78                               &spec[component.end()],
  79                               compare_to);
  80 }
  81
  82 // Returns true if the given scheme identified by |scheme| within |spec| is one
  83 // of the registered "standard" schemes.
  84 template<typename CHAR>
  85 bool DoIsStandard(const CHAR* spec, const url_parse::Component& scheme) {
  86   if (!scheme.is_nonempty())
  87     return false;  // Empty or invalid schemes are non-standard.
  88
  89   InitStandardSchemes();
  90   for (size_t i = 0; i < standard_schemes->size(); i++) {
  91     if (LowerCaseEqualsASCII(&spec[scheme.begin], &spec[scheme.end()],
  92                              standard_schemes->at(i)))
  93       return true;
  94   }
  95   return false;
  96 }
  97
  98 template<typename CHAR>
  99 bool DoFindAndCompareScheme(const CHAR* str,
 100                             int str_len,
 101                             const char* compare,
 102                             url_parse::Component* found_scheme) {
 103   // Before extracting scheme, canonicalize the URL to remove any whitespace.
 104   // This matches the canonicalization done in DoCanonicalize function.
 105   url_canon::RawCanonOutputT<CHAR> whitespace_buffer;
 106   int spec_len;
 107   const CHAR* spec = RemoveURLWhitespace(str, str_len,
 108                                          &whitespace_buffer, &spec_len);
 109
 110   url_parse::Component our_scheme;
 111   if (!url_parse::ExtractScheme(spec, spec_len, &our_scheme)) {
 112     // No scheme.
 113     if (found_scheme)
 114       *found_scheme = url_parse::Component();
 115     return false;
 116   }
 117   if (found_scheme)
 118     *found_scheme = our_scheme;
 119   return DoCompareSchemeComponent(spec, our_scheme, compare);
 120 }
 121
 122 template<typename CHAR>
 123 bool DoCanonicalize(const CHAR* in_spec, int in_spec_len,
 124                     bool trim_path_end,
 125                     url_canon::CharsetConverter* charset_converter,
 126                     url_canon::CanonOutput* output,
 127                     url_parse::Parsed* output_parsed) {
 128   // Remove any whitespace from the middle of the relative URL, possibly
 129   // copying to the new buffer.
 130   url_canon::RawCanonOutputT<CHAR> whitespace_buffer;
 131   int spec_len;
 132   const CHAR* spec = RemoveURLWhitespace(in_spec, in_spec_len,
 133                                          &whitespace_buffer, &spec_len);
 134
 135   url_parse::Parsed parsed_input;
 136 #ifdef WIN32
 137   // For Windows, we allow things that look like absolute Windows paths to be
 138   // fixed up magically to file URLs. This is done for IE compatability. For
 139   // example, this will change "c:/foo" into a file URL rather than treating
 140   // it as a URL with the protocol "c". It also works for UNC ("\\foo\bar.txt").
 141   // There is similar logic in url_canon_relative.cc for
 142   //
 143   // For Max & Unix, we don't do this (the equivalent would be "/foo/bar" which
 144   // has no meaning as an absolute path name. This is because browsers on Mac
 145   // & Unix don't generally do this, so there is no compatibility reason for
 146   // doing so.
 147   if (url_parse::DoesBeginUNCPath(spec, 0, spec_len, false) ||
 148       url_parse::DoesBeginWindowsDriveSpec(spec, 0, spec_len)) {
 149     url_parse::ParseFileURL(spec, spec_len, &parsed_input);
 150     return url_canon::CanonicalizeFileURL(spec, spec_len, parsed_input,
 151                                           charset_converter,
 152                                           output, output_parsed);
 153   }
 154 #endif
 155
 156   url_parse::Component scheme;
 157   if (!url_parse::ExtractScheme(spec, spec_len, &scheme))
 158     return false;
 159
 160   // This is the parsed version of the input URL, we have to canonicalize it
 161   // before storing it in our object.
 162   bool success;
 163   if (DoCompareSchemeComponent(spec, scheme, kFileScheme)) {
 164     // File URLs are special.
 165     url_parse::ParseFileURL(spec, spec_len, &parsed_input);
 166     success = url_canon::CanonicalizeFileURL(spec, spec_len, parsed_input,
 167                                              charset_converter, output,
 168                                              output_parsed);
 169   } else if (DoCompareSchemeComponent(spec, scheme, kFileSystemScheme)) {
 170     // Filesystem URLs are special.
 171     url_parse::ParseFileSystemURL(spec, spec_len, &parsed_input);
 172     success = url_canon::CanonicalizeFileSystemURL(spec, spec_len,
 173                                                    parsed_input,
 174                                                    charset_converter,
 175                                                    output, output_parsed);
 176
 177   } else if (DoIsStandard(spec, scheme)) {
 178     // All "normal" URLs.
 179     url_parse::ParseStandardURL(spec, spec_len, &parsed_input);
 180     success = url_canon::CanonicalizeStandardURL(spec, spec_len, parsed_input,
 181                                                  charset_converter,
 182                                                  output, output_parsed);
 183
 184   } else if (DoCompareSchemeComponent(spec, scheme, kMailtoScheme)) {
 185     // Mailto are treated like a standard url with only a scheme, path, query
 186     url_parse::ParseMailtoURL(spec, spec_len, &parsed_input);
 187     success = url_canon::CanonicalizeMailtoURL(spec, spec_len, parsed_input,
 188                                                output, output_parsed);
 189
 190   } else {
 191     // "Weird" URLs like data: and javascript:
 192     url_parse::ParsePathURL(spec, spec_len, trim_path_end, &parsed_input);
 193     success = url_canon::CanonicalizePathURL(spec, spec_len, parsed_input,
 194                                              output, output_parsed);
 195   }
 196   return success;
 197 }
 198
 199 template<typename CHAR>
 200 bool DoResolveRelative(const char* base_spec,
 201                        int base_spec_len,
 202                        const url_parse::Parsed& base_parsed,
 203                        const CHAR* in_relative,
 204                        int in_relative_length,
 205                        url_canon::CharsetConverter* charset_converter,
 206                        url_canon::CanonOutput* output,
 207                        url_parse::Parsed* output_parsed) {
 208   // Remove any whitespace from the middle of the relative URL, possibly
 209   // copying to the new buffer.
 210   url_canon::RawCanonOutputT<CHAR> whitespace_buffer;
 211   int relative_length;
 212   const CHAR* relative = RemoveURLWhitespace(in_relative, in_relative_length,
 213                                              &whitespace_buffer,
 214                                              &relative_length);
 215   bool base_is_authority_based = false;
 216   bool base_is_hierarchical = false;
 217   if (base_spec &&
 218       base_parsed.scheme.is_nonempty()) {
 219     int after_scheme = base_parsed.scheme.end() + 1;  // Skip past the colon.
 220     int num_slashes = url_parse::CountConsecutiveSlashes(
 221         base_spec, after_scheme, base_spec_len);
 222     base_is_authority_based = num_slashes > 1;
 223     base_is_hierarchical = num_slashes > 0;
 224   }
 225
 226   bool standard_base_scheme =
 227       base_parsed.scheme.is_nonempty() &&
 228       DoIsStandard(base_spec, base_parsed.scheme);
 229
 230   bool is_relative;
 231   url_parse::Component relative_component;
 232   if (!url_canon::IsRelativeURL(base_spec, base_parsed,
 233                                 relative, relative_length,
 234                                 (base_is_hierarchical || standard_base_scheme),
 235                                 &is_relative,
 236                                 &relative_component)) {
 237     // Error resolving.
 238     return false;
 239   }
 240
 241   // Pretend for a moment that |base_spec| is a standard URL. Normally
 242   // non-standard URLs are treated as PathURLs, but if the base has an
 243   // authority we would like to preserve it.
 244   if (is_relative && base_is_authority_based && !standard_base_scheme) {
 245     url_parse::Parsed base_parsed_authority;
 246     ParseStandardURL(base_spec, base_spec_len, &base_parsed_authority);
 247     if (base_parsed_authority.host.is_nonempty()) {
 248       bool did_resolve_succeed =
 249           url_canon::ResolveRelativeURL(base_spec, base_parsed_authority,
 250                                         false, relative,
 251                                         relative_component, charset_converter,
 252                                         output, output_parsed);
 253       // The output_parsed is incorrect at this point (because it was built
 254       // based on base_parsed_authority instead of base_parsed) and needs to be
 255       // re-created.
 256       ParsePathURL(output->data(), output->length(), true,
 257                    output_parsed);
 258       return did_resolve_succeed;
 259     }
 260   } else if (is_relative) {
 261     // Relative, resolve and canonicalize.
 262     bool file_base_scheme = base_parsed.scheme.is_nonempty() &&
 263         DoCompareSchemeComponent(base_spec, base_parsed.scheme, kFileScheme);
 264     return url_canon::ResolveRelativeURL(base_spec, base_parsed,
 265                                          file_base_scheme, relative,
 266                                          relative_component, charset_converter,
 267                                          output, output_parsed);
 268   }
 269
 270   // Not relative, canonicalize the input.
 271   return DoCanonicalize(relative, relative_length, true, charset_converter,
 272                         output, output_parsed);
 273 }
 274
 275 template<typename CHAR>
 276 bool DoReplaceComponents(const char* spec,
 277                          int spec_len,
 278                          const url_parse::Parsed& parsed,
 279                          const url_canon::Replacements<CHAR>& replacements,
 280                          url_canon::CharsetConverter* charset_converter,
 281                          url_canon::CanonOutput* output,
 282                          url_parse::Parsed* out_parsed) {
 283   // If the scheme is overridden, just do a simple string substitution and
 284   // reparse the whole thing. There are lots of edge cases that we really don't
 285   // want to deal with. Like what happens if I replace "http://e:8080/foo"
 286   // with a file. Does it become "file:///E:/8080/foo" where the port number
 287   // becomes part of the path? Parsing that string as a file URL says "yes"
 288   // but almost no sane rule for dealing with the components individually would
 289   // come up with that.
 290   //
 291   // Why allow these crazy cases at all? Programatically, there is almost no
 292   // case for replacing the scheme. The most common case for hitting this is
 293   // in JS when building up a URL using the location object. In this case, the
 294   // JS code expects the string substitution behavior:
 295   //   http://www.w3.org/TR/2008/WD-html5-20080610/structured.html#common3
 296   if (replacements.IsSchemeOverridden()) {
 297     // Canonicalize the new scheme so it is 8-bit and can be concatenated with
 298     // the existing spec.
 299     url_canon::RawCanonOutput<128> scheme_replaced;
 300     url_parse::Component scheme_replaced_parsed;
 301     url_canon::CanonicalizeScheme(
 302         replacements.sources().scheme,
 303         replacements.components().scheme,
 304         &scheme_replaced, &scheme_replaced_parsed);
 305
 306     // We can assume that the input is canonicalized, which means it always has
 307     // a colon after the scheme (or where the scheme would be).
 308     int spec_after_colon = parsed.scheme.is_valid() ? parsed.scheme.end() + 1
 309                                                     : 1;
 310     if (spec_len - spec_after_colon > 0) {
 311       scheme_replaced.Append(&spec[spec_after_colon],
 312                              spec_len - spec_after_colon);
 313     }
 314
 315     // We now need to completely re-parse the resulting string since its meaning
 316     // may have changed with the different scheme.
 317     url_canon::RawCanonOutput<128> recanonicalized;
 318     url_parse::Parsed recanonicalized_parsed;
 319     DoCanonicalize(scheme_replaced.data(), scheme_replaced.length(), true,
 320                    charset_converter,
 321                    &recanonicalized, &recanonicalized_parsed);
 322
 323     // Recurse using the version with the scheme already replaced. This will now
 324     // use the replacement rules for the new scheme.
 325     //
 326     // Warning: this code assumes that ReplaceComponents will re-check all
 327     // components for validity. This is because we can't fail if DoCanonicalize
 328     // failed above since theoretically the thing making it fail could be
 329     // getting replaced here. If ReplaceComponents didn't re-check everything,
 330     // we wouldn't know if something *not* getting replaced is a problem.
 331     // If the scheme-specific replacers are made more intelligent so they don't
 332     // re-check everything, we should instead recanonicalize the whole thing
 333     // after this call to check validity (this assumes replacing the scheme is
 334     // much much less common than other types of replacements, like clearing the
 335     // ref).
 336     url_canon::Replacements<CHAR> replacements_no_scheme = replacements;
 337     replacements_no_scheme.SetScheme(NULL, url_parse::Component());
 338     return DoReplaceComponents(recanonicalized.data(), recanonicalized.length(),
 339                                recanonicalized_parsed, replacements_no_scheme,
 340                                charset_converter, output, out_parsed);
 341   }
 342
 343   // If we get here, then we know the scheme doesn't need to be replaced, so can
 344   // just key off the scheme in the spec to know how to do the replacements.
 345   if (DoCompareSchemeComponent(spec, parsed.scheme, kFileScheme)) {
 346     return url_canon::ReplaceFileURL(spec, parsed, replacements,
 347                                      charset_converter, output, out_parsed);
 348   }
 349   if (DoCompareSchemeComponent(spec, parsed.scheme, kFileSystemScheme)) {
 350     return url_canon::ReplaceFileSystemURL(spec, parsed, replacements,
 351                                            charset_converter, output,
 352                                            out_parsed);
 353   }
 354   if (DoIsStandard(spec, parsed.scheme)) {
 355     return url_canon::ReplaceStandardURL(spec, parsed, replacements,
 356                                          charset_converter, output, out_parsed);
 357   }
 358   if (DoCompareSchemeComponent(spec, parsed.scheme, kMailtoScheme)) {
 359      return url_canon::ReplaceMailtoURL(spec, parsed, replacements,
 360                                         output, out_parsed);
 361   }
 362
 363   // Default is a path URL.
 364   return url_canon::ReplacePathURL(spec, parsed, replacements,
 365                                    output, out_parsed);
 366 }
 367
 368 }  // namespace
 369
 370 void Initialize() {
 371   InitStandardSchemes();
 372 }
 373
 374 void Shutdown() {
 375   if (standard_schemes) {
 376     delete standard_schemes;
 377     standard_schemes = NULL;
 378   }
 379 }
 380
 381 void AddStandardScheme(const char* new_scheme) {
 382   // If this assert triggers, it means you've called AddStandardScheme after
 383   // LockStandardSchemes have been called (see the header file for
 384   // LockStandardSchemes for more).
 385   //
 386   // This normally means you're trying to set up a new standard scheme too late
 387   // in your application's init process. Locate where your app does this
 388   // initialization and calls LockStandardScheme, and add your new standard
 389   // scheme there.
 390   DCHECK(!standard_schemes_locked) <<
 391       "Trying to add a standard scheme after the list has been locked.";
 392
 393   size_t scheme_len = strlen(new_scheme);
 394   if (scheme_len == 0)
 395     return;
 396
 397   // Dulicate the scheme into a new buffer and add it to the list of standard
 398   // schemes. This pointer will be leaked on shutdown.
 399   char* dup_scheme = new char[scheme_len + 1];
 400   memcpy(dup_scheme, new_scheme, scheme_len + 1);
 401
 402   InitStandardSchemes();
 403   standard_schemes->push_back(dup_scheme);
 404 }
 405
 406 void LockStandardSchemes() {
 407   standard_schemes_locked = true;
 408 }
 409
 410 bool IsStandard(const char* spec, const url_parse::Component& scheme) {
 411   return DoIsStandard(spec, scheme);
 412 }
 413
 414 bool IsStandard(const base::char16* spec, const url_parse::Component& scheme) {
 415   return DoIsStandard(spec, scheme);
 416 }
 417
 418 bool FindAndCompareScheme(const char* str,
 419                           int str_len,
 420                           const char* compare,
 421                           url_parse::Component* found_scheme) {
 422   return DoFindAndCompareScheme(str, str_len, compare, found_scheme);
 423 }
 424
 425 bool FindAndCompareScheme(const base::char16* str,
 426                           int str_len,
 427                           const char* compare,
 428                           url_parse::Component* found_scheme) {
 429   return DoFindAndCompareScheme(str, str_len, compare, found_scheme);
 430 }
 431
 432 bool Canonicalize(const char* spec,
 433                   int spec_len,
 434                   bool trim_path_end,
 435                   url_canon::CharsetConverter* charset_converter,
 436                   url_canon::CanonOutput* output,
 437                   url_parse::Parsed* output_parsed) {
 438   return DoCanonicalize(spec, spec_len, trim_path_end, charset_converter,
 439                         output, output_parsed);
 440 }
 441
 442 bool Canonicalize(const base::char16* spec,
 443                   int spec_len,
 444                   bool trim_path_end,
 445                   url_canon::CharsetConverter* charset_converter,
 446                   url_canon::CanonOutput* output,
 447                   url_parse::Parsed* output_parsed) {
 448   return DoCanonicalize(spec, spec_len, trim_path_end, charset_converter,
 449                         output, output_parsed);
 450 }
 451
 452 bool ResolveRelative(const char* base_spec,
 453                      int base_spec_len,
 454                      const url_parse::Parsed& base_parsed,
 455                      const char* relative,
 456                      int relative_length,
 457                      url_canon::CharsetConverter* charset_converter,
 458                      url_canon::CanonOutput* output,
 459                      url_parse::Parsed* output_parsed) {
 460   return DoResolveRelative(base_spec, base_spec_len, base_parsed,
 461                            relative, relative_length,
 462                            charset_converter, output, output_parsed);
 463 }
 464
 465 bool ResolveRelative(const char* base_spec,
 466                      int base_spec_len,
 467                      const url_parse::Parsed& base_parsed,
 468                      const base::char16* relative,
 469                      int relative_length,
 470                      url_canon::CharsetConverter* charset_converter,
 471                      url_canon::CanonOutput* output,
 472                      url_parse::Parsed* output_parsed) {
 473   return DoResolveRelative(base_spec, base_spec_len, base_parsed,
 474                            relative, relative_length,
 475                            charset_converter, output, output_parsed);
 476 }
 477
 478 bool ReplaceComponents(const char* spec,
 479                        int spec_len,
 480                        const url_parse::Parsed& parsed,
 481                        const url_canon::Replacements<char>& replacements,
 482                        url_canon::CharsetConverter* charset_converter,
 483                        url_canon::CanonOutput* output,
 484                        url_parse::Parsed* out_parsed) {
 485   return DoReplaceComponents(spec, spec_len, parsed, replacements,
 486                              charset_converter, output, out_parsed);
 487 }
 488
 489 bool ReplaceComponents(
 490       const char* spec,
 491       int spec_len,
 492       const url_parse::Parsed& parsed,
 493       const url_canon::Replacements<base::char16>& replacements,
 494       url_canon::CharsetConverter* charset_converter,
 495       url_canon::CanonOutput* output,
 496       url_parse::Parsed* out_parsed) {
 497   return DoReplaceComponents(spec, spec_len, parsed, replacements,
 498                              charset_converter, output, out_parsed);
 499 }
 500
 501 // Front-ends for LowerCaseEqualsASCII.
 502 bool LowerCaseEqualsASCII(const char* a_begin,
 503                           const char* a_end,
 504                           const char* b) {
 505   return DoLowerCaseEqualsASCII(a_begin, a_end, b);
 506 }
 507
 508 bool LowerCaseEqualsASCII(const char* a_begin,
 509                           const char* a_end,
 510                           const char* b_begin,
 511                           const char* b_end) {
 512   while (a_begin != a_end && b_begin != b_end &&
 513          ToLowerASCII(*a_begin) == *b_begin) {
 514     a_begin++;
 515     b_begin++;
 516   }
 517   return a_begin == a_end && b_begin == b_end;
 518 }
 519
 520 bool LowerCaseEqualsASCII(const base::char16* a_begin,
 521                           const base::char16* a_end,
 522                           const char* b) {
 523   return DoLowerCaseEqualsASCII(a_begin, a_end, b);
 524 }
 525
 526 void DecodeURLEscapeSequences(const char* input, int length,
 527                               url_canon::CanonOutputW* output) {
 528   url_canon::RawCanonOutputT<char> unescaped_chars;
 529   for (int i = 0; i < length; i++) {
 530     if (input[i] == '%') {
 531       unsigned char ch;
 532       if (url_canon::DecodeEscaped(input, &i, length, &ch)) {
 533         unescaped_chars.push_back(ch);
 534       } else {
 535         // Invalid escape sequence, copy the percent literal.
 536         unescaped_chars.push_back('%');
 537       }
 538     } else {
 539       // Regular non-escaped 8-bit character.
 540       unescaped_chars.push_back(input[i]);
 541     }
 542   }
 543
 544   // Convert that 8-bit to UTF-16. It's not clear IE does this at all to
 545   // JavaScript URLs, but Firefox and Safari do.
 546   for (int i = 0; i < unescaped_chars.length(); i++) {
 547     unsigned char uch = static_cast<unsigned char>(unescaped_chars.at(i));
 548     if (uch < 0x80) {
 549       // Non-UTF-8, just append directly
 550       output->push_back(uch);
 551     } else {
 552       // next_ch will point to the last character of the decoded
 553       // character.
 554       int next_character = i;
 555       unsigned code_point;
 556       if (url_canon::ReadUTFChar(unescaped_chars.data(), &next_character,
 557                                  unescaped_chars.length(), &code_point)) {
 558         // Valid UTF-8 character, convert to UTF-16.
 559         url_canon::AppendUTF16Value(code_point, output);
 560         i = next_character;
 561       } else {
 562         // If there are any sequences that are not valid UTF-8, we keep
 563         // invalid code points and promote to UTF-16. We copy all characters
 564         // from the current position to the end of the identified sequence.
 565         while (i < next_character) {
 566           output->push_back(static_cast<unsigned char>(unescaped_chars.at(i)));
 567           i++;
 568         }
 569         output->push_back(static_cast<unsigned char>(unescaped_chars.at(i)));
 570       }
 571     }
 572   }
 573 }
 574
 575 void EncodeURIComponent(const char* input, int length,
 576                         url_canon::CanonOutput* output) {
 577   for (int i = 0; i < length; ++i) {
 578     unsigned char c = static_cast<unsigned char>(input[i]);
 579     if (url_canon::IsComponentChar(c))
 580       output->push_back(c);
 581     else
 582       AppendEscapedChar(c, output);
 583   }
 584 }
 585
 586 bool CompareSchemeComponent(const char* spec,
 587                             const url_parse::Component& component,
 588                             const char* compare_to) {
 589   return DoCompareSchemeComponent(spec, component, compare_to);
 590 }
 591
 592 bool CompareSchemeComponent(const base::char16* spec,
 593                             const url_parse::Component& component,
 594                             const char* compare_to) {
 595   return DoCompareSchemeComponent(spec, component, compare_to);
 596 }
 597
 598 }  // namespace url_util