url/url_canon_pathurl.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 "path" URLs. Not to be confused with the path
   6 // of a URL, these are URLs that have no authority section, only a path. For
   7 // example, "javascript:" and "data:".
   8
   9 #include "url/url_canon.h"
  10 #include "url/url_canon_internal.h"
  11
  12 namespace url_canon {
  13
  14 namespace {
  15
  16 template<typename CHAR, typename UCHAR>
  17 bool DoCanonicalizePathURL(const URLComponentSource<CHAR>& source,
  18                            const url_parse::Parsed& parsed,
  19                            CanonOutput* output,
  20                            url_parse::Parsed* new_parsed) {
  21   // Scheme: this will append the colon.
  22   bool success = CanonicalizeScheme(source.scheme, parsed.scheme,
  23                                     output, &new_parsed->scheme);
  24
  25   // We assume there's no authority for path URLs. Note that hosts should never
  26   // have -1 length.
  27   new_parsed->username.reset();
  28   new_parsed->password.reset();
  29   new_parsed->host.reset();
  30   new_parsed->port.reset();
  31
  32   if (parsed.path.is_valid()) {
  33     // Copy the path using path URL's more lax escaping rules (think for
  34     // javascript:). We convert to UTF-8 and escape non-ASCII, but leave all
  35     // ASCII characters alone. This helps readability of JavaStript.
  36     new_parsed->path.begin = output->length();
  37     int end = parsed.path.end();
  38     for (int i = parsed.path.begin; i < end; i++) {
  39       UCHAR uch = static_cast<UCHAR>(source.path[i]);
  40       if (uch < 0x20 || uch >= 0x80)
  41         success &= AppendUTF8EscapedChar(source.path, &i, end, output);
  42       else
  43         output->push_back(static_cast<char>(uch));
  44     }
  45     new_parsed->path.len = output->length() - new_parsed->path.begin;
  46   } else {
  47     // Empty path.
  48     new_parsed->path.reset();
  49   }
  50
  51   // Assume there's no query or ref.
  52   new_parsed->query.reset();
  53   new_parsed->ref.reset();
  54
  55   return success;
  56 }
  57
  58 }  // namespace
  59
  60 bool CanonicalizePathURL(const char* spec,
  61                          int spec_len,
  62                          const url_parse::Parsed& parsed,
  63                          CanonOutput* output,
  64                          url_parse::Parsed* new_parsed) {
  65   return DoCanonicalizePathURL<char, unsigned char>(
  66       URLComponentSource<char>(spec), parsed, output, new_parsed);
  67 }
  68
  69 bool CanonicalizePathURL(const char16* spec,
  70                          int spec_len,
  71                          const url_parse::Parsed& parsed,
  72                          CanonOutput* output,
  73                          url_parse::Parsed* new_parsed) {
  74   return DoCanonicalizePathURL<char16, char16>(
  75       URLComponentSource<char16>(spec), parsed, output, new_parsed);
  76 }
  77
  78 bool ReplacePathURL(const char* base,
  79                     const url_parse::Parsed& base_parsed,
  80                     const Replacements<char>& replacements,
  81                     CanonOutput* output,
  82                     url_parse::Parsed* new_parsed) {
  83   URLComponentSource<char> source(base);
  84   url_parse::Parsed parsed(base_parsed);
  85   SetupOverrideComponents(base, replacements, &source, &parsed);
  86   return DoCanonicalizePathURL<char, unsigned char>(
  87       source, parsed, output, new_parsed);
  88 }
  89
  90 bool ReplacePathURL(const char* base,
  91                     const url_parse::Parsed& base_parsed,
  92                     const Replacements<char16>& replacements,
  93                     CanonOutput* output,
  94                     url_parse::Parsed* new_parsed) {
  95   RawCanonOutput<1024> utf8;
  96   URLComponentSource<char> source(base);
  97   url_parse::Parsed parsed(base_parsed);
  98   SetupUTF16OverrideComponents(base, replacements, &utf8, &source, &parsed);
  99   return DoCanonicalizePathURL<char, unsigned char>(
 100       source, parsed, output, new_parsed);
 101 }
 102
 103 }  // namespace url_canon