url/url_canon_mailtourl.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 "mailto:" 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 template <typename CHAR, typename UCHAR>
  17 bool DoCanonicalizeMailtoURL(const URLComponentSource<CHAR>& source,
  18                              const Parsed& parsed,
  19                              CanonOutput* output,
  20                              Parsed* new_parsed) {
  21   // mailto: only uses {scheme, path, query} -- clear the rest.
  22   new_parsed->username = Component();
  23   new_parsed->password = Component();
  24   new_parsed->host = Component();
  25   new_parsed->port = Component();
  26   new_parsed->ref = Component();
  27
  28   // Scheme (known, so we don't bother running it through the more
  29   // complicated scheme canonicalizer).
  30   new_parsed->scheme.begin = output->length();
  31   output->Append("mailto:", 7);
  32   new_parsed->scheme.len = 6;
  33
  34   bool success = true;
  35
  36   // Path
  37   if (parsed.path.is_valid()) {
  38     new_parsed->path.begin = output->length();
  39
  40     // Copy the path using path URL's more lax escaping rules.
  41     // We convert to UTF-8 and escape non-ASCII, but leave all
  42     // ASCII characters alone.
  43     int end = parsed.path.end();
  44     for (int i = parsed.path.begin; i < end; ++i) {
  45       UCHAR uch = static_cast<UCHAR>(source.path[i]);
  46       if (uch < 0x20 || uch >= 0x80)
  47         success &= AppendUTF8EscapedChar(source.path, &i, end, output);
  48       else
  49         output->push_back(static_cast<char>(uch));
  50     }
  51
  52     new_parsed->path.len = output->length() - new_parsed->path.begin;
  53   } else {
  54     // No path at all
  55     new_parsed->path.reset();
  56   }
  57
  58   // Query -- always use the default UTF8 charset converter.
  59   CanonicalizeQuery(source.query, parsed.query, NULL,
  60                     output, &new_parsed->query);
  61
  62   return success;
  63 }
  64
  65 } // namespace
  66
  67 bool CanonicalizeMailtoURL(const char* spec,
  68                            int spec_len,
  69                            const Parsed& parsed,
  70                            CanonOutput* output,
  71                            Parsed* new_parsed) {
  72   return DoCanonicalizeMailtoURL<char, unsigned char>(
  73       URLComponentSource<char>(spec), parsed, output, new_parsed);
  74 }
  75
  76 bool CanonicalizeMailtoURL(const base::char16* spec,
  77                            int spec_len,
  78                            const Parsed& parsed,
  79                            CanonOutput* output,
  80                            Parsed* new_parsed) {
  81   return DoCanonicalizeMailtoURL<base::char16, base::char16>(
  82       URLComponentSource<base::char16>(spec), parsed, output, new_parsed);
  83 }
  84
  85 bool ReplaceMailtoURL(const char* base,
  86                       const Parsed& base_parsed,
  87                       const Replacements<char>& replacements,
  88                       CanonOutput* output,
  89                       Parsed* new_parsed) {
  90   URLComponentSource<char> source(base);
  91   Parsed parsed(base_parsed);
  92   SetupOverrideComponents(base, replacements, &source, &parsed);
  93   return DoCanonicalizeMailtoURL<char, unsigned char>(
  94       source, parsed, output, new_parsed);
  95 }
  96
  97 bool ReplaceMailtoURL(const char* base,
  98                       const Parsed& base_parsed,
  99                       const Replacements<base::char16>& replacements,
 100                       CanonOutput* output,
 101                       Parsed* new_parsed) {
 102   RawCanonOutput<1024> utf8;
 103   URLComponentSource<char> source(base);
 104   Parsed parsed(base_parsed);
 105   SetupUTF16OverrideComponents(base, replacements, &utf8, &source, &parsed);
 106   return DoCanonicalizeMailtoURL<char, unsigned char>(
 107       source, parsed, output, new_parsed);
 108 }
 109
 110 }  // namespace url