• Home
  • Line#
  • Scopes#
  • Navigate#
  • Raw
  • Download
1 // Copyright 2013 The Chromium Authors
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 to canonicalize "standard" URLs, which are ones that have an
6 // authority section including a host name.
7 
8 #include "url/url_canon.h"
9 #include "url/url_canon_internal.h"
10 #include "url/url_constants.h"
11 
12 namespace url {
13 
14 namespace {
15 
16 template <typename CHAR, typename UCHAR>
DoCanonicalizeStandardURL(const URLComponentSource<CHAR> & source,const Parsed & parsed,SchemeType scheme_type,CharsetConverter * query_converter,CanonOutput * output,Parsed * new_parsed)17 bool DoCanonicalizeStandardURL(const URLComponentSource<CHAR>& source,
18                                const Parsed& parsed,
19                                SchemeType scheme_type,
20                                CharsetConverter* query_converter,
21                                CanonOutput* output,
22                                Parsed* new_parsed) {
23   // Scheme: this will append the colon.
24   bool success = CanonicalizeScheme(source.scheme, parsed.scheme,
25                                     output, &new_parsed->scheme);
26 
27   bool scheme_supports_user_info =
28       (scheme_type == SCHEME_WITH_HOST_PORT_AND_USER_INFORMATION);
29   bool scheme_supports_ports =
30       (scheme_type == SCHEME_WITH_HOST_PORT_AND_USER_INFORMATION ||
31        scheme_type == SCHEME_WITH_HOST_AND_PORT);
32 
33   // Authority (username, password, host, port)
34   bool have_authority;
35   if ((scheme_supports_user_info &&
36        (parsed.username.is_valid() || parsed.password.is_valid())) ||
37       parsed.host.is_nonempty() ||
38       (scheme_supports_ports && parsed.port.is_valid())) {
39     have_authority = true;
40 
41     // Only write the authority separators when we have a scheme.
42     if (parsed.scheme.is_valid()) {
43       output->push_back('/');
44       output->push_back('/');
45     }
46 
47     // User info: the canonicalizer will handle the : and @.
48     if (scheme_supports_user_info) {
49       success &= CanonicalizeUserInfo(
50           source.username, parsed.username, source.password, parsed.password,
51           output, &new_parsed->username, &new_parsed->password);
52     } else {
53       new_parsed->username.reset();
54       new_parsed->password.reset();
55     }
56 
57     success &= CanonicalizeHost(source.host, parsed.host,
58                                 output, &new_parsed->host);
59 
60     // Host must not be empty for standard URLs.
61     if (parsed.host.is_empty())
62       success = false;
63 
64     // Port: the port canonicalizer will handle the colon.
65     if (scheme_supports_ports) {
66       int default_port = DefaultPortForScheme(
67           &output->data()[new_parsed->scheme.begin], new_parsed->scheme.len);
68       success &= CanonicalizePort(source.port, parsed.port, default_port,
69                                   output, &new_parsed->port);
70     } else {
71       new_parsed->port.reset();
72     }
73   } else {
74     // No authority, clear the components.
75     have_authority = false;
76     new_parsed->host.reset();
77     new_parsed->username.reset();
78     new_parsed->password.reset();
79     new_parsed->port.reset();
80     success = false;  // Standard URLs must have an authority.
81   }
82 
83   // Path
84   if (parsed.path.is_valid()) {
85     success &= CanonicalizePath(source.path, parsed.path,
86                                 output, &new_parsed->path);
87   } else if (have_authority ||
88              parsed.query.is_valid() || parsed.ref.is_valid()) {
89     // When we have an empty path, make up a path when we have an authority
90     // or something following the path. The only time we allow an empty
91     // output path is when there is nothing else.
92     new_parsed->path = Component(output->length(), 1);
93     output->push_back('/');
94   } else {
95     // No path at all
96     new_parsed->path.reset();
97   }
98 
99   // Query
100   CanonicalizeQuery(source.query, parsed.query, query_converter,
101                     output, &new_parsed->query);
102 
103   // Ref: ignore failure for this, since the page can probably still be loaded.
104   CanonicalizeRef(source.ref, parsed.ref, output, &new_parsed->ref);
105 
106   // Carry over the flag for potentially dangling markup:
107   if (parsed.potentially_dangling_markup)
108     new_parsed->potentially_dangling_markup = true;
109 
110   return success;
111 }
112 
113 }  // namespace
114 
115 // Returns the default port for the given canonical scheme, or PORT_UNSPECIFIED
116 // if the scheme is unknown.
117 //
118 // Please keep blink::DefaultPortForProtocol and url::DefaultPortForProtocol in
119 // sync.
DefaultPortForScheme(const char * scheme,int scheme_len)120 int DefaultPortForScheme(const char* scheme, int scheme_len) {
121   int default_port = PORT_UNSPECIFIED;
122   switch (scheme_len) {
123     case 4:
124       if (!strncmp(scheme, kHttpScheme, scheme_len))
125         default_port = 80;
126       break;
127     case 5:
128       if (!strncmp(scheme, kHttpsScheme, scheme_len))
129         default_port = 443;
130       break;
131     case 3:
132       if (!strncmp(scheme, kFtpScheme, scheme_len))
133         default_port = 21;
134       else if (!strncmp(scheme, kWssScheme, scheme_len))
135         default_port = 443;
136       break;
137     case 2:
138       if (!strncmp(scheme, kWsScheme, scheme_len))
139         default_port = 80;
140       break;
141   }
142   return default_port;
143 }
144 
CanonicalizeStandardURL(const char * spec,int spec_len,const Parsed & parsed,SchemeType scheme_type,CharsetConverter * query_converter,CanonOutput * output,Parsed * new_parsed)145 bool CanonicalizeStandardURL(const char* spec,
146                              int spec_len,
147                              const Parsed& parsed,
148                              SchemeType scheme_type,
149                              CharsetConverter* query_converter,
150                              CanonOutput* output,
151                              Parsed* new_parsed) {
152   return DoCanonicalizeStandardURL<char, unsigned char>(
153       URLComponentSource<char>(spec), parsed, scheme_type, query_converter,
154       output, new_parsed);
155 }
156 
CanonicalizeStandardURL(const char16_t * spec,int spec_len,const Parsed & parsed,SchemeType scheme_type,CharsetConverter * query_converter,CanonOutput * output,Parsed * new_parsed)157 bool CanonicalizeStandardURL(const char16_t* spec,
158                              int spec_len,
159                              const Parsed& parsed,
160                              SchemeType scheme_type,
161                              CharsetConverter* query_converter,
162                              CanonOutput* output,
163                              Parsed* new_parsed) {
164   return DoCanonicalizeStandardURL<char16_t, char16_t>(
165       URLComponentSource<char16_t>(spec), parsed, scheme_type, query_converter,
166       output, new_parsed);
167 }
168 
169 // It might be nice in the future to optimize this so unchanged components don't
170 // need to be recanonicalized. This is especially true since the common case for
171 // ReplaceComponents is removing things we don't want, like reference fragments
172 // and usernames. These cases can become more efficient if we can assume the
173 // rest of the URL is OK with these removed (or only the modified parts
174 // recanonicalized). This would be much more complex to implement, however.
175 //
176 // You would also need to update DoReplaceComponents in url_util.cc which
177 // relies on this re-checking everything (see the comment there for why).
ReplaceStandardURL(const char * base,const Parsed & base_parsed,const Replacements<char> & replacements,SchemeType scheme_type,CharsetConverter * query_converter,CanonOutput * output,Parsed * new_parsed)178 bool ReplaceStandardURL(const char* base,
179                         const Parsed& base_parsed,
180                         const Replacements<char>& replacements,
181                         SchemeType scheme_type,
182                         CharsetConverter* query_converter,
183                         CanonOutput* output,
184                         Parsed* new_parsed) {
185   URLComponentSource<char> source(base);
186   Parsed parsed(base_parsed);
187   SetupOverrideComponents(base, replacements, &source, &parsed);
188   return DoCanonicalizeStandardURL<char, unsigned char>(
189       source, parsed, scheme_type, query_converter, output, new_parsed);
190 }
191 
192 // For 16-bit replacements, we turn all the replacements into UTF-8 so the
193 // regular code path can be used.
ReplaceStandardURL(const char * base,const Parsed & base_parsed,const Replacements<char16_t> & replacements,SchemeType scheme_type,CharsetConverter * query_converter,CanonOutput * output,Parsed * new_parsed)194 bool ReplaceStandardURL(const char* base,
195                         const Parsed& base_parsed,
196                         const Replacements<char16_t>& replacements,
197                         SchemeType scheme_type,
198                         CharsetConverter* query_converter,
199                         CanonOutput* output,
200                         Parsed* new_parsed) {
201   RawCanonOutput<1024> utf8;
202   URLComponentSource<char> source(base);
203   Parsed parsed(base_parsed);
204   SetupUTF16OverrideComponents(base, replacements, &utf8, &source, &parsed);
205   return DoCanonicalizeStandardURL<char, unsigned char>(
206       source, parsed, scheme_type, query_converter, output, new_parsed);
207 }
208 
209 }  // namespace url
210