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.
5 // Functions to canonicalize "standard" URLs, which are ones that have an
6 // authority section including a host name.
8 #include "url/url_canon.h"
9 #include "url/url_canon_internal.h"
10 #include "url/url_constants.h"
16 template <typename CHAR, typename UCHAR>
17 bool DoCanonicalizeStandardURL(const URLComponentSource<CHAR>& source,
19 SchemeType scheme_type,
20 CharsetConverter* query_converter,
23 // Scheme: this will append the colon.
24 bool success = CanonicalizeScheme(source.scheme, parsed.scheme,
25 output, &new_parsed->scheme);
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);
33 // Authority (username, password, host, port)
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;
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('/');
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);
53 new_parsed->username.reset();
54 new_parsed->password.reset();
57 success &= CanonicalizeHost(source.host, parsed.host,
58 output, &new_parsed->host);
60 // Host must not be empty for standard URLs.
61 if (!parsed.host.is_nonempty())
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);
71 new_parsed->port.reset();
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.
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('/');
96 new_parsed->path.reset();
100 CanonicalizeQuery(source.query, parsed.query, query_converter,
101 output, &new_parsed->query);
103 // Ref: ignore failure for this, since the page can probably still be loaded.
104 CanonicalizeRef(source.ref, parsed.ref, output, &new_parsed->ref);
112 // Returns the default port for the given canonical scheme, or PORT_UNSPECIFIED
113 // if the scheme is unknown.
114 int DefaultPortForScheme(const char* scheme, int scheme_len) {
115 int default_port = PORT_UNSPECIFIED;
116 switch (scheme_len) {
118 if (!strncmp(scheme, kHttpScheme, scheme_len))
122 if (!strncmp(scheme, kHttpsScheme, scheme_len))
126 if (!strncmp(scheme, kFtpScheme, scheme_len))
128 else if (!strncmp(scheme, kWssScheme, scheme_len))
132 if (!strncmp(scheme, kGopherScheme, scheme_len))
136 if (!strncmp(scheme, kWsScheme, scheme_len))
143 bool CanonicalizeStandardURL(const char* spec,
145 const Parsed& parsed,
146 SchemeType scheme_type,
147 CharsetConverter* query_converter,
149 Parsed* new_parsed) {
150 return DoCanonicalizeStandardURL<char, unsigned char>(
151 URLComponentSource<char>(spec), parsed, scheme_type, query_converter,
155 bool CanonicalizeStandardURL(const base::char16* spec,
157 const Parsed& parsed,
158 SchemeType scheme_type,
159 CharsetConverter* query_converter,
161 Parsed* new_parsed) {
162 return DoCanonicalizeStandardURL<base::char16, base::char16>(
163 URLComponentSource<base::char16>(spec), parsed, scheme_type,
164 query_converter, output, new_parsed);
167 // It might be nice in the future to optimize this so unchanged components don't
168 // need to be recanonicalized. This is especially true since the common case for
169 // ReplaceComponents is removing things we don't want, like reference fragments
170 // and usernames. These cases can become more efficient if we can assume the
171 // rest of the URL is OK with these removed (or only the modified parts
172 // recanonicalized). This would be much more complex to implement, however.
174 // You would also need to update DoReplaceComponents in url_util.cc which
175 // relies on this re-checking everything (see the comment there for why).
176 bool ReplaceStandardURL(const char* base,
177 const Parsed& base_parsed,
178 const Replacements<char>& replacements,
179 SchemeType scheme_type,
180 CharsetConverter* query_converter,
182 Parsed* new_parsed) {
183 URLComponentSource<char> source(base);
184 Parsed parsed(base_parsed);
185 SetupOverrideComponents(base, replacements, &source, &parsed);
186 return DoCanonicalizeStandardURL<char, unsigned char>(
187 source, parsed, scheme_type, query_converter, output, new_parsed);
190 // For 16-bit replacements, we turn all the replacements into UTF-8 so the
191 // regular code path can be used.
192 bool ReplaceStandardURL(const char* base,
193 const Parsed& base_parsed,
194 const Replacements<base::char16>& replacements,
195 SchemeType scheme_type,
196 CharsetConverter* query_converter,
198 Parsed* new_parsed) {
199 RawCanonOutput<1024> utf8;
200 URLComponentSource<char> source(base);
201 Parsed parsed(base_parsed);
202 SetupUTF16OverrideComponents(base, replacements, &utf8, &source, &parsed);
203 return DoCanonicalizeStandardURL<char, unsigned char>(
204 source, parsed, scheme_type, query_converter, output, new_parsed);