| OLD | NEW |
| 1 // Copyright (c) 2012 The Chromium Authors. All rights reserved. | 1 // Copyright (c) 2012 The Chromium Authors. All rights reserved. |
| 2 // Use of this source code is governed by a BSD-style license that can be | 2 // Use of this source code is governed by a BSD-style license that can be |
| 3 // found in the LICENSE file. | 3 // found in the LICENSE file. |
| 4 | 4 |
| 5 #include "extensions/common/url_pattern.h" | 5 #include "extensions/common/url_pattern.h" |
| 6 | 6 |
| 7 #include <stddef.h> | 7 #include <stddef.h> |
| 8 | 8 |
| 9 #include <ostream> | 9 #include <ostream> |
| 10 | 10 |
| (...skipping 125 matching lines...) Expand 10 before | Expand all | Expand 10 after Loading... |
| 136 int result = 0; | 136 int result = 0; |
| 137 for (size_t i = 0; i < arraysize(kValidSchemeMasks); ++i) | 137 for (size_t i = 0; i < arraysize(kValidSchemeMasks); ++i) |
| 138 result |= kValidSchemeMasks[i]; | 138 result |= kValidSchemeMasks[i]; |
| 139 return result; | 139 return result; |
| 140 } | 140 } |
| 141 | 141 |
| 142 URLPattern::URLPattern() | 142 URLPattern::URLPattern() |
| 143 : valid_schemes_(SCHEME_NONE), | 143 : valid_schemes_(SCHEME_NONE), |
| 144 match_all_urls_(false), | 144 match_all_urls_(false), |
| 145 match_subdomains_(false), | 145 match_subdomains_(false), |
| 146 match_effective_tld_(true), |
| 146 port_("*") {} | 147 port_("*") {} |
| 147 | 148 |
| 148 URLPattern::URLPattern(int valid_schemes) | 149 URLPattern::URLPattern(int valid_schemes) |
| 149 : valid_schemes_(valid_schemes), | 150 : valid_schemes_(valid_schemes), |
| 150 match_all_urls_(false), | 151 match_all_urls_(false), |
| 151 match_subdomains_(false), | 152 match_subdomains_(false), |
| 153 match_effective_tld_(true), |
| 152 port_("*") {} | 154 port_("*") {} |
| 153 | 155 |
| 154 URLPattern::URLPattern(int valid_schemes, const std::string& pattern) | 156 URLPattern::URLPattern(int valid_schemes, const std::string& pattern) |
| 155 // Strict error checking is used, because this constructor is only | 157 // Strict error checking is used, because this constructor is only |
| 156 // appropriate when we know |pattern| is valid. | 158 // appropriate when we know |pattern| is valid. |
| 157 : valid_schemes_(valid_schemes), | 159 : valid_schemes_(valid_schemes), |
| 158 match_all_urls_(false), | 160 match_all_urls_(false), |
| 159 match_subdomains_(false), | 161 match_subdomains_(false), |
| 162 match_effective_tld_(true), |
| 160 port_("*") { | 163 port_("*") { |
| 161 ParseResult result = Parse(pattern); | 164 ParseResult result = Parse(pattern); |
| 162 if (PARSE_SUCCESS != result) | 165 if (PARSE_SUCCESS != result) |
| 163 NOTREACHED() << "URLPattern invalid: " << pattern << " result " << result; | 166 NOTREACHED() << "URLPattern invalid: " << pattern << " result " << result; |
| 164 } | 167 } |
| 165 | 168 |
| 166 URLPattern::URLPattern(const URLPattern& other) = default; | 169 URLPattern::URLPattern(const URLPattern& other) = default; |
| 167 | 170 |
| 168 URLPattern::~URLPattern() { | 171 URLPattern::~URLPattern() { |
| 169 } | 172 } |
| 170 | 173 |
| 171 bool URLPattern::operator<(const URLPattern& other) const { | 174 bool URLPattern::operator<(const URLPattern& other) const { |
| 172 return GetAsString() < other.GetAsString(); | 175 return GetAsString() < other.GetAsString(); |
| 173 } | 176 } |
| 174 | 177 |
| 175 bool URLPattern::operator>(const URLPattern& other) const { | 178 bool URLPattern::operator>(const URLPattern& other) const { |
| 176 return GetAsString() > other.GetAsString(); | 179 return GetAsString() > other.GetAsString(); |
| 177 } | 180 } |
| 178 | 181 |
| 179 bool URLPattern::operator==(const URLPattern& other) const { | 182 bool URLPattern::operator==(const URLPattern& other) const { |
| 180 return GetAsString() == other.GetAsString(); | 183 return GetAsString() == other.GetAsString(); |
| 181 } | 184 } |
| 182 | 185 |
| 183 std::ostream& operator<<(std::ostream& out, const URLPattern& url_pattern) { | 186 std::ostream& operator<<(std::ostream& out, const URLPattern& url_pattern) { |
| 184 return out << '"' << url_pattern.GetAsString() << '"'; | 187 return out << '"' << url_pattern.GetAsString() << '"'; |
| 185 } | 188 } |
| 186 | 189 |
| 187 URLPattern::ParseResult URLPattern::Parse(const std::string& pattern) { | 190 URLPattern::ParseResult URLPattern::Parse(const std::string& pattern) { |
| 191 return Parse(pattern, false); |
| 192 } |
| 193 |
| 194 URLPattern::ParseResult URLPattern::Parse( |
| 195 const std::string& pattern, |
| 196 const bool allow_wildcard_effective_tld) { |
| 188 spec_.clear(); | 197 spec_.clear(); |
| 189 SetMatchAllURLs(false); | 198 SetMatchAllURLs(false); |
| 190 SetMatchSubdomains(false); | 199 SetMatchSubdomains(false); |
| 200 SetMatchEffectiveTld(true); |
| 191 SetPort("*"); | 201 SetPort("*"); |
| 192 | 202 |
| 193 // Special case pattern to match every valid URL. | 203 // Special case pattern to match every valid URL. |
| 194 if (pattern == kAllUrlsPattern) { | 204 if (pattern == kAllUrlsPattern) { |
| 195 SetMatchAllURLs(true); | 205 SetMatchAllURLs(true); |
| 196 return PARSE_SUCCESS; | 206 return PARSE_SUCCESS; |
| 197 } | 207 } |
| 198 | 208 |
| 199 // Parse out the scheme. | 209 // Parse out the scheme. |
| 200 size_t scheme_end_pos = pattern.find(url::kStandardSchemeSeparator); | 210 size_t scheme_end_pos = pattern.find(url::kStandardSchemeSeparator); |
| (...skipping 58 matching lines...) Expand 10 before | Expand all | Expand 10 after Loading... |
| 259 // Could be empty if the host only consists of whitespace characters. | 269 // Could be empty if the host only consists of whitespace characters. |
| 260 if (host_components.empty() || | 270 if (host_components.empty() || |
| 261 (host_components.size() == 1 && host_components[0].empty())) | 271 (host_components.size() == 1 && host_components[0].empty())) |
| 262 return PARSE_ERROR_EMPTY_HOST; | 272 return PARSE_ERROR_EMPTY_HOST; |
| 263 | 273 |
| 264 if (host_components[0] == "*") { | 274 if (host_components[0] == "*") { |
| 265 match_subdomains_ = true; | 275 match_subdomains_ = true; |
| 266 host_components.erase(host_components.begin(), | 276 host_components.erase(host_components.begin(), |
| 267 host_components.begin() + 1); | 277 host_components.begin() + 1); |
| 268 } | 278 } |
| 279 |
| 280 // If explicitly allowed, the last component can optionally be '*' to |
| 281 // match all effective TLDs. |
| 282 if (allow_wildcard_effective_tld && host_components.size() > 1 && |
| 283 host_components[1] == "*") { |
| 284 match_effective_tld_ = false; |
| 285 host_components.pop_back(); |
| 286 } |
| 269 host_ = base::JoinString(host_components, "."); | 287 host_ = base::JoinString(host_components, "."); |
| 270 | 288 |
| 271 path_start_pos = host_end_pos; | 289 path_start_pos = host_end_pos; |
| 272 } | 290 } |
| 273 | 291 |
| 274 SetPath(pattern.substr(path_start_pos)); | 292 SetPath(pattern.substr(path_start_pos)); |
| 275 | 293 |
| 276 size_t port_pos = host_.find(':'); | 294 size_t port_pos = host_.find(':'); |
| 277 if (port_pos != std::string::npos) { | 295 if (port_pos != std::string::npos) { |
| 278 if (!SetPort(host_.substr(port_pos + 1))) | 296 if (!SetPort(host_.substr(port_pos + 1))) |
| (...skipping 34 matching lines...) Expand 10 before | Expand all | Expand 10 after Loading... |
| 313 host_.clear(); | 331 host_.clear(); |
| 314 SetPath("/*"); | 332 SetPath("/*"); |
| 315 } | 333 } |
| 316 } | 334 } |
| 317 | 335 |
| 318 void URLPattern::SetMatchSubdomains(bool val) { | 336 void URLPattern::SetMatchSubdomains(bool val) { |
| 319 spec_.clear(); | 337 spec_.clear(); |
| 320 match_subdomains_ = val; | 338 match_subdomains_ = val; |
| 321 } | 339 } |
| 322 | 340 |
| 341 void URLPattern::SetMatchEffectiveTld(bool val) { |
| 342 spec_.clear(); |
| 343 match_effective_tld_ = val; |
| 344 } |
| 345 |
| 323 bool URLPattern::SetScheme(const std::string& scheme) { | 346 bool URLPattern::SetScheme(const std::string& scheme) { |
| 324 spec_.clear(); | 347 spec_.clear(); |
| 325 scheme_ = scheme; | 348 scheme_ = scheme; |
| 326 if (scheme_ == "*") { | 349 if (scheme_ == "*") { |
| 327 valid_schemes_ &= (SCHEME_HTTP | SCHEME_HTTPS); | 350 valid_schemes_ &= (SCHEME_HTTP | SCHEME_HTTPS); |
| 328 } else if (!IsValidScheme(scheme_)) { | 351 } else if (!IsValidScheme(scheme_)) { |
| 329 return false; | 352 return false; |
| 330 } | 353 } |
| 331 return true; | 354 return true; |
| 332 } | 355 } |
| (...skipping 79 matching lines...) Expand 10 before | Expand all | Expand 10 after Loading... |
| 412 | 435 |
| 413 bool URLPattern::MatchesHost(const std::string& host) const { | 436 bool URLPattern::MatchesHost(const std::string& host) const { |
| 414 std::string test(url::kHttpScheme); | 437 std::string test(url::kHttpScheme); |
| 415 test += url::kStandardSchemeSeparator; | 438 test += url::kStandardSchemeSeparator; |
| 416 test += host; | 439 test += host; |
| 417 test += "/"; | 440 test += "/"; |
| 418 return MatchesHost(GURL(test)); | 441 return MatchesHost(GURL(test)); |
| 419 } | 442 } |
| 420 | 443 |
| 421 bool URLPattern::MatchesHost(const GURL& test) const { | 444 bool URLPattern::MatchesHost(const GURL& test) const { |
| 422 const base::StringPiece test_host( | 445 base::StringPiece test_host(CanonicalizeHostForMatching(test.host_piece())); |
| 423 CanonicalizeHostForMatching(test.host_piece())); | |
| 424 const base::StringPiece pattern_host(CanonicalizeHostForMatching(host_)); | 446 const base::StringPiece pattern_host(CanonicalizeHostForMatching(host_)); |
| 425 | 447 |
| 448 // If we don't care about matching the effective TLD, remove it. |
| 449 if (!match_effective_tld_) { |
| 450 int reg_length = net::registry_controlled_domains::GetRegistryLength( |
| 451 test, net::registry_controlled_domains::EXCLUDE_UNKNOWN_REGISTRIES, |
| 452 net::registry_controlled_domains::EXCLUDE_PRIVATE_REGISTRIES); |
| 453 if (reg_length > 0) { |
| 454 test_host = test_host.substr(0, test_host.size() - reg_length - 1); |
| 455 } |
| 456 } |
| 457 |
| 426 // If the hosts are exactly equal, we have a match. | 458 // If the hosts are exactly equal, we have a match. |
| 427 if (test_host == pattern_host) | 459 if (test_host == pattern_host) |
| 428 return true; | 460 return true; |
| 429 | 461 |
| 430 // If we're matching subdomains, and we have no host in the match pattern, | 462 // If we're matching subdomains, and we have no host in the match pattern, |
| 431 // that means that we're matching all hosts, which means we have a match no | 463 // that means that we're matching all hosts, which means we have a match no |
| 432 // matter what the test host is. | 464 // matter what the test host is. |
| 433 if (match_subdomains_ && pattern_host.empty()) | 465 if (match_subdomains_ && pattern_host.empty()) |
| 434 return true; | 466 return true; |
| 435 | 467 |
| (...skipping 79 matching lines...) Expand 10 before | Expand all | Expand 10 after Loading... |
| 515 if (scheme_ != url::kFileScheme && standard_scheme) { | 547 if (scheme_ != url::kFileScheme && standard_scheme) { |
| 516 if (match_subdomains_) { | 548 if (match_subdomains_) { |
| 517 spec += "*"; | 549 spec += "*"; |
| 518 if (!host_.empty()) | 550 if (!host_.empty()) |
| 519 spec += "."; | 551 spec += "."; |
| 520 } | 552 } |
| 521 | 553 |
| 522 if (!host_.empty()) | 554 if (!host_.empty()) |
| 523 spec += host_; | 555 spec += host_; |
| 524 | 556 |
| 557 if (!match_effective_tld_) { |
| 558 if (!host_.empty()) |
| 559 spec += "."; |
| 560 spec += "*"; |
| 561 } |
| 562 |
| 525 if (port_ != "*") { | 563 if (port_ != "*") { |
| 526 spec += ":"; | 564 spec += ":"; |
| 527 spec += port_; | 565 spec += port_; |
| 528 } | 566 } |
| 529 } | 567 } |
| 530 | 568 |
| 531 if (!path_.empty()) | 569 if (!path_.empty()) |
| 532 spec += path_; | 570 spec += path_; |
| 533 | 571 |
| 534 spec_ = spec; | 572 spec_ = spec; |
| (...skipping 88 matching lines...) Expand 10 before | Expand all | Expand 10 after Loading... |
| 623 } | 661 } |
| 624 | 662 |
| 625 return result; | 663 return result; |
| 626 } | 664 } |
| 627 | 665 |
| 628 // static | 666 // static |
| 629 const char* URLPattern::GetParseResultString( | 667 const char* URLPattern::GetParseResultString( |
| 630 URLPattern::ParseResult parse_result) { | 668 URLPattern::ParseResult parse_result) { |
| 631 return kParseResultMessages[parse_result]; | 669 return kParseResultMessages[parse_result]; |
| 632 } | 670 } |
| OLD | NEW |