OLD | NEW |
(Empty) | |
| 1 // Copyright 2007, Google Inc. |
| 2 // All rights reserved. |
| 3 // |
| 4 // Redistribution and use in source and binary forms, with or without |
| 5 // modification, are permitted provided that the following conditions are |
| 6 // met: |
| 7 // |
| 8 // * Redistributions of source code must retain the above copyright |
| 9 // notice, this list of conditions and the following disclaimer. |
| 10 // * Redistributions in binary form must reproduce the above |
| 11 // copyright notice, this list of conditions and the following disclaimer |
| 12 // in the documentation and/or other materials provided with the |
| 13 // distribution. |
| 14 // * Neither the name of Google Inc. nor the names of its |
| 15 // contributors may be used to endorse or promote products derived from |
| 16 // this software without specific prior written permission. |
| 17 // |
| 18 // THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS |
| 19 // "AS IS" AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT |
| 20 // LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR |
| 21 // A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT |
| 22 // OWNER OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, |
| 23 // SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT |
| 24 // LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, |
| 25 // DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY |
| 26 // THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT |
| 27 // (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE |
| 28 // OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE. |
| 29 #ifndef GOOGLEURL_SRC_URL_CANON_H__ |
| 30 #define GOOGLEURL_SRC_URL_CANON_H__ |
| 31 |
| 32 #include <string.h> |
| 33 #include <stdlib.h> |
| 34 |
| 35 #include "base/string16.h" |
| 36 #include "googleurl/src/url_common.h" |
| 37 #include "googleurl/src/url_parse.h" |
| 38 |
| 39 namespace url_canon { |
| 40 |
| 41 // Canonicalizer output ------------------------------------------------------- |
| 42 |
| 43 // Base class for the canonicalizer output, this maintains a buffer and |
| 44 // supports simple resizing and append operations on it. |
| 45 // |
| 46 // It is VERY IMPORTANT that no virtual function calls be made on the common |
| 47 // code path. We only have two virtual function calls, the destructor and a |
| 48 // resize function that is called when the existing buffer is not big enough. |
| 49 // The derived class is then in charge of setting up our buffer which we will |
| 50 // manage. |
| 51 template<typename T> |
| 52 class CanonOutputT { |
| 53 public: |
| 54 CanonOutputT() : buffer_(NULL), buffer_len_(0), cur_len_(0) { |
| 55 } |
| 56 virtual ~CanonOutputT() { |
| 57 } |
| 58 |
| 59 // Implemented to resize the buffer. This function should update the buffer |
| 60 // pointer to point to the new buffer, and any old data up to |cur_len_| in |
| 61 // the buffer must be copied over. |
| 62 // |
| 63 // The new size |sz| must be larger than buffer_len_. |
| 64 virtual void Resize(int sz) = 0; |
| 65 |
| 66 // Accessor for returning a character at a given position. The input offset |
| 67 // must be in the valid range. |
| 68 inline char at(int offset) const { |
| 69 return buffer_[offset]; |
| 70 } |
| 71 |
| 72 // Sets the character at the given position. The given position MUST be less |
| 73 // than the length(). |
| 74 inline void set(int offset, int ch) { |
| 75 buffer_[offset] = ch; |
| 76 } |
| 77 |
| 78 // Returns the number of characters currently in the buffer. |
| 79 inline int length() const { |
| 80 return cur_len_; |
| 81 } |
| 82 |
| 83 // Returns the current capacity of the buffer. The length() is the number of |
| 84 // characters that have been declared to be written, but the capacity() is |
| 85 // the number that can be written without reallocation. If the caller must |
| 86 // write many characters at once, it can make sure there is enough capacity, |
| 87 // write the data, then use set_size() to declare the new length(). |
| 88 int capacity() const { |
| 89 return buffer_len_; |
| 90 } |
| 91 |
| 92 // Called by the user of this class to get the output. The output will NOT |
| 93 // be NULL-terminated. Call length() to get the |
| 94 // length. |
| 95 const T* data() const { |
| 96 return buffer_; |
| 97 } |
| 98 T* data() { |
| 99 return buffer_; |
| 100 } |
| 101 |
| 102 // Shortens the URL to the new length. Used for "backing up" when processing |
| 103 // relative paths. This can also be used if an external function writes a lot |
| 104 // of data to the buffer (when using the "Raw" version below) beyond the end, |
| 105 // to declare the new length. |
| 106 // |
| 107 // This MUST NOT be used to expand the size of the buffer beyond capacity(). |
| 108 void set_length(int new_len) { |
| 109 cur_len_ = new_len; |
| 110 } |
| 111 |
| 112 // This is the most performance critical function, since it is called for |
| 113 // every character. |
| 114 void push_back(T ch) { |
| 115 // In VC2005, putting this common case first speeds up execution |
| 116 // dramatically because this branch is predicted as taken. |
| 117 if (cur_len_ < buffer_len_) { |
| 118 buffer_[cur_len_] = ch; |
| 119 cur_len_++; |
| 120 return; |
| 121 } |
| 122 |
| 123 // Grow the buffer to hold at least one more item. Hopefully we won't have |
| 124 // to do this very often. |
| 125 if (!Grow(1)) |
| 126 return; |
| 127 |
| 128 // Actually do the insertion. |
| 129 buffer_[cur_len_] = ch; |
| 130 cur_len_++; |
| 131 } |
| 132 |
| 133 // Appends the given string to the output. |
| 134 void Append(const T* str, int str_len) { |
| 135 if (cur_len_ + str_len > buffer_len_) { |
| 136 if (!Grow(cur_len_ + str_len - buffer_len_)) |
| 137 return; |
| 138 } |
| 139 for (int i = 0; i < str_len; i++) |
| 140 buffer_[cur_len_ + i] = str[i]; |
| 141 cur_len_ += str_len; |
| 142 } |
| 143 |
| 144 protected: |
| 145 // Grows the given buffer so that it can fit at least |min_additional| |
| 146 // characters. Returns true if the buffer could be resized, false on OOM. |
| 147 bool Grow(int min_additional) { |
| 148 static const int kMinBufferLen = 16; |
| 149 int new_len = (buffer_len_ == 0) ? kMinBufferLen : buffer_len_; |
| 150 do { |
| 151 if (new_len >= (1 << 30)) // Prevent overflow below. |
| 152 return false; |
| 153 new_len *= 2; |
| 154 } while (new_len < buffer_len_ + min_additional); |
| 155 Resize(new_len); |
| 156 return true; |
| 157 } |
| 158 |
| 159 T* buffer_; |
| 160 int buffer_len_; |
| 161 |
| 162 // Used characters in the buffer. |
| 163 int cur_len_; |
| 164 }; |
| 165 |
| 166 // Simple implementation of the CanonOutput using new[]. This class |
| 167 // also supports a static buffer so if it is allocated on the stack, most |
| 168 // URLs can be canonicalized with no heap allocations. |
| 169 template<typename T, int fixed_capacity = 1024> |
| 170 class RawCanonOutputT : public CanonOutputT<T> { |
| 171 public: |
| 172 RawCanonOutputT() : CanonOutputT<T>() { |
| 173 this->buffer_ = fixed_buffer_; |
| 174 this->buffer_len_ = fixed_capacity; |
| 175 } |
| 176 virtual ~RawCanonOutputT() { |
| 177 if (this->buffer_ != fixed_buffer_) |
| 178 delete[] this->buffer_; |
| 179 } |
| 180 |
| 181 virtual void Resize(int sz) { |
| 182 T* new_buf = new T[sz]; |
| 183 memcpy(new_buf, this->buffer_, |
| 184 sizeof(T) * (this->cur_len_ < sz ? this->cur_len_ : sz)); |
| 185 if (this->buffer_ != fixed_buffer_) |
| 186 delete[] this->buffer_; |
| 187 this->buffer_ = new_buf; |
| 188 this->buffer_len_ = sz; |
| 189 } |
| 190 |
| 191 protected: |
| 192 T fixed_buffer_[fixed_capacity]; |
| 193 }; |
| 194 |
| 195 // Normally, all canonicalization output is in narrow characters. We support |
| 196 // the templates so it can also be used internally if a wide buffer is |
| 197 // required. |
| 198 typedef CanonOutputT<char> CanonOutput; |
| 199 typedef CanonOutputT<char16> CanonOutputW; |
| 200 |
| 201 template<int fixed_capacity> |
| 202 class RawCanonOutput : public RawCanonOutputT<char, fixed_capacity> {}; |
| 203 template<int fixed_capacity> |
| 204 class RawCanonOutputW : public RawCanonOutputT<char16, fixed_capacity> {}; |
| 205 |
| 206 // Character set converter ---------------------------------------------------- |
| 207 // |
| 208 // Converts query strings into a custom encoding. The embedder can supply an |
| 209 // implementation of this class to interface with their own character set |
| 210 // conversion libraries. |
| 211 // |
| 212 // Embedders will want to see the unit test for the ICU version. |
| 213 |
| 214 class CharsetConverter { |
| 215 public: |
| 216 CharsetConverter() {} |
| 217 virtual ~CharsetConverter() {} |
| 218 |
| 219 // Converts the given input string from UTF-16 to whatever output format the |
| 220 // converter supports. This is used only for the query encoding conversion, |
| 221 // which does not fail. Instead, the converter should insert "invalid |
| 222 // character" characters in the output for invalid sequences, and do the |
| 223 // best it can. |
| 224 // |
| 225 // If the input contains a character not representable in the output |
| 226 // character set, the converter should append the HTML entity sequence in |
| 227 // decimal, (such as "你") with escaping of the ampersand, number |
| 228 // sign, and semicolon (in the previous example it would be |
| 229 // "%26%2320320%3B"). This rule is based on what IE does in this situation. |
| 230 virtual void ConvertFromUTF16(const char16* input, |
| 231 int input_len, |
| 232 CanonOutput* output) = 0; |
| 233 }; |
| 234 |
| 235 // Whitespace ----------------------------------------------------------------- |
| 236 |
| 237 // Searches for whitespace that should be removed from the middle of URLs, and |
| 238 // removes it. Removed whitespace are tabs and newlines, but NOT spaces. Spaces |
| 239 // are preserved, which is what most browsers do. A pointer to the output will |
| 240 // be returned, and the length of that output will be in |output_len|. |
| 241 // |
| 242 // This should be called before parsing if whitespace removal is desired (which |
| 243 // it normally is when you are canonicalizing). |
| 244 // |
| 245 // If no whitespace is removed, this function will not use the buffer and will |
| 246 // return a pointer to the input, to avoid the extra copy. If modification is |
| 247 // required, the given |buffer| will be used and the returned pointer will |
| 248 // point to the beginning of the buffer. |
| 249 // |
| 250 // Therefore, callers should not use the buffer, since it may actuall be empty, |
| 251 // use the computed pointer and |*output_len| instead. |
| 252 GURL_API const char* RemoveURLWhitespace(const char* input, int input_len, |
| 253 CanonOutputT<char>* buffer, |
| 254 int* output_len); |
| 255 GURL_API const char16* RemoveURLWhitespace(const char16* input, int input_len, |
| 256 CanonOutputT<char16>* buffer, |
| 257 int* output_len); |
| 258 |
| 259 // IDN ------------------------------------------------------------------------ |
| 260 |
| 261 // Converts the Unicode input representing a hostname to ASCII using IDN rules. |
| 262 // The output must fall in the ASCII range, but will be encoded in UTF-16. |
| 263 // |
| 264 // On success, the output will be filled with the ASCII host name and it will |
| 265 // return true. Unlike most other canonicalization functions, this assumes that |
| 266 // the output is empty. The beginning of the host will be at offset 0, and |
| 267 // the length of the output will be set to the length of the new host name. |
| 268 // |
| 269 // On error, returns false. The output in this case is undefined. |
| 270 GURL_API bool IDNToASCII(const char16* src, int src_len, CanonOutputW* output); |
| 271 |
| 272 // Piece-by-piece canonicalizers ---------------------------------------------- |
| 273 // |
| 274 // These individual canonicalizers append the canonicalized versions of the |
| 275 // corresponding URL component to the given std::string. The spec and the |
| 276 // previously-identified range of that component are the input. The range of |
| 277 // the canonicalized component will be written to the output component. |
| 278 // |
| 279 // These functions all append to the output so they can be chained. Make sure |
| 280 // the output is empty when you start. |
| 281 // |
| 282 // These functions returns boolean values indicating success. On failure, they |
| 283 // will attempt to write something reasonable to the output so that, if |
| 284 // displayed to the user, they will recognise it as something that's messed up. |
| 285 // Nothing more should ever be done with these invalid URLs, however. |
| 286 |
| 287 // Scheme: Appends the scheme and colon to the URL. The output component will |
| 288 // indicate the range of characters up to but not including the colon. |
| 289 // |
| 290 // Canonical URLs always have a scheme. If the scheme is not present in the |
| 291 // input, this will just write the colon to indicate an empty scheme. Does not |
| 292 // append slashes which will be needed before any authority components for most |
| 293 // URLs. |
| 294 // |
| 295 // The 8-bit version requires UTF-8 encoding. |
| 296 GURL_API bool CanonicalizeScheme(const char* spec, |
| 297 const url_parse::Component& scheme, |
| 298 CanonOutput* output, |
| 299 url_parse::Component* out_scheme); |
| 300 GURL_API bool CanonicalizeScheme(const char16* spec, |
| 301 const url_parse::Component& scheme, |
| 302 CanonOutput* output, |
| 303 url_parse::Component* out_scheme); |
| 304 |
| 305 // User info: username/password. If present, this will add the delimiters so |
| 306 // the output will be "<username>:<password>@" or "<username>@". Empty |
| 307 // username/password pairs, or empty passwords, will get converted to |
| 308 // nonexistant in the canonical version. |
| 309 // |
| 310 // The components for the username and password refer to ranges in the |
| 311 // respective source strings. Usually, these will be the same string, which |
| 312 // is legal as long as the two components don't overlap. |
| 313 // |
| 314 // The 8-bit version requires UTF-8 encoding. |
| 315 GURL_API bool CanonicalizeUserInfo(const char* username_source, |
| 316 const url_parse::Component& username, |
| 317 const char* password_source, |
| 318 const url_parse::Component& password, |
| 319 CanonOutput* output, |
| 320 url_parse::Component* out_username, |
| 321 url_parse::Component* out_password); |
| 322 GURL_API bool CanonicalizeUserInfo(const char16* username_source, |
| 323 const url_parse::Component& username, |
| 324 const char16* password_source, |
| 325 const url_parse::Component& password, |
| 326 CanonOutput* output, |
| 327 url_parse::Component* out_username, |
| 328 url_parse::Component* out_password); |
| 329 |
| 330 |
| 331 // This structure holds detailed state exported from the IP/Host canonicalizers. |
| 332 // Additional fields may be added as callers require them. |
| 333 struct CanonHostInfo { |
| 334 CanonHostInfo() : family(NEUTRAL), num_ipv4_components(0), out_host() {} |
| 335 |
| 336 // Convenience function to test if family is an IP address. |
| 337 bool IsIPAddress() const { return family == IPV4 || family == IPV6; } |
| 338 |
| 339 // This field summarizes how the input was classified by the canonicalizer. |
| 340 enum Family { |
| 341 NEUTRAL, // - Doesn't resemble an IP address. As far as the IP |
| 342 // canonicalizer is concerned, it should be treated as a |
| 343 // hostname. |
| 344 BROKEN, // - Almost an IP, but was not canonicalized. This could be an |
| 345 // IPv4 address where truncation occurred, or something |
| 346 // containing the special characters :[] which did not parse |
| 347 // as an IPv6 address. Never attempt to connect to this |
| 348 // address, because it might actually succeed! |
| 349 IPV4, // - Successfully canonicalized as an IPv4 address. |
| 350 IPV6, // - Successfully canonicalized as an IPv6 address. |
| 351 }; |
| 352 Family family; |
| 353 |
| 354 // If |family| is IPV4, then this is the number of nonempty dot-separated |
| 355 // components in the input text, from 1 to 4. If |family| is not IPV4, |
| 356 // this value is undefined. |
| 357 int num_ipv4_components; |
| 358 |
| 359 // Location of host within the canonicalized output. |
| 360 // CanonicalizeIPAddress() only sets this field if |family| is IPV4 or IPV6. |
| 361 // CanonicalizeHostVerbose() always sets it. |
| 362 url_parse::Component out_host; |
| 363 |
| 364 // |address| contains the parsed IP Address (if any) in its first |
| 365 // AddressLength() bytes, in network order. If IsIPAddress() is false |
| 366 // AddressLength() will return zero and the content of |address| is undefined. |
| 367 unsigned char address[16]; |
| 368 |
| 369 // Convenience function to calculate the length of an IP address corresponding |
| 370 // to the current IP version in |family|, if any. For use with |address|. |
| 371 int AddressLength() const { |
| 372 return family == IPV4 ? 4 : (family == IPV6 ? 16 : 0); |
| 373 } |
| 374 }; |
| 375 |
| 376 |
| 377 // Host. |
| 378 // |
| 379 // The 8-bit version requires UTF-8 encoding. Use this version when you only |
| 380 // need to know whether canonicalization succeeded. |
| 381 GURL_API bool CanonicalizeHost(const char* spec, |
| 382 const url_parse::Component& host, |
| 383 CanonOutput* output, |
| 384 url_parse::Component* out_host); |
| 385 GURL_API bool CanonicalizeHost(const char16* spec, |
| 386 const url_parse::Component& host, |
| 387 CanonOutput* output, |
| 388 url_parse::Component* out_host); |
| 389 |
| 390 // Extended version of CanonicalizeHost, which returns additional information. |
| 391 // Use this when you need to know whether the hostname was an IP address. |
| 392 // A successful return is indicated by host_info->family != BROKEN. See the |
| 393 // definition of CanonHostInfo above for details. |
| 394 GURL_API void CanonicalizeHostVerbose(const char* spec, |
| 395 const url_parse::Component& host, |
| 396 CanonOutput* output, |
| 397 CanonHostInfo* host_info); |
| 398 GURL_API void CanonicalizeHostVerbose(const char16* spec, |
| 399 const url_parse::Component& host, |
| 400 CanonOutput* output, |
| 401 CanonHostInfo* host_info); |
| 402 |
| 403 |
| 404 // IP addresses. |
| 405 // |
| 406 // Tries to interpret the given host name as an IPv4 or IPv6 address. If it is |
| 407 // an IP address, it will canonicalize it as such, appending it to |output|. |
| 408 // Additional status information is returned via the |*host_info| parameter. |
| 409 // See the definition of CanonHostInfo above for details. |
| 410 // |
| 411 // This is called AUTOMATICALLY from the host canonicalizer, which ensures that |
| 412 // the input is unescaped and name-prepped, etc. It should not normally be |
| 413 // necessary or wise to call this directly. |
| 414 GURL_API void CanonicalizeIPAddress(const char* spec, |
| 415 const url_parse::Component& host, |
| 416 CanonOutput* output, |
| 417 CanonHostInfo* host_info); |
| 418 GURL_API void CanonicalizeIPAddress(const char16* spec, |
| 419 const url_parse::Component& host, |
| 420 CanonOutput* output, |
| 421 CanonHostInfo* host_info); |
| 422 |
| 423 // Port: this function will add the colon for the port if a port is present. |
| 424 // The caller can pass url_parse::PORT_UNSPECIFIED as the |
| 425 // default_port_for_scheme argument if there is no default port. |
| 426 // |
| 427 // The 8-bit version requires UTF-8 encoding. |
| 428 GURL_API bool CanonicalizePort(const char* spec, |
| 429 const url_parse::Component& port, |
| 430 int default_port_for_scheme, |
| 431 CanonOutput* output, |
| 432 url_parse::Component* out_port); |
| 433 GURL_API bool CanonicalizePort(const char16* spec, |
| 434 const url_parse::Component& port, |
| 435 int default_port_for_scheme, |
| 436 CanonOutput* output, |
| 437 url_parse::Component* out_port); |
| 438 |
| 439 // Returns the default port for the given canonical scheme, or PORT_UNSPECIFIED |
| 440 // if the scheme is unknown. |
| 441 GURL_API int DefaultPortForScheme(const char* scheme, int scheme_len); |
| 442 |
| 443 // Path. If the input does not begin in a slash (including if the input is |
| 444 // empty), we'll prepend a slash to the path to make it canonical. |
| 445 // |
| 446 // The 8-bit version assumes UTF-8 encoding, but does not verify the validity |
| 447 // of the UTF-8 (i.e., you can have invalid UTF-8 sequences, invalid |
| 448 // characters, etc.). Normally, URLs will come in as UTF-16, so this isn't |
| 449 // an issue. Somebody giving us an 8-bit path is responsible for generating |
| 450 // the path that the server expects (we'll escape high-bit characters), so |
| 451 // if something is invalid, it's their problem. |
| 452 GURL_API bool CanonicalizePath(const char* spec, |
| 453 const url_parse::Component& path, |
| 454 CanonOutput* output, |
| 455 url_parse::Component* out_path); |
| 456 GURL_API bool CanonicalizePath(const char16* spec, |
| 457 const url_parse::Component& path, |
| 458 CanonOutput* output, |
| 459 url_parse::Component* out_path); |
| 460 |
| 461 // Canonicalizes the input as a file path. This is like CanonicalizePath except |
| 462 // that it also handles Windows drive specs. For example, the path can begin |
| 463 // with "c|\" and it will get properly canonicalized to "C:/". |
| 464 // The string will be appended to |*output| and |*out_path| will be updated. |
| 465 // |
| 466 // The 8-bit version requires UTF-8 encoding. |
| 467 GURL_API bool FileCanonicalizePath(const char* spec, |
| 468 const url_parse::Component& path, |
| 469 CanonOutput* output, |
| 470 url_parse::Component* out_path); |
| 471 GURL_API bool FileCanonicalizePath(const char16* spec, |
| 472 const url_parse::Component& path, |
| 473 CanonOutput* output, |
| 474 url_parse::Component* out_path); |
| 475 |
| 476 // Query: Prepends the ? if needed. |
| 477 // |
| 478 // The 8-bit version requires the input to be UTF-8 encoding. Incorrectly |
| 479 // encoded characters (in UTF-8 or UTF-16) will be replaced with the Unicode |
| 480 // "invalid character." This function can not fail, we always just try to do |
| 481 // our best for crazy input here since web pages can set it themselves. |
| 482 // |
| 483 // This will convert the given input into the output encoding that the given |
| 484 // character set converter object provides. The converter will only be called |
| 485 // if necessary, for ASCII input, no conversions are necessary. |
| 486 // |
| 487 // The converter can be NULL. In this case, the output encoding will be UTF-8. |
| 488 GURL_API void CanonicalizeQuery(const char* spec, |
| 489 const url_parse::Component& query, |
| 490 CharsetConverter* converter, |
| 491 CanonOutput* output, |
| 492 url_parse::Component* out_query); |
| 493 GURL_API void CanonicalizeQuery(const char16* spec, |
| 494 const url_parse::Component& query, |
| 495 CharsetConverter* converter, |
| 496 CanonOutput* output, |
| 497 url_parse::Component* out_query); |
| 498 |
| 499 // Ref: Prepends the # if needed. The output will be UTF-8 (this is the only |
| 500 // canonicalizer that does not produce ASCII output). The output is |
| 501 // guaranteed to be valid UTF-8. |
| 502 // |
| 503 // This function will not fail. If the input is invalid UTF-8/UTF-16, we'll use |
| 504 // the "Unicode replacement character" for the confusing bits and copy the rest. |
| 505 GURL_API void CanonicalizeRef(const char* spec, |
| 506 const url_parse::Component& path, |
| 507 CanonOutput* output, |
| 508 url_parse::Component* out_path); |
| 509 GURL_API void CanonicalizeRef(const char16* spec, |
| 510 const url_parse::Component& path, |
| 511 CanonOutput* output, |
| 512 url_parse::Component* out_path); |
| 513 |
| 514 // Full canonicalizer --------------------------------------------------------- |
| 515 // |
| 516 // These functions replace any string contents, rather than append as above. |
| 517 // See the above piece-by-piece functions for information specific to |
| 518 // canonicalizing individual components. |
| 519 // |
| 520 // The output will be ASCII except the reference fragment, which may be UTF-8. |
| 521 // |
| 522 // The 8-bit versions require UTF-8 encoding. |
| 523 |
| 524 // Use for standard URLs with authorities and paths. |
| 525 GURL_API bool CanonicalizeStandardURL(const char* spec, |
| 526 int spec_len, |
| 527 const url_parse::Parsed& parsed, |
| 528 CharsetConverter* query_converter, |
| 529 CanonOutput* output, |
| 530 url_parse::Parsed* new_parsed); |
| 531 GURL_API bool CanonicalizeStandardURL(const char16* spec, |
| 532 int spec_len, |
| 533 const url_parse::Parsed& parsed, |
| 534 CharsetConverter* query_converter, |
| 535 CanonOutput* output, |
| 536 url_parse::Parsed* new_parsed); |
| 537 |
| 538 // Use for file URLs. |
| 539 GURL_API bool CanonicalizeFileURL(const char* spec, |
| 540 int spec_len, |
| 541 const url_parse::Parsed& parsed, |
| 542 CharsetConverter* query_converter, |
| 543 CanonOutput* output, |
| 544 url_parse::Parsed* new_parsed); |
| 545 GURL_API bool CanonicalizeFileURL(const char16* spec, |
| 546 int spec_len, |
| 547 const url_parse::Parsed& parsed, |
| 548 CharsetConverter* query_converter, |
| 549 CanonOutput* output, |
| 550 url_parse::Parsed* new_parsed); |
| 551 |
| 552 // Use for filesystem URLs. |
| 553 GURL_API bool CanonicalizeFileSystemURL(const char* spec, |
| 554 int spec_len, |
| 555 const url_parse::Parsed& parsed, |
| 556 CharsetConverter* query_converter, |
| 557 CanonOutput* output, |
| 558 url_parse::Parsed* new_parsed); |
| 559 GURL_API bool CanonicalizeFileSystemURL(const char16* spec, |
| 560 int spec_len, |
| 561 const url_parse::Parsed& parsed, |
| 562 CharsetConverter* query_converter, |
| 563 CanonOutput* output, |
| 564 url_parse::Parsed* new_parsed); |
| 565 |
| 566 // Use for path URLs such as javascript. This does not modify the path in any |
| 567 // way, for example, by escaping it. |
| 568 GURL_API bool CanonicalizePathURL(const char* spec, |
| 569 int spec_len, |
| 570 const url_parse::Parsed& parsed, |
| 571 CanonOutput* output, |
| 572 url_parse::Parsed* new_parsed); |
| 573 GURL_API bool CanonicalizePathURL(const char16* spec, |
| 574 int spec_len, |
| 575 const url_parse::Parsed& parsed, |
| 576 CanonOutput* output, |
| 577 url_parse::Parsed* new_parsed); |
| 578 |
| 579 // Use for mailto URLs. This "canonicalizes" the url into a path and query |
| 580 // component. It does not attempt to merge "to" fields. It uses UTF-8 for |
| 581 // the query encoding if there is a query. This is because a mailto URL is |
| 582 // really intended for an external mail program, and the encoding of a page, |
| 583 // etc. which would influence a query encoding normally are irrelevant. |
| 584 GURL_API bool CanonicalizeMailtoURL(const char* spec, |
| 585 int spec_len, |
| 586 const url_parse::Parsed& parsed, |
| 587 CanonOutput* output, |
| 588 url_parse::Parsed* new_parsed); |
| 589 GURL_API bool CanonicalizeMailtoURL(const char16* spec, |
| 590 int spec_len, |
| 591 const url_parse::Parsed& parsed, |
| 592 CanonOutput* output, |
| 593 url_parse::Parsed* new_parsed); |
| 594 |
| 595 // Part replacer -------------------------------------------------------------- |
| 596 |
| 597 // Internal structure used for storing separate strings for each component. |
| 598 // The basic canonicalization functions use this structure internally so that |
| 599 // component replacement (different strings for different components) can be |
| 600 // treated on the same code path as regular canonicalization (the same string |
| 601 // for each component). |
| 602 // |
| 603 // A url_parse::Parsed structure usually goes along with this. Those |
| 604 // components identify offsets within these strings, so that they can all be |
| 605 // in the same string, or spread arbitrarily across different ones. |
| 606 // |
| 607 // This structures does not own any data. It is the caller's responsibility to |
| 608 // ensure that the data the pointers point to stays in scope and is not |
| 609 // modified. |
| 610 template<typename CHAR> |
| 611 struct URLComponentSource { |
| 612 // Constructor normally used by callers wishing to replace components. This |
| 613 // will make them all NULL, which is no replacement. The caller would then |
| 614 // override the components they want to replace. |
| 615 URLComponentSource() |
| 616 : scheme(NULL), |
| 617 username(NULL), |
| 618 password(NULL), |
| 619 host(NULL), |
| 620 port(NULL), |
| 621 path(NULL), |
| 622 query(NULL), |
| 623 ref(NULL) { |
| 624 } |
| 625 |
| 626 // Constructor normally used internally to initialize all the components to |
| 627 // point to the same spec. |
| 628 explicit URLComponentSource(const CHAR* default_value) |
| 629 : scheme(default_value), |
| 630 username(default_value), |
| 631 password(default_value), |
| 632 host(default_value), |
| 633 port(default_value), |
| 634 path(default_value), |
| 635 query(default_value), |
| 636 ref(default_value) { |
| 637 } |
| 638 |
| 639 const CHAR* scheme; |
| 640 const CHAR* username; |
| 641 const CHAR* password; |
| 642 const CHAR* host; |
| 643 const CHAR* port; |
| 644 const CHAR* path; |
| 645 const CHAR* query; |
| 646 const CHAR* ref; |
| 647 }; |
| 648 |
| 649 // This structure encapsulates information on modifying a URL. Each component |
| 650 // may either be left unchanged, replaced, or deleted. |
| 651 // |
| 652 // By default, each component is unchanged. For those components that should be |
| 653 // modified, call either Set* or Clear* to modify it. |
| 654 // |
| 655 // The string passed to Set* functions DOES NOT GET COPIED AND MUST BE KEPT |
| 656 // IN SCOPE BY THE CALLER for as long as this object exists! |
| 657 // |
| 658 // Prefer the 8-bit replacement version if possible since it is more efficient. |
| 659 template<typename CHAR> |
| 660 class Replacements { |
| 661 public: |
| 662 Replacements() { |
| 663 } |
| 664 |
| 665 // Scheme |
| 666 void SetScheme(const CHAR* s, const url_parse::Component& comp) { |
| 667 sources_.scheme = s; |
| 668 components_.scheme = comp; |
| 669 } |
| 670 // Note: we don't have a ClearScheme since this doesn't make any sense. |
| 671 bool IsSchemeOverridden() const { return sources_.scheme != NULL; } |
| 672 |
| 673 // Username |
| 674 void SetUsername(const CHAR* s, const url_parse::Component& comp) { |
| 675 sources_.username = s; |
| 676 components_.username = comp; |
| 677 } |
| 678 void ClearUsername() { |
| 679 sources_.username = Placeholder(); |
| 680 components_.username = url_parse::Component(); |
| 681 } |
| 682 bool IsUsernameOverridden() const { return sources_.username != NULL; } |
| 683 |
| 684 // Password |
| 685 void SetPassword(const CHAR* s, const url_parse::Component& comp) { |
| 686 sources_.password = s; |
| 687 components_.password = comp; |
| 688 } |
| 689 void ClearPassword() { |
| 690 sources_.password = Placeholder(); |
| 691 components_.password = url_parse::Component(); |
| 692 } |
| 693 bool IsPasswordOverridden() const { return sources_.password != NULL; } |
| 694 |
| 695 // Host |
| 696 void SetHost(const CHAR* s, const url_parse::Component& comp) { |
| 697 sources_.host = s; |
| 698 components_.host = comp; |
| 699 } |
| 700 void ClearHost() { |
| 701 sources_.host = Placeholder(); |
| 702 components_.host = url_parse::Component(); |
| 703 } |
| 704 bool IsHostOverridden() const { return sources_.host != NULL; } |
| 705 |
| 706 // Port |
| 707 void SetPort(const CHAR* s, const url_parse::Component& comp) { |
| 708 sources_.port = s; |
| 709 components_.port = comp; |
| 710 } |
| 711 void ClearPort() { |
| 712 sources_.port = Placeholder(); |
| 713 components_.port = url_parse::Component(); |
| 714 } |
| 715 bool IsPortOverridden() const { return sources_.port != NULL; } |
| 716 |
| 717 // Path |
| 718 void SetPath(const CHAR* s, const url_parse::Component& comp) { |
| 719 sources_.path = s; |
| 720 components_.path = comp; |
| 721 } |
| 722 void ClearPath() { |
| 723 sources_.path = Placeholder(); |
| 724 components_.path = url_parse::Component(); |
| 725 } |
| 726 bool IsPathOverridden() const { return sources_.path != NULL; } |
| 727 |
| 728 // Query |
| 729 void SetQuery(const CHAR* s, const url_parse::Component& comp) { |
| 730 sources_.query = s; |
| 731 components_.query = comp; |
| 732 } |
| 733 void ClearQuery() { |
| 734 sources_.query = Placeholder(); |
| 735 components_.query = url_parse::Component(); |
| 736 } |
| 737 bool IsQueryOverridden() const { return sources_.query != NULL; } |
| 738 |
| 739 // Ref |
| 740 void SetRef(const CHAR* s, const url_parse::Component& comp) { |
| 741 sources_.ref = s; |
| 742 components_.ref = comp; |
| 743 } |
| 744 void ClearRef() { |
| 745 sources_.ref = Placeholder(); |
| 746 components_.ref = url_parse::Component(); |
| 747 } |
| 748 bool IsRefOverridden() const { return sources_.ref != NULL; } |
| 749 |
| 750 // Getters for the itnernal data. See the variables below for how the |
| 751 // information is encoded. |
| 752 const URLComponentSource<CHAR>& sources() const { return sources_; } |
| 753 const url_parse::Parsed& components() const { return components_; } |
| 754 |
| 755 private: |
| 756 // Returns a pointer to a static empty string that is used as a placeholder |
| 757 // to indicate a component should be deleted (see below). |
| 758 const CHAR* Placeholder() { |
| 759 static const CHAR empty_string = 0; |
| 760 return &empty_string; |
| 761 } |
| 762 |
| 763 // We support three states: |
| 764 // |
| 765 // Action | Source Component |
| 766 // -----------------------+-------------------------------------------------- |
| 767 // Don't change component | NULL (unused) |
| 768 // Replace component | (replacement string) (replacement component) |
| 769 // Delete component | (non-NULL) (invalid component: (0,-1)) |
| 770 // |
| 771 // We use a pointer to the empty string for the source when the component |
| 772 // should be deleted. |
| 773 URLComponentSource<CHAR> sources_; |
| 774 url_parse::Parsed components_; |
| 775 }; |
| 776 |
| 777 // The base must be an 8-bit canonical URL. |
| 778 GURL_API bool ReplaceStandardURL(const char* base, |
| 779 const url_parse::Parsed& base_parsed, |
| 780 const Replacements<char>& replacements, |
| 781 CharsetConverter* query_converter, |
| 782 CanonOutput* output, |
| 783 url_parse::Parsed* new_parsed); |
| 784 GURL_API bool ReplaceStandardURL(const char* base, |
| 785 const url_parse::Parsed& base_parsed, |
| 786 const Replacements<char16>& replacements, |
| 787 CharsetConverter* query_converter, |
| 788 CanonOutput* output, |
| 789 url_parse::Parsed* new_parsed); |
| 790 |
| 791 // Filesystem URLs can only have the path, query, or ref replaced. |
| 792 // All other components will be ignored. |
| 793 GURL_API bool ReplaceFileSystemURL(const char* base, |
| 794 const url_parse::Parsed& base_parsed, |
| 795 const Replacements<char>& replacements, |
| 796 CharsetConverter* query_converter, |
| 797 CanonOutput* output, |
| 798 url_parse::Parsed* new_parsed); |
| 799 GURL_API bool ReplaceFileSystemURL(const char* base, |
| 800 const url_parse::Parsed& base_parsed, |
| 801 const Replacements<char16>& replacements, |
| 802 CharsetConverter* query_converter, |
| 803 CanonOutput* output, |
| 804 url_parse::Parsed* new_parsed); |
| 805 |
| 806 // Replacing some parts of a file URL is not permitted. Everything except |
| 807 // the host, path, query, and ref will be ignored. |
| 808 GURL_API bool ReplaceFileURL(const char* base, |
| 809 const url_parse::Parsed& base_parsed, |
| 810 const Replacements<char>& replacements, |
| 811 CharsetConverter* query_converter, |
| 812 CanonOutput* output, |
| 813 url_parse::Parsed* new_parsed); |
| 814 GURL_API bool ReplaceFileURL(const char* base, |
| 815 const url_parse::Parsed& base_parsed, |
| 816 const Replacements<char16>& replacements, |
| 817 CharsetConverter* query_converter, |
| 818 CanonOutput* output, |
| 819 url_parse::Parsed* new_parsed); |
| 820 |
| 821 // Path URLs can only have the scheme and path replaced. All other components |
| 822 // will be ignored. |
| 823 GURL_API bool ReplacePathURL(const char* base, |
| 824 const url_parse::Parsed& base_parsed, |
| 825 const Replacements<char>& replacements, |
| 826 CanonOutput* output, |
| 827 url_parse::Parsed* new_parsed); |
| 828 GURL_API bool ReplacePathURL(const char* base, |
| 829 const url_parse::Parsed& base_parsed, |
| 830 const Replacements<char16>& replacements, |
| 831 CanonOutput* output, |
| 832 url_parse::Parsed* new_parsed); |
| 833 |
| 834 // Mailto URLs can only have the scheme, path, and query replaced. |
| 835 // All other components will be ignored. |
| 836 GURL_API bool ReplaceMailtoURL(const char* base, |
| 837 const url_parse::Parsed& base_parsed, |
| 838 const Replacements<char>& replacements, |
| 839 CanonOutput* output, |
| 840 url_parse::Parsed* new_parsed); |
| 841 GURL_API bool ReplaceMailtoURL(const char* base, |
| 842 const url_parse::Parsed& base_parsed, |
| 843 const Replacements<char16>& replacements, |
| 844 CanonOutput* output, |
| 845 url_parse::Parsed* new_parsed); |
| 846 |
| 847 // Relative URL --------------------------------------------------------------- |
| 848 |
| 849 // Given an input URL or URL fragment |fragment|, determines if it is a |
| 850 // relative or absolute URL and places the result into |*is_relative|. If it is |
| 851 // relative, the relevant portion of the URL will be placed into |
| 852 // |*relative_component| (there may have been trimmed whitespace, for example). |
| 853 // This value is passed to ResolveRelativeURL. If the input is not relative, |
| 854 // this value is UNDEFINED (it may be changed by the function). |
| 855 // |
| 856 // Returns true on success (we successfully determined the URL is relative or |
| 857 // not). Failure means that the combination of URLs doesn't make any sense. |
| 858 // |
| 859 // The base URL should always be canonical, therefore is ASCII. |
| 860 GURL_API bool IsRelativeURL(const char* base, |
| 861 const url_parse::Parsed& base_parsed, |
| 862 const char* fragment, |
| 863 int fragment_len, |
| 864 bool is_base_hierarchical, |
| 865 bool* is_relative, |
| 866 url_parse::Component* relative_component); |
| 867 GURL_API bool IsRelativeURL(const char* base, |
| 868 const url_parse::Parsed& base_parsed, |
| 869 const char16* fragment, |
| 870 int fragment_len, |
| 871 bool is_base_hierarchical, |
| 872 bool* is_relative, |
| 873 url_parse::Component* relative_component); |
| 874 |
| 875 // Given a canonical parsed source URL, a URL fragment known to be relative, |
| 876 // and the identified relevant portion of the relative URL (computed by |
| 877 // IsRelativeURL), this produces a new parsed canonical URL in |output| and |
| 878 // |out_parsed|. |
| 879 // |
| 880 // It also requires a flag indicating whether the base URL is a file: URL |
| 881 // which triggers additional logic. |
| 882 // |
| 883 // The base URL should be canonical and have a host (may be empty for file |
| 884 // URLs) and a path. If it doesn't have these, we can't resolve relative |
| 885 // URLs off of it and will return the base as the output with an error flag. |
| 886 // Becausee it is canonical is should also be ASCII. |
| 887 // |
| 888 // The query charset converter follows the same rules as CanonicalizeQuery. |
| 889 // |
| 890 // Returns true on success. On failure, the output will be "something |
| 891 // reasonable" that will be consistent and valid, just probably not what |
| 892 // was intended by the web page author or caller. |
| 893 GURL_API bool ResolveRelativeURL(const char* base_url, |
| 894 const url_parse::Parsed& base_parsed, |
| 895 bool base_is_file, |
| 896 const char* relative_url, |
| 897 const url_parse::Component& relative_component, |
| 898 CharsetConverter* query_converter, |
| 899 CanonOutput* output, |
| 900 url_parse::Parsed* out_parsed); |
| 901 GURL_API bool ResolveRelativeURL(const char* base_url, |
| 902 const url_parse::Parsed& base_parsed, |
| 903 bool base_is_file, |
| 904 const char16* relative_url, |
| 905 const url_parse::Component& relative_component, |
| 906 CharsetConverter* query_converter, |
| 907 CanonOutput* output, |
| 908 url_parse::Parsed* out_parsed); |
| 909 |
| 910 } // namespace url_canon |
| 911 |
| 912 #endif // GOOGLEURL_SRC_URL_CANON_H__ |
OLD | NEW |