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#include "base/logging.h" 6#include "url/url_canon.h" 7#include "url/url_canon_internal.h" 8#include "url/url_parse_internal.h" 9 10namespace url { 11 12namespace { 13 14enum CharacterFlags { 15 // Pass through unchanged, whether escaped or unescaped. This doesn't 16 // actually set anything so you can't OR it to check, it's just to make the 17 // table below more clear when neither ESCAPE or UNESCAPE is set. 18 PASS = 0, 19 20 // This character requires special handling in DoPartialPath. Doing this test 21 // first allows us to filter out the common cases of regular characters that 22 // can be directly copied. 23 SPECIAL = 1, 24 25 // This character must be escaped in the canonical output. Note that all 26 // escaped chars also have the "special" bit set so that the code that looks 27 // for this is triggered. Not valid with PASS or ESCAPE 28 ESCAPE_BIT = 2, 29 ESCAPE = ESCAPE_BIT | SPECIAL, 30 31 // This character must be unescaped in canonical output. Not valid with 32 // ESCAPE or PASS. We DON'T set the SPECIAL flag since if we encounter these 33 // characters unescaped, they should just be copied. 34 UNESCAPE = 4, 35 36 // This character is disallowed in URLs. Note that the "special" bit is also 37 // set to trigger handling. 38 INVALID_BIT = 8, 39 INVALID = INVALID_BIT | SPECIAL, 40}; 41 42// This table contains one of the above flag values. Note some flags are more 43// than one bits because they also turn on the "special" flag. Special is the 44// only flag that may be combined with others. 45// 46// This table is designed to match exactly what IE does with the characters. 47// 48// Dot is even more special, and the escaped version is handled specially by 49// IsDot. Therefore, we don't need the "escape" flag, and even the "unescape" 50// bit is never handled (we just need the "special") bit. 51const unsigned char kPathCharLookup[0x100] = { 52// NULL control chars... 53 INVALID, ESCAPE, ESCAPE, ESCAPE, ESCAPE, ESCAPE, ESCAPE, ESCAPE, ESCAPE, ESCAPE, ESCAPE, ESCAPE, ESCAPE, ESCAPE, ESCAPE, ESCAPE, 54// control chars... 55 ESCAPE, ESCAPE, ESCAPE, ESCAPE, ESCAPE, ESCAPE, ESCAPE, ESCAPE, ESCAPE, ESCAPE, ESCAPE, ESCAPE, ESCAPE, ESCAPE, ESCAPE, ESCAPE, 56// ' ' ! " # $ % & ' ( ) * + , - . / 57 ESCAPE, PASS, ESCAPE, ESCAPE, PASS, ESCAPE, PASS, PASS, PASS, PASS, PASS, PASS, PASS, UNESCAPE,SPECIAL, PASS, 58// 0 1 2 3 4 5 6 7 8 9 : ; < = > ? 59 UNESCAPE,UNESCAPE,UNESCAPE,UNESCAPE,UNESCAPE,UNESCAPE,UNESCAPE,UNESCAPE,UNESCAPE,UNESCAPE,PASS, PASS, ESCAPE, PASS, ESCAPE, ESCAPE, 60// @ A B C D E F G H I J K L M N O 61 PASS, UNESCAPE,UNESCAPE,UNESCAPE,UNESCAPE,UNESCAPE,UNESCAPE,UNESCAPE,UNESCAPE,UNESCAPE,UNESCAPE,UNESCAPE,UNESCAPE,UNESCAPE,UNESCAPE,UNESCAPE, 62// P Q R S T U V W X Y Z [ \ ] ^ _ 63 UNESCAPE,UNESCAPE,UNESCAPE,UNESCAPE,UNESCAPE,UNESCAPE,UNESCAPE,UNESCAPE,UNESCAPE,UNESCAPE,UNESCAPE,PASS, ESCAPE, PASS, ESCAPE, UNESCAPE, 64// ` a b c d e f g h i j k l m n o 65 ESCAPE, UNESCAPE,UNESCAPE,UNESCAPE,UNESCAPE,UNESCAPE,UNESCAPE,UNESCAPE,UNESCAPE,UNESCAPE,UNESCAPE,UNESCAPE,UNESCAPE,UNESCAPE,UNESCAPE,UNESCAPE, 66// p q r s t u v w x y z { | } ~ <NBSP> 67 UNESCAPE,UNESCAPE,UNESCAPE,UNESCAPE,UNESCAPE,UNESCAPE,UNESCAPE,UNESCAPE,UNESCAPE,UNESCAPE,UNESCAPE,ESCAPE, ESCAPE, ESCAPE, UNESCAPE,ESCAPE, 68// ...all the high-bit characters are escaped 69 ESCAPE, ESCAPE, ESCAPE, ESCAPE, ESCAPE, ESCAPE, ESCAPE, ESCAPE, ESCAPE, ESCAPE, ESCAPE, ESCAPE, ESCAPE, ESCAPE, ESCAPE, ESCAPE, 70 ESCAPE, ESCAPE, ESCAPE, ESCAPE, ESCAPE, ESCAPE, ESCAPE, ESCAPE, ESCAPE, ESCAPE, ESCAPE, ESCAPE, ESCAPE, ESCAPE, ESCAPE, ESCAPE, 71 ESCAPE, ESCAPE, ESCAPE, ESCAPE, ESCAPE, ESCAPE, ESCAPE, ESCAPE, ESCAPE, ESCAPE, ESCAPE, ESCAPE, ESCAPE, ESCAPE, ESCAPE, ESCAPE, 72 ESCAPE, ESCAPE, ESCAPE, ESCAPE, ESCAPE, ESCAPE, ESCAPE, ESCAPE, ESCAPE, ESCAPE, ESCAPE, ESCAPE, ESCAPE, ESCAPE, ESCAPE, ESCAPE, 73 ESCAPE, ESCAPE, ESCAPE, ESCAPE, ESCAPE, ESCAPE, ESCAPE, ESCAPE, ESCAPE, ESCAPE, ESCAPE, ESCAPE, ESCAPE, ESCAPE, ESCAPE, ESCAPE, 74 ESCAPE, ESCAPE, ESCAPE, ESCAPE, ESCAPE, ESCAPE, ESCAPE, ESCAPE, ESCAPE, ESCAPE, ESCAPE, ESCAPE, ESCAPE, ESCAPE, ESCAPE, ESCAPE, 75 ESCAPE, ESCAPE, ESCAPE, ESCAPE, ESCAPE, ESCAPE, ESCAPE, ESCAPE, ESCAPE, ESCAPE, ESCAPE, ESCAPE, ESCAPE, ESCAPE, ESCAPE, ESCAPE, 76 ESCAPE, ESCAPE, ESCAPE, ESCAPE, ESCAPE, ESCAPE, ESCAPE, ESCAPE, ESCAPE, ESCAPE, ESCAPE, ESCAPE, ESCAPE, ESCAPE, ESCAPE, ESCAPE}; 77 78enum DotDisposition { 79 // The given dot is just part of a filename and is not special. 80 NOT_A_DIRECTORY, 81 82 // The given dot is the current directory. 83 DIRECTORY_CUR, 84 85 // The given dot is the first of a double dot that should take us up one. 86 DIRECTORY_UP 87}; 88 89// When the path resolver finds a dot, this function is called with the 90// character following that dot to see what it is. The return value 91// indicates what type this dot is (see above). This code handles the case 92// where the dot is at the end of the input. 93// 94// |*consumed_len| will contain the number of characters in the input that 95// express what we found. 96// 97// If the input is "../foo", |after_dot| = 1, |end| = 6, and 98// at the end, |*consumed_len| = 2 for the "./" this function consumed. The 99// original dot length should be handled by the caller. 100template<typename CHAR> 101DotDisposition ClassifyAfterDot(const CHAR* spec, int after_dot, 102 int end, int* consumed_len) { 103 if (after_dot == end) { 104 // Single dot at the end. 105 *consumed_len = 0; 106 return DIRECTORY_CUR; 107 } 108 if (IsURLSlash(spec[after_dot])) { 109 // Single dot followed by a slash. 110 *consumed_len = 1; // Consume the slash 111 return DIRECTORY_CUR; 112 } 113 114 int second_dot_len = IsDot(spec, after_dot, end); 115 if (second_dot_len) { 116 int after_second_dot = after_dot + second_dot_len; 117 if (after_second_dot == end) { 118 // Double dot at the end. 119 *consumed_len = second_dot_len; 120 return DIRECTORY_UP; 121 } 122 if (IsURLSlash(spec[after_second_dot])) { 123 // Double dot followed by a slash. 124 *consumed_len = second_dot_len + 1; 125 return DIRECTORY_UP; 126 } 127 } 128 129 // The dots are followed by something else, not a directory. 130 *consumed_len = 0; 131 return NOT_A_DIRECTORY; 132} 133 134// Rewinds the output to the previous slash. It is assumed that the output 135// ends with a slash and this doesn't count (we call this when we are 136// appending directory paths, so the previous path component has and ending 137// slash). 138// 139// This will stop at the first slash (assumed to be at position 140// |path_begin_in_output| and not go any higher than that. Some web pages 141// do ".." too many times, so we need to handle that brokenness. 142// 143// It searches for a literal slash rather than including a backslash as well 144// because it is run only on the canonical output. 145// 146// The output is guaranteed to end in a slash when this function completes. 147void BackUpToPreviousSlash(int path_begin_in_output, 148 CanonOutput* output) { 149 DCHECK(output->length() > 0); 150 151 int i = output->length() - 1; 152 DCHECK(output->at(i) == '/'); 153 if (i == path_begin_in_output) 154 return; // We're at the first slash, nothing to do. 155 156 // Now back up (skipping the trailing slash) until we find another slash. 157 i--; 158 while (output->at(i) != '/' && i > path_begin_in_output) 159 i--; 160 161 // Now shrink the output to just include that last slash we found. 162 output->set_length(i + 1); 163} 164 165// Appends the given path to the output. It assumes that if the input path 166// starts with a slash, it should be copied to the output. If no path has 167// already been appended to the output (the case when not resolving 168// relative URLs), the path should begin with a slash. 169// 170// If there are already path components (this mode is used when appending 171// relative paths for resolving), it assumes that the output already has 172// a trailing slash and that if the input begins with a slash, it should be 173// copied to the output. 174// 175// We do not collapse multiple slashes in a row to a single slash. It seems 176// no web browsers do this, and we don't want incompababilities, even though 177// it would be correct for most systems. 178template<typename CHAR, typename UCHAR> 179bool DoPartialPath(const CHAR* spec, 180 const Component& path, 181 int path_begin_in_output, 182 CanonOutput* output) { 183 int end = path.end(); 184 185 bool success = true; 186 for (int i = path.begin; i < end; i++) { 187 UCHAR uch = static_cast<UCHAR>(spec[i]); 188 if (sizeof(CHAR) > sizeof(char) && uch >= 0x80) { 189 // We only need to test wide input for having non-ASCII characters. For 190 // narrow input, we'll always just use the lookup table. We don't try to 191 // do anything tricky with decoding/validating UTF-8. This function will 192 // read one or two UTF-16 characters and append the output as UTF-8. This 193 // call will be removed in 8-bit mode. 194 success &= AppendUTF8EscapedChar(spec, &i, end, output); 195 } else { 196 // Normal ASCII character or 8-bit input, use the lookup table. 197 unsigned char out_ch = static_cast<unsigned char>(uch); 198 unsigned char flags = kPathCharLookup[out_ch]; 199 if (flags & SPECIAL) { 200 // Needs special handling of some sort. 201 int dotlen; 202 if ((dotlen = IsDot(spec, i, end)) > 0) { 203 // See if this dot was preceeded by a slash in the output. We 204 // assume that when canonicalizing paths, they will always 205 // start with a slash and not a dot, so we don't have to 206 // bounds check the output. 207 // 208 // Note that we check this in the case of dots so we don't have to 209 // special case slashes. Since slashes are much more common than 210 // dots, this actually increases performance measurably (though 211 // slightly). 212 DCHECK(output->length() > path_begin_in_output); 213 if (output->length() > path_begin_in_output && 214 output->at(output->length() - 1) == '/') { 215 // Slash followed by a dot, check to see if this is means relative 216 int consumed_len; 217 switch (ClassifyAfterDot<CHAR>(spec, i + dotlen, end, 218 &consumed_len)) { 219 case NOT_A_DIRECTORY: 220 // Copy the dot to the output, it means nothing special. 221 output->push_back('.'); 222 i += dotlen - 1; 223 break; 224 case DIRECTORY_CUR: // Current directory, just skip the input. 225 i += dotlen + consumed_len - 1; 226 break; 227 case DIRECTORY_UP: 228 BackUpToPreviousSlash(path_begin_in_output, output); 229 i += dotlen + consumed_len - 1; 230 break; 231 } 232 } else { 233 // This dot is not preceeded by a slash, it is just part of some 234 // file name. 235 output->push_back('.'); 236 i += dotlen - 1; 237 } 238 239 } else if (out_ch == '\\') { 240 // Convert backslashes to forward slashes 241 output->push_back('/'); 242 243 } else if (out_ch == '%') { 244 // Handle escape sequences. 245 unsigned char unescaped_value; 246 if (DecodeEscaped(spec, &i, end, &unescaped_value)) { 247 // Valid escape sequence, see if we keep, reject, or unescape it. 248 char unescaped_flags = kPathCharLookup[unescaped_value]; 249 250 if (unescaped_flags & UNESCAPE) { 251 // This escaped value shouldn't be escaped, copy it. 252 output->push_back(unescaped_value); 253 } else if (unescaped_flags & INVALID_BIT) { 254 // Invalid escaped character, copy it and remember the error. 255 output->push_back('%'); 256 output->push_back(static_cast<char>(spec[i - 1])); 257 output->push_back(static_cast<char>(spec[i])); 258 success = false; 259 } else { 260 // Valid escaped character but we should keep it escaped. We 261 // don't want to change the case of any hex letters in case 262 // the server is sensitive to that, so we just copy the two 263 // characters without checking (DecodeEscape will have advanced 264 // to the last character of the pair). 265 output->push_back('%'); 266 output->push_back(static_cast<char>(spec[i - 1])); 267 output->push_back(static_cast<char>(spec[i])); 268 } 269 } else { 270 // Invalid escape sequence. IE7 rejects any URLs with such 271 // sequences, while Firefox, IE6, and Safari all pass it through 272 // unchanged. We are more permissive unlike IE7. I don't think this 273 // can cause significant problems, if it does, we should change 274 // to be more like IE7. 275 output->push_back('%'); 276 } 277 278 } else if (flags & INVALID_BIT) { 279 // For NULLs, etc. fail. 280 AppendEscapedChar(out_ch, output); 281 success = false; 282 283 } else if (flags & ESCAPE_BIT) { 284 // This character should be escaped. 285 AppendEscapedChar(out_ch, output); 286 } 287 } else { 288 // Nothing special about this character, just append it. 289 output->push_back(out_ch); 290 } 291 } 292 } 293 return success; 294} 295 296template<typename CHAR, typename UCHAR> 297bool DoPath(const CHAR* spec, 298 const Component& path, 299 CanonOutput* output, 300 Component* out_path) { 301 bool success = true; 302 out_path->begin = output->length(); 303 if (path.len > 0) { 304 // Write out an initial slash if the input has none. If we just parse a URL 305 // and then canonicalize it, it will of course have a slash already. This 306 // check is for the replacement and relative URL resolving cases of file 307 // URLs. 308 if (!IsURLSlash(spec[path.begin])) 309 output->push_back('/'); 310 311 success = DoPartialPath<CHAR, UCHAR>(spec, path, out_path->begin, output); 312 } else { 313 // No input, canonical path is a slash. 314 output->push_back('/'); 315 } 316 out_path->len = output->length() - out_path->begin; 317 return success; 318} 319 320} // namespace 321 322bool CanonicalizePath(const char* spec, 323 const Component& path, 324 CanonOutput* output, 325 Component* out_path) { 326 return DoPath<char, unsigned char>(spec, path, output, out_path); 327} 328 329bool CanonicalizePath(const base::char16* spec, 330 const Component& path, 331 CanonOutput* output, 332 Component* out_path) { 333 return DoPath<base::char16, base::char16>(spec, path, output, out_path); 334} 335 336bool CanonicalizePartialPath(const char* spec, 337 const Component& path, 338 int path_begin_in_output, 339 CanonOutput* output) { 340 return DoPartialPath<char, unsigned char>(spec, path, path_begin_in_output, 341 output); 342} 343 344bool CanonicalizePartialPath(const base::char16* spec, 345 const Component& path, 346 int path_begin_in_output, 347 CanonOutput* output) { 348 return DoPartialPath<base::char16, base::char16>(spec, path, 349 path_begin_in_output, 350 output); 351} 352 353} // namespace url 354