pegasus/src/Pegasus/Common/LanguageParser.cpp - annotate

Return to LanguageParser.cpp CVS log

Up to [Pegasus] / pegasus / src / Pegasus / Common

1 martin 1.35 //%LICENSE////////////////////////////////////////////////////////////////
2 martin 1.36 //
3 martin 1.35 // Licensed to The Open Group (TOG) under one or more contributor license 4 // agreements. Refer to the OpenPegasusNOTICE.txt file distributed with 5 // this work for additional information regarding copyright ownership. 6 // Each contributor licenses this file to you under the OpenPegasus Open 7 // Source License; you may not use this file except in compliance with the 8 // License.
9 martin 1.36 //
10 martin 1.35 // Permission is hereby granted, free of charge, to any person obtaining a 11 // copy of this software and associated documentation files (the "Software"), 12 // to deal in the Software without restriction, including without limitation 13 // the rights to use, copy, modify, merge, publish, distribute, sublicense, 14 // and/or sell copies of the Software, and to permit persons to whom the 15 // Software is furnished to do so, subject to the following conditions:
16 martin 1.36 //
17 martin 1.35 // The above copyright notice and this permission notice shall be included 18 // in all copies or substantial portions of the Software.
19 martin 1.36 //
20 martin 1.35 // THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS
21 martin 1.36 // OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF
22 martin 1.35 // MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. 23 // IN NO EVENT SHALL THE AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY 24 // CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, 25 // TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE 26 // SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.
27 martin 1.36 //
28 martin 1.35 //////////////////////////////////////////////////////////////////////////
29 humberto 1.8 // 30 //%///////////////////////////////////////////////////////////////////////////// 31 32 #include <Pegasus/Common/LanguageParser.h> 33 #include <Pegasus/Common/InternalException.h> 34 #include <Pegasus/Common/Tracer.h>
35 kumpf 1.30 #include <Pegasus/Common/MessageLoader.h>
36 humberto 1.8 #include <cstring> 37
38 kumpf 1.21 #ifdef PEGASUS_HAS_ICU 39 # include <unicode/locid.h> 40 #endif 41
42 humberto 1.8 PEGASUS_NAMESPACE_BEGIN 43
44 kumpf 1.21 static char LANGUAGE_TAG_SEPARATOR_CHAR = '-'; 45
46 kumpf 1.23 AcceptLanguageList LanguageParser::parseAcceptLanguageHeader(
47 kumpf 1.21 const String& acceptLanguageHeader) 48 { 49 PEG_METHOD_ENTER(TRC_L10N, "LanguageParser::parseAcceptLanguageHeader"); 50
51 kumpf 1.23 AcceptLanguageList acceptLanguages;
52 kumpf 1.21 53 try 54 { 55 Array<String> languageElements; 56 LanguageParser::_parseLanguageHeader( 57 acceptLanguageHeader, 58 languageElements); 59 60 for (Uint32 i = 0; i < languageElements.size(); i++) 61 { 62 String languageTagString; 63 Real32 qualityValue; 64 LanguageParser::_parseAcceptLanguageElement( 65 languageElements[i], languageTagString, qualityValue);
66 kumpf 1.30 acceptLanguages.insert( 67 LanguageTag(languageTagString), qualityValue);
68 kumpf 1.21 } 69 } 70 catch (Exception& e) 71 { 72 throw InvalidAcceptLanguageHeader(e.getMessage()); 73 } 74 75 PEG_METHOD_EXIT(); 76 return acceptLanguages; 77 } 78
79 kumpf 1.23 ContentLanguageList LanguageParser::parseContentLanguageHeader(
80 kumpf 1.21 const String& contentLanguageHeader) 81 { 82 PEG_METHOD_ENTER(TRC_L10N, "LanguageParser::parseContentLanguageHeader"); 83
84 kumpf 1.23 ContentLanguageList contentLanguages;
85 kumpf 1.21 86 try 87 { 88 Array<String> languageElements; 89 LanguageParser::_parseLanguageHeader( 90 contentLanguageHeader, 91 languageElements); 92 93 for (Uint32 i = 0; i < languageElements.size(); i++) 94 { 95 contentLanguages.append(LanguageTag(languageElements[i])); 96 } 97 } 98 catch (Exception& e) 99 { 100 throw InvalidContentLanguageHeader(e.getMessage()); 101 } 102 103 PEG_METHOD_EXIT(); 104 return contentLanguages; 105 } 106 kumpf 1.21 107 void LanguageParser::parseLanguageTag(
108 kumpf 1.24 const String& languageTagString,
109 kumpf 1.21 String& language, 110 String& country, 111 String& variant) 112 { 113 PEG_METHOD_ENTER(TRC_L10N, "LanguageParser::parseLanguageTag"); 114 115 language.clear(); 116 country.clear(); 117 variant.clear(); 118 119 if (languageTagString == "*") 120 { 121 // Parsing and validation is complete 122 PEG_METHOD_EXIT(); 123 return; 124 } 125 126 Boolean isStandardFormat = true; // RFC 3066 (ISO 639, ISO 3166) 127 Array<String> subtags; 128 129 _parseLanguageSubtags(subtags, languageTagString); 130 kumpf 1.21 131 // _parseLanguageSubtags() always returns at least one subtag. 132 PEGASUS_ASSERT(subtags.size() > 0); 133 134 // Validate the primary subtag. 135 // Given a languageTagString "en-US-mn" the language is "en". 136 137 language = subtags[0]; 138 139 if ((language == "i") \|\| (language == "x")) 140 { 141 // These primary tags are allowed, but are not ISO 639 compliant 142 isStandardFormat = false; 143 language.clear(); 144 } 145 else if ((language.size() != 2) && (language.size() != 3)) 146 { 147 // Except for "i" and "x", primary tags must be 2 or 3 characters, 148 // according to RFC 3066.
149 kumpf 1.33 150 // Do not localize this message; it could cause recursion.
151 kumpf 1.21 PEG_METHOD_EXIT();
152 kumpf 1.33 throw Exception(Formatter::format( 153 "Invalid language tag \"$0\".", 154 languageTagString));
155 kumpf 1.21 } 156 157 if (subtags.size() == 1) 158 { 159 // If only the primary subtag is present, we are done! 160 PEG_METHOD_EXIT(); 161 return; 162 } 163 164 // Validate the second subtag. 165 // Given a languageTagString "en-US-mn" the country is "US". 166 167 if (subtags[1].size() == 1) 168 { 169 // The second subtag may not be a single character according to 170 // RFC 3066.
171 kumpf 1.33 172 // Do not localize this message; it could cause recursion.
173 kumpf 1.21 PEG_METHOD_EXIT();
174 kumpf 1.33 throw Exception(Formatter::format( 175 "Invalid language tag \"$0\".", 176 languageTagString));
177 kumpf 1.21 } 178 179 if (isStandardFormat) 180 { 181 Uint32 variantIndex = 1; 182 183 if (subtags[1].size() == 2) 184 { 185 country = subtags[1]; 186 variantIndex = 2; 187 } 188 189 Uint32 numSubtags = subtags.size(); 190 191 if (variantIndex < numSubtags) 192 { 193 variant = subtags[variantIndex++]; 194 195 while (variantIndex < numSubtags) 196 { 197 variant.append(LANGUAGE_TAG_SEPARATOR_CHAR); 198 kumpf 1.21 variant.append(subtags[variantIndex++]); 199 } 200 } 201 } 202 203 PEG_METHOD_EXIT(); 204 } 205 206 void LanguageParser::validateQualityValue(Real32 quality) 207 { 208 if ((quality > 1.0) \|\| (quality < 0.0)) 209 { 210 MessageLoaderParms parms( 211 "Common.LanguageParser.INVALID_QUALITY_VALUE", 212 "AcceptLanguage contains an invalid quality value"); 213 throw InvalidAcceptLanguageHeader(MessageLoader::getMessage(parms)); 214 } 215 } 216 217 String LanguageParser::buildAcceptLanguageHeader(
218 kumpf 1.23 const AcceptLanguageList& acceptLanguages)
219 kumpf 1.21 { 220 String alString; 221 Uint32 numAcceptLanguages = acceptLanguages.size(); 222 223 for (Uint32 i = 0; i < numAcceptLanguages; i++) 224 { 225 alString.append(acceptLanguages.getLanguageTag(i).toString()); 226 227 Real32 q = acceptLanguages.getQualityValue(i); 228 if (q != 1.0) 229 { 230 char qValueString[6]; 231 sprintf(qValueString, "%4.3f", q); 232 alString.append(";q="); 233 alString.append(qValueString); 234 } 235 236 if (i < numAcceptLanguages - 1) 237 { 238 alString.append(","); 239 } 240 kumpf 1.21 } 241 242 return alString; 243 } 244 245 String LanguageParser::buildContentLanguageHeader(
246 kumpf 1.23 const ContentLanguageList& contentLanguages)
247 kumpf 1.21 { 248 String clString; 249 Uint32 numContentLanguages = contentLanguages.size(); 250 251 for (Uint32 i = 0; i < numContentLanguages; i++) 252 { 253 clString.append(contentLanguages.getLanguageTag(i).toString()); 254 255 if (i < numContentLanguages - 1) 256 { 257 clString.append(","); 258 } 259 } 260 261 return clString; 262 } 263
264 kumpf 1.26 #ifdef PEGASUS_HAS_ICU 265 String& LanguageParser::convertLocaleIdToLanguageTag(String& localeId) 266 {
267 marek 1.37 static char LOCALE_ID_SEPARATOR_CHAR = '_';
268 kumpf 1.26 Uint32 index = 0; 269 while ((index = localeId.find(index, LOCALE_ID_SEPARATOR_CHAR)) != 270 PEG_NOT_FOUND) 271 { 272 localeId[index] = LANGUAGE_TAG_SEPARATOR_CHAR; 273 } 274 return localeId; 275 } 276 #endif 277
278 kumpf 1.23 AcceptLanguageList LanguageParser::getDefaultAcceptLanguages()
279 kumpf 1.21 { 280 #if defined(PEGASUS_HAS_MESSAGES) && defined(PEGASUS_HAS_ICU) 281 Locale default_loc = Locale::getDefault(); 282
283 kumpf 1.26 String localeId = default_loc.getName(); 284
285 kumpf 1.21 try 286 {
287 kumpf 1.26 return LanguageParser::parseAcceptLanguageHeader( 288 convertLocaleIdToLanguageTag(localeId));
289 kumpf 1.21 } 290 catch (const InvalidAcceptLanguageHeader& e) 291 { 292 Logger::put_l(Logger::ERROR_LOG, System::CIMSERVER, Logger::SEVERE,
293 kumpf 1.34 MessageLoaderParms( 294 "src.Server.cimserver.FAILED_TO_SET_PROCESS_LOCALE", 295 "Cannot convert the system process locale into a valid " 296 "Accept-Language format."));
297 kumpf 1.21 Logger::put(Logger::ERROR_LOG, System::CIMSERVER, Logger::SEVERE, 298 e.getMessage());
299 kumpf 1.23 AcceptLanguageList al;
300 kumpf 1.22 al.insert(LanguageTag("*"), 1); 301 return al;
302 kumpf 1.21 }
303 kumpf 1.22 #else
304 kumpf 1.23 return AcceptLanguageList();
305 kumpf 1.21 #endif 306 } 307 308 void LanguageParser::_parseLanguageHeader( 309 const String& languageHeaderValue, 310 Array<String>& languageElements) 311 { 312 PEG_METHOD_ENTER(TRC_L10N, "LanguageParser::_parseLanguageHeader"); 313 314 languageElements.clear(); 315 String element; 316 317 for (Uint32 i=0, len=languageHeaderValue.size(); i<len; i++) 318 { 319 Char16 nextChar = languageHeaderValue[i]; 320 321 if (isascii(nextChar) && isspace(nextChar)) 322 { 323 // Ignore whitespace 324 } 325 else if (nextChar == '(') 326 kumpf 1.21 { 327 // Ignore comments 328 while (i < len) 329 { 330 // Search for the closing parenthesis 331 if (languageHeaderValue[i] == ')') 332 { 333 break; 334 } 335 336 // Skip over escape characters 337 if (languageHeaderValue[i] == '\\') 338 { 339 i++; 340 } 341 342 i++; 343 } 344 345 // Check for a missing closing parenthesis 346 if (i >= len) 347 kumpf 1.21 { 348 MessageLoaderParms parms( 349 "Common.LanguageParser.DOES_NOT_CONTAIN_TERMINATING", 350 "Closing \")\" character is missing."); 351 throw Exception(MessageLoader::getMessage(parms)); 352 } 353 } 354 else if (nextChar == ',') 355 { 356 // Check for the end of the element 357 languageElements.append(element); 358 element.clear(); 359 } 360 else 361 { 362 // Unescape an escape character 363 if ((nextChar == '\\') && (i < len-1)) 364 { 365 nextChar = languageHeaderValue[++i]; 366 } 367 368 kumpf 1.21 // Include this character in the value 369 element.append(nextChar); 370 } 371 } 372 373 // Include the last element in the languageElements array 374 languageElements.append(element); 375 376 PEG_METHOD_EXIT(); 377 } 378 379 void LanguageParser::_parseAcceptLanguageElement( 380 const String& acceptLanguageElement, 381 String& languageTag, 382 Real32& quality) 383 { 384 PEG_METHOD_ENTER(TRC_L10N, "LanguageParser::_parseAcceptLanguageElement"); 385 386 // look for ';' in acceptLanguageElement, that means we have a 387 // quality value to capture. If not present, we only have a language tag. 388 389 kumpf 1.21 Uint32 semicolonIndex = acceptLanguageElement.find(";"); 390 if (semicolonIndex != PEG_NOT_FOUND) 391 { 392 // Separate the language tag and quality value 393 394 String qualityString = 395 acceptLanguageElement.subString(semicolonIndex+1); 396 languageTag = acceptLanguageElement.subString(0, semicolonIndex); 397 398 // Parse the quality value
399 humberto 1.8
400 kumpf 1.21 char dummyChar; 401 int scanfConversions = sscanf( 402 qualityString.getCString(), 403 "q=%f%c", &quality, &dummyChar); 404 405 if ((scanfConversions != 1) \|\| 406 (qualityString.size() > 7)) 407 { 408 MessageLoaderParms parms( 409 "Common.LanguageParser.INVALID_QUALITY_VALUE", 410 "AcceptLanguage contains an invalid quality value"); 411 PEG_METHOD_EXIT(); 412 throw Exception(MessageLoader::getMessage(parms)); 413 } 414 } 415 else 416 { 417 languageTag = acceptLanguageElement; 418 quality = 1.0; 419 } 420 421 kumpf 1.21 PEG_METHOD_EXIT(); 422 } 423 424 void LanguageParser::_parseLanguageSubtags( 425 Array<String>& subtags, 426 const String& languageTagString) 427 { 428 PEG_METHOD_ENTER(TRC_L10N, "LanguageParser::_parseLanguageSubtags"); 429 430 // Parse the language tag into subtags 431 432 Uint32 subtagIndex = 0; 433 Uint32 separatorIndex; 434 while ((separatorIndex = languageTagString.find( 435 subtagIndex, LANGUAGE_TAG_SEPARATOR_CHAR)) != PEG_NOT_FOUND) 436 { 437 subtags.append(languageTagString.subString( 438 subtagIndex, separatorIndex - subtagIndex)); 439 subtagIndex = separatorIndex + 1; 440 } 441 subtags.append(languageTagString.subString(subtagIndex)); 442 kumpf 1.21 443 // Validate the syntax of each of the subtags 444 445 for (Uint32 i = 0, n = subtags.size(); i < n; i++) 446 { 447 if (((i == 0) && !_isValidPrimarySubtagSyntax(subtags[i])) \|\| 448 ((i > 0) && !_isValidSubtagSyntax(subtags[i]))) 449 {
450 marek 1.38 // throw Exception(MessageLoader::getMessage(parms)); 451 // do not localize message, requires a language tag for this 452 // localization can cause recursion here 453 // MessageLoaderParms::toString adds 5kb static code size, Do NOT 454 // include in non-debug builds 455 #ifdef PEGASUS_DEBUG
456 kumpf 1.21 MessageLoaderParms parms( 457 "Common.LanguageParser.MALFORMED_LANGUAGE_TAG", 458 "Malformed language tag \"$0\".", languageTagString); 459 PEG_METHOD_EXIT();
460 marek 1.28 throw Exception(parms.toString());
461 marek 1.38 #else 462 String malFormed("Malformed language tag:"); 463 malFormed.append(languageTagString); 464 PEG_METHOD_EXIT(); 465 throw Exception(malFormed); 466 #endif
467 kumpf 1.21 } 468 } 469 470 PEG_METHOD_EXIT();
471 humberto 1.8 } 472
473 kumpf 1.21 Boolean LanguageParser::_isValidPrimarySubtagSyntax(const String& subtag) 474 { 475 if ((subtag.size() == 0) \|\| (subtag.size() > 8)) 476 { 477 return false; 478 } 479 480 for (Uint32 i = 0, n = subtag.size(); i < n; i++) 481 { 482 if (!(isascii(subtag[i]) && isalpha(subtag[i]))) 483 { 484 return false; 485 } 486 } 487 488 return true; 489 } 490 491 Boolean LanguageParser::_isValidSubtagSyntax(const String& subtag) 492 { 493 if ((subtag.size() == 0) \|\| (subtag.size() > 8)) 494 kumpf 1.21 { 495 return false; 496 } 497 498 for (Uint32 i = 0, n = subtag.size(); i < n; i++) 499 { 500 if (!(isascii(subtag[i]) && isalnum(subtag[i]))) 501 { 502 return false; 503 } 504 } 505 506 return true; 507 }
508 humberto 1.8 509 PEGASUS_NAMESPACE_END

No CVS admin address has been configured