(C) An X-Ryl669 project 2007
This document describes Unlimited Zooming Interface source code. UZI stands for Unlimited Zooming Interface, and source code license is
00001 #ifndef hpp_CPP_Parser_CPP_hpp 00002 #define hpp_CPP_Parser_CPP_hpp 00003 00004 00005 /* 00006 ============================================================================== 00007 00008 HTML Parser for UZI 00009 00010 ============================================================================== 00011 */ 00012 00013 // We need DTD types 00014 #include "DTDTypes.hpp" 00015 // We need generic elements 00016 #include "GenericElements.hpp" 00017 // We need generic attributes 00018 #include "GenericAttributes.hpp" 00019 // We need streams 00020 #include "../Streams/Streams.hpp" 00021 // We need trees 00022 #include "../Tree/NTree.hpp" 00023 // We need allocators too 00024 #include "Allocator.hpp" 00025 // We need HTML entity tagging too 00026 #include "HTMLEscape.hpp" 00027 00029 namespace HTML 00030 { 00032 class BuildRadix 00033 { 00034 private: 00036 uint32 radix; 00037 public: 00041 inline void addToRadix(tchar ch) throw() { if (ch >= 'A' && ch <= 'Z') radix |= 1 << (ch - 'A'); 00042 if (ch >= 'a' && ch <= 'z') radix |= 1 << (ch - 'a'); } 00043 00048 inline bool checkAgainstLowercase(tchar ch) const throw() { return (radix & (1 << (ch - 'a'))) > 0; } 00049 00054 inline bool checkAgainstUppercase(tchar ch) const throw() { return (radix & (1 << (ch - 'A'))) > 0; } 00055 00057 BuildRadix() : radix(0) {} 00058 }; 00059 00138 class Parser 00139 { 00140 public: 00142 enum ParsingTime 00143 { 00144 InstantParsing = 0, 00145 DelayedParsing = 1, 00146 }; 00147 00151 public: 00153 static const BuildRadix radixElements; 00154 private: 00156 Stream::MemoryBufferedStream inputStream; 00161 struct ParsingError 00162 { 00164 enum ErrorCode 00165 { 00166 Valid = 0, 00167 00169 WarningBadStructure = 0x00000001, 00170 WarningBadEntity = 0x00000002, 00171 WarningUnknownAttribute = 0x00000003, 00172 WarningBadElement = 0x00000004, 00173 WarningBadAttributeValue = 0x00000005, 00174 WarningDifferentLengths = 0x00000006, 00175 WarningNotHTML = 0x00000007, 00176 WarningEndTagExpected = 0x00000008, 00177 WarningEndTagUnexpected = 0x00000009, 00178 WarningEndTagMismatch = 0x0000000A, 00179 WarningStartTagInstead = 0x0000000B, 00180 WarningStartTagUnexpected = 0x0000000C, 00181 00183 ErrorBadStructure = 0x80000001, 00184 ErrorBadEntity = 0x80000002, 00185 ErrorUnknownAttribute = 0x80000003, 00186 ErrorBadElement = 0x80000004, 00187 ErrorBadAttributeValue = 0x80000005, 00188 ErrorDocumentToBig = 0x80000006, 00189 ErrorInvalidInputStream = 0x80000700, 00190 ErrorNotEnoughMemory = 0x80000701, 00191 00192 ErrorNotHTML = 0x80000711, 00193 00195 ErrorMask = 0x80000000, 00196 00197 }; 00198 00200 ErrorCode errorCode; 00202 uint32 errorPosition; 00204 uint32 errorLength; 00206 ParsingError * nextError; 00207 00208 00210 ParsingError(const ErrorCode & code, uint32 pos, uint32 len) : errorCode(code), errorPosition(pos), errorLength(len), nextError(0) {} 00212 /*virtual*/ ~ParsingError() { if (nextError) delete nextError; nextError = 0; } 00214 void chainError(ParsingError * newError) 00215 { 00216 if (errorCode == Valid) 00217 { 00218 errorCode = newError->errorCode; 00219 errorPosition = newError->errorPosition; 00220 errorLength = newError->errorLength; 00221 nextError = newError->nextError; 00222 newError->nextError = 0; 00223 delete newError; 00224 } 00225 else if (nextError == 0) 00226 nextError = newError; 00227 else 00228 nextError->chainError(newError); 00229 } 00231 const tchar * getErrorString() const 00232 { 00233 // The following regular expression make the conversion 00234 // ^\:b+\(\:w\)\:b+= 0x\:h,\:b+//!<\ (.+\)$ 00235 // case \1 : return "\2"; 00236 switch(errorCode) 00237 { 00238 case Valid: 00239 return "No error, the document is valid"; 00240 case WarningBadStructure: 00241 return "The document is badly structured, see the content of errorPosition for details where"; 00242 case WarningBadEntity: 00243 return "The document contains a bad entity (not declared in current standard), see content of errorPosition for details where"; 00244 case WarningUnknownAttribute: 00245 return "One attribute for an element is unknown, see the content of errorPosition for details where"; 00246 case WarningBadElement: 00247 return "The document contains a bad element (not declared in current standard), see content of errorPosition for details where"; 00248 case WarningBadAttributeValue: 00249 return "The document contains a bad attribute value (not declared in current standard), see content of errorPosition for details where"; 00250 case WarningDifferentLengths: 00251 return "The stream length is not the same as the detected text length, see errorPosition for detected text length and errorLength for declared stream length"; 00252 case WarningNotHTML: 00253 return "The document doesn't start by an HTML element as root"; 00254 case WarningEndTagExpected: 00255 return "An end tag was expected, see the content of errorPosition for details where"; 00256 case WarningEndTagUnexpected: 00257 return "An unexpected end tag has been found, see the content of errorPosition for details where"; 00258 case WarningEndTagMismatch: 00259 return "An mismatching end tag has been found, see the content of errorPosition for details where"; 00260 case WarningStartTagInstead: 00261 return "An end tag was expected but a start tag has been found, see the content of errorPosition for details where"; 00262 case WarningStartTagUnexpected: 00263 return "An unexpected start tag has been found, see the content of errorPosition for details where"; 00264 00265 00266 case ErrorBadStructure: 00267 return "The document is badly structured, see the content of errorPosition for details where"; 00268 case ErrorBadEntity: 00269 return "The document contains a bad entity (not declared in current standard), see content of errorPosition for details where"; 00270 case ErrorUnknownAttribute: 00271 return "One attribute for an element is unknown, see the content of errorPosition for details where"; 00272 case ErrorBadElement: 00273 return "The document contains a bad element (not declared in current standard), see content of errorPosition for details where"; 00274 case ErrorBadAttributeValue: 00275 return "The document contains a bad attribute value (not declared in current standard), see content of errorPosition for details where"; 00276 case ErrorDocumentToBig: 00277 return "The document is far too big to be parsed, set errorLength for declared document size"; 00278 case ErrorInvalidInputStream: 00279 return "The input stream is invalid or can't be read"; 00280 case ErrorNotEnoughMemory: 00281 return "Not enough memory to allocate elements"; 00282 case ErrorNotHTML: 00283 return "The given document is clearly not HTML"; 00284 default: 00285 return "This error code is unknown"; 00286 } 00287 } 00289 inline bool isError() const 00290 { 00291 if ((uint32)errorCode & (uint32)ErrorMask) return true; 00292 else if (nextError) return nextError->isError(); 00293 return false; 00294 } 00296 inline bool isWarning() const 00297 { 00298 if (((uint32)errorCode & (uint32)ErrorMask) == 0 && errorCode != Valid ) return true; 00299 else if (nextError) return nextError->isWarning(); 00300 return false; 00301 } 00303 inline uint32 chainedErrorCount() const 00304 { 00305 if (nextError) return nextError->chainedErrorCount() + 1; 00306 return 1; 00307 } 00308 }; 00309 00311 DOMTree DOMtree; 00313 const CharsetFunctions & charsetFunctions; 00317 DOM::NodeDeleter parserDeleter; 00319 ParsingTime parsingTime; 00321 HTML::DTDType chosenDTD; 00323 ParsingError lastParsingError; 00324 00325 00326 public: 00333 Parser(Stream::InputStream & inputStreamRef, HTML::Elements::Allocators::BaseAllocator & allocatorRef, const ParsingTime & whenToParse = InstantParsing, const HTML::DTDType & dtd = HTML::StandardDTD); 00335 ~Parser() 00336 { 00337 // The DOM tree is automaticcally deleted by the DOM::NodeDeleter instance 00338 } 00339 00340 00344 const ParsingError & Parse(); 00345 00347 inline const ParsingError & getLastParsingError() const { return lastParsingError; } 00348 00350 inline HTML::DOMTree & getDOMTree() { return DOMtree; } 00352 inline const HTML::DOMTree & getDOMTree() const { return DOMtree; } 00353 00357 const unsigned char * getUnboundedAccessToStream(const uint32 startPos) { if (startPos < inputStream.fullSize()) return inputStream.getBuffer() + startPos; return 0; } 00358 00359 // Helpers 00360 private: 00362 uint32 parseAttribute(const tchar * buffer, uint32 & currentPosition, int & remainingLength, int & attribStartPos, int & attribEndPos, int & attribContentStart, int & attribContentEnd); 00364 void parseAttributeIntoNode(DOMTree::Node * newNode, const tchar * buffer, uint32 & currentPosition, int & remainingLength, int & attribStartPos, int & attribEndPos, int & attribContentStart, int & attribContentEnd); 00365 }; 00366 00367 00368 } 00369 00370 00371 00372 /* All CSS2 properties below 00373 TODO: Write a CSS 2.0 parser 00374 accelerator 00375 azimuth 00376 background 00377 background-attachment 00378 background-color 00379 background-image 00380 background-position 00381 background-position-x 00382 background-position-y 00383 background-repeat 00384 behavior 00385 border 00386 border-bottom 00387 border-bottom-color 00388 border-bottom-style 00389 border-bottom-width 00390 border-collapse 00391 border-color 00392 border-left 00393 border-left-color 00394 border-left-style 00395 border-left-width 00396 border-right 00397 border-right-color 00398 border-right-style 00399 border-right-width 00400 border-spacing 00401 border-style 00402 border-top 00403 border-top-color 00404 border-top-style 00405 border-top-width 00406 border-width 00407 bottom 00408 caption-side 00409 clear 00410 clip 00411 color 00412 content 00413 counter-increment 00414 counter-reset 00415 cue 00416 cue-after 00417 cue-before 00418 cursor 00419 direction 00420 display 00421 elevation 00422 empty-cells 00423 filter 00424 float 00425 font 00426 font-family 00427 font-size 00428 font-size-adjust 00429 font-stretch 00430 font-style 00431 font-variant 00432 font-weight 00433 height 00434 ime-mode 00435 include-source 00436 layer-background-color 00437 layer-background-image 00438 layout-flow 00439 layout-grid 00440 layout-grid-char 00441 layout-grid-char-spacing 00442 layout-grid-line 00443 layout-grid-mode 00444 layout-grid-type 00445 left 00446 letter-spacing 00447 line-break 00448 line-height 00449 list-style 00450 list-style-image 00451 list-style-position 00452 list-style-type 00453 margin 00454 margin-bottom 00455 margin-left 00456 margin-right 00457 margin-top 00458 marker-offset 00459 marks 00460 max-height 00461 max-width 00462 min-height 00463 min-width 00464 -moz-binding 00465 -moz-border-radius 00466 -moz-border-radius-topleft 00467 -moz-border-radius-topright 00468 -moz-border-radius-bottomright 00469 -moz-border-radius-bottomleft 00470 -moz-border-top-colors 00471 -moz-border-right-colors 00472 -moz-border-bottom-colors 00473 -moz-border-left-colors 00474 -moz-opacity 00475 -moz-outline 00476 -moz-outline-color 00477 -moz-outline-style 00478 -moz-outline-width 00479 -moz-user-focus 00480 -moz-user-input 00481 -moz-user-modify 00482 -moz-user-select 00483 orphans 00484 outline 00485 outline-color 00486 outline-style 00487 outline-width 00488 overflow 00489 overflow-X 00490 overflow-Y 00491 padding 00492 padding-bottom 00493 padding-left 00494 padding-right 00495 padding-top 00496 page 00497 page-break-after 00498 page-break-before 00499 page-break-inside 00500 pause 00501 pause-after 00502 pause-before 00503 pitch 00504 pitch-range 00505 play-during 00506 position 00507 quotes 00508 -replace 00509 richness 00510 right 00511 ruby-align 00512 ruby-overhang 00513 ruby-position 00514 -set-link-source 00515 size 00516 speak 00517 speak-header 00518 speak-numeral 00519 speak-punctuation 00520 speech-rate 00521 stress 00522 scrollbar-arrow-color 00523 scrollbar-base-color 00524 scrollbar-dark-shadow-color 00525 scrollbar-face-color 00526 scrollbar-highlight-color 00527 scrollbar-shadow-color 00528 scrollbar-3d-light-color 00529 scrollbar-track-color 00530 table-layout 00531 text-align 00532 text-align-last 00533 text-decoration 00534 text-indent 00535 text-justify 00536 text-overflow 00537 text-shadow 00538 text-transform 00539 text-autospace 00540 text-kashida-space 00541 text-underline-position 00542 top 00543 unicode-bidi 00544 -use-link-source 00545 vertical-align 00546 visibility 00547 voice-family 00548 volume 00549 white-space 00550 widows 00551 width 00552 word-break 00553 word-spacing 00554 word-wrap 00555 writing-mode 00556 z-index 00557 zoom 00558 00559 */ 00560 /* 00561 class HTMLElement 00562 { 00563 String element; 00564 String attribute; 00565 00566 00567 00568 }; 00569 00570 00571 class HTMLParser 00572 { 00573 00574 00575 00576 }; 00577 */ 00578 00579 #endif 00580