/extensions/ParserWiki/ParserEngine.php

https://github.com/ChuguluGames/mediawiki-svn · PHP · 236 lines · 24 code · 2 blank · 210 comment · 5 complexity · fede7dc38e98b02e4e8b30273dda162b MD5 · raw file 

    

  1. <?php
    2. 

  2. /**
    3. 

  3.  * Acts as the primary interface between the world and the parser.
    4. 

  4.  * mStartRule - the first rule to use while parsing
    5. 

  5.  * mRules - The list of rules to use while parsing
    6. 

  6.  * mDom - Used to create Dom objects and get's returned at the end of parsing
    7. 

  7.  * mIter - Keeps track of how many times the parser recurses to stop endless loops
    8. 

  8.  */
    9. 

  9. class ParseEngine {
    10. 

  10. 	const maxIter = 2048;
    11. 

  11. 	private $mGrammars;
    12. 

  12. 

  13. 	function __construct() {
    14. 

  14. 		$this->mGrammars = array();
    15. 

  15. 	}
    16. 

  16. 

  17. 	function parse($grammarName, &$text) {
    18. 

  18. 		wfDebugLog("ParseEngine", "==========Start Parse Engine==========\n");
    19. 

  19. 		$grammar = isset($this->mGrammars[$grammarName]) ? $this->mGrammars[$grammarName] : NULL;
    20. 

  20. 		if ($grammar == NULL) {
    21. 

  21. 			$revision = Revision::newFromTitle(Title::newFromText($grammarName, NS_GRAMMAR));
    22. 

  22. 			$grammar = new DOMDocument();
    23. 

  23. 			if ($revision == NULL || ! $grammar->loadXML($revision->getText(), LIBXML_NOBLANKS)) {
    24. 

  24. 				return TRUE;
    25. 

  25. 			}
    26. 

  26. 			$this->pushTags($grammar->documentElement, NULL);
    27. 

  27. 			$this->mGrammars[$grammarName] = $grammar;
    28. 

  28. 		}
    29. 

  29. 		$doc = new DOMDocument();
    30. 

  30. 		$rootTag = $doc->createElement($grammar->documentElement->getAttribute("rootTag"));
    31. 

  31. 		$startRule = $grammar->documentElement->getAttribute("startRule");
    32. 

  32. 		$xpath = new DOMXPath($grammar);
    33. 

  33. 		$startRule = $xpath->query("/Grammar/*[@name='$startRule']")->item(0);
    34. 

  34. 		$refText = $text;
    35. 

  35. 		if (! $this->parseRec($startRule, "", "", $iter, $refText, $rootTag)) {
    36. 

  36. 			return TRUE;
    37. 

  37. 		}
    38. 

  38. 		$doc->appendChild($rootTag);
    39. 

  39. 		$text = $doc->saveXML();
    40. 

  40. 		wfDebugLog("ParseEngine", "Parsed text - $text\n");
    41. 

  41. 		return TRUE;
    42. 

  42. 	}
    43. 

  43. 

  44. 	static function unparse($inNodes) {
    45. 

  45. 		$retStr = "";
    46. 

  46. 		foreach ($inNodes as $child) {
    47. 

  47. 			if ($child instanceof DOMText) {
    48. 

  48. 				$retStr .= $child->data;
    49. 

  49. 			} else {
    50. 

  50. 				$retStr .= $child->getAttribute("tag") . self::unparse($child->childNodes);
    51. 

  51. 			}
    52. 

  52. 		}
    53. 

  53. 		return $retStr;
    54. 

  54. 	}
    55. 

  55. 

  56. 	private function parseRec($rule, $replaceStr, $saveTags, &$iter, &$text, &$outNode) {
    57. 

  57. 		wfDebugLog("ParseEngine", "Entering {$rule->nodeName}, {$rule->getAttribute("name")}\n");
    58. 

  58. 		$iter ++;
    59. 

  59. 		if ($iter > ParseEngine::maxIter) {
    60. 

  60. 			throw new MWException("Parser iterated too many times. Probable loop in grammar.");
    61. 

  61. 		}
    62. 

  62. 		if ($rule->nodeName == "Assignment" || $rule->nodeName == "Reference" || $rule->nodeName == "Text") {
    63. 

  63. 			$saveTags = str_replace("~r", preg_quote($replaceStr, "/"), $saveTags);
    64. 

  64. 			$newTags = $rule->getAttribute("saveTags");
    65. 

  65. 			if ($saveTags == "") {
    66. 

  66. 				$saveTags = $newTags;
    67. 

  67. 			} elseif ($newTags != "") {
    68. 

  68. 				$saveTags .= "|" . $newTags;
    69. 

  69. 			}
    70. 

  70. 		}
    71. 

  71. 		$dom = $outNode->ownerDocument;
    72. 

  72. 		$retCode = FALSE;
    73. 

  73. 		if ($rule->nodeName == "Assignment") {
    74. 

  74. 			$tag = $rule->getAttribute("tag");
    75. 

  75. 			$foundTag = $tag == NULL;
    76. 

  76. 			if (! $foundTag) {
    77. 

  77. 				if ($rule->getAttribute("regex") != NULL) {
    78. 

  78. 					$tag = str_replace("~r", preg_quote($replaceStr, "/"), $tag);
    79. 

  79. 					$foundTag = preg_match("/^$tag/s", $text, $matches);
    80. 

  80. 					if ($foundTag) {
    81. 

  81. 						$tag = $matches[0];
    82. 

  82. 						if (isset($matches[1])) {
    83. 

  83. 							$replaceStr = $matches[1];
    84. 

  84. 						}
    85. 

  85. 					}
    86. 

  86. 				} else {
    87. 

  87. 					$tag = str_replace("~r", $replaceStr, $tag);
    88. 

  88. 					$foundTag = strncmp($tag, $text, strlen($tag)) == 0;
    89. 

  89. 				}
    90. 

  90. 			}
    91. 

  91. 			if ($foundTag) {
    92. 

  92. 				$newText = $text;
    93. 

  93. 				$newElement = $dom->createElement($rule->getAttribute("tagName"));
    94. 

  94. 				if ($tag != NULL) {
    95. 

  95. 					$newText = substr($newText, strlen($tag));
    96. 

  96. 					$newElement->setAttribute("tag", $tag);
    97. 

  97. 				}
    98. 

  98. 				$retCode = $rule->firstChild == NULL || $this->parseRec($rule->firstChild, $replaceStr, $saveTags, $iter, $newText, $newElement);
    99. 

  99. 				if ($retCode) {
    100. 

  100. 					$outNode->appendChild($newElement);
    101. 

  101. 					$text = $newText;
    102. 

  102. 				}
    103. 

  103. 			}
    104. 

  104. 		} elseif ($rule->nodeName == "Sequence") {
    105. 

  105. 			$saveText = $text;
    106. 

  106. 			$saveNode = $outNode->cloneNode(TRUE);
    107. 

  107. 			$pushInd = $rule->getAttribute("pushInd");
    108. 

  108. 			foreach ($rule->childNodes as $i => $crrnt) {
    109. 

  109. 				$pushTags = $i >= $pushInd ? $saveTags : "";
    110. 

  110. 				$retCode = $this->parseRec($crrnt, $replaceStr, $pushTags, $iter, $text, $outNode);
    111. 

  111. 				if (! $retCode) {
    112. 

  112. 					$text = $saveText;
    113. 

  113. 					$outNode = $saveNode;
    114. 

  114. 					break;
    115. 

  115. 				}
    116. 

  116. 			}
    117. 

  117. 		} elseif ($rule->nodeName == "Choice") {
    118. 

  118. 			foreach ($rule->childNodes as $crrnt) {
    119. 

  119. 				$retCode = $this->parseRec($crrnt, $replaceStr, $saveTags, $iter, $text, $outNode);
    120. 

  120. 				if ($retCode) {
    121. 

  121. 					break;
    122. 

  122. 				}
    123. 

  123. 			}
    124. 

  124. 			$retCode |= $rule->getAttribute("failSafe") != NULL;
    125. 

  125. 		} elseif ($rule->nodeName == "Reference") {
    126. 

  126. 			$newVar = $rule->hasAttribute("var") ? str_replace("~r", $replaceStr, $rule->getAttribute("var")) : $replaceStr;
    127. 

  127. 			$xpath = new DOMXPath($rule->ownerDocument);
    128. 

  128. 			$refRule = $xpath->query("/Grammar/*[@name='{$rule->getAttribute("name")}']")->item(0);
    129. 

  129. 			$retCode = $this->parseRec($refRule, $newVar, $saveTags, $iter, $text, $outNode);
    130. 

  130. 		} elseif ($rule->nodeName == "Text") {
    131. 

  131. 			$tagSearch = $rule->getAttribute("childTags");
    132. 

  132. 			if ($tagSearch == "") {
    133. 

  133. 				$tagSearch = $saveTags;
    134. 

  134. 			} elseif ($saveTags != "") {
    135. 

  135. 				$tagSearch .= "|" . $saveTags;
    136. 

  136. 			}
    137. 

  137. 			while ($text != "" && ($saveTags == "" || ! preg_match("/^($saveTags)/s", $text))) {
    138. 

  138. 				$offset = $rule->firstChild != NULL && $this->parseRec($rule->firstChild, $replaceStr, "", $iter, $text, $outNode) ? 0 : 1;
    139. 

  139. 				if (preg_match("/$tagSearch/s", $text, $matches, PREG_OFFSET_CAPTURE, $offset)) {
    140. 

  140. 					if ($matches[0][1] > 0) {
    141. 

  141. 						$outNode->appendChild($dom->createTextNode(substr($text, 0, $matches[0][1])));
    142. 

  142. 						$text = substr($text, $matches[0][1]);
    143. 

  143. 					}
    144. 

  144. 				} else {
    145. 

  145. 					$outNode->appendChild($dom->createTextNode($text));
    146. 

  146. 					$text = "";
    147. 

  147. 				}
    148. 

  148. 			}
    149. 

  149. 			$retCode = true;
    150. 

  150. 		}
    151. 

  151. 		wfDebugLog("ParseEngine", "Exiting {$rule->nodeName}, Return Code - $retCode\n");
    152. 

  152. 		wfDebugLog("ParseEngine", "Text - $text\n");
    153. 

  153. 		return $retCode;
    154. 

  154. 	}
    155. 

  155. 

  156. 	private function pushTags($rule, $tagStr) {
    157. 

  157. 		if ($rule->nodeName == "Sequence") {
    158. 

  158. 			$pushInd = $rule->childNodes->length - 1;
    159. 

  159. 			$shouldPush = true;
    160. 

  160. 			for ($child = $rule->lastChild; $child != NULL; $child = $child->previousSibling) {
    161. 

  161. 				$this->pushTags($child, $tagStr);
    162. 

  162. 				if ($child->previousSibling != NULL) {
    163. 

  163. 					if ($this->pullTags($child, $iter, $childTag)) {
    164. 

  164. 						if ($shouldPush) {
    165. 

  165. 							$pushInd --;
    166. 

  166. 						}
    167. 

  167. 						if ($tagStr == "") {
    168. 

  168. 							$tagStr = $childTag;
    169. 

  169. 						} elseif ($childTag != "") {
    170. 

  170. 							$tagStr .= "|" . $childTag;
    171. 

  171. 						}
    172. 

  172. 					} else {
    173. 

  173. 						$shouldPush = false;
    174. 

  174. 						$tagStr = $childTag;
    175. 

  175. 					}
    176. 

  176. 				}
    177. 

  177. 			}
    178. 

  178. 			$rule->setAttribute("pushInd", $pushInd);
    179. 

  179. 		} else {
    180. 

  180. 			if ($rule->nodeName != "Choice") {
    181. 

  181. 				$rule->setAttribute("saveTags", $tagStr);
    182. 

  182. 				$tagStr = NULL;
    183. 

  183. 				if ($rule->nodeName == "Text") {
    184. 

  184. 					$childTags = "";
    185. 

  185. 					foreach ($rule->childNodes as $crrnt) {
    186. 

  186. 						if ($childTags != "") {
    187. 

  187. 							$childTags .= "|";
    188. 

  188. 						}
    189. 

  189. 						$this->pullTags($crrnt, $iter, $childTag);
    190. 

  190. 						$childTags .= $childTag;
    191. 

  191. 					}
    192. 

  192. 					$rule->setAttribute("childTags", $childTags);
    193. 

  193. 				}
    194. 

  194. 			}
    195. 

  195. 			foreach ($rule->childNodes as $crrnt) {
    196. 

  196. 				$this->pushTags($crrnt, $tagStr);
    197. 

  197. 			}
    198. 

  198. 		}
    199. 

  199. 	}
    200. 

  200. 

  201. 	private function pullTags($rule, &$iter, &$childTags) {
    202. 

  202. 		$iter ++;
    203. 

  203. 		if ($iter > ParseEngine::maxIter) {
    204. 

  204. 			throw new MWException("Collecter iterated too many times. Probable loop in grammar.");
    205. 

  205. 		}
    206. 

  206. 		$childTags = "";
    207. 

  207. 		$failSafe = TRUE;
    208. 

  208. 		if ($rule->nodeName == "Assignment") {
    209. 

  209. 			$childTags = $rule->getAttribute("tag");
    210. 

  210. 			if ($rule->getAttribute("regex") == NULL) {
    211. 

  211. 				$childTags = preg_quote($childTags, "/");
    212. 

  212. 			}
    213. 

  213. 			$failSafe = FALSE;
    214. 

  214. 		} elseif ($rule->nodeName == "Choice" || $rule->nodeName == "Sequence") {
    215. 

  215. 			$failSafe = $rule->nodeName == "Sequence";
    216. 

  216. 			foreach ($rule->childNodes as $child) {
    217. 

  217. 				$failSafe = $this->pullTags($child, $iter, $newTags);
    218. 

  218. 				if ($childTags == "") {
    219. 

  219. 					$childTags = $newTags;
    220. 

  220. 				} elseif ($newTags != "") {
    221. 

  221. 					$childTags .= "|" . $newTags;
    222. 

  222. 				}
    223. 

  223. 				if (($failSafe && $rule->nodeName == "Choice") || (! $failSafe && $rule->nodeName == "Sequence")) {
    224. 

  224. 					break;
    225. 

  225. 				}
    226. 

  226. 			}
    227. 

  227. 			$failSafe |= $rule->nodeName == "Choice" && $rule->getAttribute("failSafe") != NULL;
    228. 

  228. 		} elseif ($rule->nodeName == "Reference") {
    229. 

  229. 			$xpath = new DOMXPath($rule->ownerDocument);
    230. 

  230. 			$refRule = $xpath->query("/Grammar/*[@name='{$rule->getAttribute("name")}']")->item(0);
    231. 

  231. 			$failSafe = $this->pullTags($refRule, $iter, $childTags);
    232. 

  232. 		}
    233. 

  233. 		return $failSafe;
    234. 

  234. 	}
    235. 

  235. }
    236.