| Index: branches/parser-work/phase3/docs/hooks.txt |
| — | — | @@ -541,6 +541,9 @@ |
| 542 | 542 | &$parser: Parser object |
| 543 | 543 | &$ig: ImageGallery object |
| 544 | 544 | |
| | 545 | +'BeforePreSaveTransform': before wikitext is prepared for saving |
| | 546 | +&$text: Text that will be transformed |
| | 547 | + |
| 545 | 548 | 'BlockIp': before an IP address or user is blocked |
| 546 | 549 | $block: the Block object about to be saved |
| 547 | 550 | $user: the user _doing_ the block (not the one being blocked) |
| Index: branches/parser-work/phase3/includes/parser/ParseEngine.php |
| — | — | @@ -1,244 +0,0 @@ |
| 2 | | -<?php |
| 3 | | -/** |
| 4 | | - * Acts as the primary interface between the world and the parser. |
| 5 | | - * mStartRule - the first rule to use while parsing |
| 6 | | - * mRules - The list of rules to use while parsing |
| 7 | | - * mDom - Used to create Dom objects and get's returned at the end of parsing |
| 8 | | - * mIter - Keeps track of how many times the parser recurses to stop endless loops |
| 9 | | - */ |
| 10 | | -class ParseEngine { |
| 11 | | - const maxIter = 2048; |
| 12 | | - private $mGrammar; |
| 13 | | - |
| 14 | | - function __construct($grammarFile) { |
| 15 | | - global $IP; |
| 16 | | - $this->mGrammar = new DOMDocument(); |
| 17 | | - if (! $this->mGrammar->load("$IP/$grammarFile", LIBXML_NOBLANKS)) { |
| 18 | | - throw new MWException("Failed to load $grammarFile."); |
| 19 | | - } |
| 20 | | - foreach ($this->mGrammar->documentElement->childNodes as $crrnt) { |
| 21 | | - $this->pushTags($crrnt, NULL); |
| 22 | | - } |
| 23 | | - } |
| 24 | | - |
| 25 | | - function parse($text) { |
| 26 | | - global $wgDebugParserLog; |
| 27 | | - if ($wgDebugParserLog != '') { |
| 28 | | - wfErrorLog("==========Start Parsing==========\n", $wgDebugParserLog); |
| 29 | | - } |
| 30 | | - $doc = new DOMDocument(); |
| 31 | | - $rule = $this->mGrammar->documentElement; |
| 32 | | - $rootTag = $doc->createElement($rule->getAttribute("rootTag")); |
| 33 | | - $xpath = new DOMXPath($this->mGrammar); |
| 34 | | - $startRule = $xpath->query("/Grammar/*[@name='{$rule->getAttribute("startRule")}']")->item(0); |
| 35 | | - $iter = 0; |
| 36 | | - if (! $this->parseRec($startRule, "", $saveTags, $iter, $text, $rootTag)) { |
| 37 | | - throw new MWException("Failed to parse the given text."); |
| 38 | | - } |
| 39 | | - $doc->appendChild($rootTag); |
| 40 | | - $doc->normalizeDocument(); |
| 41 | | - if ($wgDebugParserLog != '') { |
| 42 | | - wfErrorLog("XML - {$doc->saveXML()}\n", $wgDebugParserLog); |
| 43 | | - } |
| 44 | | - return $doc; |
| 45 | | - } |
| 46 | | - |
| 47 | | - static function unparse($inNodes) { |
| 48 | | - $retStr = ""; |
| 49 | | - foreach ($inNodes as $child) { |
| 50 | | - if ($child instanceof DOMText) { |
| 51 | | - $retStr .= $child->data; |
| 52 | | - } else { |
| 53 | | - $retStr .= $child->getAttribute("tag") . self::unparse($child->childNodes); |
| 54 | | - } |
| 55 | | - } |
| 56 | | - return $retStr; |
| 57 | | - } |
| 58 | | - |
| 59 | | - private function parseRec($rule, $replaceStr, $saveTags, &$iter, &$text, &$outNode) { |
| 60 | | - global $wgDebugParserLog; |
| 61 | | - if ($wgDebugParserLog != '') { |
| 62 | | - wfErrorLog("Entering {$rule->nodeName}, {$rule->getAttribute("name")}\n", $wgDebugParserLog); |
| 63 | | - } |
| 64 | | - $iter ++; |
| 65 | | - if ($iter > ParseEngine::maxIter) { |
| 66 | | - throw new MWException("Parser iterated too many times. Probable loop in grammar."); |
| 67 | | - } |
| 68 | | - if ($rule->nodeName == "Assignment" || $rule->nodeName == "Reference" || $rule->nodeName == "Text") { |
| 69 | | - $saveTags = str_replace("~r", preg_quote($replaceStr, "/"), $saveTags); |
| 70 | | - $newTags = $rule->getAttribute("saveTags"); |
| 71 | | - if ($saveTags == "") { |
| 72 | | - $saveTags = $newTags; |
| 73 | | - } elseif ($newTags != "") { |
| 74 | | - $saveTags .= "|" . $newTags; |
| 75 | | - } |
| 76 | | - } |
| 77 | | - $dom = $outNode->ownerDocument; |
| 78 | | - $retCode = FALSE; |
| 79 | | - if ($rule->nodeName == "Assignment") { |
| 80 | | - $tag = $rule->getAttribute("tag"); |
| 81 | | - $foundTag = $tag == NULL; |
| 82 | | - if (! $foundTag) { |
| 83 | | - if ($rule->getAttribute("regex") != NULL) { |
| 84 | | - $tag = str_replace("~r", preg_quote($replaceStr, "/"), $tag); |
| 85 | | - $foundTag = preg_match("/^$tag/s", $text, $matches); |
| 86 | | - if ($foundTag) { |
| 87 | | - $tag = $matches[0]; |
| 88 | | - if (isset($matches[1])) { |
| 89 | | - $replaceStr = $matches[1]; |
| 90 | | - } |
| 91 | | - } |
| 92 | | - } else { |
| 93 | | - $tag = str_replace("~r", $replaceStr, $tag); |
| 94 | | - $foundTag = strncmp($tag, $text, strlen($tag)) == 0; |
| 95 | | - } |
| 96 | | - } |
| 97 | | - if ($foundTag) { |
| 98 | | - $newText = $text; |
| 99 | | - $newElement = $dom->createElement($rule->getAttribute("tagName")); |
| 100 | | - if ($tag != NULL) { |
| 101 | | - $newText = substr($newText, strlen($tag)); |
| 102 | | - $newElement->setAttribute("tag", $tag); |
| 103 | | - } |
| 104 | | - $retCode = $rule->firstChild == NULL || $this->parseRec($rule->firstChild, $replaceStr, $saveTags, $iter, $newText, $newElement); |
| 105 | | - if ($retCode) { |
| 106 | | - $outNode->appendChild($newElement); |
| 107 | | - $text = $newText; |
| 108 | | - } |
| 109 | | - } |
| 110 | | - } elseif ($rule->nodeName == "Sequence") { |
| 111 | | - $saveText = $text; |
| 112 | | - $saveNode = $outNode->cloneNode(TRUE); |
| 113 | | - $pushInd = $rule->getAttribute("pushInd"); |
| 114 | | - foreach ($rule->childNodes as $i => $crrnt) { |
| 115 | | - $pushTags = $i >= $pushInd ? $saveTags : ""; |
| 116 | | - $retCode = $this->parseRec($crrnt, $replaceStr, $pushTags, $iter, $text, $outNode); |
| 117 | | - if (! $retCode) { |
| 118 | | - $text = $saveText; |
| 119 | | - $outNode = $saveNode; |
| 120 | | - break; |
| 121 | | - } |
| 122 | | - } |
| 123 | | - } elseif ($rule->nodeName == "Choice") { |
| 124 | | - foreach ($rule->childNodes as $crrnt) { |
| 125 | | - $retCode = $this->parseRec($crrnt, $replaceStr, $saveTags, $iter, $text, $outNode); |
| 126 | | - if ($retCode) { |
| 127 | | - break; |
| 128 | | - } |
| 129 | | - } |
| 130 | | - $retCode |= $rule->getAttribute("failSafe") != NULL; |
| 131 | | - } elseif ($rule->nodeName == "Reference") { |
| 132 | | - $newVar = $rule->hasAttribute("var") ? str_replace("~r", $replaceStr, $rule->getAttribute("var")) : $replaceStr; |
| 133 | | - $xpath = new DOMXPath($this->mGrammar); |
| 134 | | - $refRule = $xpath->query("/Grammar/*[@name='{$rule->getAttribute("name")}']")->item(0); |
| 135 | | - $retCode = $this->parseRec($refRule, $newVar, $saveTags, $iter, $text, $outNode); |
| 136 | | - } elseif ($rule->nodeName == "Text") { |
| 137 | | - $tagSearch = $rule->getAttribute("childTags"); |
| 138 | | - if ($tagSearch == "") { |
| 139 | | - $tagSearch = $saveTags; |
| 140 | | - } elseif ($saveTags != "") { |
| 141 | | - $tagSearch .= "|" . $saveTags; |
| 142 | | - } |
| 143 | | - while ($text != "" && ($saveTags == "" || ! preg_match("/^($saveTags)/s", $text))) { |
| 144 | | - $offset = $rule->firstChild != NULL && $this->parseRec($rule->firstChild, $replaceStr, "", $iter, $text, $outNode) ? 0 : 1; |
| 145 | | - if (preg_match("/$tagSearch/s", $text, $matches, PREG_OFFSET_CAPTURE, $offset)) { |
| 146 | | - if ($matches[0][1] > 0) { |
| 147 | | - $outNode->appendChild($dom->createTextNode(substr($text, 0, $matches[0][1]))); |
| 148 | | - $text = substr($text, $matches[0][1]); |
| 149 | | - } |
| 150 | | - } else { |
| 151 | | - $outNode->appendChild($dom->createTextNode($text)); |
| 152 | | - $text = ""; |
| 153 | | - } |
| 154 | | - } |
| 155 | | - $retCode = true; |
| 156 | | - } |
| 157 | | - if ($wgDebugParserLog != '') { |
| 158 | | - wfErrorLog("Exiting {$rule->nodeName}, Return Code - $retCode\n", $wgDebugParserLog); |
| 159 | | - wfErrorLog("Text - $text\n", $wgDebugParserLog); |
| 160 | | - } |
| 161 | | - return $retCode; |
| 162 | | - } |
| 163 | | - |
| 164 | | - private function pushTags($rule, $tagStr) { |
| 165 | | - $iter = 0; |
| 166 | | - if ($rule->nodeName == "Sequence") { |
| 167 | | - $pushInd = $rule->childNodes->length - 1; |
| 168 | | - $shouldPush = true; |
| 169 | | - for ($child = $rule->lastChild; $child != NULL; $child = $child->previousSibling) { |
| 170 | | - $this->pushTags($child, $tagStr); |
| 171 | | - if ($child->previousSibling != NULL) { |
| 172 | | - if ($this->pullTags($child, $iter, $childTag)) { |
| 173 | | - if ($shouldPush) { |
| 174 | | - $pushInd --; |
| 175 | | - } |
| 176 | | - if ($tagStr == "") { |
| 177 | | - $tagStr = $childTag; |
| 178 | | - } elseif ($childTag != "") { |
| 179 | | - $tagStr .= "|" . $childTag; |
| 180 | | - } |
| 181 | | - } else { |
| 182 | | - $shouldPush = false; |
| 183 | | - $tagStr = $childTag; |
| 184 | | - } |
| 185 | | - } |
| 186 | | - } |
| 187 | | - $rule->setAttribute("pushInd", $pushInd); |
| 188 | | - } else { |
| 189 | | - if ($rule->nodeName != "Choice") { |
| 190 | | - $rule->setAttribute("saveTags", $tagStr); |
| 191 | | - $tagStr = NULL; |
| 192 | | - if ($rule->nodeName == "Text") { |
| 193 | | - $childTags = ""; |
| 194 | | - foreach ($rule->childNodes as $crrnt) { |
| 195 | | - if ($childTags != "") { |
| 196 | | - $childTags .= "|"; |
| 197 | | - } |
| 198 | | - $this->pullTags($crrnt, $iter, $childTag); |
| 199 | | - $childTags .= $childTag; |
| 200 | | - } |
| 201 | | - $rule->setAttribute("childTags", $childTags); |
| 202 | | - } |
| 203 | | - } |
| 204 | | - foreach ($rule->childNodes as $crrnt) { |
| 205 | | - $this->pushTags($crrnt, $tagStr); |
| 206 | | - } |
| 207 | | - } |
| 208 | | - } |
| 209 | | - |
| 210 | | - private function pullTags($rule, &$iter, &$childTags) { |
| 211 | | - $iter ++; |
| 212 | | - if ($iter > ParseEngine::maxIter) { |
| 213 | | - throw new MWException("Collecter iterated too many times. Probable loop in grammar."); |
| 214 | | - } |
| 215 | | - $childTags = ""; |
| 216 | | - $failSafe = TRUE; |
| 217 | | - if ($rule->nodeName == "Assignment") { |
| 218 | | - $childTags = $rule->getAttribute("tag"); |
| 219 | | - if ($rule->getAttribute("regex") == NULL) { |
| 220 | | - $childTags = preg_quote($childTags, "/"); |
| 221 | | - } |
| 222 | | - $failSafe = FALSE; |
| 223 | | - } elseif ($rule->nodeName == "Choice" || $rule->nodeName == "Sequence") { |
| 224 | | - $failSafe = $rule->nodeName == "Sequence"; |
| 225 | | - foreach ($rule->childNodes as $child) { |
| 226 | | - $failSafe = $this->pullTags($child, $iter, $newTags); |
| 227 | | - if ($childTags == "") { |
| 228 | | - $childTags = $newTags; |
| 229 | | - } elseif ($newTags != "") { |
| 230 | | - $childTags .= "|" . $newTags; |
| 231 | | - } |
| 232 | | - if (($failSafe && $rule->nodeName == "Choice") || (! $failSafe && $rule->nodeName == "Sequence")) { |
| 233 | | - break; |
| 234 | | - } |
| 235 | | - } |
| 236 | | - $failSafe |= $rule->nodeName == "Choice" && $rule->getAttribute("failSafe") != NULL; |
| 237 | | - } elseif ($rule->nodeName == "Reference") { |
| 238 | | - $xpath = new DOMXPath($this->mGrammar); |
| 239 | | - $refRule = $xpath->query("/Grammar/*[@name='{$rule->getAttribute("name")}']")->item(0); |
| 240 | | - $failSafe = $this->pullTags($refRule, $iter, $childTags); |
| 241 | | - } |
| 242 | | - return $failSafe; |
| 243 | | - } |
| 244 | | -} |
| 245 | | - |
| Index: branches/parser-work/phase3/includes/parser/WikiTextGrammar.xml |
| — | — | @@ -1,145 +0,0 @@ |
| 2 | | -<?xml version="1.0"?> |
| 3 | | -<Grammar rootTag="root" startRule="start" version="1.0"> |
| 4 | | - <Sequence name="start" > |
| 5 | | - <Reference name="postNewLine" /> |
| 6 | | - <Reference name="main" /> |
| 7 | | - </Sequence> |
| 8 | | - <Text name="main"> |
| 9 | | - <Choice> |
| 10 | | - <Sequence> |
| 11 | | - <Reference name="newLine" /> |
| 12 | | - <Reference name="postNewLine" /> |
| 13 | | - </Sequence> |
| 14 | | - <Assignment tagName="link" tag="[["> |
| 15 | | - <Reference name="endText" var="]]" /> |
| 16 | | - </Assignment> |
| 17 | | - <Assignment tagName="tplArg" tag="{{{(?!{)" regex="true"> |
| 18 | | - <Sequence> |
| 19 | | - <Reference name="name" /> |
| 20 | | - <Choice failSafe="true"> |
| 21 | | - <Assignment tagName="default" tag="|"> |
| 22 | | - <Reference name="main" /> |
| 23 | | - </Assignment> |
| 24 | | - </Choice> |
| 25 | | - <Assignment tagName="endTag" tag="}}}" /> |
| 26 | | - </Sequence> |
| 27 | | - </Assignment> |
| 28 | | - <Assignment tagName="template" tag="{{"> |
| 29 | | - <Sequence> |
| 30 | | - <Reference name="name" /> |
| 31 | | - <Choice failSafe="true"> |
| 32 | | - <Assignment tagName="name2" tag=":"> |
| 33 | | - <Reference name="main" /> |
| 34 | | - </Assignment> |
| 35 | | - </Choice> |
| 36 | | - <Reference name="partList" /> |
| 37 | | - <Assignment tagName="endTag" tag="}}" /> |
| 38 | | - </Sequence> |
| 39 | | - </Assignment> |
| 40 | | - <Reference name="comment" /> |
| 41 | | - <Assignment tagName="noWiki" tag="<nowiki>"> |
| 42 | | - <Sequence> |
| 43 | | - <Text /> |
| 44 | | - <Assignment tagName="endTag" tag="<\/nowiki>" /> |
| 45 | | - </Sequence> |
| 46 | | - </Assignment> |
| 47 | | - <Assignment tagName="xmlTag" tag="<(?=(\w+)[\s\/>])" regex="true"> |
| 48 | | - <Sequence> |
| 49 | | - <Assignment tagName="name" tag="~r" /> |
| 50 | | - <Reference name="attrList" /> |
| 51 | | - <Choice> |
| 52 | | - <Assignment tagName="endTag" tag="\s*\/>" regex="true" /> |
| 53 | | - <Sequence> |
| 54 | | - <Assignment tagName="inner" tag="\s*>" regex="true"> |
| 55 | | - <Reference name="main" /> |
| 56 | | - </Assignment> |
| 57 | | - <Assignment tagName="endTag" tag="</~r>" /> |
| 58 | | - </Sequence> |
| 59 | | - </Choice> |
| 60 | | - </Sequence> |
| 61 | | - </Assignment> |
| 62 | | - </Choice> |
| 63 | | - </Text> |
| 64 | | - <Sequence name="endText"> |
| 65 | | - <Reference name="main" /> |
| 66 | | - <Assignment tagName="endTag" tag="~r" /> |
| 67 | | - </Sequence> |
| 68 | | - <Assignment name="newLine" tagName="newLine" tag="\r?\n" regex="true" /> |
| 69 | | - <Assignment name="eol" tagName="eol" tag="(?=\n|$)" regex="true" /> |
| 70 | | - <Choice name="ignoreList" failSafe="true"> |
| 71 | | - <Sequence> |
| 72 | | - <Choice> |
| 73 | | - <Assignment tag="[ \t]+" regex="true" /> |
| 74 | | - <Reference name="comment" /> |
| 75 | | - </Choice> |
| 76 | | - <Reference name="ignoreList" /> |
| 77 | | - </Sequence> |
| 78 | | - </Choice> |
| 79 | | - <Choice name="postNewLine" failSafe="true"> |
| 80 | | - <Sequence> |
| 81 | | - <Assignment tagName="h" tag="(={1,6})" regex="true"> |
| 82 | | - <Reference name="endText" /> |
| 83 | | - </Assignment> |
| 84 | | - <Reference name="ignoreList" /> |
| 85 | | - <Reference name="eol" /> |
| 86 | | - </Sequence> |
| 87 | | - <Reference name="listChoice" var="" /> |
| 88 | | - </Choice> |
| 89 | | - <Choice name="listChoice"> |
| 90 | | - <Assignment tagName="orderedList" tag="(?=(~r#))" regex="true"> |
| 91 | | - <Reference name="itemList" /> |
| 92 | | - </Assignment> |
| 93 | | - <Assignment tagName="unorderedList" tag="(?=(~r\*))" regex="true"> |
| 94 | | - <Reference name="itemList" /> |
| 95 | | - </Assignment> |
| 96 | | - </Choice> |
| 97 | | - <Sequence name="itemList"> |
| 98 | | - <Choice> |
| 99 | | - <Reference name="listChoice" /> |
| 100 | | - <Assignment tagName="listItem" tag="~r"> |
| 101 | | - <Sequence> |
| 102 | | - <Reference name="main" /> |
| 103 | | - <Reference name="eol" /> |
| 104 | | - </Sequence> |
| 105 | | - </Assignment> |
| 106 | | - </Choice> |
| 107 | | - <Choice failSafe="true"> |
| 108 | | - <Sequence> |
| 109 | | - <Reference name="newLine" /> |
| 110 | | - <Reference name="itemList" /> |
| 111 | | - </Sequence> |
| 112 | | - </Choice> |
| 113 | | - </Sequence> |
| 114 | | - <Assignment name="comment" tagName="comment" tag="<!--.*?(?:-->|$)" regex="true" /> |
| 115 | | - <Assignment name="name" tagName="name"> |
| 116 | | - <Reference name="main" /> |
| 117 | | - </Assignment> |
| 118 | | - <Choice name="partList" failSafe="true"> |
| 119 | | - <Sequence> |
| 120 | | - <Assignment tagName="part" tag="|"> |
| 121 | | - <Sequence> |
| 122 | | - <Reference name="name" /> |
| 123 | | - <Choice failSafe="true"> |
| 124 | | - <Assignment tagName="value" tag="="> |
| 125 | | - <Reference name="main" /> |
| 126 | | - </Assignment> |
| 127 | | - </Choice> |
| 128 | | - </Sequence> |
| 129 | | - </Assignment> |
| 130 | | - <Reference name="partList" /> |
| 131 | | - </Sequence> |
| 132 | | - </Choice> |
| 133 | | - <Choice name="attrList" failSafe="true"> |
| 134 | | - <Sequence> |
| 135 | | - <Assignment tagName="attribute" tag="\s+(?!\/?>)" regex="true"> |
| 136 | | - <Sequence> |
| 137 | | - <Reference name="name" /> |
| 138 | | - <Assignment tagName="value" tag="\s*=\s*("|')" regex="true"> |
| 139 | | - <Reference name="endText" /> |
| 140 | | - </Assignment> |
| 141 | | - </Sequence> |
| 142 | | - </Assignment> |
| 143 | | - <Reference name="attrList" /> |
| 144 | | - </Sequence> |
| 145 | | - </Choice> |
| 146 | | -</Grammar> |
| Index: branches/parser-work/phase3/includes/parser/Parser.php |
| — | — | @@ -3978,6 +3978,8 @@ |
| 3979 | 3979 | $this->clearState(); |
| 3980 | 3980 | } |
| 3981 | 3981 | |
| | 3982 | + wfRunHooks( 'BeforePreSaveTransform', array( &$text ) ); |
| | 3983 | + |
| 3982 | 3984 | $pairs = array( |
| 3983 | 3985 | "\r\n" => "\n", |
| 3984 | 3986 | ); |