r67702 MediaWiki - Code Review archive

Repository:	MediaWiki
Revision:	< r67701‎ \| r67702 \| r67703 >
Date:	11:25, 9 June 2010
Author:	reedy
Status:	deferred
Tags:
Comment:	Revert r65412 to do the move properly
Modified paths:	/trunk/extensions/DataTransclusion/WebDataTransclusionSource.php (modified) (history) /trunk/extensions/ParseEngine (added) (history) /trunk/extensions/ParserWiki (deleted) (history)

Diff [purge]

Index: trunk/extensions/ParseEngine/ParseEngine.body.php
—	—	@@ -0,0 +1,237 @@
	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 $mGrammars;
	13	+
	14	+ function __construct() {
	15	+ $this->mGrammars = array();
	16	+ }
	17	+
	18	+ function parse($grammarName, &$text) {
	19	+ global $IP;
	20	+ wfDebugLog("ParseEngine", "==========Start Parse Engine==========\n");
	21	+ $grammar = isset($this->mGrammars[$grammarName]) ? $this->mGrammars[$grammarName] : NULL;
	22	+ if ($grammar == NULL) {
	23	+ $revision = Revision::newFromTitle(Title::newFromText($grammarName, NS_GRAMMAR));
	24	+ $grammar = new DOMDocument();
	25	+ if ($revision == NULL \|\| ! $grammar->loadXML($revision->getText(), LIBXML_NOBLANKS)) {
	26	+ return TRUE;
	27	+ }
	28	+ $this->pushTags($grammar->documentElement, NULL);
	29	+ $this->mGrammars[$grammarName] = $grammar;
	30	+ }
	31	+ $doc = new DOMDocument();
	32	+ $rootTag = $doc->createElement($grammar->documentElement->getAttribute("rootTag"));
	33	+ $startRule = $grammar->documentElement->getAttribute("startRule");
	34	+ $xpath = new DOMXPath($grammar);
	35	+ $startRule = $xpath->query("/Grammar/*[@name='$startRule']")->item(0);
	36	+ $refText = $text;
	37	+ if (! $this->parseRec($startRule, "", "", $iter, $refText, $rootTag)) {
	38	+ return TRUE;
	39	+ }
	40	+ $doc->appendChild($rootTag);
	41	+ $text = $doc->saveXML();
	42	+ wfDebugLog("ParseEngine", "Parsed text - $text\n");
	43	+ return TRUE;
	44	+ }
	45	+
	46	+ static function unparse($inNodes) {
	47	+ $retStr = "";
	48	+ foreach ($inNodes as $child) {
	49	+ if ($child instanceof DOMText) {
	50	+ $retStr .= $child->data;
	51	+ } else {
	52	+ $retStr .= $child->getAttribute("tag") . self::unparse($child->childNodes);
	53	+ }
	54	+ }
	55	+ return $retStr;
	56	+ }
	57	+
	58	+ private function parseRec($rule, $replaceStr, $saveTags, &$iter, &$text, &$outNode) {
	59	+ wfDebugLog("ParseEngine", "Entering {$rule->nodeName}, {$rule->getAttribute("name")}\n");
	60	+ $iter ++;
	61	+ if ($iter > ParseEngine::maxIter) {
	62	+ throw new MWException("Parser iterated too many times. Probable loop in grammar.");
	63	+ }
	64	+ if ($rule->nodeName == "Assignment" \|\| $rule->nodeName == "Reference" \|\| $rule->nodeName == "Text") {
	65	+ $saveTags = str_replace("~r", preg_quote($replaceStr, "/"), $saveTags);
	66	+ $newTags = $rule->getAttribute("saveTags");
	67	+ if ($saveTags == "") {
	68	+ $saveTags = $newTags;
	69	+ } elseif ($newTags != "") {
	70	+ $saveTags .= "\|" . $newTags;
	71	+ }
	72	+ }
	73	+ $dom = $outNode->ownerDocument;
	74	+ $retCode = FALSE;
	75	+ if ($rule->nodeName == "Assignment") {
	76	+ $tag = $rule->getAttribute("tag");
	77	+ $foundTag = $tag == NULL;
	78	+ if (! $foundTag) {
	79	+ if ($rule->getAttribute("regex") != NULL) {
	80	+ $tag = str_replace("~r", preg_quote($replaceStr, "/"), $tag);
	81	+ $foundTag = preg_match("/^$tag/s", $text, $matches);
	82	+ if ($foundTag) {
	83	+ $tag = $matches[0];
	84	+ if (isset($matches[1])) {
	85	+ $replaceStr = $matches[1];
	86	+ }
	87	+ }
	88	+ } else {
	89	+ $tag = str_replace("~r", $replaceStr, $tag);
	90	+ $foundTag = strncmp($tag, $text, strlen($tag)) == 0;
	91	+ }
	92	+ }
	93	+ if ($foundTag) {
	94	+ $newText = $text;
	95	+ $newElement = $dom->createElement($rule->getAttribute("tagName"));
	96	+ if ($tag != NULL) {
	97	+ $newText = substr($newText, strlen($tag));
	98	+ $newElement->setAttribute("tag", $tag);
	99	+ }
	100	+ $retCode = $rule->firstChild == NULL \|\| $this->parseRec($rule->firstChild, $replaceStr, $saveTags, $iter, $newText, $newElement);
	101	+ if ($retCode) {
	102	+ $outNode->appendChild($newElement);
	103	+ $text = $newText;
	104	+ }
	105	+ }
	106	+ } elseif ($rule->nodeName == "Sequence") {
	107	+ $saveText = $text;
	108	+ $saveNode = $outNode->cloneNode(TRUE);
	109	+ $pushInd = $rule->getAttribute("pushInd");
	110	+ foreach ($rule->childNodes as $i => $crrnt) {
	111	+ $pushTags = $i >= $pushInd ? $saveTags : "";
	112	+ $retCode = $this->parseRec($crrnt, $replaceStr, $pushTags, $iter, $text, $outNode);
	113	+ if (! $retCode) {
	114	+ $text = $saveText;
	115	+ $outNode = $saveNode;
	116	+ break;
	117	+ }
	118	+ }
	119	+ } elseif ($rule->nodeName == "Choice") {
	120	+ foreach ($rule->childNodes as $crrnt) {
	121	+ $retCode = $this->parseRec($crrnt, $replaceStr, $saveTags, $iter, $text, $outNode);
	122	+ if ($retCode) {
	123	+ break;
	124	+ }
	125	+ }
	126	+ $retCode \|= $rule->getAttribute("failSafe") != NULL;
	127	+ } elseif ($rule->nodeName == "Reference") {
	128	+ $newVar = $rule->hasAttribute("var") ? str_replace("~r", $replaceStr, $rule->getAttribute("var")) : $replaceStr;
	129	+ $xpath = new DOMXPath($rule->ownerDocument);
	130	+ $refRule = $xpath->query("/Grammar/*[@name='{$rule->getAttribute("name")}']")->item(0);
	131	+ $retCode = $this->parseRec($refRule, $newVar, $saveTags, $iter, $text, $outNode);
	132	+ } elseif ($rule->nodeName == "Text") {
	133	+ $tagSearch = $rule->getAttribute("childTags");
	134	+ if ($tagSearch == "") {
	135	+ $tagSearch = $saveTags;
	136	+ } elseif ($saveTags != "") {
	137	+ $tagSearch .= "\|" . $saveTags;
	138	+ }
	139	+ while ($text != "" && ($saveTags == "" \|\| ! preg_match("/^($saveTags)/s", $text))) {
	140	+ $offset = $rule->firstChild != NULL && $this->parseRec($rule->firstChild, $replaceStr, "", $iter, $text, $outNode) ? 0 : 1;
	141	+ if (preg_match("/$tagSearch/s", $text, $matches, PREG_OFFSET_CAPTURE, $offset)) {
	142	+ if ($matches[0][1] > 0) {
	143	+ $outNode->appendChild($dom->createTextNode(substr($text, 0, $matches[0][1])));
	144	+ $text = substr($text, $matches[0][1]);
	145	+ }
	146	+ } else {
	147	+ $outNode->appendChild($dom->createTextNode($text));
	148	+ $text = "";
	149	+ }
	150	+ }
	151	+ $retCode = true;
	152	+ }
	153	+ wfDebugLog("ParseEngine", "Exiting {$rule->nodeName}, Return Code - $retCode\n");
	154	+ wfDebugLog("ParseEngine", "Text - $text\n");
	155	+ return $retCode;
	156	+ }
	157	+
	158	+ private function pushTags($rule, $tagStr) {
	159	+ if ($rule->nodeName == "Sequence") {
	160	+ $pushInd = $rule->childNodes->length - 1;
	161	+ $shouldPush = true;
	162	+ for ($child = $rule->lastChild; $child != NULL; $child = $child->previousSibling) {
	163	+ $this->pushTags($child, $tagStr);
	164	+ if ($child->previousSibling != NULL) {
	165	+ if ($this->pullTags($child, $iter, $childTag)) {
	166	+ if ($shouldPush) {
	167	+ $pushInd --;
	168	+ }
	169	+ if ($tagStr == "") {
	170	+ $tagStr = $childTag;
	171	+ } elseif ($childTag != "") {
	172	+ $tagStr .= "\|" . $childTag;
	173	+ }
	174	+ } else {
	175	+ $shouldPush = false;
	176	+ $tagStr = $childTag;
	177	+ }
	178	+ }
	179	+ }
	180	+ $rule->setAttribute("pushInd", $pushInd);
	181	+ } else {
	182	+ if ($rule->nodeName != "Choice") {
	183	+ $rule->setAttribute("saveTags", $tagStr);
	184	+ $tagStr = NULL;
	185	+ if ($rule->nodeName == "Text") {
	186	+ $childTags = "";
	187	+ foreach ($rule->childNodes as $crrnt) {
	188	+ if ($childTags != "") {
	189	+ $childTags .= "\|";
	190	+ }
	191	+ $this->pullTags($crrnt, $iter, $childTag);
	192	+ $childTags .= $childTag;
	193	+ }
	194	+ $rule->setAttribute("childTags", $childTags);
	195	+ }
	196	+ }
	197	+ foreach ($rule->childNodes as $crrnt) {
	198	+ $this->pushTags($crrnt, $tagStr);
	199	+ }
	200	+ }
	201	+ }
	202	+
	203	+ private function pullTags($rule, &$iter, &$childTags) {
	204	+ $iter ++;
	205	+ if ($iter > ParseEngine::maxIter) {
	206	+ throw new MWException("Collecter iterated too many times. Probable loop in grammar.");
	207	+ }
	208	+ $childTags = "";
	209	+ $failSafe = TRUE;
	210	+ if ($rule->nodeName == "Assignment") {
	211	+ $childTags = $rule->getAttribute("tag");
	212	+ if ($rule->getAttribute("regex") == NULL) {
	213	+ $childTags = preg_quote($childTags, "/");
	214	+ }
	215	+ $failSafe = FALSE;
	216	+ } elseif ($rule->nodeName == "Choice" \|\| $rule->nodeName == "Sequence") {
	217	+ $failSafe = $rule->nodeName == "Sequence";
	218	+ foreach ($rule->childNodes as $child) {
	219	+ $failSafe = $this->pullTags($child, $iter, $newTags);
	220	+ if ($childTags == "") {
	221	+ $childTags = $newTags;
	222	+ } elseif ($newTags != "") {
	223	+ $childTags .= "\|" . $newTags;
	224	+ }
	225	+ if (($failSafe && $rule->nodeName == "Choice") \|\| (! $failSafe && $rule->nodeName == "Sequence")) {
	226	+ break;
	227	+ }
	228	+ }
	229	+ $failSafe \|= $rule->nodeName == "Choice" && $rule->getAttribute("failSafe") != NULL;
	230	+ } elseif ($rule->nodeName == "Reference") {
	231	+ $xpath = new DOMXPath($rule->ownerDocument);
	232	+ $refRule = $xpath->query("/Grammar/*[@name='{$rule->getAttribute("name")}']")->item(0);
	233	+ $failSafe = $this->pullTags($refRule, $iter, $childTags);
	234	+ }
	235	+ return $failSafe;
	236	+ }
	237	+}
	238	+
Property changes on: trunk/extensions/ParseEngine/ParseEngine.body.php
___________________________________________________________________
Name: svn:eol-style
1	239	+ native
Index: trunk/extensions/ParseEngine/ParseEngine.php
—	—	@@ -0,0 +1,38 @@
	2	+<?php
	3	+/**
	4	+ * Allows people to define a grammar in a wiki then use that grammar to input information to the same wiki
	5	+ * @file
	6	+ * @ingroup Extensions
	7	+ * @author Nathanael Thompson <than4213@gmail.com>
	8	+ * @copyright Copyright © 2010 Nathanael Thompson
	9	+ * @license http://www.gnu.org/copyleft/gpl.html GNU General Public License
	10	+ */
	11	+if ( !defined( "MEDIAWIKI" ) ) {
	12	+ die( "This is not a valid entry point.\n" );
	13	+}
	14	+
	15	+$wgExtensionCredits["other"][] = array(
	16	+ "path" => __FILE__,
	17	+ "name" => "ParseEngine",
	18	+ "author" => "Nathanael Thompson",
	19	+ "url" => "http://www.mediawiki.org/wiki/Extension:ParseEngine",
	20	+ "version" => "1.0",
	21	+ "descriptionmsg" => "parseengine-desc",
	22	+);
	23	+
	24	+$dir = dirname( __FILE__ );
	25	+$wgAutoloadClasses["ParseEngine"] = "$dir/ParseEngine.body.php";
	26	+
	27	+$wgTheParseEngine = new ParseEngine();
	28	+$wgHooks["BeforePreSaveTransform"][] = array($wgTheParseEngine, "parse", $wgParseEngineGrammar);
	29	+$wgHooks["ParserBeforeStrip"][] = "wfParseEngineCallFromParse";
	30	+
	31	+define ( "NS_GRAMMAR" , 91628);
	32	+define ( "NS_GRAMMAR_TALK" , 91629);
	33	+$wgExtraNamespaces[NS_GRAMMAR] = "Grammar";
	34	+$wgExtraNamespaces[NS_GRAMMAR_TALK] = "Grammar_talk";
	35	+
	36	+function wfParseEngineCallFromParse($unUsed, $text) {
	37	+ global $wgTheParseEngine, $wgParseEngineGrammar;
	38	+ return $wgTheParseEngine->parse($wgParseEngineGrammar, $text);
	39	+}
Property changes on: trunk/extensions/ParseEngine/ParseEngine.php
___________________________________________________________________
Name: svn:eol-style
1	40	+ native
Index: trunk/extensions/ParseEngine/WikiTextGrammar.xml
—	—	@@ -0,0 +1,145 @@
	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>
Property changes on: trunk/extensions/ParseEngine/WikiTextGrammar.xml
___________________________________________________________________
Name: svn:eol-style
1	147	+ native
Property changes on: trunk/extensions/DataTransclusion/WebDataTransclusionSource.php
___________________________________________________________________
Name: svn:mergeinfo
2	148	-

Past revisions this follows-up on

Revision	Commit summary	Author	Date
r65412	Rename extension to ParserWiki. Added test case to see if my parser language...	than4213	19:40, 21 April 2010

Status & tagging log

11:34, 6 July 2010 Reedy (talk | contribs) changed the status of r67702 [removed: new added: deferred]