r65437 MediaWiki - Code Review archive

Repository:	MediaWiki
Revision:	< r65436‎ \| r65437 \| r65438 >
Date:	19:52, 22 April 2010
Author:	raymond
Status:	ok
Tags:
Comment:	Run stylize.php
Modified paths:	/trunk/extensions/ParserWiki/ParseEngine.php (modified) (history) /trunk/extensions/ParserWiki/ParserWiki.php (modified) (history)

Diff [purge]

Index: trunk/extensions/ParserWiki/ParseEngine.php
—	—	@@ -10,136 +10,136 @@
11	11	const maxIter = 4096;
12	12	private $mGrammar;
13	13
14		~~- function __construct($grammar) {~~
15		~~- $xpath = new DOMXPath($grammar);~~
16		~~- $rootRules = $xpath->query("/grammar/parseObject \| /grammar/rule/parseObject");~~
17		~~- foreach ($rootRules as $child) {~~
18		~~- $this->pushTags($child, NULL);~~
	14	+ function __construct( $grammar ) {
	15	+ $xpath = new DOMXPath( $grammar );
	16	+ $rootRules = $xpath->query( "/grammar/parseObject \| /grammar/rule/parseObject" );
	17	+ foreach ( $rootRules as $child ) {
	18	+ $this->pushTags( $child, NULL );
19	19	}
20	20	$this->mGrammar = $grammar;
21	21	}
22	22
23		~~- function parse($text) {~~
24		~~- wfDebugLog("ParseEngine", "==========Start Parse Engine==========\n");~~
25		~~- $xpath = new DOMXPath($this->mGrammar);~~
26		~~- $rootAssign = $xpath->query("/grammar/parseObject")->item(0);~~
	23	+ function parse( $text ) {
	24	+ wfDebugLog( "ParseEngine", "==========Start Parse Engine==========\n" );
	25	+ $xpath = new DOMXPath( $this->mGrammar );
	26	+ $rootAssign = $xpath->query( "/grammar/parseObject" )->item( 0 );
27	27	$doc = new DOMDocument();
28		~~- if (! $this->parseRec($rootAssign, "", "", $iter, $text, $doc)) {~~
	28	+ if ( ! $this->parseRec( $rootAssign, "", "", $iter, $text, $doc ) ) {
29	29	$doc = NULL;
30	30	}
31	31	return $doc;
32	32	}
33	33
34		~~- static function unparse($inNodes) {~~
	34	+ static function unparse( $inNodes ) {
35	35	$retStr = "";
36		~~- foreach ($inNodes as $child) {~~
37		~~- if ($child instanceof DOMText) {~~
	36	+ foreach ( $inNodes as $child ) {
	37	+ if ( $child instanceof DOMText ) {
38	38	$retStr .= $child->data;
39	39	} else {
40		~~- $retStr .= $child->getAttribute("tag") . self::unparse($child->childNodes);~~
	40	+ $retStr .= $child->getAttribute( "tag" ) . self::unparse( $child->childNodes );
41	41	}
42	42	}
43	43	return $retStr;
44	44	}
45	45
46		~~- private function parseRec($rule, $replaceStr, $saveTags, &$iter, &$text, &$outNode) {~~
	46	+ private function parseRec( $rule, $replaceStr, $saveTags, &$iter, &$text, &$outNode ) {
47	47	$iter ++;
48		~~- if ($iter > ParseEngine::maxIter) {~~
49		~~- throw new MWException("Parser iterated too many times. Probable loop in grammar.");~~
	48	+ if ( $iter > ParseEngine::maxIter ) {
	49	+ throw new MWException( "Parser iterated too many times. Probable loop in grammar." );
50	50	}
51	51	$rule = $rule->firstChild;
52		~~- if ($rule->nodeName == "assignment" \|\| $rule->nodeName == "reference" \|\| $rule->nodeName == "text") {~~
53		~~- $saveTags = str_replace("~r", preg_quote($replaceStr, "/"), $saveTags);~~
54		~~- $newTags = $rule->getAttribute("saveTags");~~
55		~~- if ($saveTags == "") {~~
	52	+ if ( $rule->nodeName == "assignment" \|\| $rule->nodeName == "reference" \|\| $rule->nodeName == "text" ) {
	53	+ $saveTags = str_replace( "~r", preg_quote( $replaceStr, "/" ), $saveTags );
	54	+ $newTags = $rule->getAttribute( "saveTags" );
	55	+ if ( $saveTags == "" ) {
56	56	$saveTags = $newTags;
57		~~- } elseif ($newTags != "") {~~
	57	+ } elseif ( $newTags != "" ) {
58	58	$saveTags .= "\|" . $newTags;
59	59	}
60	60	}
61	61	$dom = $outNode->ownerDocument == NULL ? $outNode : $outNode->ownerDocument;
62		~~- $xpath = new DOMXPath($rule->ownerDocument);~~
63		~~- $childRules = $xpath->query("parseObject", $rule);~~
	62	+ $xpath = new DOMXPath( $rule->ownerDocument );
	63	+ $childRules = $xpath->query( "parseObject", $rule );
64	64	$retCode = TRUE;
65		~~- if ($rule->nodeName == "assignment") {~~
66		~~- $patterns = $xpath->query("pattern", $rule);~~
	65	+ if ( $rule->nodeName == "assignment" ) {
	66	+ $patterns = $xpath->query( "pattern", $rule );
67	67	$tag = "";
68		~~- if ($patterns->length > 0) {~~
69		~~- $pattern = str_replace("~r", $replaceStr, $patterns->item(0)->getAttribute("tag"));~~
70		~~- $retCode = preg_match("/^$pattern/s", $text, $matches);~~
71		~~- if ($retCode) {~~
	68	+ if ( $patterns->length > 0 ) {
	69	+ $pattern = str_replace( "~r", $replaceStr, $patterns->item( 0 )->getAttribute( "tag" ) );
	70	+ $retCode = preg_match( "/^$pattern/s", $text, $matches );
	71	+ if ( $retCode ) {
72	72	$tag = $matches[0];
73		~~- if (isset($matches[1])) {~~
	73	+ if ( isset( $matches[1] ) ) {
74	74	$replaceStr = $matches[1];
75	75	}
76	76	}
77	77	}
78		~~- if ($retCode) {~~
	78	+ if ( $retCode ) {
79	79	$newText = $text;
80		~~- $newElement = $dom->createElement($rule->getAttribute("tag"));~~
81		~~- if ($tag != "") {~~
82		~~- $newText = substr($newText, strlen($tag));~~
83		~~- $newElement->setAttribute("tag", $tag);~~
	80	+ $newElement = $dom->createElement( $rule->getAttribute( "tag" ) );
	81	+ if ( $tag != "" ) {
	82	+ $newText = substr( $newText, strlen( $tag ) );
	83	+ $newElement->setAttribute( "tag", $tag );
84	84	}
85		~~- $retCode = $childRules->length <= 0 \|\| $this->parseRec($childRules->item(0), $replaceStr, $saveTags, $iter, $newText, $newElement);~~
86		~~- if ($retCode) {~~
87		~~- $outNode->appendChild($newElement);~~
	85	+ $retCode = $childRules->length <= 0 \|\| $this->parseRec( $childRules->item( 0 ), $replaceStr, $saveTags, $iter, $newText, $newElement );
	86	+ if ( $retCode ) {
	87	+ $outNode->appendChild( $newElement );
88	88	$text = $newText;
89	89	}
90	90	}
91		~~- } elseif ($rule->nodeName == "sequence") {~~
92		~~- $pushInd = $rule->getAttribute("pushInd");~~
93		~~- if ($pushInd > 0) {~~
	91	+ } elseif ( $rule->nodeName == "sequence" ) {
	92	+ $pushInd = $rule->getAttribute( "pushInd" );
	93	+ if ( $pushInd > 0 ) {
94	94	$saveText = $text;
95		~~- $saveNode = $outNode->cloneNode(TRUE);~~
	95	+ $saveNode = $outNode->cloneNode( TRUE );
96	96	}
97		~~- foreach ($childRules as $i => $child) {~~
	97	+ foreach ( $childRules as $i => $child ) {
98	98	$pushTags = $i >= $pushInd ? $saveTags : "";
99		~~- $retCode = $this->parseRec($child, $replaceStr, $pushTags, $iter, $text, $outNode);~~
100		~~- if (! $retCode) {~~
101		~~- if ($i > 0) {~~
	99	+ $retCode = $this->parseRec( $child, $replaceStr, $pushTags, $iter, $text, $outNode );
	100	+ if ( ! $retCode ) {
	101	+ if ( $i > 0 ) {
102	102	$text = $saveText;
103	103	$outNode = $saveNode;
104	104	}
105	105	break;
106	106	}
107	107	}
108		~~- } elseif ($rule->nodeName == "choice") {~~
109		~~- foreach ($childRules as $child) {~~
110		~~- $retCode = $this->parseRec($child, $replaceStr, $saveTags, $iter, $text, $outNode);~~
111		~~- if ($retCode) {~~
	108	+ } elseif ( $rule->nodeName == "choice" ) {
	109	+ foreach ( $childRules as $child ) {
	110	+ $retCode = $this->parseRec( $child, $replaceStr, $saveTags, $iter, $text, $outNode );
	111	+ if ( $retCode ) {
112	112	break;
113	113	}
114	114	}
115		~~- $retCode \|= $rule->hasAttribute("tag");~~
116		~~- } elseif ($rule->nodeName == "reference") {~~
117		~~- $childRule = $rule->getAttribute("tag");~~
118		~~- wfDebugLog("ParseEngine", "Entering $childRule\n");~~
119		~~- $varNode = $xpath->query("pattern", $rule);~~
120		~~- if ($varNode->length > 0) {~~
121		~~- $replaceStr = str_replace("~r", $replaceStr, $varNode->item(0)->getAttribute("tag"));~~
	115	+ $retCode \|= $rule->hasAttribute( "tag" );
	116	+ } elseif ( $rule->nodeName == "reference" ) {
	117	+ $childRule = $rule->getAttribute( "tag" );
	118	+ wfDebugLog( "ParseEngine", "Entering $childRule\n" );
	119	+ $varNode = $xpath->query( "pattern", $rule );
	120	+ if ( $varNode->length > 0 ) {
	121	+ $replaceStr = str_replace( "~r", $replaceStr, $varNode->item( 0 )->getAttribute( "tag" ) );
122	122	}
123		~~- $refRule = $xpath->query("/grammar/rule[@tag='$childRule']/parseObject")->item(0);~~
124		~~- $retCode = $this->parseRec($refRule, $replaceStr, $saveTags, $iter, $text, $outNode);~~
125		~~- wfDebugLog("ParseEngine", "Exiting $childRule, Return Code - $retCode\n");~~
126		~~- wfDebugLog("ParseEngine", "text - $text\n");~~
127		~~- } elseif ($rule->nodeName == "text") {~~
128		~~- $tagSearch = $rule->getAttribute("childTags");~~
129		~~- if ($tagSearch == "") {~~
	123	+ $refRule = $xpath->query( "/grammar/rule[@tag='$childRule']/parseObject" )->item( 0 );
	124	+ $retCode = $this->parseRec( $refRule, $replaceStr, $saveTags, $iter, $text, $outNode );
	125	+ wfDebugLog( "ParseEngine", "Exiting $childRule, Return Code - $retCode\n" );
	126	+ wfDebugLog( "ParseEngine", "text - $text\n" );
	127	+ } elseif ( $rule->nodeName == "text" ) {
	128	+ $tagSearch = $rule->getAttribute( "childTags" );
	129	+ if ( $tagSearch == "" ) {
130	130	$tagSearch = $saveTags;
131		~~- } elseif ($saveTags != "") {~~
	131	+ } elseif ( $saveTags != "" ) {
132	132	$tagSearch .= "\|" . $saveTags;
133	133	}
134		~~- $childNode = $childRules->length <= 0 ? NULL : $childRules->item(0);~~
135		~~- while ($text != "" && ($saveTags == "" \|\| ! preg_match("/^($saveTags)/s", $text))) {~~
136		~~- $offset = $childNode != NULL && $this->parseRec($childNode, $replaceStr, "", $iter, $text, $outNode) ? 0 : 1;~~
137		~~- if (preg_match("/$tagSearch/s", $text, $matches, PREG_OFFSET_CAPTURE, $offset)) {~~
138		~~- if ($matches[0][1] > 0) {~~
139		~~- $outNode->appendChild($dom->createTextNode(substr($text, 0, $matches[0][1])));~~
140		~~- $text = substr($text, $matches[0][1]);~~
	134	+ $childNode = $childRules->length <= 0 ? NULL : $childRules->item( 0 );
	135	+ while ( $text != "" && ( $saveTags == "" \|\| ! preg_match( "/^($saveTags)/s", $text ) ) ) {
	136	+ $offset = $childNode != NULL && $this->parseRec( $childNode, $replaceStr, "", $iter, $text, $outNode ) ? 0 : 1;
	137	+ if ( preg_match( "/$tagSearch/s", $text, $matches, PREG_OFFSET_CAPTURE, $offset ) ) {
	138	+ if ( $matches[0][1] > 0 ) {
	139	+ $outNode->appendChild( $dom->createTextNode( substr( $text, 0, $matches[0][1] ) ) );
	140	+ $text = substr( $text, $matches[0][1] );
141	141	}
142	142	} else {
143		~~- $outNode->appendChild($dom->createTextNode($text));~~
	143	+ $outNode->appendChild( $dom->createTextNode( $text ) );
144	144	$text = "";
145	145	}
146	146	}
—	—	@@ -147,85 +147,85 @@
148	148	return $retCode;
149	149	}
150	150
151		~~- private function pushTags($rule, $tagStr) {~~
	151	+ private function pushTags( $rule, $tagStr ) {
152	152	$rule = $rule->firstChild;
153		~~- $xpath = new DOMXPath($rule->ownerDocument);~~
154		~~- $childRules = $xpath->query("parseObject", $rule);~~
155		~~- if ($rule->nodeName == "sequence") {~~
	153	+ $xpath = new DOMXPath( $rule->ownerDocument );
	154	+ $childRules = $xpath->query( "parseObject", $rule );
	155	+ if ( $rule->nodeName == "sequence" ) {
156	156	$pushInd = 0;
157		~~- for ($i = $childRules->length - 1; $i >= 0; $i --) {~~
158		~~- $this->pushTags($childRules->item($i), $tagStr);~~
159		~~- if ($i > 0) {~~
160		~~- if ($this->pullTags($childRules->item($i), $iter, $childTag)) {~~
161		~~- if ($tagStr == "") {~~
	157	+ for ( $i = $childRules->length - 1; $i >= 0; $i -- ) {
	158	+ $this->pushTags( $childRules->item( $i ), $tagStr );
	159	+ if ( $i > 0 ) {
	160	+ if ( $this->pullTags( $childRules->item( $i ), $iter, $childTag ) ) {
	161	+ if ( $tagStr == "" ) {
162	162	$tagStr = $childTag;
163		~~- } elseif ($childTag != "") {~~
	163	+ } elseif ( $childTag != "" ) {
164	164	$tagStr .= "\|" . $childTag;
165	165	}
166	166	} else {
167		~~- if ($pushInd < $i) {~~
	167	+ if ( $pushInd < $i ) {
168	168	$pushInd = $i;
169	169	}
170	170	$tagStr = $childTag;
171	171	}
172	172	}
173	173	}
174		~~- $rule->setAttribute("pushInd", $pushInd);~~
	174	+ $rule->setAttribute( "pushInd", $pushInd );
175	175	} else {
176		~~- if ($rule->nodeName != "choice") {~~
177		~~- $rule->setAttribute("saveTags", $tagStr);~~
	176	+ if ( $rule->nodeName != "choice" ) {
	177	+ $rule->setAttribute( "saveTags", $tagStr );
178	178	$tagStr = NULL;
179		~~- if ($rule->nodeName == "text") {~~
	179	+ if ( $rule->nodeName == "text" ) {
180	180	$childTags = "";
181		~~- foreach ($childRules as $child) {~~
182		~~- if ($childTags != "") {~~
	181	+ foreach ( $childRules as $child ) {
	182	+ if ( $childTags != "" ) {
183	183	$childTags .= "\|";
184	184	}
185		~~- $this->pullTags($child, $iter, $childTag);~~
	185	+ $this->pullTags( $child, $iter, $childTag );
186	186	$childTags .= $childTag;
187	187	}
188		~~- $rule->setAttribute("childTags", $childTags);~~
	188	+ $rule->setAttribute( "childTags", $childTags );
189	189	}
190	190	}
191		~~- foreach ($childRules as $child) {~~
192		~~- $this->pushTags($child, $tagStr);~~
	191	+ foreach ( $childRules as $child ) {
	192	+ $this->pushTags( $child, $tagStr );
193	193	}
194	194	}
195	195	}
196	196
197		~~- private function pullTags($rule, &$iter, &$childTags) {~~
	197	+ private function pullTags( $rule, &$iter, &$childTags ) {
198	198	$iter ++;
199		~~- if ($iter > ParseEngine::maxIter) {~~
200		~~- throw new MWException("Collecter iterated too many times. Probable loop in grammar.");~~
	199	+ if ( $iter > ParseEngine::maxIter ) {
	200	+ throw new MWException( "Collecter iterated too many times. Probable loop in grammar." );
201	201	}
202	202	$rule = $rule->firstChild;
203		~~- $xpath = new DOMXPath($rule->ownerDocument);~~
	203	+ $xpath = new DOMXPath( $rule->ownerDocument );
204	204	$childTags = "";
205	205	$failSafe = TRUE;
206		~~- if ($rule->nodeName == "assignment") {~~
207		~~- $patterns = $xpath->query("pattern", $rule);~~
208		~~- if ($patterns->length > 0) {~~
209		~~- $childTags = $patterns->item(0)->getAttribute("tag");~~
	206	+ if ( $rule->nodeName == "assignment" ) {
	207	+ $patterns = $xpath->query( "pattern", $rule );
	208	+ if ( $patterns->length > 0 ) {
	209	+ $childTags = $patterns->item( 0 )->getAttribute( "tag" );
210	210	}
211	211	$failSafe = FALSE;
212		~~- } elseif ($rule->nodeName == "choice" \|\| $rule->nodeName == "sequence") {~~
213		~~- $childRules = $xpath->query("parseObject", $rule);~~
	212	+ } elseif ( $rule->nodeName == "choice" \|\| $rule->nodeName == "sequence" ) {
	213	+ $childRules = $xpath->query( "parseObject", $rule );
214	214	$failSafe = $rule->nodeName == "sequence";
215		~~- foreach ($childRules as $child) {~~
216		~~- $failSafe = $this->pullTags($child, $iter, $newTags);~~
217		~~- if ($childTags == "") {~~
	215	+ foreach ( $childRules as $child ) {
	216	+ $failSafe = $this->pullTags( $child, $iter, $newTags );
	217	+ if ( $childTags == "" ) {
218	218	$childTags = $newTags;
219		~~- } elseif ($newTags != "") {~~
	219	+ } elseif ( $newTags != "" ) {
220	220	$childTags .= "\|" . $newTags;
221	221	}
222		~~- if (($failSafe && $rule->nodeName == "choice") \|\| (! $failSafe && $rule->nodeName == "sequence")) {~~
	222	+ if ( ( $failSafe && $rule->nodeName == "choice" ) \|\| ( ! $failSafe && $rule->nodeName == "sequence" ) ) {
223	223	break;
224	224	}
225	225	}
226		~~- $failSafe \|= $rule->hasAttribute("tag");~~
227		~~- } elseif ($rule->nodeName == "reference") {~~
228		~~- $refRule = $xpath->query("/grammar/rule[@tag='{$rule->getAttribute("tag")}']/parseObject")->item(0);~~
229		~~- $failSafe = $this->pullTags($refRule, $iter, $childTags);~~
	226	+ $failSafe \|= $rule->hasAttribute( "tag" );
	227	+ } elseif ( $rule->nodeName == "reference" ) {
	228	+ $refRule = $xpath->query( "/grammar/rule[@tag='{$rule->getAttribute("tag")}']/parseObject" )->item( 0 );
	229	+ $failSafe = $this->pullTags( $refRule, $iter, $childTags );
230	230	}
231	231	return $failSafe;
232	232	}
Index: trunk/extensions/ParserWiki/ParserWiki.php
—	—	@@ -22,10 +22,10 @@
23	23	$wgAutoloadClasses["ParseEngine"] = dirname( __FILE__ ) . "/ParseEngine.php";
24	24
25	25	$wgTheParserWiki = new ParserWiki();
26		~~-$wgHooks["ParserBeforeStrip"][] = array($wgTheParserWiki, "callFromParse");~~
	26	+$wgHooks["ParserBeforeStrip"][] = array( $wgTheParserWiki, "callFromParse" );
27	27
28		~~-define ( "NS_GRAMMAR" , 91628);~~
29		~~-define ( "NS_GRAMMAR_TALK" , 91629);~~
	28	+define ( "NS_GRAMMAR" , 91628 );
	29	+define ( "NS_GRAMMAR_TALK" , 91629 );
30	30	$wgExtraNamespaces[NS_GRAMMAR] = "Grammar";
31	31	$wgExtraNamespaces[NS_GRAMMAR_TALK] = "Grammar_talk";
32	32
—	—	@@ -36,20 +36,20 @@
37	37	$this->mEngines = array();
38	38	}
39	39
40		~~- function callFromParse($unUsed, &$text) {~~
	40	+ function callFromParse( $unUsed, &$text ) {
41	41	global $wgParserWikiGrammar;
42	42	$engine = $this->mEngines[$wgParserWikiGrammar];
43		~~- if ($engine == NULL) {~~
44		~~- $revision = Revision::newFromTitle(Title::newFromText($wgParserWikiGrammar, NS_GRAMMAR));~~
	43	+ if ( $engine == NULL ) {
	44	+ $revision = Revision::newFromTitle( Title::newFromText( $wgParserWikiGrammar, NS_GRAMMAR ) );
45	45	$grammar = new DOMDocument();
46		~~- if ($revision == NULL \|\| ! $grammar->loadXML($revision->getText(), LIBXML_NOBLANKS)) {~~
	46	+ if ( $revision == NULL \|\| ! $grammar->loadXML( $revision->getText(), LIBXML_NOBLANKS ) ) {
47	47	return TRUE;
48	48	}
49		~~- $engine = new ParseEngine($grammar);~~
	49	+ $engine = new ParseEngine( $grammar );
50	50	$this->mEngines[$wgParserWikiGrammar] = $engine;
51	51	}
52		~~- $parseTree = $engine->parse($text);~~
53		~~- if ($parseTree == NULL) {~~
	52	+ $parseTree = $engine->parse( $text );
	53	+ if ( $parseTree == NULL ) {
54	54	return TRUE;
55	55	}
56	56	$text = $parseTree->saveXML();

Status & tagging log

21:11, 1 May 2010 Siebrand (talk | contribs) changed the status of r65437 [removed: new added: ok]