r13883 MediaWiki - Code Review archive

Repository:	MediaWiki
Revision:	< r13882‎ \| r13883 \| r13884 >
Date:	09:22, 27 April 2006
Author:	magnusmanske
Status:	old
Tags:
Comment:	Adding MediaWiki extension capability
Modified paths:	/trunk/wiki2xml/php/default.php (modified) (history) /trunk/wiki2xml/php/extension.php (added) (history) /trunk/wiki2xml/php/filter_named_entities.php (modified) (history) /trunk/wiki2xml/php/mediawiki_converter.php (modified) (history) /trunk/wiki2xml/php/w2x.php (modified) (history) /trunk/wiki2xml/php/xml2docbook_xml.php (modified) (history) /trunk/wiki2xml/php/xml2odt.php (modified) (history) /trunk/wiki2xml/php/xml2txt.php (modified) (history)

Diff [purge]

Index: trunk/wiki2xml/php/xml2odt.php
—	—	@@ -658,6 +658,6 @@
659	659	}
660	660	}
661	661
662		~~-require_once ( "./xml2tree.php" ) ; # Uses the "element" class defined above~~
	662	+require_once ( "xml2tree.php" ) ; # Uses the "element" class defined above
663	663
664	664	?>
Index: trunk/wiki2xml/php/mediawiki_converter.php
—	—	@@ -60,7 +60,7 @@
61	61	*/
62	62	function articles2text ( &$xml , $params = array () ) {
63	63	global $wiki2xml_authors ;
64		~~- require_once ( "./xml2txt.php" ) ;~~
	64	+ require_once ( "xml2txt.php" ) ;
65	65
66	66	$wiki2xml_authors = array () ;
67	67	$x2t = new xml2php ;
—	—	@@ -91,7 +91,7 @@
92	92	*/
93	93	function articles2xhtml ( &$xml , $params = array () ) {
94	94	global $xml2xhtml ;
95		~~- require_once ( "./xml2xhtml.php" ) ;~~
	95	+ require_once ( "xml2xhtml.php" ) ;
96	96	$lang = "EN" ; # Dummy
97	97
98	98	$ret = "" ;
—	—	@@ -127,7 +127,7 @@
128	128	*/
129	129	function articles2odt ( &$xml , $params = array () , $use_gfdl = false ) {
130	130	global $wiki2xml_authors , $xml2odt ;
131		~~- require_once ( "./xml2odt.php" ) ;~~
	131	+ require_once ( "xml2odt.php" ) ;
132	132
133	133	# XML text to tree
134	134	$xml2odt = new XML2ODT ;
—	—	@@ -151,7 +151,7 @@
152	152	*/
153	153	function articles2docbook_xml ( &$xml , $params = array () , $use_gfdl = false ) {
154	154	global $wiki2xml_authors ;
155		~~- require_once ( "./xml2docbook_xml.php" ) ;~~
	155	+ require_once ( "xml2docbook_xml.php" ) ;
156	156
157	157	$wiki2xml_authors = array () ;
158	158	$x2t = new xml2php ;
—	—	@@ -214,6 +214,7 @@
215	215	* Uses articles2docbook_xml
216	216	*/
217	217	function articles2docbook_pdf ( &$xml , $params = array () , $mode = "PDF" ) {
	218	+ global $xmlg ;
218	219	$docbook_xml = $this->articles2docbook_xml ( $xml , $params , $params['add_gfdl'] ) ;
219	220
220	221	# Create temporary directory
—	—	@@ -230,7 +231,7 @@
231	232	fwrite ( $handle , utf8_encode ( $docbook_xml ) ) ;
232	233	fclose ( $handle ) ;
233	234	if ( $params['add_gfdl'] ) {
234		~~- copy ( "./gfdl.xml" , $temp_dir . "/gfdl.xml" ) ;~~
	235	+ copy ( $xmlg['sourcedir'] . "/gfdl.xml" , $temp_dir . "/gfdl.xml" ) ;
235	236	}
236	237
237	238	# Call converter
Index: trunk/wiki2xml/php/filter_named_entities.php
—	—	@@ -5,7 +5,7 @@
6	6	* which do not replace some entities correctly
7	7	*/
8	8
9		~~-$html_named_entities_mapping = array (~~
	9	+$html_named_entities_mapping_mine = array (
10	10	// Obtained with:
11	11	// less /usr/share/xml/entities/xhtml/.ent\|grep '^<!ENTITY'\|sed -e 's/^<\!ENTITY[ \t]$[A-Za-z0-9]$[ \t]"&#$[0-9]$;".$/"\1"=>\2,/' > /home/dom/data/2005/04/entities-table
12	12	"nbsp"=>160,
—	—	@@ -272,8 +272,8 @@
273	273	}
274	274
275	275	function filter_named_entities(&$content) {
276		~~- global $html_named_entities_mapping;~~
277		~~- foreach($html_named_entities_mapping as $name => $value) {~~
	276	+ global $html_named_entities_mapping_mine;
	277	+ foreach($html_named_entities_mapping_mine as $name => $value) {
278	278	$content=str_replace('&'.$name.';',utf8_chr ( $value ),$content);
279	279	}
280	280	$content=str_replace('�','i',$content); # Ugly hack
Index: trunk/wiki2xml/php/xml2docbook_xml.php
—	—	@@ -451,6 +451,6 @@
452	452	}
453	453	}
454	454
455		~~-require_once ( "./xml2tree.php" ) ; # Uses the "element" class defined above~~
	455	+require_once ( "xml2tree.php" ) ; # Uses the "element" class defined above
456	456
457	457	?>
Index: trunk/wiki2xml/php/extension.php
—	—	@@ -0,0 +1,68 @@
	2	+<?php
	3	+/*
	4	+To enable this extension, put all files in this directory into a "wiki2xml" subdirectory of your MediaWiki extensions directory
	5	+Also, add
	6	+ require_once ( "extensions/wiki2xml/extension.php" ) ;
	7	+to your LocalSettings.php
	8	+The extension can then be accessed as [[Special:Wiki2XML]]
	9	+*/
	10	+
	11	+if( !defined( 'MEDIAWIKI' ) ) die();
	12	+
	13	+# Integrating into the MediaWiki environment
	14	+
	15	+$wgExtensionCredits['Wiki2XML'][] = array(
	16	+ 'name' => 'Wiki2XML',
	17	+ 'description' => 'An extension to convert wiki markup into XML.',
	18	+ 'author' => 'Magnus Manske'
	19	+);
	20	+
	21	+$wgExtensionFunctions[] = 'wfWiki2XMLExtension';
	22	+
	23	+
	24	+#_____________________________________________________________________________
	25	+
	26	+/**
	27	+* The special page
	28	+*/
	29	+function wfWiki2XMLExtension() { # Checked for HTML and MySQL insertion attacks
	30	+ global $IP, $wgMessageCache;
	31	+# wfTasksAddCache();
	32	+
	33	+ // FIXME : i18n
	34	+ $wgMessageCache->addMessage( 'wiki2xml', 'Wiki2XML' );
	35	+
	36	+ require_once $IP.'/includes/SpecialPage.php';
	37	+
	38	+ class SpecialWiki2XML extends SpecialPage {
	39	+
	40	+ /**
	41	+ * Constructor
	42	+ */
	43	+ function SpecialWiki2XML() { # Checked for HTML and MySQL insertion attacks
	44	+ SpecialPage::SpecialPage( 'Wiki2XML' );
	45	+ $this->includable( true );
	46	+ }
	47	+
	48	+ /**
	49	+ * Special page main function
	50	+ */
	51	+ function execute( $par = null ) { # Checked for HTML and MySQL insertion attacks
	52	+ global $wgOut, $wgRequest, $wgUser, $wgTitle, $IP;
	53	+ $fname = 'Special::Tasks:execute';
	54	+ global $xmlg , $html_named_entities_mapping_mine;
	55	+ include_once ( "default.php" ) ;
	56	+ $xmlg['sourcedir'] = $IP.'/extensions/wiki2xml' ;
	57	+ include_once ( "w2x.php" ) ;
	58	+
	59	+ $this->setHeaders();
	60	+ $wgOut->addHtml( $out );
	61	+ }
	62	+
	63	+ } # end of class
	64	+
	65	+ SpecialPage::addPage( new SpecialWiki2XML );
	66	+}
	67	+
	68	+
	69	+?>
Index: trunk/wiki2xml/php/w2x.php
—	—	@@ -2,7 +2,10 @@
3	3	# Copyright by Magnus Manske (2005)
4	4	# Released under GPL
5	5
6		~~-include_once ( "default.php" ) ; # Which will include local.php, if available~~
	6	+if( !defined( 'MEDIAWIKI' ) ) { # Stand-alone
	7	+ include_once ( "default.php" ) ; # Which will include local.php, if available
	8	+}
	9	+
7	10	require_once ( "mediawiki_converter.php" ) ;
8	11
9	12	@set_time_limit ( 0 ) ; # No time limit
—	—	@@ -16,6 +19,67 @@
17	20	return ((float)$usec + (float)$sec);
18	21	}
19	22
	23	+function get_form () {
	24	+ global $xmlg ;
	25	+ $optional = array () ;
	26	+ if ( isset ( $xmlg['docbook']['command_pdf'] ) ) {
	27	+ $optional[] = "<INPUT type='radio' name='output_format' value='docbook_pdf'>DocBook PDF" ;
	28	+ }
	29	+ if ( isset ( $xmlg['docbook']['command_html'] ) ) {
	30	+ $optional[] = "<INPUT type='radio' name='output_format' value='docbook_html'>DocBook HTML" ;
	31	+ }
	32	+ if ( isset ( $xmlg['zip_odt'] ) ) {
	33	+ $optional[] = "<INPUT type='radio' name='output_format' value='odt_xml'>OpenOffice XML" ;
	34	+ $optional[] = "<INPUT type='radio' name='output_format' value='odt'>OpenOffice ODT" ;
	35	+ }
	36	+ $optional = "<br/>" . implode ( "<br/>" , $optional ) ;
	37	+
	38	+
	39	+
	40	+return "<form method='post'>
	41	+<h2>Paste article list or wikitext here</h2>
	42	+<table border='0' width='100%'><tr>
	43	+<td valign='top'><textarea rows='20' cols='80' style='width:100%' name='text'></textarea></td>
	44	+<td width='200px' valign='top' nowrap>
	45	+<INPUT checked type='radio' name='use_templates' value='all'>Use all templates<br/>
	46	+<INPUT type='radio' name='use_templates' value='none'>Do not use templates<br/>
	47	+<INPUT type='radio' name='use_templates' value='these'>Use these templates<br/>
	48	+<INPUT type='radio' name='use_templates' value='notthese'>Use all but these templates<br/>
	49	+<textarea rows='15' cols='30' style='width:100%' name='templates'></textarea>
	50	+</td></tr></table>
	51	+<table border='0'><tr>
	52	+<td valign='top'>
	53	+This is
	54	+<INPUT type='radio' name='whatsthis' value='wikitext'>raw wikitext
	55	+<INPUT checked type='radio' name='whatsthis' value='articlelist'>a list of articles
	56	+<br/>
	57	+
	58	+Site : http://<input type='text' name='site' value='".$xmlg["site_base_url"]."'/>/index.php<br/>
	59	+Title : <input type='text' name='document_title' value='' size=40/><br/>
	60	+<input type='checkbox' name='add_gfdl' value='1' checked>Include GFDL (for some output formats)</input><br/>
	61	+<input type='checkbox' name='keep_categories' value='1' checked>Keep categories</input><br/>
	62	+<input type='checkbox' name='keep_interlanguage' value='1' checked>Keep interlanguage links</input><br/>
	63	+<input type='submit' name='doit' value='Convert'/>
	64	+</td><td valign='top' style='border-left:1px black solid'>
	65	+<b>Output</b>
	66	+<br/><INPUT checked type='radio' name='output_format' value='xml'>XML
	67	+<br/><INPUT type='radio' name='output_format' value='text'>Plain text
	68	+ <input type='checkbox' name='plaintext_markup' value='1' checked>Use *_/ markup</input>
	69	+ <input type='checkbox' name='plaintext_prelink' value='1' checked>Put → before internal links</input>
	70	+<br/><INPUT type='radio' name='output_format' value='xhtml'>XHTML
	71	+<br/><INPUT type='radio' name='output_format' value='docbook_xml'>DocBook XML
	72	+{$optional}
	73	+</tr></table>
	74	+</form>
	75	+<p>
	76	+Known issues:
	77	+<ul>
	78	+<li>In templates, {{{variables}}} used within <nowiki> tags will be replaced as well (too lazy to strip them)</li>
	79	+<li>HTML comments are removed (instead of converted into XML tags)</li>
	80	+</ul>
	81	+</p>" ;
	82	+}
	83	+
20	84	## MAIN PROGRAM
21	85
22	86	if ( isset ( $_POST['doit'] ) ) { # Process
—	—	@@ -83,7 +147,8 @@
84	148	# header("Content-type: application/xhtml+xml");
85	149	echo $converter->articles2xhtml ( $xml , $xmlg ) ;
86	150	} else if ( $format == "odt" \|\| $format == "odt_xml" ) {
87		~~- $cwd = getcwd() ;~~
	151	+ if ( $xmlg['sourcedir'] == '.' ) $cwd = getcwd() ;
	152	+ else $cwd = $xmlg['sourcedir'] ;
88	153	$template_file = $cwd . '/template.odt' ;
89	154
90	155	$dir_file = tempnam($xmlg["temp_dir"], "ODD");
—	—	@@ -169,69 +234,20 @@
170	235	SureRemoveDir ( $pdf_dir ) ;
171	236	@rmdir ( $pdf_dir ) ;
172	237	}
173		-
	238	+ exit ;
174	239	} else { # Show the form
175		~~- header('Content-type: text/html; charset=utf-8');~~
176		-
177		~~- $optional = array () ;~~
178		~~- if ( isset ( $xmlg['docbook']['command_pdf'] ) ) {~~
179		~~- $optional[] = "<INPUT type='radio' name='output_format' value='docbook_pdf'>DocBook PDF" ;~~
	240	+ if( !defined( 'MEDIAWIKI' ) ) { # Stand-alone
	241	+ header('Content-type: text/html; charset=utf-8');
	242	+ print "
	243	+<html><head></head><body>
	244	+<h1>Magnus' magic MediaWiki-to-XML-to-stuff converter</h1>
	245	+<p>All written in PHP - so portable, <s>so incredibly slow...</s> <i>about as fast as the original MediaWiki parser!</i></p>" ;
	246	+ print get_form () ;
	247	+ print "</body></html>" ;
	248	+ } else { # MediaWiki extension
	249	+ $out = get_form () ;
180	250	}
181		~~- if ( isset ( $xmlg['docbook']['command_html'] ) ) {~~
182		~~- $optional[] = "<INPUT type='radio' name='output_format' value='docbook_html'>DocBook HTML" ;~~
183		~~- }~~
184		~~- if ( isset ( $xmlg['zip_odt'] ) ) {~~
185		~~- $optional[] = "<INPUT type='radio' name='output_format' value='odt_xml'>OpenOffice XML" ;~~
186		~~- $optional[] = "<INPUT type='radio' name='output_format' value='odt'>OpenOffice ODT" ;~~
187		~~- }~~
188		~~- $optional = "<br/>" . implode ( "<br/>" , $optional ) ;~~
189	251
190		~~- print "~~
191		~~-<html><head></head><body><form method='post'>~~
192		~~-<h1>Magnus' magic MediaWiki-to-XML-to-stuff converter</h1>~~
193		~~-<p>All written in PHP - so portable, <s>so incredibly slow...</s> <i>about as fast as the original MediaWiki parser!</i></p>~~
194		~~-<h2>Paste article list or wikitext here</h2>~~
195		~~-<table border='0' width='100%'><tr>~~
196		~~-<td valign='top'><textarea rows='20' cols='80' style='width:100%' name='text'></textarea></td>~~
197		~~-<td width='200px' valign='top' nowrap>~~
198		~~-<INPUT checked type='radio' name='use_templates' value='all'>Use all templates<br/>~~
199		~~-<INPUT type='radio' name='use_templates' value='none'>Do not use templates<br/>~~
200		~~-<INPUT type='radio' name='use_templates' value='these'>Use these templates<br/>~~
201		~~-<INPUT type='radio' name='use_templates' value='notthese'>Use all but these templates~~
202		~~-<textarea rows='15' cols='30' style='width:100%' name='templates'></textarea>~~
203		~~-</td></tr></table>~~
204		~~-<table border='0'><tr>~~
205		~~-<td valign='top'>~~
206		~~-This is~~
207		~~-<INPUT type='radio' name='whatsthis' value='wikitext'>raw wikitext~~
208		~~-<INPUT checked type='radio' name='whatsthis' value='articlelist'>a list of articles~~
209		~~-<br/>~~
210		-
211		~~-Site : http://<input type='text' name='site' value='".$xmlg["site_base_url"]."'/>/index.php<br/>~~
212		~~-Title : <input type='text' name='document_title' value='' size=40/><br/>~~
213		~~-<input type='checkbox' name='add_gfdl' value='1' checked>Include GFDL (for some output formats)</input><br/>~~
214		~~-<input type='checkbox' name='keep_categories' value='1' checked>Keep categories</input><br/>~~
215		~~-<input type='checkbox' name='keep_interlanguage' value='1' checked>Keep interlanguage links</input><br/>~~
216		~~-<input type='submit' name='doit' value='Convert'/>~~
217		~~-</td><td valign='top' style='border-left:1px black solid'>~~
218		~~-<b>Output</b>~~
219		~~-<br/><INPUT checked type='radio' name='output_format' value='xml'>XML~~
220		~~-<br/><INPUT type='radio' name='output_format' value='text'>Plain text~~
221		~~- <input type='checkbox' name='plaintext_markup' value='1' checked>Use *_/ markup</input>~~
222		~~- <input type='checkbox' name='plaintext_prelink' value='1' checked>Put → before internal links</input>~~
223		~~-<br/><INPUT type='radio' name='output_format' value='xhtml'>XHTML~~
224		~~-<br/><INPUT type='radio' name='output_format' value='docbook_xml'>DocBook XML~~
225		~~-{$optional}~~
226		~~-</tr></table>~~
227		~~-</form>~~
228		~~-<p>~~
229		~~-Known issues:~~
230		~~-<ul>~~
231		~~-<li>In templates, {{{variables}}} used within <nowiki> tags will be replaced as well (too lazy to strip them)</li>~~
232		~~-<li>HTML comments are removed (instead of converted into XML tags)</li>~~
233		~~-</ul>~~
234		~~-</p>~~
235		~~-</body></html>" ;~~
236	252	}
237	253
238	254	#<input type='checkbox' name='resolvetemplates' value='1' checked>Automatically resolve templates</input><br/>
Index: trunk/wiki2xml/php/xml2txt.php
—	—	@@ -147,7 +147,7 @@
148	148	}
149	149	}
150	150
151		~~-require_once ( "./xml2tree.php" ) ;~~
	151	+require_once ( "xml2tree.php" ) ;
152	152
153	153
154	154
Index: trunk/wiki2xml/php/default.php
—	—	@@ -3,6 +3,7 @@
4	4	$xmlg["namespace_template"] = "Template" ;
5	5	$xmlg["site_base_url"] = "en.wikipedia.org/w" ;
6	6	$xmlg["book_title"] = "No title" ;
	7	+$xmlg['sourcedir'] = "." ;
7	8	$xmlg["temp_dir"] = "/tmp" ;
8	9
9	10	@include ( "local.php" ) ;

Status & tagging log

01:58, 13 October 2010 😂 (talk | contribs) changed the status of r13883 [removed: new added: old]