| 1 | <?PHP
|
|---|
| 2 |
|
|---|
| 3 | error_reporting ( E_ALL ) ;
|
|---|
| 4 |
|
|---|
| 5 | require_once ( "mediawiki_converter.php" ) ;
|
|---|
| 6 |
|
|---|
| 7 | function treat ( $s ) {
|
|---|
| 8 | $arr = array ( 'li','p','dd' ) ;
|
|---|
| 9 | foreach ( $arr AS $a ) $s = str_replace ( "</$a>" , "\n</$a>" , $s ) ;
|
|---|
| 10 | $s = htmlentities ( $s ) ;
|
|---|
| 11 | # $s = "<pre>$s</pre>" ;
|
|---|
| 12 | $s = str_replace ( "\n" , "<br/>\n" , $s ) ;
|
|---|
| 13 | return $s ;
|
|---|
| 14 | }
|
|---|
| 15 |
|
|---|
| 16 | $lines = explode ( "\n" , str_replace ( "\r" , "" , file_get_contents ( "../../phase3/maintenance/parserTests.txt" ) ) ) ; # This path is for trunk
|
|---|
| 17 |
|
|---|
| 18 | $tests = array () ;
|
|---|
| 19 | $articles = array () ;
|
|---|
| 20 | $data = array () ;
|
|---|
| 21 | $cmds = array () ;
|
|---|
| 22 | $command = '' ;
|
|---|
| 23 | $d = '' ;
|
|---|
| 24 | foreach ( $lines AS $l ) {
|
|---|
| 25 | if ( $command == '' ) {
|
|---|
| 26 | if ( $l == '' ) continue ;
|
|---|
| 27 | if ( substr ( $l , 0 , 1 ) == '#' ) continue ;
|
|---|
| 28 | }
|
|---|
| 29 | if ( substr ( $l , 0 , 2 ) == '!!' ) {
|
|---|
| 30 | $new_command = strtolower ( trim ( substr ( $l , 2 ) ) ) ;
|
|---|
| 31 | if ( $new_command == 'end' || $new_command == 'endarticle' ) $new_command = '' ; # Simplify
|
|---|
| 32 |
|
|---|
| 33 | if ( $new_command == '' ) {
|
|---|
| 34 | if ( $cmds[1] == 'article') {
|
|---|
| 35 | $articles[trim($data['article'])] = $d ;
|
|---|
| 36 | } else if ( $cmds[1] == 'test') {
|
|---|
| 37 | $t = '' ;
|
|---|
| 38 | $t->name = trim ( $data['test'] ) ;
|
|---|
| 39 | $t->input = $data['input'] ;
|
|---|
| 40 | if ( isset ( $data['options'] ) ) $t->options = $data['options'] ;
|
|---|
| 41 | else $t->options = '' ;
|
|---|
| 42 | $t->result = $d ;
|
|---|
| 43 | $tests[] = $t ;
|
|---|
| 44 | }
|
|---|
| 45 | $data = array () ;
|
|---|
| 46 | $cmds = array () ;
|
|---|
| 47 | } else {
|
|---|
| 48 | $data[$command] = $d ;
|
|---|
| 49 | $cmds[] = $command ;
|
|---|
| 50 | }
|
|---|
| 51 | $d = '' ;
|
|---|
| 52 | $command = $new_command ;
|
|---|
| 53 | } else {
|
|---|
| 54 | if ( $d != '' ) $d .= "\n" ;
|
|---|
| 55 | $d .= $l ;
|
|---|
| 56 | }
|
|---|
| 57 | }
|
|---|
| 58 |
|
|---|
| 59 | # Run tests
|
|---|
| 60 | $xmlg["useapi"] = false ;
|
|---|
| 61 | $xmlg["book_title"] = 'Title';
|
|---|
| 62 | $xmlg["site_base_url"] = 'en.wikipedia.org/w' ;
|
|---|
| 63 | $xmlg["resolvetemplates"] = 'all' ;
|
|---|
| 64 | $xmlg['templates'] = array () ;
|
|---|
| 65 | $xmlg['add_gfdl'] = false ;
|
|---|
| 66 | $xmlg['keep_interlanguage'] = true ;
|
|---|
| 67 | $xmlg['keep_categories'] = true ;
|
|---|
| 68 | $xmlg['xml_articles_header'] = "<articles>" ;
|
|---|
| 69 |
|
|---|
| 70 | $xmlg['xhtml_justify'] = false ;
|
|---|
| 71 | $xmlg['xhtml_logical_markup'] = false ;
|
|---|
| 72 | $xmlg['xhtml_source'] = false ;
|
|---|
| 73 |
|
|---|
| 74 |
|
|---|
| 75 | $cnt = 1 ;
|
|---|
| 76 | print "<table border=1 width='100%'><tr><th>Test</th><th>Result</th><th>wiki2xml</th><th>Input</th><th>XML</th></tr>" ;
|
|---|
| 77 | foreach ( $tests AS $t ) {
|
|---|
| 78 | $res = $t->result ;
|
|---|
| 79 | $col = '' ;
|
|---|
| 80 | $content_provider = new ContentProviderHTTP ;
|
|---|
| 81 | $converter = new MediaWikiConverter ;
|
|---|
| 82 | $xml = $converter->article2xml ( "" , $t->input , $xmlg ) ;
|
|---|
| 83 | $nr = $converter->articles2xhtml ( $xml , $xmlg ) ;
|
|---|
| 84 | $nr = array_pop ( explode ( '<body>' , $nr , 2 ) ) ;
|
|---|
| 85 | $nr = array_shift ( explode ( '</body>' , $nr , 2 ) ) ;
|
|---|
| 86 |
|
|---|
| 87 | # Fixing things to compare to the stupid parser test formatting
|
|---|
| 88 | $res = trim ( $res ) ;
|
|---|
| 89 | $res = str_replace ( "<li> " , "<li>" , $res ) ;
|
|---|
| 90 | $res = str_replace ( "<dd> " , "<dd>" , $res ) ;
|
|---|
| 91 | $res = str_replace ( "\n<" , "<" , $res ) ;
|
|---|
| 92 | $res = str_replace ( "\n" , " " , $res ) ;
|
|---|
| 93 | $res = str_replace ( " </p>" , "</p>" , $res ) ;
|
|---|
| 94 | do { $o = $res ; $res = str_replace ( " " , " " , $res ) ; } while ( $o != $res ) ;
|
|---|
| 95 |
|
|---|
| 96 | $nr = trim ( $nr ) ;
|
|---|
| 97 | $nr = str_replace ( "<li> " , "<li>" , $nr ) ;
|
|---|
| 98 | $nr = str_replace ( "<dd> " , "<dd>" , $nr ) ;
|
|---|
| 99 | $nr = str_replace ( "\n<" , "<" , $nr ) ;
|
|---|
| 100 | $nr = str_replace ( "\r" , "" , $nr ) ;
|
|---|
| 101 | $nr = str_replace ( "\n" , " " , $nr ) ;
|
|---|
| 102 | do { $o = $nr ; $nr = str_replace ( " " , " " , $nr ) ; } while ( $o != $nr ) ;
|
|---|
| 103 |
|
|---|
| 104 |
|
|---|
| 105 | # $arr = array ( 'li' ) ;
|
|---|
| 106 | # foreach ( $arr AS $a ) $nr = str_replace ( "<$a>" , "<$a> " , $nr ) ;
|
|---|
| 107 |
|
|---|
| 108 |
|
|---|
| 109 | # Indicator color
|
|---|
| 110 | $col = 'red' ;
|
|---|
| 111 | if ( $res == $nr ) $col = 'green' ;
|
|---|
| 112 | # $nr = str_replace ( '</' , "\n</" , $xml ) ;
|
|---|
| 113 |
|
|---|
| 114 |
|
|---|
| 115 |
|
|---|
| 116 | print "<tr><th bgcolor='$col'>" . treat ( $t->name ) . "</th>" ;
|
|---|
| 117 | print "<td>" . treat ( $res ) . "</td>" ;
|
|---|
| 118 | print "<td>" . treat ( $nr ) . "</td>" ;
|
|---|
| 119 | print "<td>" . treat ( $t->input ) . "</td>" ;
|
|---|
| 120 | print "<td>" . treat ( $xml ) . "</td>" ;
|
|---|
| 121 |
|
|---|
| 122 | $cnt++ ;
|
|---|
| 123 | if ( $cnt > 40 ) break ;
|
|---|
| 124 | }
|
|---|
| 125 | print "</table>" ;
|
|---|
| 126 |
|
|---|
| 127 | ?> |
|---|
