| Index: trunk/tools/robots.txt/robots.php |
| — | — | @@ -0,0 +1,92 @@ |
| | 2 | +<?php |
| | 3 | +//$lang = "meta"; |
| | 4 | +define( "MEDIAWIKI", true ); |
| | 5 | +#include_once("CommonSettings.php"); |
| | 6 | +#include_once("/apache/common/wmf-deployment/includes/ProfileStub.php" ); |
| | 7 | +#include_once("/apache/common/wmf-deployment/includes/Defines.php" ); |
| | 8 | +#include_once("/apache/common/wmf-deployment/wmf-config/CommonSettings.php"); |
| | 9 | +#include_once("Setup.php"); |
| | 10 | + |
| | 11 | +include "/apache/common/live-1.5/MWVersion.php"; |
| | 12 | +include getMediaWiki("includes/WebStart.php"); |
| | 13 | + |
| | 14 | +// determine language code |
| | 15 | +$secure = getenv( 'MW_SECURE_HOST' ); |
| | 16 | +if ( (@$_SERVER['SCRIPT_NAME']) == '/w/thumb.php' && (@$_SERVER['SERVER_NAME']) == 'upload.wikimedia.org' ) { |
| | 17 | + $pathBits = explode( '/', $_SERVER['PATH_INFO'] ); |
| | 18 | + $site = $pathBits[1]; |
| | 19 | + $lang = $pathBits[2]; |
| | 20 | +} elseif (php_sapi_name() == 'cgi-fcgi') { |
| | 21 | + if (!preg_match('/^([^.]+).([^.]+).*$/', $_SERVER['SERVER_NAME'], $m)) |
| | 22 | + die("invalid hostname"); |
| | 23 | + |
| | 24 | + $lang = $m[1]; |
| | 25 | + $site = $m[2]; |
| | 26 | + |
| | 27 | + if (in_array($lang, array("commons", "grants", "sources", "wikimania", "wikimania2006", "foundation", "meta"))) |
| | 28 | + $site = "wikipedia"; |
| | 29 | +} elseif( $secure ) { |
| | 30 | + if (!preg_match('/^([^.]+).([^.]+).*$/', $secure, $m)) |
| | 31 | + die("invalid hostname"); |
| | 32 | + |
| | 33 | + $lang = $m[1]; |
| | 34 | + $site = $m[2]; |
| | 35 | + |
| | 36 | + if (in_array($lang, array("commons", "grants", "sources", "wikimania", "wikimania2006", "foundation", "meta"))) |
| | 37 | + $site = "wikipedia"; |
| | 38 | +} else { |
| | 39 | + if ( !isset( $site ) ) { |
| | 40 | + $site = "wikipedia"; |
| | 41 | + if ( !isset( $lang ) ) { |
| | 42 | + |
| | 43 | + $server = $_SERVER['SERVER_NAME']; |
| | 44 | + $docRoot = $_SERVER['DOCUMENT_ROOT']; |
| | 45 | + if ( preg_match( '/^\/usr\/local\/apache\/(?:htdocs|common\/docroot)\/([a-z]+)\.org/', $docRoot, $matches ) ) { |
| | 46 | + $site = $matches[1]; |
| | 47 | + if ( preg_match( '/^(.*)\.' . preg_quote( $site ) . '\.org$/', $server, $matches ) ) { |
| | 48 | + $lang = $matches[1]; |
| | 49 | + // For some special subdomains, like pa.us |
| | 50 | + $lang = str_replace( '.', '-', $lang ); |
| | 51 | + } else { |
| | 52 | + die( "Invalid host name, can't determine language" ); |
| | 53 | + } |
| | 54 | + } elseif ( preg_match( "/^\/usr\/local\/apache\/(?:htdocs|common\/docroot)\/([a-z0-9\-_]*)$/", $docRoot, $matches ) ) { |
| | 55 | + $site = "wikipedia"; |
| | 56 | + $lang = $matches[1]; |
| | 57 | + } else { |
| | 58 | + die( "Invalid host name (docroot=" . $_SERVER['DOCUMENT_ROOT'] . "), can't determine language" ); |
| | 59 | + } |
| | 60 | + } |
| | 61 | + } |
| | 62 | +} |
| | 63 | + |
| | 64 | + |
| | 65 | +$wgTitle = Title::newFromText( 'Mediawiki:robots.txt' ); |
| | 66 | +$wgArticle = new Article( $wgTitle ); |
| | 67 | + |
| | 68 | +header( 'Content-Type: text/plain; charset=utf-8'); |
| | 69 | +header( 'X-Article-ID: ' . $wgArticle->getID() ); |
| | 70 | +header( 'X-Language: ' . $lang ); |
| | 71 | +header( 'X-Site: ' . $site ); |
| | 72 | + |
| | 73 | +$robotsfile = "/apache/common/robots.txt"; |
| | 74 | +$robots = fopen( $robotsfile, 'rb' ); |
| | 75 | +$text=''; |
| | 76 | + |
| | 77 | +header( 'Cache-Control: s-maxage=3600, must-revalidate, max-age=0' ); |
| | 78 | + |
| | 79 | +if ( $wgArticle->getID() != 0 ) { |
| | 80 | + $text = $wgArticle->getContent( false ) ; |
| | 81 | + $lastmod = gmdate( 'D, j M Y H:i:s', wfTimestamp(TS_UNIX, $wgArticle->mTouched ) ) . ' GMT'; |
| | 82 | + header( "Last-modified: $lastmod" ); |
| | 83 | +} else { |
| | 84 | + $stats = fstat( $robots ); |
| | 85 | + |
| | 86 | + $lastmod = gmdate( 'D, j M Y H:i:s', $stats['mtime'] ) . ' GMT'; |
| | 87 | + header( "Last-modified: $lastmod" ); |
| | 88 | + header("Content-Length: " . filesize($robotsfile)); |
| | 89 | +} |
| | 90 | +fpassthru( $robots ); |
| | 91 | + |
| | 92 | +echo "#\n#\n#----------------------------------------------------------#\n#\n#\n#\n" . $text; |
| | 93 | +?> |