r87725 MediaWiki - Code Review archive

Repository:MediaWiki
Revision:< r87724‎ | r87725 | r87726 >
Date:15:35, 9 May 2011
Author:neilk
Status:ok (Comments)
Tags:todo 
Comment:
precheck commons blacklist (in a hacky way, sans API)
Modified paths:
  • /trunk/extensions/UploadWizard/UploadWizard.i18n.php (modified) (history)
  • /trunk/extensions/UploadWizard/UploadWizardHooks.php (modified) (history)
  • /trunk/extensions/UploadWizard/resources/jquery/jquery.validate.wmCommonsBlacklist.js (added) (history)
  • /trunk/extensions/UploadWizard/resources/mw.UploadWizardDetails.js (modified) (history)
  • /trunk/extensions/UploadWizard/resources/uploadWizard.css (modified) (history)

Diff [purge]

Index: trunk/extensions/UploadWizard/UploadWizardHooks.php
——@@ -42,6 +42,9 @@
4343 // mediawiki-specific interface helper (relies on mediawiki globals)
4444 'resources/jquery/jquery.mwCoolCats.js',
4545
 46+ // wikimedia-comons specific title checker
 47+ 'resources/jquery/jquery.validate.wmCommonsBlacklist.js',
 48+
4649 // common utilities
4750 'resources/mw.Log.js',
4851 'resources/mw.Utilities.js',
——@@ -254,6 +257,11 @@
255258 'mwe-upwiz-error-too-short',
256259 'mwe-upwiz-error-bad-chars',
257260 'mwe-upwiz-error-date',
 261+ 'mwe-upwiz-error-title-blacklisted',
 262+ 'mwe-upwiz-error-title-badchars',
 263+ 'mwe-upwiz-error-title-senselessimagename',
 264+ 'mwe-upwiz-error-title-hosting',
 265+ 'mwe-upwiz-error-title-thumbnail',
258266 'mwe-upwiz-license-cc-by-sa-3.0',
259267 'mwe-upwiz-license-cc-by-3.0',
260268 'mwe-upwiz-license-cc-zero',
Index: trunk/extensions/UploadWizard/UploadWizard.i18n.php
——@@ -177,6 +177,11 @@
178178 'mwe-upwiz-error-bad-chars' => 'This field contains symbols that are not allowed.
179179 Please do not use wikitext or HTML here.',
180180 'mwe-upwiz-error-date' => 'Please enter a valid date in YYYY-MM-DD format, or pick a date from the popup calendar.',
 181+ 'mwe-upwiz-error-title-blacklisted' => 'This title contains some undesirable text. Please revise it',
 182+ 'mwe-upwiz-error-title-badchars' => 'This title contains some undesirable characters. Please remove them',
 183+ 'mwe-upwiz-error-title-senselessimagename' => 'Please make this title more meaningful.',
 184+ 'mwe-upwiz-error-title-hosting' => 'This looks like a file you obtained from another imagehost. Please make the title more meaningful. Also, double check that you have the rights to publish it on {{SITENAME}}.',
 185+ 'mwe-upwiz-error-title-thumbnail' => 'This looks like a thumbnail title. Please do not upload thumbnails back to the same wiki. Otherwise, please fix the filename so it is more meaningful, and does not have the thumbnail prefix.',
181186
182187 /* LICENSES & combinations of licenses */
183188 /* may be a good idea to shift to WikimediaLicenseTexts? */
Index: trunk/extensions/UploadWizard/resources/jquery/jquery.validate.wmCommonsBlacklist.js
——@@ -0,0 +1,117 @@
 2+/**
 3+ * One day this will look up an API method to determine if a filename is blacklisted by your local MediaWiki
 4+ * For now we check with some regexes locally, with a Commons-specific blacklist
 5+ * This is an incomplete rendering of the meta.wikimedia.org and commons.wikimedia.org blacklist as they existed on 2011-05-05, and
 6+ * ignores cases that are irrelevant to uploading new media images.
 7+ * - all regexes are case INsensitive by default
 8+ * - casesensitive is considered
 9+ * - errmsg is considered
 10+ * - namespaces and File: prefix are removed since everything we upload is under File: anyway
 11+ * - noedit, moveonly, repuload is irrelevant
 12+ * - we can't check autoconfirmed-ness of users here, so we ignore it
 13+ * - Javascript doesn't have a standard way to access unicode character properties in regexes, so \p{PROPERTY}, \P{PROPERTY}, and [[:PROPERTY:]] have been changed when possible
 14+ * or the associated regex removed
 15+*/
 16+( function( $ ) {
 17+
 18+ var regexSets = {
 19+
 20+ 'titleBlacklist': [
 21+ /(?:suck|his|your|my) penis/i,
 22+ /\bnimp\.org./i,
 23+ /Lawl,/i,
 24+ /HAGG[EA]R[^A-Z]*/i,
 25+ /[–ùHŒó][E–ïŒï]R[M–úŒú][YŒ•]/,
 26+ /[–ùHŒó][AŒëŒë]GG[E–ïŒï]R/,
 27+ /[–ùH]\W*[AŒëŒë]\W*G\W*G\W*[E–ïŒï]\W*R(?!ston)/i,
 28+ /\bHERMY/,
 29+ / on wheels/i,
 30+ /(?:.*?\/)?index\.php(?:\/.*)?/i,
 31+ /(?:http|https|ftp|mailto|torrent|ed2k)\:\/\/[\w\d:@\-]+\.[\w\d\-]+/i,
 32+ /http:\/\//i,
 33+ /\/wiki\//i,
 34+ /\bis\s+(?:a|an)\s+(?:dick|cunt|fag|bitch|shit|fuck|fucker|loser|ass|gay|ghey|moron|retard|stupid|slut|pa?edo)/i,
 35+ /\sprefix:/i, // search from inputboxes (f.e. in the village pumps)
 36+ /.*[!?]{3,}.* /i,
 37+ /(?:[^\/i]+[\/:])?(index\.php|w\/wiki)(?:\/.+)?/,
 38+ /[–ùHŒóÔº®‚±ß][E–ïŒï√ãÔº•ƒñ][RÔº≤–Ø][M–úŒúÔº≠][YŒ•Ôºπ]/i,
 39+ /JEWS DID/i,
 40+ /ON WHE/,
 41+ /(?:Moulton|Barsoom Tork|Pocoyo Albatross|Anything Muppets|Moosey Mouse|Inbloomed Muppets).*/,
 42+ /Jorge Queirolo Bravo/i,
 43+ /Vasilisa(\d)+/i,
 44+ /google.*\.html/i
 45+ ],
 46+
 47+ 'titleBadchars': [
 48+ /[\u00A0\u1680\u180E\u2000-\u200B\u2028\u2029\u202F\u205F\u3000]/, // NBSP and other unusual spaces
 49+ /[\u202A-\u202E]/, // BiDi overrides
 50+ /[\x00-\x1f]/, // Control characters
 51+ /\uFEFF/, // Byte order mark
 52+ /\u00AD/, // Soft-hyphen
 53+ /[\uD800-\uDFFF\uE000-\uF8FF\uFFF0-\uFFFF]/, // Surrogates, Private Use Area and Specials, including the Replacement Character U+FFFD
 54+ /[^\0-\uFFFF]/, // Very few characters outside the Basic Multilingual Plane are useful in titles
 55+ /''/
 56+ ],
 57+
 58+ // note lack of extension, since we test title without extension.
 59+ 'titleSenselessimagename': [
 60+ /^DCP[\d\s]+$/i, // Kodak
 61+ /^DSC.[\d\s]+$/i, // [[w:Design rule for Camera File system]] (Nikon, Fuji, Polaroid)
 62+ /^MVC-?[\d\s]+$/i, // Sony Mavica
 63+ /^P[\dA-F][\d\s]+$/, // Olympus, Kodak
 64+ /^I?MG[P_]?[\d\s]+$/, // Canon, Pentax
 65+ /^1\d+-\d+(_IMG)?$/, // Canon
 66+ /^(IM|EX)[\d\s]+$/, // HP Photosmart
 67+ /^DC[\d\s]+[SML]$/, // Kodak
 68+ /^PIC[T_]?[\d\s]+$/, // Minolta
 69+ /^PANA[\d\s]+$/, // Panasonic
 70+ /^DUW[\d\s]+$/, // some mobile phones
 71+ /^CIMG[\d\s]+$/, // Casio
 72+ /^JD[\d\s]+$/, // Jenoptik
 73+ /^SDC[\d\s]+$/, // Samsung
 74+ /^DVC[\d\s]+$/, // DoCoMo
 75+ /^SANY[\d\s]+$/, // Sanyo
 76+ ],
 77+
 78+ // filename from elsewhere
 79+ 'titleHosting': [
 80+ /\d{9}[A-Z]{6}_[A-Z]{2}[^A-Za-z]*/, // some image hosting site?
 81+ /\d{8,}_[\dA-F]{10}(_[A-Z])?[^A-Za-z]*/i, // http://www.flickr.com/services/api/misc.urls.html/i,
 82+ /([\dA-F]{8}-)?[\dA-F]{4}-[\dA-F]{4}-[\dA-F]{4}-?[\dA-F]{12}.*/, // [[w:UUID]] (with some variations included)
 83+ /([SML]|\d+)_[\dA-F]{10,}(-\d+-|_?(\w\w?|full))?/, // L_9173c67eae58edc35ba7f2df08a7d5c6.jpg, 2421601587_abaf4e3e81.jpg, 1_bf38bcd9c5512a5ab99ca2219a4b1e2f_full.gif, etc.
 84+ /AT[AEIMQUYcgkosw048]AAA[A-D][-_A-Za-z0-9]+/, // see Commons:Village pump#File ATgAAA... (Jan 20, 2009, full link in edit comment)
 85+ /(\d+_){2,}[qtsn]/, // Facebook: \d+_\d+_(\d+)_(\d+)_\d+_[qtsn]\.jpg -> photo.php?id=$1&pid=$2
 86+ /[qtsn]\d+(_\d+)+/, // Facebook (older?): [qtsn](\d+)_(\d+)_\d+\.jpg -> photo.php?id=$1&pid=$2
 87+ /^Tumblr_[a-z\d]{19}_\d+\.w+/, // Tumblr
 88+ /^File$/,
 89+ /^[^A-Za-z]*(small|medium|large)\)?/,
 90+ /^(Untitled|No[-_]?name|Picture|Pict?|Image[mn]?|Img|Immagine|Photo|Foto|Bild|Scan|Panorama|Sin_t√≠tulo)[^A-Za-z]*/,
 91+ /^(January|Jan|February|Febr?|March|Mar|April|Apr|May|June?|July?|August|Aug|September|Sept?|October|Oct|November|Nov|December|Dec)[^A-Za-z]*/,
 92+ /^[0-9 ]*([A-Z][0-9 ]*){30,}/,
 93+ ],
 94+
 95+ 'titleThumbnail': [
 96+ /^\d+px-.*/
 97+ ]
 98+
 99+ };
 100+
 101+ $j.each( regexSets, function( name, regexes ) {
 102+ var tester = ( function( regexes ) {
 103+ return function( value, element, params ) {
 104+ var ok = true;
 105+ $.each( regexes, function( i, regex ) {
 106+ if ( value.match( regex ) ) {
 107+ ok = false;
 108+ return false;
 109+ }
 110+ } );
 111+ return ok;
 112+ };
 113+ } )( regexes );
 114+ $.validator.addMethod( name, tester, "This title is not allowed" );
 115+ } );
 116+
 117+
 118+} )( jQuery );
Property changes on: trunk/extensions/UploadWizard/resources/jquery/jquery.validate.wmCommonsBlacklist.js
___________________________________________________________________
Added: svn:eol-style
1119 + native
Index: trunk/extensions/UploadWizard/resources/uploadWizard.css
——@@ -597,12 +597,12 @@
598598 cursor: pointer;
599599 }
600600
601 -.mwe-error, .mwe-validator-error {
 601+.mwe-error, .mwe-validator-error, .errorTitleUnique {
602602 color: #ff0000;
603603 }
604604
605605 .mwe-upwiz-deed-thirdparty .mwe-upwiz-deed-form-internal label.mwe-error,
606 -.mwe-upwiz-deed-thirdparty .mwe-upwiz-deed-form-internal label.mwe-validator-error {
 606+.mwe-upwiz-deed-thirdparty .mwe-upwiz-deed-form-internal label.mwe-validator-error, label.errorTitleUnique {
607607 margin-left: 0;
608608 }
609609
Index: trunk/extensions/UploadWizard/resources/mw.UploadWizardDetails.js
——@@ -1,4 +1,3 @@
2 -
32 /**
43 * Object that represents the Details (step 2) portion of the UploadWizard
54 * n.b. each upload gets its own details.
——@@ -62,7 +61,10 @@
6362 processResult: function( result ) { _this.processDestinationCheck( result ); }
6463 } );
6564
66 - _this.titleErrorDiv = $j('<div class="mwe-upwiz-details-input-error"><label class="mwe-validator-error" for="' + _this.titleId + '" generated="true"/></div>');
 65+ _this.titleErrorDiv = $j('<div class="mwe-upwiz-details-input-error">'
 66+ + '<label class="mwe-validator-error" for="' + _this.titleId + '" generated="true"/>'
 67+ + '<label class="errorTitleUnique" for="' + _this.titleId + '" generated="true"/>'
 68+ + '</div>');
6769
6870 var titleHintId = 'mwe-upwiz-title-hint-' + _this.upload.index;
6971 var $titleDialog = $('<div>')
——@@ -229,12 +231,22 @@
230232 _this.addDescription( true, mw.config.get( 'wgUserLanguage' ) );
231233 $j( containerDiv ).append( _this.div );
232234
233 - // make the title field required
 235+ // make the title field required, and non-blacklisted
234236 _this.$form.find( '.mwe-title' )
235237 .rules( "add", {
236238 required: true,
 239+ titleBlacklist: true,
 240+ titleBadchars: true,
 241+ titleSenselessimagename: true,
 242+ titleHosting: true,
 243+ titleThumbnail: true,
237244 messages: {
238 - required: gM( 'mwe-upwiz-error-blank' )
 245+ required: gM( 'mwe-upwiz-error-blank' ),
 246+ titleBlacklist: gM( 'mwe-upwiz-error-title-blacklisted' ),
 247+ titleBadchars: gM( 'mwe-upwiz-error-title-badchars' ),
 248+ titleSenselessimagename: gM( 'mwe-upwiz-error-title-senselessimagename' ),
 249+ titleHosting: gM( 'mwe-upwiz-error-title-hosting' ),
 250+ titleThumbnail: gM( 'mwe-upwiz-error-title-thumbnail' )
239251 }
240252 } );
241253
——@@ -323,9 +335,11 @@
324336 */
325337 processDestinationCheck: function( result ) {
326338 var _this = this;
 339+ var $errorEl = _this.$form.find( 'label[for=' + _this.titleId + '].errorTitleUnique' );
 340+
327341 if ( result.isUnique ) {
328342 $j( _this.titleInput ).data( 'valid', true );
329 - _this.$form.find( 'label[for=' + _this.titleId + ']' ).hide().empty();
 343+ $errorEl.hide().empty();
330344 _this.ignoreWarningsInput = undefined;
331345 return;
332346 }
——@@ -348,9 +362,7 @@
349363 errHtml = gM( 'mwe-upwiz-fileexists-replace-no-link', titleString );
350364 }
351365
352 - _this.$form.find( 'label[for=' + _this.titleId + ']' )
353 - .html( errHtml )
354 - .show();
 366+ $errorEl.html( errHtml ).show();
355367 },
356368
357369 /**

Comments

#Comment by Brion VIBBER (talk | contribs)   19:58, 9 May 2011

The hardcoded regex sets definitely will need to be replaced with using an API call, but it looks functional for now.

Adding todo tag.

Status & tagging log