GANReviewTool: Convert the rest to TemplateFinder (#278)

* GANReviewTool: use TemplateFinder for GARCloserWikicodeGenerator

* fix CI

* fix CI

* diagnose CI error

* diagnose CI error

* fix CI

* add docblock

* add tests

---------

Co-authored-by: NovemLinguae <novemlinguae@gmail.com>
Co-authored-by: NovemLinguae <79697282+NovemLinguae@users.noreply.github.com>

Author: bhsd-harry

GANReviewTool/modules/GARCloserWikicodeGenerator.js

@@ -97,16 +97,27 @@ __TOC__`;
 
 	processDelistForArticle( wikicode ) {
 		const gaTemplateNames = [ 'ga icon', 'ga article', 'good article' ];
+		const templateFinder = new TemplateFinder( wikicode );
 		for ( const templateName of gaTemplateNames ) {
+			const template = templateFinder.firstTemplate( templateName );
+			if ( !template ) {
+				continue;
+			}
+			const { previousSibling, nextSibling } = template;
+			template.remove();
 			// handle lots of line breaks: \n\n{{templateName}}\n\n -> \n\n
-			let regex = new RegExp( '\\n\\n\\{\\{' + templateName + '\\}\\}\\n\\n', 'i' );
-			wikicode = wikicode.replace( regex, '\n\n' );
-
-			// handle normal: {{templateName}}\n -> '', {{templateName}} -> ''
-			regex = new RegExp( '\\{\\{' + templateName + '\\}\\}\\n?', 'i' );
-			wikicode = wikicode.replace( regex, '' );
+			if (
+				previousSibling && previousSibling.type === 'text' && previousSibling.data.endsWith( '\n\n' ) &&
+				nextSibling && nextSibling.type === 'text' && nextSibling.data.startsWith( '\n\n' )
+			) {
+				nextSibling.deleteData( 0, 2 );
+
+				// handle normal: {{templateName}}\n -> '', {{templateName}} -> ''
+			} else if ( nextSibling && nextSibling.type === 'text' && nextSibling.data.startsWith( '\n' ) ) {
+				nextSibling.deleteData( 0, 1 );
+			}
 		}
-		return wikicode;
+		return templateFinder.getWikitext();
 	}
 
 	processDelistForGAList( wikicode, articleToRemove ) {
@@ -288,8 +299,14 @@ __TOC__`;
 			'weapons and buildings': 'Wikipedia:Good articles/Warfare',
 			weapons: 'Wikipedia:Good articles/Warfare'
 		};
-		let topic = wikicode.match( /(?:\{\{Article ?history|\{\{GA\s*(?=\|)).*?\|\s*(?:sub)?topic\s*=\s*([^|}\n]+)/is )[ 1 ];
-		topic = topic.toLowerCase().trim();
+		const templateFinder = new TemplateFinder( wikicode );
+		const templates = templateFinder.getTemplates( [ 'Article history', 'Articlehistory', 'GA' ] );
+		const template = templates.find( ( t ) => t.getArgs( 'topic' ).size || t.getArgs( 'subtopic' ).size );
+		let topic = template.getValue( 'topic' );
+		if ( topic === undefined ) {
+			topic = template.getValue( 'subtopic' );
+		}
+		topic = topic.toLowerCase();
 		const gaListTitle = dictionary[ topic ];
 		// throw the error a little later rather than now. that way it doesn't interrupt modifying the article talk page.
 		return gaListTitle;
@@ -374,16 +391,9 @@ __TOC__`;
 	}
 
 	removeTemplate( templateName, wikicode ) {
-		const regex = new RegExp( `\\{\\{${ this.regExEscape( templateName ) }[^\\}]*\\}\\}\\n?`, 'i' );
-		return wikicode.replace( regex, '' );
-	}
-
-	regexGetFirstMatchString( regex, haystack ) {
-		const matches = haystack.match( regex );
-		if ( matches !== null && matches[ 1 ] !== undefined ) {
-			return matches[ 1 ];
-		}
-		return null;
+		const templateFinder = new TemplateFinder( wikicode );
+		templateFinder.deleteTemplate( templateName );
+		return templateFinder.getWikitext();
 	}
 
 	/**
@@ -392,15 +402,15 @@ __TOC__`;
 	convertGATemplateToArticleHistoryIfPresent( talkPageTitle, wikicode ) {
 		const templateFinder = new TemplateFinder( wikicode );
 		const hasArticleHistory = templateFinder.hasTemplate( 'Article ?history' );
-		const gaTemplateWikicode = this.regexGetFirstMatchString( /(\{\{GA[^}]*\}\})/i, wikicode );
-		if ( !hasArticleHistory && gaTemplateWikicode ) {
+		const gaTemplate = templateFinder.firstTemplate( 'GA' );
+		if ( !hasArticleHistory && gaTemplate ) {
 			// delete {{ga}} template
 			templateFinder.deleteTemplate( 'GA' );
 			wikicode = templateFinder.getWikitext().trim();
 
 			// parse its parameters
 			// example: |21:00, 12 March 2017 (UTC)|topic=Sports and recreation|page=1|oldid=769997774
-			const parameters = this.getParametersFromTemplateWikicode( gaTemplateWikicode );
+			const parameters = this.getParametersFromTemplateWikicode( gaTemplate );
 
 			// if no page specified, assume page is 1. so then the good article review link will be parsed as /GA1
 			const noPageSpecified = parameters.page === undefined;
@@ -547,28 +557,18 @@ __TOC__`;
 	/**
 	 * @return {Object} Parameters, with keys being equivalent to the template parameter names. Unnamed parameters will be 1, 2, 3, etc.
 	 */
-	getParametersFromTemplateWikicode( wikicodeOfSingleTemplate ) {
-		wikicodeOfSingleTemplate = wikicodeOfSingleTemplate.slice( 2, -2 ); // remove {{ and }}
-		// TODO: explode without exploding | inside of inner templates
-		const strings = wikicodeOfSingleTemplate.split( '|' );
+	getParametersFromTemplateWikicode( template ) {
+		if ( typeof template === 'string' ) {
+			const templateFinder = new TemplateFinder( template );
+			template = templateFinder.firstTemplate();
+		}
+		if ( template.type !== 'template' ) {
+			throw new Error( `InvalidArgumentException: ${ template.type }
+${ template }` );
+		}
 		const parameters = {};
-		let unnamedParameterCount = 1;
-		let i = 0;
-		for ( const string of strings ) {
-			i++;
-			if ( i == 1 ) {
-				continue; // skip the template name, this is not a parameter
-			}
-			const hasEquals = string.indexOf( '=' );
-			if ( hasEquals === -1 ) {
-				parameters[ unnamedParameterCount ] = string;
-				unnamedParameterCount++;
-			} else {
-				const matches = string.match( /^([^=]*)=(.*)/s ); // isolate param name and param value by looking for first equals sign
-				const paramName = matches[ 1 ].trim().toLowerCase();
-				const paramValue = matches[ 2 ].trim();
-				parameters[ paramName ] = paramValue;
-			}
+		for ( const parameter of template.getAllArgs() ) {
+			parameters[ parameter.name.toLowerCase() ] = parameter.getValue();
 		}
 		return parameters;
 	}

GANReviewTool/modules/MassGARWikicodeGenerator.js

@@ -1,3 +1,5 @@
+import { TemplateFinder } from './TemplateFinder.js';
+
 export class MassGARWikicodeGenerator {
 	hasGoodArticleTemplate( mainArticleWikicode ) {
 		const gaTemplateNames = [ 'ga icon', 'ga article', 'good article' ];
@@ -12,12 +14,14 @@ export class MassGARWikicodeGenerator {
 		}
 
 		// Check for {{Article history|currentstatus=GA}}
-		// TODO: currently just checks for |currentstatus=GA anywhere on the page. Could improve this algorithm if there end up being false positives.
-		const matches = talkPageWikicode.match( /\|\s*currentstatus\s*=\s*GA\b/i );
-		if ( matches ) {
-			return true;
+		const templateFinder = new TemplateFinder( talkPageWikicode );
+		const aliases = [ 'Articlehistory', 'Article milestones', 'Articlemilestones', 'Article History', 'ArticleHistory' ];
+		const articleHistoryTemplate = templateFinder.firstTemplate( aliases );
+		if ( !articleHistoryTemplate ) {
+			return false;
 		}
-		return false;
+		const value = articleHistoryTemplate.getValue( 'currentstatus' );
+		return Boolean( value ) && value.toUpperCase() === 'GA';
 	}
 
 	hasOpenGAR( talkPageWikicode ) {
@@ -30,21 +34,7 @@ export class MassGARWikicodeGenerator {
 	 * @param {Array} listOfTemplates Case insensitive.
 	 */
 	_wikicodeHasTemplate( wikicode, listOfTemplates ) {
-		const stringForRegEx = listOfTemplates
-			.map( ( v ) => this._regExEscape( v ) )
-			.join( '|' );
-		const regex = new RegExp( `{{(?:${ stringForRegEx })\\b`, 'i' );
-		const matches = wikicode.match( regex );
-		if ( matches ) {
-			return true;
-		}
-		return false;
-	}
-
-	/**
-	 * @copyright coolaj86, CC BY-SA 4.0 https://stackoverflow.com/a/6969486/3480193
-	 */
-	_regExEscape( string ) {
-		return string.replace( /[.*+?^${}()|[\]\\]/g, '\\$&' ); // $& means the whole matched string
+		const templateFinder = new TemplateFinder( wikicode );
+		return templateFinder.hasTemplate( listOfTemplates );
 	}
 }

GANReviewTool/modules/TemplateFinder.js

@@ -69,10 +69,18 @@ export class TemplateFinder {
 		}
 	}
 
-	getTemplates( templateNameCaseInsensitive ) {
-		const templateName = `template:${ templateNameCaseInsensitive.toLowerCase().replace( / /g, '_' ) }`;
+	/**
+	 * @param {string|Array} templateNamesCaseInsensitive
+	 */
+	getTemplates( templateNamesCaseInsensitive ) {
+		if ( typeof templateNamesCaseInsensitive === 'string' ) {
+			templateNamesCaseInsensitive = [ templateNamesCaseInsensitive ];
+		}
+		const templateNames = templateNamesCaseInsensitive.map(
+			( t ) => `template:${ t.toLowerCase().replace( / /g, '_' ) }`
+		);
 		return this.wikiPage.querySelectorAll( 'template' )
-			.filter( ( { name } ) => name.toLowerCase() === templateName );
+			.filter( ( { name } ) => templateNames.includes( name.toLowerCase() ) );
 	}
 
 	deleteTemplate( templateNameRegExOrArrayCaseInsensitive ) {

GANReviewTool/tests/GARCloserWikicodeGenerator.test.js

@@ -746,7 +746,7 @@ describe( 'getGAListTitleFromTalkPageWikicode(wikicode)', () => {
 	} );
 
 	it( 'Should be case insensitive', () => {
-		const wikicode = '{{aRTiClE HiStOrY|ToPiC=SpOrTs}}';
+		const wikicode = '{{aRTiClE HiStOrY|topic=SpOrTs}}';
 		const output = 'Wikipedia:Good articles/Sports and recreation';
 		expect( wg.getGAListTitleFromTalkPageWikicode( wikicode ) ).toBe( output );
 	} );
@@ -1459,36 +1459,6 @@ describe( 'deleteMiddleOfString(string, deleteStartPosition, deleteEndPosition)'
 	} );
 } );
 
-describe( 'regexGetFirstMatchString(regex, haystack)', () => {
-	test( 'match', () => {
-		const regex = /hello ([^ ]+)/;
-		const haystack = 'hello test goodbye';
-		const result = wg.regexGetFirstMatchString( regex, haystack );
-		expect( result ).toBe( 'test' );
-	} );
-
-	test( 'no match', () => {
-		const regex = /hello (bob)/;
-		const haystack = 'hello test goodbye';
-		const result = wg.regexGetFirstMatchString( regex, haystack );
-		expect( result ).toBe( null );
-	} );
-
-	test( 'no capture group, no match', () => {
-		const regex = /hello bob/;
-		const haystack = 'hello test goodbye';
-		const result = wg.regexGetFirstMatchString( regex, haystack );
-		expect( result ).toBe( null );
-	} );
-
-	test( 'no capture group, yes match', () => {
-		const regex = /hello bob/;
-		const haystack = 'hello bob';
-		const result = wg.regexGetFirstMatchString( regex, haystack );
-		expect( result ).toBe( null );
-	} );
-} );
-
 describe( 'convertGATemplateToArticleHistoryIfPresent(talkPageTitle, wikicode)', () => {
 	it( 'should default to page=1 when no page parameter', () => {
 		const talkPageTitle = 'Talk:Test';
@@ -1895,4 +1865,11 @@ describe( 'removeTemplate( templateName, wikicode )', () => {
 		const expected = `Test `;
 		expect( wg.removeTemplate( templateName, wikicode ) ).toBe( expected );
 	} );
+
+	it( '1 template with params + nested template', () => {
+		const templateName = 'GA';
+		const wikicode = `Test {{GA|date={{formatdate|2025-12-20}}}}`;
+		const expected = `Test `;
+		expect( wg.removeTemplate( templateName, wikicode ) ).toBe( expected );
+	} );
 } );

GANReviewTool/tests/MassGARWikicodeGenerator.test.js

@@ -276,6 +276,13 @@ describe( '_wikicodeHasTemplate( wikicode, listOfTemplates )', () => {
 		expect( wg._wikicodeHasTemplate( wikicode, listOfTemplates ) ).toBe( expected );
 	} );
 
+	it( '1 template in haystack, 0 templates to search for', () => {
+		const wikicode = `Test {{GA}}`;
+		const listOfTemplates = [];
+		const expected = false;
+		expect( wg._wikicodeHasTemplate( wikicode, listOfTemplates ) ).toBe( expected );
+	} );
+
 	it( '1 template in haystack, 1 template to search for', () => {
 		const wikicode = `Test {{GA}}`;
 		const listOfTemplates = [ 'GA' ];