GANReviewTool: TemplateFinder

GANReviewTool/modules/GANReviewWikicodeGenerator.js

@@ -1,4 +1,5 @@
 /* eslint-disable indent */
+import { TemplateFinder } from './TemplateFinder.js';
 
 export class GANReviewWikicodeGenerator {
 	getPassWikicodeForGANPage( reviewWikicode ) {
@@ -255,11 +256,12 @@ export class GANReviewWikicodeGenerator {
 	}
 
 	getFirstTemplateNameFromWikicode( wikicode ) {
-		const match = wikicode.match( /(?<=\{\{)[^|}]+/ );
-		if ( !match ) {
+		const templateFinder = new TemplateFinder( wikicode );
+		const template = templateFinder.firstTemplate();
+		if ( !template ) {
 			throw new Error( 'getFirstTemplateNameFromWikicode: No template found in Wikicode.' );
 		}
-		return match[ 0 ];
+		return TemplateFinder.removePrefix( template.name );
 	}
 
 	/**
@@ -350,7 +352,7 @@ export class GANReviewWikicodeGenerator {
 		const topicString = `\n|topic = ${ topic }`;
 
 		// https://en.wikipedia.org/wiki/Template:Article_history#How_to_use_in_practice
-		const existingStatus = this.firstTemplateGetParameterValue( talkWikicode, 'Artricle history', 'currentstatus' );
+		const existingStatus = this.firstTemplateGetParameterValue( talkWikicode, 'Article ?history', 'currentstatus' );
 		talkWikicode = this.firstTemplateDeleteParameter( talkWikicode, 'Article ?history', 'currentstatus' );
 		const currentStatusString = this.getArticleHistoryNewStatus( existingStatus, listedOrFailed );
 
@@ -393,84 +395,20 @@ export class GANReviewWikicodeGenerator {
 	}
 
 	firstTemplateInsertCode( wikicode, templateNameRegExNoDelimiters, codeToInsert ) {
-		// TODO: handle nested templates
-		const regex = new RegExp( `(\\{\\{${ templateNameRegExNoDelimiters }[^\\}]*)(\\}\\})`, 'i' );
-		return wikicode.replace( regex, `$1\n${ codeToInsert }\n$2` );
-	}
-
-	firstTemplateGetParameterValue( wikicode, template, parameter ) {
-		// TODO: rewrite to be more robust. currently using a simple algorithm that is prone to failure
-		// new algorithm:
-			// find start of template. use regex /i (ignore case)
-			// iterate using loops until end of template found
-				// handle <nowiki>
-				// handle triple {{{
-				// handle nested
-
-		const regex = new RegExp( `\\|\\s*${ parameter }\\s*=\\s*([^\\n\\|\\}]*)\\s*`, '' );
-		const result = wikicode.match( regex );
-		if ( wikicode.match( regex ) === null ) {
-			return null;
-		}
-		return result[ 1 ];
+		const templateFinder = new TemplateFinder( wikicode );
+		templateFinder.firstTemplateInsertCode( templateNameRegExNoDelimiters, codeToInsert );
+		return templateFinder.getWikitext();
 	}
 
-	/**
-	 * @param {RegExp} regex
-	 */
-	preg_position( regex, haystack ) {
-		const matches = [ ...haystack.matchAll( regex ) ];
-		const hasMatches = matches.length;
-		if ( hasMatches ) {
-			return matches[ 0 ].index;
-		}
-		return false;
-	}
-
-	findEndOfTemplate( wikicode, templateStartPosition ) {
-		// TODO: handle triple braces, handle <nowiki> tags
-		let nesting = 0;
-		let templateEndPosition = -1;
-		// +1 to skip the first {{, will throw off our nesting count
-		for ( let i = templateStartPosition + 1; i < wikicode.length; i++ ) {
-			const nextTwoChars = wikicode.slice( i, i + 2 );
-			if ( nextTwoChars === '{{' ) {
-				nesting++;
-				continue;
-			} else if ( nextTwoChars === '}}' ) {
-				if ( nesting > 0 ) {
-					nesting--;
-					continue;
-				} else {
-					templateEndPosition = i + 2;
-					break;
-				}
-			}
-		}
-		return templateEndPosition;
+	firstTemplateGetParameterValue( wikicode, templateRegEx, parameter ) {
+		const templateFinder = new TemplateFinder( wikicode );
+		return templateFinder.firstTemplateGetParameterValue( templateRegEx, parameter );
 	}
 
 	firstTemplateDeleteParameter( wikicode, templateRegEx, parameter ) {
-		// templateStartPosition
-		const regex = new RegExp( '{{' + templateRegEx, 'gi' );
-		const templateStartPosition = this.preg_position( regex, wikicode );
-
-		// templateEndPosition
-		const templateEndPosition = this.findEndOfTemplate( wikicode, templateStartPosition );
-
-		// slice
-		const firstPiece = wikicode.slice( 0, templateStartPosition );
-		let secondPiece = wikicode.slice( templateStartPosition, templateEndPosition );
-		const thirdPiece = wikicode.slice( templateEndPosition );
-
-		// replace only inside the slice
-		const regex2 = new RegExp( `\\|\\s*${ parameter }\\s*=\\s*([^\\n\\|\\}]*)\\s*`, '' );
-		secondPiece = secondPiece.replace( regex2, '' );
-
-		// glue back together
-		wikicode = firstPiece + secondPiece + thirdPiece;
-
-		return wikicode;
+		const templateFinder = new TemplateFinder( wikicode );
+		templateFinder.firstTemplateDeleteParameter( templateRegEx, parameter );
+		return templateFinder.getWikitext();
 	}
 
 	removeFormattingThatInterferesWithSort( str ) {

GANReviewTool/modules/GARCloserWikicodeGenerator.js

@@ -1,3 +1,5 @@
+import { TemplateFinder } from './TemplateFinder.js';
+
 export class GARCloserWikicodeGenerator {
 	processKeepForGARPage( garPageWikicode, message, isCommunityAssessment ) {
 		return this.processGARPage( garPageWikicode, message, isCommunityAssessment, 'Kept.', 'green' );
@@ -583,15 +585,15 @@ __TOC__`;
 			throw new Error( 'InvalidArgumentException' );
 		}
 
-		const topic = this.firstTemplateGetParameterValue( wikicode, 'Artricle history', 'topic' );
+		const topic = this.firstTemplateGetParameterValue( wikicode, 'Article ?history', 'topic' );
 		let topicString = '';
 		if ( !topic ) {
 			topicString = `\n|topic = ${ topic }`;
 		}
 
 		// https://en.wikipedia.org/wiki/Template:Article_history#How_to_use_in_practice
-		const existingStatus = this.firstTemplateGetParameterValue( wikicode, 'Artricle history', 'currentstatus' );
-		wikicode = this.firstTemplateDeleteParameter( wikicode, 'Article history', 'currentstatus' );
+		const existingStatus = this.firstTemplateGetParameterValue( wikicode, 'Article ?history', 'currentstatus' );
+		wikicode = this.firstTemplateDeleteParameter( wikicode, 'Article ?history', 'currentstatus' );
 		const currentStatusString = this.getArticleHistoryNewStatus( existingStatus, keepOrDelist );
 
 		const result = this.getKeepOrDelistPastTense( keepOrDelist );
@@ -634,21 +636,9 @@ __TOC__`;
 		}
 	}
 
-	firstTemplateGetParameterValue( wikicode, template, parameter ) {
-		// TODO: rewrite to be more robust. currently using a simple algorithm that is prone to failure
-		// new algorithm:
-			// find start of template. use regex /i (ignore case)
-			// iterate using loops until end of template found
-				// handle <nowiki>
-				// handle triple {{{
-				// handle nested
-
-		const regex = new RegExp( `\\|\\s*${ parameter }\\s*=\\s*([^\\n\\|\\}]*)\\s*`, '' );
-		const result = wikicode.match( regex );
-		if ( wikicode.match( regex ) === null ) {
-			return null;
-		}
-		return result[ 1 ];
+	firstTemplateGetParameterValue( wikicode, templateRegEx, parameter ) {
+		const templateFinder = new TemplateFinder( wikicode );
+		return templateFinder.firstTemplateGetParameterValue( templateRegEx, parameter );
 	}
 
 	getArticleHistoryNewStatus( existingStatus, keepOrDelist ) {
@@ -663,63 +653,18 @@ __TOC__`;
 	 * @param {Array} templateNameArrayCaseInsensitive
 	 */
 	firstTemplateInsertCode( wikicode, templateNameArrayCaseInsensitive, codeToInsert ) {
-		for ( const templateName of templateNameArrayCaseInsensitive ) {
-			const strPosOfEndOfFirstTemplate = this.getStrPosOfEndOfFirstTemplateFound( wikicode, templateName );
-			if ( strPosOfEndOfFirstTemplate !== null ) {
-				const insertPosition = strPosOfEndOfFirstTemplate - 2; // 2 characters from the end, right before }}
-				const result = this.insertStringIntoStringAtPosition( wikicode, `\n${ codeToInsert }\n`, insertPosition );
-				return result;
-			}
-		}
-	}
-
-	/**
-	 * CC BY-SA 4.0, jAndy, https://stackoverflow.com/a/4364902/3480193
-	 */
-	insertStringIntoStringAtPosition( bigString, insertString, position ) {
-		return [
-			bigString.slice( 0, position ),
-			insertString,
-			bigString.slice( position )
-		].join( '' );
-	}
-
-	/**
-	 * Grabs string position of the END of first {{template}} contained in wikicode. Case insensitive. Returns null if no template found. Handles nested templates.
-	 *
-	 * @return {number|null}
-	 */
-	getStrPosOfEndOfFirstTemplateFound( wikicode, templateName ) {
-		const starting_position = wikicode.toLowerCase().indexOf( '{{' + templateName.toLowerCase() );
-		if ( starting_position === -1 ) {
-			return null;
-		}
-		let counter = 0;
-		const length = wikicode.length;
-		for ( let i = starting_position + 2; i < length; i++ ) {
-			const next_two = wikicode.substr( i, 2 );
-			if ( next_two == '{{' ) {
-				counter++;
-				continue;
-			} else if ( next_two == '}}' ) {
-				if ( counter == 0 ) {
-					return i + 2; // +2 to account for next_two being }} (2 characters)
-				} else {
-					counter--;
-					continue;
-				}
-			}
-		}
-		return null;
+		const templateFinder = new TemplateFinder( wikicode );
+		templateFinder.firstTemplateInsertCode( templateNameArrayCaseInsensitive, codeToInsert );
+		return templateFinder.getWikitext();
 	}
 
 	removeGAStatusFromWikiprojectBanners( wikicode ) {
 		return wikicode.replace( /(\|\s*class\s*=\s*)([^}|\s]*)/gi, '$1' );
 	}
 
 	firstTemplateDeleteParameter( wikicode, template, parameter ) {
-		// TODO: rewrite to be more robust. currently using a simple algorithm that is prone to failure
-		const regex = new RegExp( `\\|\\s*${ parameter }\\s*=\\s*([^\\n\\|\\}]*)\\s*`, '' );
-		return wikicode.replace( regex, '' );
+		const templateFinder = new TemplateFinder( wikicode );
+		templateFinder.firstTemplateDeleteParameter( template, parameter );
+		return templateFinder.getWikitext();
 	}
 }

GANReviewTool/modules/Parser.js

@@ -0,0 +1,36 @@
+import Parser from 'wikiparser-template';
+
+Parser.config = {
+	ext: [
+		'pre',
+		'nowiki',
+		'gallery',
+		'indicator',
+		'langconvert',
+		'graph',
+		'timeline',
+		'hiero',
+		'charinsert',
+		'ref',
+		'references',
+		'inputbox',
+		'imagemap',
+		'source',
+		'syntaxhighlight',
+		'poem',
+		'categorytree',
+		'section',
+		'score',
+		'templatestyles',
+		'templatedata',
+		'math',
+		'ce',
+		'chem',
+		'maplink',
+		'mapframe',
+		'page-collection',
+		'phonos'
+	]
+};
+
+export default Parser;

GANReviewTool/modules/TemplateFinder.js

@@ -1,10 +1,57 @@
 // TODO: A couple of recent bugs will require a lexer or template parser type class to solve.
+import Parser from './Parser.js';
 
-class TemplateFinder {
-	// getTemplateList()
-	// appendParameter()
-	// addWikitextAfterTemplate()
-	// getWikitext()
+export class TemplateFinder {
+	constructor( wikicode ) {
+		this.root = Parser.parse( wikicode, false, 2 );
+	}
 
-	// appends/adds need to shift all the position variables by the length of the append/add
-}
+	static removePrefix( templateName ) {
+		return templateName.replace( /^Template:/, '' );
+	}
+
+	getWikitext() {
+		return String( this.root );
+	}
+
+	firstTemplate( templateNameRegExOrArrayCaseInsensitive ) {
+		let filter;
+		if ( !templateNameRegExOrArrayCaseInsensitive ) {
+			filter = () => true;
+		} else if ( Array.isArray( templateNameRegExOrArrayCaseInsensitive ) ) {
+			const templateNameArray = templateNameRegExOrArrayCaseInsensitive
+				.map( ( name ) => name.toLowerCase().replace( /\s/g, '_' ) );
+			filter = ( { name } ) => templateNameArray.includes( TemplateFinder.removePrefix( name ).toLowerCase() );
+		} else {
+			const regEx = new RegExp( `^Template:${ templateNameRegExOrArrayCaseInsensitive }$`, 'i' );
+			filter = ( { name } ) => regEx.test( name.replace( /_/g, ' ' ) );
+		}
+		return this.root.querySelectorAll( 'template' ).find( filter );
+	}
+
+	firstTemplateInsertCode( templateNameRegExOrArrayCaseInsensitive, codeToInsert ) {
+		const template = this.firstTemplate( templateNameRegExOrArrayCaseInsensitive );
+		if ( template ) {
+			template.append( `${ codeToInsert.replace( /^\|/, '' ) }\n` );
+		}
+	}
+
+	firstTemplateGetParameterValue( templateNameRegExOrArrayCaseInsensitive, parameter ) {
+		const template = this.firstTemplate( templateNameRegExOrArrayCaseInsensitive );
+		if ( !template ) {
+			return null;
+		}
+		return template.getValue( parameter ) ?? null;
+	}
+
+	firstTemplateDeleteParameter( templateNameRegExOrArrayCaseInsensitive, parameter ) {
+		const template = this.firstTemplate( templateNameRegExOrArrayCaseInsensitive );
+		if ( template ) {
+			for ( const token of template.getAllArgs() ) {
+				if ( token.name.toLowerCase() === parameter ) {
+					token.remove();
+				}
+			}
+		}
+	}
+}

GANReviewTool/package.json

@@ -1,7 +1,8 @@
 {
 	"dependencies": {
 		"jest": "^28.1.0",
-		"types-mediawiki": "github:wikimedia-gadgets/types-mediawiki"
+		"types-mediawiki": "github:wikimedia-gadgets/types-mediawiki",
+		"wikiparser-template": "^1.24.1"
 	},
 	"scripts": {
 		"test": "jest"