Download requested pages, extract requested paragraphs, send them to AI

.eslintrc.json

@@ -1,7 +1,7 @@
 {
 	"root": true,
 	"extends": [
-		"wikimedia/client",
+		"wikimedia/client-es6",
 		"wikimedia/jquery",
 		"wikimedia/mediawiki"
 	],
@@ -11,6 +11,7 @@
 	"rules": {
 		"no-jquery/no-event-shorthand": "off",
 		"no-jquery/no-global-selector": "off",
-		"no-console": "off"
+		"no-console": "off",
+		"compat/compat": "off"
 	}
 }

Gruntfile.js

@@ -1,6 +1,6 @@
 /* eslint-env node */
 module.exports = function ( grunt ) {
-	var conf = grunt.file.readJSON( 'extension.json' );
+	const conf = grunt.file.readJSON( 'extension.json' );
 
 	grunt.loadNpmTasks( 'grunt-banana-checker' );
 	grunt.loadNpmTasks( 'grunt-eslint' );

extension.json

@@ -57,6 +57,9 @@
 			],
 			"messages": [
 				"askai-submit-failed"
+			],
+			"dependencies": [
+				"mediawiki.Title"
 			]
 		}
 	},

includes/SpecialAI.php

@@ -60,6 +60,9 @@ protected function getFormFields() {
 				'type' => 'text',
 				'label-message' => 'askai-field-prompt',
 				'required' => true
+			],
+			'Extract' => [
+				'type' => 'hidden'
 			]
 		];
 	}
@@ -77,7 +80,7 @@ public function onSubmit( array $data ) {
 			return Status::newFatal( 'askai-unknown-service' );
 		}
 
-		$response = $ai->query( $data['Prompt'], $data['Pages'] );
+		$response = $ai->query( $data['Prompt'], $data['Extract'] );
 
 		$this->getOutput()->disable();
 		echo Xml::element( 'div', [

modules/ext.askai.js

@@ -1,29 +1,106 @@
 /* Submits the form [[Special:AI]] and displays results without reloading the page. */
 
 $( function () {
-	var $form = $( '#mw-askai' ),
+	const $form = $( '#mw-askai' ),
 		$response = $form.find( '[name="wpResponse"]' ),
 		$pages = $form.find( '[name="wpPages"]' ),
 		$prompt = $form.find( '[name="wpPrompt"]' ),
 		token = $( '#wpEditToken' ).val(),
 		url = $form[ 0 ].action;
 
-	function onsubmit( ev ) {
-		ev.preventDefault();
+	function extractParagraphs() {
+		// List of pages isn't useful to the AI (it doesn't know what to do with it),
+		// we need to retrieve the text of paragraphs (e.g. [[Some page#p6-8]])
+		// and send this text to AI as a part of instructions (not the user-chosen Prompt).
+		const promises = $pages.val().split( '\n' ).map( function ( pageName ) {
+			let title;
+			try {
+				title = new mw.Title( pageName );
+			} catch ( error ) {
+				// Invalid title.
+				return [];
+			}
+
+			const fragment = title.fragment,
+				parNumbers = new Set();
+
+			if ( fragment && fragment.match( /^p[0-9\-,]+$/ ) ) {
+				// Anchor is the list of paragraphs, e.g. "p4", or "p6-8", or "p3,5,7".
+				fragment.slice( 1 ).split( ',' ).forEach( function ( pair ) {
+					const range = pair.split( '-' ),
+						start = range[ 0 ],
+						end = range.length > 1 ? range[ 1 ] : start;
+
+					for ( let idx = start; idx <= end; idx++ ) {
+						parNumbers.add( idx );
+					}
+				} );
+			}
+
+			const $d = $.Deferred();
+
+			$.get( title.getUrl() ).done( function ( html ) {
+				const $paragraphs = $( '<div>' ).append( html ).find( '.mw-parser-output > p' );
+
+				let extract;
+				if ( parNumbers.size === 0 ) {
+					// Use the entire page (no paragraph numbers were selected).
+					extract = $paragraphs.toArray();
+				} else {
+					extract = [];
+					Array.from( parNumbers ).sort().forEach( function ( idx ) {
+						const p = $paragraphs[ idx ];
+						if ( p ) {
+							extract.push( p );
+						}
+					} );
+				}
+
+				$d.resolve( {
+					title: title,
+					extract: extract.map( function ( p ) {
+						return p.innerText.trim();
+					} ).join( '\n\n' )
+				} );
+			} );
+
+			return $d.promise();
+		} );
+
+		// Accumulate the results into 1 string.
+		return Promise.all( promises ).then( function ( pageResults ) {
+			return pageResults.map( function ( ret, idx ) {
+				let text = '(Source #' + ( idx + 1 ) + ') [' +
+					ret.title.getPrefixedText();
+
+				if ( ret.title.fragment ) {
+					text += '#' + ret.title.fragment;
+				}
+
+				text += ']\n\n' + ret.extract;
+				return text;
+			} ).join( '\n\n' );
+		} );
+	}
 
+	function sendPrompt( extract ) {
 		$.post( url, {
-			wpPages: $pages.val(),
+			wpExtract: extract,
 			wpPrompt: $prompt.val(),
 			wpEditToken: token
 		} ).fail( function ( xhr ) {
 			$response.val( mw.msg( 'askai-submit-failed',
 				xhr.statusText + ' (' + url + ')'
 			) );
 		} ).done( function ( ret ) {
-			var responseText = $( '<div>' ).append( ret ).find( '#mw-askai-response' ).text();
-			$response.val( responseText );
+			$response.val( $( '<div>' ).append( ret ).find( '#mw-askai-response' ).text() );
 		} );
 	}
 
+	function onsubmit( ev ) {
+		ev.preventDefault();
+		extractParagraphs().then( sendPrompt );
+	}
+
 	$form.on( 'submit', onsubmit );
 }() );