MediaWiki:Gadget-ocr.js

/*jshint boss:true*/
/*global $, mw*/

/*
 * Query an ocr for a given Page:, first try to get the hocr text layer as it's available
 * for most book, fast and of a better quality. If it fails (or returns empty string for
 * the page), try the older and slower ocr method. hocr fail around 1/5000 books. ocr should
 * never fails as it use the image visible on the Page:.
 */

var lang = mw.config.get( 'wgContentLanguage' );

function disable_input(set)
{
	if (set) {
		$(document).keyup(function(e) {
			if (e.which == 27) { disable_input(false); }
		});
	}

	set ? $('#wsOcr1').off('click') : $('#wsOcr1').on('click', do_hocr);
	set ? $('#wsOcr2').off('click') : $('#wsOcr1').on('click', fraktur_ocr);

	$('#wpTextbox1').prop('disabled', set);
}

function ocr_callback(data) {
	if (data.error) {
		alert(data.text);
	} else {
		// Checking if tb is disabled is required with chrome as ESC doesn't kill
		// the query.
		var tb = document.getElementById("wpTextbox1");
		if (tb.disabled)
			tb.value = data.text;
	}

	disable_input(false);
}

function hocr_callback(data) {
	if ((data.error) || (data.text.search("An error occurred during ocr processing")===0) || ($.trim($(data.text).text()) === '') ) {
		// Fallback to the slow way.
		disable_input(false);
		do_ocr();
		return;
	} else {
		// Checking if tb is disabled is required with chrome as ESC doesn't kill
		// the query.
		var tb = document.getElementById("wpTextbox1");
		if (tb.disabled) {
                        localStorage.ws_hOCR = data.text;

			var text = $(data.text).text();
			// Ugly as hell.
			text = text.replace(/^ +/mg, '')
				.replace(/\n{4,}/g, '@_@_@_@')
				.replace(/\n{2,}/g, '____SPACE____')
				.replace(/\n/g, ' ')
				.replace(/____SPACE____/g, '\n')
				.replace(/@_@_@_@/g, '\n\n');
			tb.value = $.trim(text);
		}
	}

	disable_input(false);
}

function do_hocr() {
	disable_input(true);

	var request_url = '//phetools.toolforge.org/hocr_cgi.py?cmd=hocr&book='
		+ encodeURIComponent(mw.config.get('wgTitle')) + '&lang=' + lang + '&user=' + mw.config.get('wgUserName');

	$.getJSON(request_url)
		.done(hocr_callback)
		.fail(do_ocr);
}

function do_ocr() {
	if ($( '.prp-page-image img' ).length) {
		disable_input(true);

		// server side can't use protocol relative url, request it as https:
		var url_image = 'https:' + $( '.prp-page-image img' ).attr('src');

		var request_url = "//phetools.toolforge.org/ocr.php?cmd=ocr&url="+url_image+"&lang="+lang+"&user="+mw.config.get('wgUserName');

		$.getJSON( request_url ).done( ocr_callback );
	}
}

function fraktur_ocr()
{
	lang = 'de-f';
	// For fraktur we need to use the slow way, all hocr for 'de'
	// are done with non-fraktur.
	do_ocr();
	lang = mw.config.get( 'wgContentLanguage' );
}

if ( mw.config.get( 'wgPageContentModel' ) === 'proofread-page' &&
	$.inArray( mw.config.get( 'wgAction' ), [ 'edit', 'submit' ] ) !== -1 &&
	!self.proofreadpage_disable_ocr
) {
	$.when(
			mw.loader.using( [ 'ext.gadget.lib-toolbar' ] )
			).then(function() {
				toolbarGadget.addButton( {
				'oldIcon': "//upload.wikimedia.org/wikipedia/commons/e/e0/Button_ocr.png",
				'newIcon': "//upload.wikimedia.org/wikipedia/commons/c/c9/Toolbaricon_OCR.png",
				'title': "Pobierz tekst przez OCR",
				'alt': "Pobierz tekst przez OCR",
				'id': "wsOcr1",
				'add_style': "width: 42px;",
				'section': "main",
				'group': "other-ocr",
				'onclick': do_hocr
		});
	});
}