Wikisource

Utente:Alex brollo/OCR.js

< Utente:Alex brollo

Nota: dopo aver pubblicato, potrebbe essere necessario pulire la cache del proprio browser per vedere i cambiamenti.

  • Firefox / Safari: tieni premuto il tasto delle maiuscole Shift e fai clic su Ricarica, oppure premi Ctrl-F5 o Ctrl-R (⌘-R su Mac)
  • Google Chrome: premi Ctrl-Shift-R (⌘-Shift-R su un Mac)
  • Internet Explorer / Edge: tieni premuto il tasto Ctrl e fai clic su Aggiorna, oppure premi Ctrl-F5
  • Opera: premi Ctrl-F5.
/*local copy of wikisource.org/MediaWiki:OCR.js 8/11/2014
See discussion into original Talk page

*/

/*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, 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:.
 */

// alex change to use test routines 
importScript("User:Alex brollo/hOCRlab.js");

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);
}
/* Original Phe callback function
function hocr_callback(data) {
	if (data.error) {
		// 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(/[ ]*\n[ ]*/g, '\n')
/*				.replace(/\n\n\n\n/g, '@_@_@_@_@_@')
				.replace(/\n\n/g, '\n')
				.replace(/@_@_@_@_@_@/g, '\n\n')
				.replace(/\n\n\n/g, '\n\n');
			tb.value = $.trim(text);
		}
	}
 
	disable_input(false);
}

*/

function ocr_callback(data) {
	console.log("chiamata a ocr_callback");
	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);
}
// local modified callback function
function hocr_callback(data) {
	console.log("chiamata a hocr_callback");
	if (data.error) {
		console.log(JSON.stringify("Errore hOCR: "+data.error));
		// 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;
                        // alex change: linking hOCR to a specific page, needed for persistent data
                        localStorage.ws_hOCR_pageName=mw.config.get("wgTitle");
			var text=analisiPagina(); 	// presently into User:Alex brollo/hOCRlab.js
			console.log("chiamata a analisiPagina");
										// it builds too a mw.pagina object and saves it into localStorage.ws_hOCR_page
			if (mw.config.get("wgArticleId")===0 || leggiBox()==="") 	tb.value = $.trim(text); 	// posts text into edit box only if the page is new
																				// else it only saves data into localStorage
			
			
			if (mw.ocr_show !== undefined)	{
				console.log("chiamata a mw.ocr_show()")
				mw.ocr_show(); 
				mw.ocr_showAllLines();
				
			}
			
		}
	}

	disable_input(false);
	$("#p-tb ul").append($('<li id="t-crop"><a href="javascript:mediaWiki.textSelect()">textSelect</a></li>'));
}

function hocr_callback_view(data) {
	if (data.error) {
		
		return;
	} else {
		
        localStorage.ws_hOCR = data.text;
        // alex change: linking hOCR to a specific page, needed for persistent data
        localStorage.ws_hOCR_pageName=mw.config.get("wgTitle");
		var text=analisiPagina(); 	// presently into User:Alex brollo/hOCRlab.js
									// it builds too a mw.pagina object and saves it into localStorage.ws_hOCR_page
		}
}

function do_hocr() {
	if (mw.config.get("wgAction")!=="view") disable_input(true);

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

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

function do_hocr_view() {
	var request_url = '//tools.wmflabs.org/phetools/hocr_cgi.py?cmd=hocr&book='
		+ encodeURIComponent(mw.config.get('wgTitle')) + '&lang=' + lang + '&user=' + mw.config.get('wgUserName');
	$.getJSON(request_url).done(hocr_callback_view);	
}


function do_hocr_page(page) {
	disable_input(true);
	page=page.replace(/Pagina:|Page:/,"");
	var request_url = '//tools.wmflabs.org/phetools/hocr_cgi.py?cmd=hocr&book='
		+ encodeURIComponent(page) + '&lang=' + lang + '&user=' + mw.config.get('wgUserName');

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

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 = "//tools.wmflabs.org/phetools/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' );
}

function addButtonToWikiEditorToolbar( b ){
	var tools = {};
	tools[ b.imageId ] = {
		label: b.speedTip,
		type: 'button',
		icon: b.imageFile,
		action: {
			type: 'callback',
			execute: b.onClick
		}
	};
	$( '#wpTextbox1' ).wikiEditor( 'addToToolbar', {
		section: 'advanced',
		group: 'format',
		tools: tools
	} );
	$( '[rel="' + b.imageId + '"]' ).width( 42 );
}

function addButtonToClassicToolbar( b ){
	mw.toolbar.addButton( {
		imageFile: b.imageFile,
		speedTip: b.speedTip,
		imageId: b.imageId
	} );
	$( '#' + b.imageId ).off( 'click' ).click( function () {
		b.onClick();
		return false;
	} ).width( 46 );
}

function customizeToolbar()
{
	var modules, add, img;
	// This can be the string "0" if the user disabled the preference ([[bugzilla:52542#c3]])
	if( mw.user.options.get( 'usebetatoolbar' ) == 1 ){
		modules = [ 'ext.wikiEditor', 'ext.proofreadpage.page.edit' ];
		img = '//upload.wikimedia.org/wikipedia/commons/c/c9/Toolbaricon_OCR.png';
		add = addButtonToWikiEditorToolbar;
	} else if ( mw.user.options.get( 'showtoolbar' ) == 1 ){
		modules = 'mediawiki.toolbar';
		img = '//upload.wikimedia.org/wikipedia/commons/e/e0/Button_ocr.png';
		add = addButtonToClassicToolbar;
	} else {
		return;
	}
	$.when(
		mw.loader.using( modules ),
		$.ready
	).then( function(){
		if( mw.config.get( 'wgContentLanguage' ) === 'de' ){
			add( {
				imageFile: img,
				speedTip: 'Normale OCR',
				imageId: 'wsOcr1',
				onClick: do_hocr
			} );
			add( {
				imageFile: '//upload.wikimedia.org/wikipedia/commons/a/af/Button_Fractur_OCR.png',
				speedTip: 'Fraktur OCR',
				imageId: 'wsOcr2',
				onClick: fraktur_ocr
			} );
		} else {
			add( {
				imageFile: img,
				speedTip: 'Get the text by OCR',
				imageId: 'wsOcr1',
				onClick: do_hocr
			} );
		}
	} );
}

if ( mw.config.get( 'wgCanonicalNamespace' ) === 'Page' &&
	$.inArray( mw.config.get( 'wgAction' ), [ 'edit', 'submit' ] ) !== -1 &&
	!self.proofreadpage_disable_ocr
) {
	mw.loader.using( 'user.options' ).done( customizeToolbar );
}
Estratto da "https://it.wikisource.org/w/index.php?title=Utente:Alex_brollo/OCR.js&oldid=2252695"