Utilisateur:Seudo/getOCR.js

/* Récupéré de Utilisateur:Phe/OCR.js pour faire quelques tests */
/* Le problème de ce script est qu'il tente de charger un script situé sur un autre site
 (tools.wmflabs.org/phetools), ce qui est apparemment bloqué par les paramètres de sécurité
 de la page. Donc il échoue et applique l’OCR (assez mauvais) au lieu de récupérer la couche
 texte existante. Mais de toute manière, en appelant directement tools.wmflabs.org/phetools, j'ai
 l’impression qu’il ne récupère pas non plus la couche texte (qu’on peut obtenir avec le script
 djvutxt dans DjVuLibre). */

/*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:.
 */

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) {
        // 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) {
           var text = $(data.text).text();
           text = text.replace(/[ ]*\n[ ]*/g, '\n');
           text = text.replace(/\n\n\n\n/g, '@_@_@_@_@_@');
           text = text.replace(/\n\n/g, '\n');
           text = text.replace(/@_@_@_@_@_@/g, '\n\n');
           tb.value = $.trim(text);
       }
   }

   disable_input(false);
}

function do_hocr() {
    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');
console.log("Requête : " + request_url);

     $.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 addOCRButton2(id,comment,source,onclick){
 
	var image = document.createElement("img");
	image.width = 46;
	image.id = id;
	image.height = 22;
	image.border = 0;
	image.className = "mw-toolbar-editbutton";
	image.style.cursor = "pointer";
	image.alt = "OCR";
	image.title = comment;
	image.src = source;
	image.onclick = onclick;

	var tb  = document.getElementById("toolbar"); 
	if(tb) {
		tb.appendChild(image);
        } else {
                if (id === 'wsOcr1')
                        image.src = '//upload.wikimedia.org/wikipedia/commons/c/c9/Toolbaricon_OCR.png';

                //append the button only after the toolbar is ready
                mw.loader.using('ext.wikiEditor', function () {
                    var new_tb  = document.getElementById("wikiEditor-ui-toolbar"); 
        	    if(new_tb) {
	        	new_tb.childNodes[0].childNodes[1].appendChild(image);
                    }
                });
        }
}

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 addOCRButton()
{
    if(mw.config.get( 'wgCanonicalNamespace' ) !== 'Page' ||  $.inArray(mw.config.get('wgAction'), ['edit', 'submit']) === -1)
        return;

    if(mw.config.get('wgContentLanguage') == "de"){
       addOCRButton2("wsOcr1","Normale OCR","//upload.wikimedia.org/wikipedia/commons/e/e0/Button_ocr.png", do_hocr);
       addOCRButton2("wsOcr2","Fraktur OCR","//upload.wikimedia.org/wikipedia/commons/a/af/Button_Fractur_OCR.png", fraktur_ocr);
    } else {
        addOCRButton2("wsOcr1","Seudo - Get the text by OCR","//upload.wikimedia.org/wikipedia/commons/e/e0/Button_ocr.png", do_hocr);
    }
}

if (!self.proofreadpage_disable_ocr) {
    $(addOCRButton);
}