mirrors/zotero
2
0
Fork 0
Code Issues Pull requests Projects Releases Packages 1 Wiki Activity Actions

Adding Sylvain's 15mar10 changes.

Browse source
This commit is contained in:
Matt Burton 2010-03-29 15:01:48 +00:00
parent 1caa7ac359
commit 466cf794e1
1 changed files with 92 additions and 72 deletions
Show stats Download patch file Download diff file Expand all files Collapse all files

78
translators/Gallica.js
View file

@ -8,7 +8,7 @@
"maxVersion":"",
"priority":100,
"inRepository":true,
"lastUpdated":"2009-10-08 17:40:00"
"lastUpdated":"2010-02-20 10:40:00"
}
function detectWeb(doc, url) {
@ -35,9 +35,9 @@ function detectWeb(doc, url) {
}
else if (indexArk == 0)
{
var iconxpath = '//div[@id="Infos"]/img';
if (elt = doc.evaluate(iconxpath, doc, nsResolver,
XPathResult.ANY_TYPE, null).iterateNext()) {
var iconxpath = '//div[@class="contenu1"]/img';
if (elt = doc.evaluate(iconxpath, doc, nsResolver, XPathResult.ANY_TYPE, null).iterateNext())
{
var icon = elt.getAttribute('src');
return getDoctypeGallica(icon);
}
@ -45,9 +45,9 @@ XPathResult.ANY_TYPE, null).iterateNext()) {
// For some biblio, the icon picture is located in another div ...
var iconxpath = '//div[@class="titrePeriodiqueGauche"]/img';
if (elt = doc.evaluate(iconxpath, doc, nsResolver,
XPathResult.ANY_TYPE, null).iterateNext()) {
XPathResult.ANY_TYPE, null).iterateNext())
{
var icon = elt.getAttribute('src');
return getDoctypeGallica(icon);
}
}
@ -62,19 +62,29 @@ function getDoctypeGallica(img)
{
var iconname = img.substring(img.lastIndexOf('/') + 1);
if ( (iconname =='doc_livre_ocr.png') || (iconname == 'doc_livre.png') )
if (iconname =='livre_a.png')
{
return "book";
}
else if (iconname == 'doc_carte.png')
else if (iconname == 'carte.png')
{
return "map";
}
else if (iconname == 'doc_image.png')
else if (iconname == 'images.png')
{
return "artwork";
}
else if ( (iconname == 'doc_periodique.png') || (iconname == 'doc_perio_vol_ocr.png') )
else if (iconname == 'docsonore.png')
{
return "audioRecording";
}
else if (iconname == 'musiquenotee.png')
{
// This icon is for Sheet music type. But no Zotero type matches
// as of today (2010-02)
return "book";
}
else if ( (iconname == 'picto_type_document1.png') || (iconname == 'perio_vol_ocr.png') )
{
return "book";
}
@ -95,23 +105,18 @@ function doWeb(doc, url) {
if (detectWeb(doc, url) == "multiple")
{
var availableItems = new Array();
var xpath = '//td[@class="ResultatsRechercheInfos"]/a';
var xpath = '//div[@class="resultats_line"]';
var elmts = doc.evaluate(xpath, doc, nsResolver, XPathResult.ANY_TYPE, null);
var elmt = elmts.iterateNext();
var itemsId = new Array();
var i = 0;
var i = 1;
do {
var id = doc.evaluate('../../..//a[@id]', elmt, nsResolver, XPathResult.ANY_TYPE, null).iterateNext();
// This id looks like idN00000. We need the information after id to get the informations about
// the title. We need to store it in an array, we leave the starting id.
var cleanId = id.getAttribute('id').substring(2);
itemsId[i] = cleanId;
var searchTitle = elmt.textContent;
availableItems[i] = searchTitle;
var id = doc.evaluate('div[@class="resultat_id"]', elmt, nsResolver, XPathResult.ANY_TYPE, null).iterateNext().textContent;
var this_result = doc.evaluate('div[@class="resultat_desc"]/div[@class="titre"]/a', elmt, nsResolver, XPathResult.ANY_TYPE, null).iterateNext();
availableItems[i] = Zotero.Utilities.cleanTags(this_result.getAttribute('title'));
i++;
} while (elmt = elmts.iterateNext());
@ -122,25 +127,33 @@ function doWeb(doc, url) {
// All informations are available on search result page. We don't need to query
// every subpage with scrape. We'are going to call the special Gallica scrape function
// This function (scrapeGallica) is reused in scrape.
var fullpath = '//div[@id="noticeComplete' + itemsId[i] + '"]/div';
var detail = doc.evaluate(fullpath, doc, nsResolver, XPathResult.ANY_TYPE, null).iterateNext();
Zotero.debug(itemsId[i]);
var iconType = doc.evaluate('//a[@id="id' + itemsId[i] + '"]/..//span[@class="typedoc"]/img', doc, nsResolver, XPathResult.ANY_TYPE, null).iterateNext();
var fullpath = '//div[@class="resultats_line"][' + i + ']';
var item_element = doc.evaluate(fullpath, doc, nsResolver, XPathResult.ANY_TYPE, null).iterateNext();
if (item_element != undefined)
{
var detail = doc.evaluate('.//div[@class="notice"]', item_element, nsResolver, XPathResult.ANY_TYPE, null).iterateNext();
var iconType = doc.evaluate('.//div[@class="picto"]/img', item_element, nsResolver, XPathResult.ANY_TYPE, null).iterateNext();
var docType = getDoctypeGallica(iconType.getAttribute('src'));
Zotero.debug( itemsId[i]);
scrapeGallica(doc, nsResolver, detail, docType);
var docUrl = doc.evaluate('.//div[@class="liens"]/a', item_element, nsResolver, XPathResult.ANY_TYPE, null).iterateNext();
docUrl = docUrl.getAttribute("href");
scrapeGallica(doc, nsResolver, detail, docType, docUrl);
}
}
}
else
{
var docType = detectWeb(doc, url);
var xpath = '//div[@id="Popup1"]/div[@class="data"]';
var xpath = '//div[@class="notice"]';
var detail = doc.evaluate(xpath, doc, nsResolver, XPathResult.ANY_TYPE, null).iterateNext();
scrapeGallica(doc, nsResolver, detail, docType);
scrapeGallica(doc, nsResolver, detail, docType, "");
}
}
function scrapeGallica(doc, nsResolver, div, type)
function scrapeGallica(doc, nsResolver, div, type, direct_url)
{
var item = new Zotero.Item;
item.itemType = type;
@ -239,8 +252,15 @@ function scrapeGallica(doc, nsResolver, div, type)
} while (elmt = elmts.iterateNext());
if ( (item.url == "") || (item.url == undefined) )
{
if (direct_url != "")
{
item.url = "http://gallica.bnf.fr" + direct_url;
}
else
{
item.url = doc.location.href;
}
}
item.complete();
}