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

adding keiths changes from dev list

Browse source
This commit is contained in:
Matt Burton 2009-07-16 15:11:24 +00:00
parent 3a5da5e6df
commit 1de4502a1a
1 changed files with 23 additions and 26 deletions
Download patch file Download diff file Expand all files Collapse all files

49
translators/National Archives of South Africa.js
View file

@ -32,20 +32,20 @@ function scrape(doc, url) {
var namespace = doc.documentElement.namespaceURI; var namespace = doc.documentElement.namespaceURI;
var nsResolver = namespace ? function(prefix) { var nsResolver = namespace ? function(prefix) {
if (prefix == 'x') return namespace; else return null; if (prefix == 'x') return namespace; else return null;
} : null; } : null;
var dataTags = new Object(); var dataTags = new Object();
var tagsContent = new Array(); var tagsContent = new Array();
var fieldTitle; var fieldTitle;
var newItem = new Zotero.Item("book"); var newItem = new Zotero.Item("book");
var headers = doc.evaluate('//td[2]/pre/b', doc, nsResolver, XPathResult.ANY_TYPE, null); var headers = doc.evaluate('//td[2]/pre/b', doc, nsResolver,XPathResult.ANY_TYPE, null);
var xPathCount = doc.evaluate('count (//td[2]/pre/b)', doc, nsResolver, XPathResult.ANY_TYPE, null); var xPathCount = doc.evaluate('count (//td[2]/pre/b)', doc,nsResolver, XPathResult.ANY_TYPE, null);
var contents = doc.evaluate('//td[2]/pre', doc, nsResolver, XPathResult.ANY_TYPE, null).iterateNext().textContent; var contents = doc.evaluate('//td[2]/pre', doc, nsResolver,XPathResult.ANY_TYPE, null).iterateNext().textContent;
var headersArray = new Array(); var headersArray = new Array();
var oneHeader = ''; var oneHeader = '';
if (xPathCount.numberValue > 1) { if (xPathCount.numberValue > 1) {
@ -56,25 +56,25 @@ function scrape(doc, url) {
} else { } else {
oneHeader = (headers.iterateNext().textContent); oneHeader = (headers.iterateNext().textContent);
} }
var contentsArray = new Array(); var contentsArray = new Array();
var j = 0; var j = 0;
if (oneHeader.length<1) { if (oneHeader.length<1) {
for (var i = headersArray.length-1; i> -1; i--) { for (var i = headersArray.length-1; i> -1; i--) {
var fieldIndex = contents.indexOf(headersArray[i]); var fieldIndex = contents.indexOf(headersArray[i]);
var shorten = headersArray[i].length; var shorten = headersArray[i].length;
contentsArray.push(contents.substr(fieldIndex)); contentsArray.push(contents.substr(fieldIndex));
contents = contents.substr(0, fieldIndex); contents = contents.substr(0, fieldIndex);
fieldTitle = headersArray[i].replace(/\s+/g, ''); fieldTitle = headersArray[i].replace(/\s+/g, '');
dataTags[fieldTitle] = contentsArray[j].substr(shorten).replace(/^\s*|\s+$/g, ''); dataTags[fieldTitle] = contentsArray[j].substr(shorten).replace(/^\s*|\s+$/g, '');
j++; j++;
} }
} }
associateData (newItem, dataTags, "DEPOT", "repository"); associateData (newItem, dataTags, "DEPOT", "repository");
associateData (newItem, dataTags, "REFERENCE", "callNumber"); associateData (newItem, dataTags, "REFERENCE", "callNumber");
@ -83,15 +83,12 @@ function scrape(doc, url) {
associateData (newItem, dataTags, "VOLUME_NO", "volume"); associateData (newItem, dataTags, "VOLUME_NO", "volume");
associateData (newItem, dataTags, "REMARKS", "extra"); associateData (newItem, dataTags, "REMARKS", "extra");
associateData (newItem, dataTags, "SUMMARY", "abstractNote"); associateData (newItem, dataTags, "SUMMARY", "abstractNote");
associateData (newItem, dataTags, "SOURCE", "series");
if (dataTags["DESCRIPTION"]) { if (dataTags["DESCRIPTION"]) {
associateData (newItem, dataTags, "DESCRIPTION", "title"); associateData (newItem, dataTags, "DESCRIPTION", "title");
} else { } else {
newItem.title = "No Title Found"; newItem.title = "No Title Found";
} }
newItem.url = doc.location.href;
newItem.complete(); newItem.complete();
} }
@ -100,18 +97,18 @@ function doWeb(doc, url) {
var nsResolver = namespace ? function(prefix) { var nsResolver = namespace ? function(prefix) {
if (prefix == 'x') return namespace; else return null; if (prefix == 'x') return namespace; else return null;
} : null; } : null;
var articles = new Array(); var articles = new Array();
if (detectWeb(doc, url) == "multiple") { if (detectWeb(doc, url) == "multiple") {
var items = new Object(); var items = new Object();
var titles = doc.evaluate('//td/a', doc, nsResolver, XPathResult.ANY_TYPE, null); var titles = doc.evaluate('//td/a', doc, nsResolver, XPathResult.ANY_TYPE, null);
var lastLink; var lastLink;
var next_title; var next_title;
while (next_title = titles.iterateNext()) { while (next_title = titles.iterateNext()) {
if (!next_title.textContent.match(/^\d\d\d\d/) && !next_title.textContent.match(/\\/) && next_title.textContent.length>3 && next_title.textContent.match(/\w/)) { if (!next_title.textContent.match(/^\d\d\d\d/) && !next_title.textContent.match(/\\/) && next_title.textContent.length>3 && next_title.textContent.match(/\w/)) {
Zotero.debug(next_title.textContent); Zotero.debug(next_title.textContent);
items[next_title.href] = next_title.textContent; items[next_title.href] = next_title.textContent;