Support Web of Science / Web of Knowledge.

master
Alinson S. Xavier 11 years ago
parent 08a1a5debe
commit 9a35ac52a3

@ -1,3 +1,6 @@
var scholarius_options = {
};
var visjs_options = {
stabilize: false,
nodes: {

@ -34,12 +34,12 @@ ScholarCrawler.prototype.add_citations = function(parent_node, levels)
var crawler = this;
if(parent_node.is_dummy)
return this._add_citations_from_scopus(parent_node, levels);
return this._add_citations_from_parser(parent_node, levels);
articles_db.findOne({_id: parent_node._id}, function(err, parent_article_db)
{
if(parent_article_db === null || !parent_article_db.is_cached)
crawler._add_citations_from_scopus(parent_node, levels);
crawler._add_citations_from_parser(parent_node, levels);
else
crawler._add_citations_from_db(crawler.article_to_node(parent_node.article), levels);
});
@ -65,7 +65,7 @@ ScholarCrawler.prototype._add_citations_from_db = function(parent_node, levels)
});
}
ScholarCrawler.prototype._add_citations_from_scopus = function(parent_node, levels)
ScholarCrawler.prototype._add_citations_from_parser = function(parent_node, levels)
{
assert(levels >= 0, "levels should be non-negative");

@ -0,0 +1,15 @@
function GenericParser()
{
this.scopus_parser = new ScopusParser();
this.web_of_science_parser = new WebOfScienceParser();
}
GenericParser.prototype.parse = function(url, callback)
{
if(url.indexOf("scopus.com") >= 0)
this.scopus_parser.parse(url, callback);
if(url.indexOf("webofknowledge.com") >= 0)
this.web_of_science_parser.parse(url, callback);
}

@ -0,0 +1,72 @@
function WebOfScienceParser()
{
};
WebOfScienceParser.prototype._parse_results_page = function(url, callback)
{
console.log("parsing");
var iframe = document.createElement("iframe");
$(iframe).hide();
document.body.appendChild(iframe);
iframe.src = url;
$(iframe).load(function()
{
var articles = [];
var ibody = $(iframe).contents()[0];
var select_pages = $(ibody.getElementsByName("pageSize").item(0));
if(select_pages.val() != 50) {
select_pages.val(50);
select_pages.change();
return;
}
$(ibody).find(".search-results-content").each(function(index, li)
{
var article = {};
$(li).find("a[href^='/full_record']").each(function(index, tag) {
article.url = tag.href;
article.title = $.trim($(tag).text());
});
$(li).next().find("a[href^='/CitingArticles']").each(function(index, tag) {
article.citations_url = tag.href;
article.n_citations = parseInt($(tag).text());
});
$(li).find('span.label').each(function(index, tag) {
//if($(tag).text().indexOf("By:") == 0)
// article.year = parseInt($.trim($(tag).next().text()));
if($(tag).text().indexOf("By:") == 0) {
article.authors = $.trim($(tag).parent().text()).replace(/^By: /,"");
var source_year = $.trim($(tag).parent().next().text());
article.source = source_year.replace(/Published.*/, "").replace(/\s\s+/g, ", ").replace(/, $/, "");
article.year = parseInt(source_year.substr(source_year.length-4));
}
});
if(!('n_citations' in article)) {
article.citations_url = undefined;
article.n_citations = 0;
}
article._id = $.md5(article.title + article.authors);
articles.push(article);
console.log(article);
});
callback(articles);
document.body.removeChild(iframe);
});
}
WebOfScienceParser.prototype.parse = function(url, callback)
{
if(!url) return;
this._parse_results_page(url, callback);
}

@ -14,7 +14,7 @@ ShowMapView.prototype.render = function(container)
this.nodes.add(this.map.nodes);
this.edges.add(this.map.edges);
var parser = new ScopusParser();
var parser = new GenericParser();
var crawler = new ScholarCrawler(parser, this.nodes, this.edges);
crawler.start();

@ -10,7 +10,9 @@
<script type="text/javascript" src="app://root/js/config.js"></script>
<script type="text/javascript" src="app://root/js/crawler.js"></script>
<script type="text/javascript" src="app://root/js/model/map.js"></script>
<script type="text/javascript" src="app://root/js/parsers/scopus.js"></script>
<script type="text/javascript" src="app://root/js/parsers/generic_parser.js"></script>
<script type="text/javascript" src="app://root/js/parsers/scopus_parser.js"></script>
<script type="text/javascript" src="app://root/js/parsers/web_of_science_parser.js"></script>
<script type="text/javascript" src="app://root/js/views/main_menu.js"></script>
<script type="text/javascript" src="app://root/js/views/show_map.js"></script>