update for grabbing

This commit is contained in:
Martin Donnelly 2016-03-02 15:55:38 +00:00
parent cbf598cbb5
commit 256ee34366

View File

@ -22,13 +22,13 @@ var dbCouch = nano.use(db_name);
var bodyfile = __dirname + '/' + 'body.html';
var htmlfile = __dirname + '/' + 'testoutput.html';
var generics = ['ARTICLE', 'div.content_column', 'div.post'];
var generics = ['ARTICLE', 'div.content_column', 'div.post','div.page'];
function cleaner(b) {
var _b = b;
var unwanted = ['div#disqus_thread', 'SCRIPT', 'FOOTER', 'div.ssba', '.shareaholic-canvas', '.yarpp-related', 'div.dfad', 'div.postFooterShare', 'div#nextPrevLinks', '.post-comments'];
var unwanted = ['div#disqus_thread', 'SCRIPT', 'FOOTER', 'div.ssba', '.shareaholic-canvas', '.yarpp-related', 'div.dfad', 'div.postFooterShare', 'div#nextPrevLinks', '.post-comments','HEADER'];
for (var i = 0; i < unwanted.length; i++) {
_b.find(unwanted[i]).remove();