diff options
-rw-r--r-- | scrapersources/postliste-universitetet-i-tromso | 7 |
1 files changed, 5 insertions, 2 deletions
diff --git a/scrapersources/postliste-universitetet-i-tromso b/scrapersources/postliste-universitetet-i-tromso index 5e623f5..eb086d3 100644 --- a/scrapersources/postliste-universitetet-i-tromso +++ b/scrapersources/postliste-universitetet-i-tromso @@ -228,13 +228,16 @@ parser = postlistelib.PDFJournalParser(agency=agency) #test_small_pdfs(parser) +process_page_queue(parser, errors) + +process_journal_pdfs(parser, "http://uit.no/om/enhet/artikkel?p_document_id=382893&p_dimension_id=88216", errors) + urls = [] +urls.extend(find_day_urls(parser, 2013)) urls.extend(find_day_urls(parser, 2014)) urls.extend(find_day_urls(parser, 2015)) for url in urls: html = scraperwiki.scrape(url).decode('utf-8') fetch_postjournal_day(parser, url, html, saver=saver) -process_journal_pdfs(parser, "http://uit.no/om/enhet/artikkel?p_document_id=382893&p_dimension_id=88216", errors) -process_page_queue(parser, errors) report_errors(errors) |