diff --git a/hepcrawl/spiders/wsp_spider.py b/hepcrawl/spiders/wsp_spider.py index 1f0786d3..5400621c 100644 --- a/hepcrawl/spiders/wsp_spider.py +++ b/hepcrawl/spiders/wsp_spider.py @@ -14,7 +14,6 @@ import os import urlparse import tempfile -import shutil from scrapy import Request from scrapy.spiders import XMLFeedSpider @@ -214,10 +213,12 @@ def parse_node(self, response, node): return parsed_record - def _get_collections(self, node, article_type, current_journal_title): + @staticmethod + def _get_collections(node, article_type, current_journal_title): """Return this articles' collection.""" conference = node.xpath('.//conference').extract() - if conference or current_journal_title == "International Journal of Modern Physics: Conference Series": + if conference or current_journal_title == "International Journal of Modern Physics:" \ + " Conference Series": return ['HEP', 'ConferencePaper'] elif article_type == "review-article": return ['HEP', 'Review']