#112 Parse epub insed of mix of ppub and epub

pubmed_parser/pubmed_oa_parser.py

@@ -161,11 +161,11 @@ def parse_pubmed_xml(path, include_path=False, nxml=False):
         journal = ""
 
     dict_article_meta = parse_article_meta(tree)
-    pub_year_node = tree.find(".//pub-date/year")
+    pub_year_node = tree.find(".//pub-date[@pub-type='epub']/year")
     pub_year = pub_year_node.text if pub_year_node is not None else ""
-    pub_month_node = tree.find(".//pub-date/month")
+    pub_month_node = tree.find(".//pub-date[@pub-type='epub']/month")
     pub_month = pub_month_node.text if pub_month_node is not None else "01"
-    pub_day_node = tree.find(".//pub-date/day")
+    pub_day_node = tree.find(".//pub-date[@pub-type='epub']/day")
     pub_day = pub_day_node.text if pub_day_node is not None else "01"
 
     subjects_node = tree.findall(".//article-categories//subj-group/subject")

tests/test_pubmed_oa_parser.py

@@ -39,6 +39,7 @@ def test_parse_pubmed_xml():
     assert parsed_xml.get("doi") == "10.1371/journal.pone.0046493"
     assert parsed_xml.get("subjects") == "Research Article; Biology; Biochemistry; Enzymes; Enzyme Metabolism; Lipids; Fatty Acids; Glycerides; Lipid Metabolism; Neutral Lipids; Metabolism; Lipid Metabolism; Proteins; Globular Proteins; Protein Classes; Recombinant Proteins; Biotechnology; Microbiology; Bacterial Pathogens; Bacteriology; Emerging Infectious Diseases; Host-Pathogen Interaction; Microbial Growth and Development; Microbial Metabolism; Microbial Pathogens; Microbial Physiology; Proteomics; Sequence Analysis; Spectrometric Identification of Proteins"  # noqa
     assert "Competing Interests: " in parsed_xml.get("coi_statement")
+    assert parsed_xml.get('publication_date') == '28-9-2012'
 
 
 def test_parse_pubmed_paragraph():