Parse pii in the web parser

titipata · Apr 22, 2024 · e49a78f · e49a78f
2 parents 5943a6e + bf5e771
commit e49a78f
Show file tree

Hide file tree

Showing 2 changed files with 18 additions and 1 deletion.
diff --git a/pubmed_parser/pubmed_web_parser.py b/pubmed_parser/pubmed_web_parser.py
@@ -137,7 +137,13 @@ def parse_pubmed_web_tree(tree):
         doi = doi[0].text
     except IndexError:
         doi = None
-
+
+    pii = tree.xpath('//elocationid[@eidtype="pii"]')
+    try:
+        pii = pii[0].text
+    except IndexError:
+        pii = None
+
     language = tree.xpath("//language")
     try:
         language = language[0].text
@@ -152,6 +158,7 @@ def parse_pubmed_web_tree(tree):
         "authors": authors_text,
         "keywords": keywords,
         "doi": doi,
+        "pii": pii,
         "year": year,
         "language": language
     }

diff --git a/tests/test_pubmed_web_parser.py b/tests/test_pubmed_web_parser.py
@@ -15,6 +15,7 @@ def test_pubmed_web_parser_all_fields_content():
             "authors": "Rieka von der Warth; Isabelle Hempler",
             "keywords": "",
             "doi": "10.1016/j.zefq.2023.11.002",
+            "pii": "S1865-9217(23)00212-X",
             "year": "2024",
             "language": "ger",
             "pmid": "38218666",
@@ -27,6 +28,7 @@ def test_pubmed_web_parser_all_fields_content():
             "authors": "Andreas Leimbach; Jörg Hacker; Ulrich Dobrindt",
             "keywords": "D000818:Animals;D004926:Escherichia coli;D004927:Escherichia coli Infections;D023281:Genomics;D006801:Humans;D007413:Intestinal Mucosa;D007422:Intestines;D010802:Phylogeny;D013559:Symbiosis",
             "doi": "10.1007/82_2012_303",
+            "pii": None,
             "year": "2013",
             "language": "eng",
             "pmid": "23340801",
@@ -49,6 +51,7 @@ def test_pubmed_web_parser_all_fields_existence():
         "authors",
         "keywords",
         "doi",
+        "pii",
         "year",
         "language",
         "pmid",
@@ -66,7 +69,14 @@ def test_pubmed_web_parser_save_xml():
 
     assert "xml" in pubmed_dict
 
+
 def test_doi():
     """Test the correct parsing of the doi."""
     pubmed_dict = pp.parse_xml_web("32145645", save_xml=False)
     assert pubmed_dict['doi'] == "10.1016/j.ejmech.2020.112186"
+
+
+def test_pii():
+    """Test the correct parsing of the pii."""
+    pubmed_dict = pp.parse_xml_web("32145645", save_xml=False)
+    assert pubmed_dict['pii'] == "S0223-5234(20)30153-7"