updates to pubmed importer

1 files changed, 45 insertions, 4 deletions

diff --git a/python/tests/import_pubmed.py b/python/tests/import_pubmed.py
index 05a77599..0185c8c4 100644
--- a/python/tests/import_pubmed.py
+++ b/python/tests/import_pubmed.py
@@ -9,12 +9,12 @@ from bs4 import BeautifulSoup
 @pytest.fixture(scope="function")
 def pubmed_importer(api):
     with open('tests/files/ISSN-to-ISSN-L.snip.txt', 'r') as issn_file:
-        yield PubmedImporter(api, issn_file, extid_map_file='tests/files/example_map.sqlite3', bezerk_mode=True)
+        yield PubmedImporter(api, issn_file, extid_map_file='tests/files/example_map.sqlite3', bezerk_mode=True, lookup_refs=True)
 
 @pytest.fixture(scope="function")
 def pubmed_importer_existing(api):
     with open('tests/files/ISSN-to-ISSN-L.snip.txt', 'r') as issn_file:
-        yield PubmedImporter(api, issn_file, extid_map_file='tests/files/example_map.sqlite3', bezerk_mode=False)
+        yield PubmedImporter(api, issn_file, extid_map_file='tests/files/example_map.sqlite3', bezerk_mode=False, lookup_refs=True)
 
 def test_pubmed_importer(pubmed_importer):
     last_index = pubmed_importer.api.get_changelog(limit=1)[0].index
@@ -73,8 +73,49 @@ def test_pubmed_xml_parse(pubmed_importer):
     assert r1.contribs[0].surname == "Blume"
 
     print(r1.extra)
-    # TODO: assert r1.extra['pubmed']['mesh_topics'] == ['Accounting', 'Economics, Hospital', 'Hospital Administration']
     assert r1.extra['pubmed']['pub_types'] == ['Journal Article']
     assert not r1.refs
 
-    # XXX: r2 tests
+    assert r2.title == "Synthesis and Antibacterial Activity of Metal(loid) Nanostructures by Environmental Multi-Metal(loid) Resistant Bacteria and Metal(loid)-Reducing Flavoproteins"
+    assert r2.subtitle == None
+    assert r2.original_title == None
+    assert r2.publisher == None
+    assert r2.release_type == "article-journal"
+    assert r2.release_stage == "published"
+    assert r2.license_slug == None
+    assert r2.ext_ids.doi == "10.3389/fmicb.2018.00959"
+    assert r2.ext_ids.pmid == "29869640"
+    assert r2.ext_ids.pmcid == "PMC5962736"
+    assert r2.language == "en"
+    assert r2.volume == "9"
+    assert r2.issue == None
+    assert r2.pages == "959"
+    assert str(r2.release_date) == "2018-05-15"
+    assert r2.release_year == 2018
+    # matched by ISSN, so shouldn't be in there?
+    #assert extra['container_name'] == "Frontiers in microbiology"
+
+    assert len(r2.contribs) > 3
+    assert r2.contribs[0].raw_name == "Maximiliano Figueroa"
+    assert r2.contribs[0].given_name == "Maximiliano"
+    assert r2.contribs[0].surname == "Figueroa"
+    assert r2.contribs[0].raw_affiliation == "Laboratorio Microbiología Molecular, Departamento de Biología, Facultad de Química y Biología, Universidad de Santiago de Chile, Santiago, Chile."
+    assert r2.contribs[4].surname == "Muñoz-Villagrán"
+    assert r2.contribs[7].surname == "Latorre"
+    assert r2.contribs[7].raw_affiliation == "Mathomics, Centro de Modelamiento Matemático, Universidad de Chile, Beauchef, Santiago, Chile."
+    assert r2.contribs[7].extra['more_affiliations'] == [
+        "Fondap-Center of Genome Regulation, Facultad de Ciencias, Universidad de Chile, Santiago, Chile.",
+        "Laboratorio de Bioinformática y Expresión Génica, INTA, Universidad de Chile, Santiago, Chile.",
+        "Instituto de Ciencias de la Ingeniería, Universidad de O'Higgins, Rancagua, Chile.",
+    ]
+    assert r2.contribs[-1].raw_name == "Felipe Arenas"
+
+    assert r2.abstracts[0].content.startswith("Microbes are suitable candidates to recover and decontaminate different environments from soluble metal ions, either via reduction")
+    assert r2.abstracts[0].lang == "en"
+
+    print(r2.extra)
+    assert r2.extra['pubmed']['pub_types'] == ['Journal Article']
+
+    assert r2.refs[0].extra['unstructured'] == "Microbiology. 2009 Jun;155(Pt 6):1840-6"
+    assert r2.refs[0].extra['pmid'] == "19383690"
+