summaryrefslogtreecommitdiffstats
path: root/tests/test_parse.py
diff options
context:
space:
mode:
authorBryan Newbold <bnewbold@archive.org>2021-10-25 17:02:14 -0700
committerBryan Newbold <bnewbold@archive.org>2021-10-25 17:02:14 -0700
commit8e68f3cd4cc3d2b6b166bf87811bb4db82cc7573 (patch)
tree3b71338fcae115c834097b16cbfe051fb88c7684 /tests/test_parse.py
parentfcdb271193ca2c6b90eeeb5f4af4bbc15083319a (diff)
downloadgrobid_tei_xml-8e68f3cd4cc3d2b6b166bf87811bb4db82cc7573.tar.gz
grobid_tei_xml-8e68f3cd4cc3d2b6b166bf87811bb4db82cc7573.zip
more test coverage and comments
Diffstat (limited to 'tests/test_parse.py')
-rw-r--r--tests/test_parse.py38
1 files changed, 36 insertions, 2 deletions
diff --git a/tests/test_parse.py b/tests/test_parse.py
index eb4b46e..976d1b1 100644
--- a/tests/test_parse.py
+++ b/tests/test_parse.py
@@ -215,14 +215,34 @@ def test_citation_list_xml() -> None:
tei_xml = f.read()
citations = parse_citations_xml(tei_xml)
- assert len(citations) == 10
- assert citations[7].title == "Global Hunger Index: The Challenge of Hidden Hunger"
+ assert len(citations) == 13
assert citations[3].note == "The Research Handbook on International Environmental Law"
assert citations[3].authors[0].surname == "Uhlířová"
+ assert citations[3].authors[1].surname == "Drumbl"
+ assert citations[3].editors
+ assert citations[3].editors[0].surname == "Fitzmaurice"
+ # TODO: multiple persName under a single <editor> (https://github.com/kermitt2/grobid/issues/845)
+ # assert citations[3].editors[1].surname == "Brus"
+
assert citations[4].authors[0].surname == "Sleytr"
assert citations[4].authors[0].middle_name == "B"
+ assert citations[7].title == "Global Hunger Index: The Challenge of Hidden Hunger"
+
+ assert citations[10].doi == "10.1093/eurheartj/ehi890"
+ assert citations[10].url is None
+
+ assert citations[11].title == "Devices, Measurements and Properties"
+ assert citations[11].series_title == "Handbook of Optics"
+ assert citations[11].publisher == "McGRAW-HILL"
+
+ assert citations[
+ 12].title == "Implications of abandoned shoreline features above Glacial Lake Duluth levels along the north shore of the Superior Basin in the vicinity of the Brule River"
+ assert citations[
+ 12].book_title == "Paper presented at the 13th Biennial Meeting of the American Quaternary Association"
+ assert citations[12].institution == "University of Minnesota"
+
def test_grobid_070_document() -> None:
# more recent GROBID v0.7.0 output
@@ -245,6 +265,20 @@ def test_grobid_070_document() -> None:
assert cite_b6.volume == "574"
assert cite_b6.issue == "1"
+ cite_b3 = doc.citations[3]
+ assert cite_b3.url == "http://unesdoc.unesco.org/ulis/"
+ assert cite_b3.title == "Requirements for Global Implementation of the Strategic Plan for Coastal GOOS"
+ assert cite_b3.authors
+ assert cite_b3.authors[0].surname == "Ioc-Unesco"
+ assert cite_b3.date == "2012"
+
+ cite_b18 = doc.citations[18]
+ assert cite_b18.note == "TriOS GmbH [Internet"
+ assert cite_b18.date == "2017-01-05"
+
+ cite_b29 = doc.citations[29]
+ assert cite_b29.note == "PhD dissertation"
+
# run these methods over some more examples
for c in doc.citations:
c.to_csl_dict()