summaryrefslogtreecommitdiffstats
path: root/python/tests/import_grobid_metadata.py
diff options
context:
space:
mode:
authorBryan Newbold <bnewbold@robocracy.org>2019-01-23 15:02:03 -0800
committerBryan Newbold <bnewbold@robocracy.org>2019-01-23 15:02:03 -0800
commit1443f05faebd9e697086132694401f6a6c42d9b5 (patch)
tree8da8b8e7f4c957c5edccefe9188741c15697cd46 /python/tests/import_grobid_metadata.py
parent1fa8f820fd3b7c64d424f55796d2b860d22e4b22 (diff)
downloadfatcat-1443f05faebd9e697086132694401f6a6c42d9b5.tar.gz
fatcat-1443f05faebd9e697086132694401f6a6c42d9b5.zip
more tests; fix some importer behavior
Diffstat (limited to 'python/tests/import_grobid_metadata.py')
-rw-r--r--python/tests/import_grobid_metadata.py16
1 files changed, 14 insertions, 2 deletions
diff --git a/python/tests/import_grobid_metadata.py b/python/tests/import_grobid_metadata.py
index 698b36be..feb604ce 100644
--- a/python/tests/import_grobid_metadata.py
+++ b/python/tests/import_grobid_metadata.py
@@ -48,12 +48,15 @@ def test_file_metadata_parse(grobid_metadata_importer):
assert fe.urls[0].rel == "webarchive"
assert len(fe.release_ids) == 0
-# TODO: use API to check that entities actually created...
def test_grobid_metadata_importer(grobid_metadata_importer):
last_index = grobid_metadata_importer.api.get_changelog(limit=1)[0].index
with open('tests/files/example_grobid_metadata_lines.tsv', 'r') as f:
grobid_metadata_importer.bezerk_mode = True
- LinePusher(grobid_metadata_importer, f).run()
+ counts = LinePusher(grobid_metadata_importer, f).run()
+ assert counts['insert'] == 10
+ assert counts['inserted.release'] == 10
+ assert counts['exists'] == 0
+ assert counts['skip'] == 0
# fetch most recent editgroup
change = grobid_metadata_importer.api.get_changelog_entry(index=last_index+1)
@@ -62,3 +65,12 @@ def test_grobid_metadata_importer(grobid_metadata_importer):
assert "grobid" in eg.description.lower()
assert eg.extra['git_rev']
assert "fatcat_tools.GrobidMetadataImporter" in eg.extra['agent']
+
+ with open('tests/files/example_grobid_metadata_lines.tsv', 'r') as f:
+ grobid_metadata_importer.reset()
+ grobid_metadata_importer.bezerk_mode = False
+ counts = LinePusher(grobid_metadata_importer, f).run()
+ assert counts['insert'] == 0
+ assert counts['inserted.release'] == 0
+ assert counts['exists'] == 10
+ assert counts['skip'] == 0