diff options
author | Bryan Newbold <bnewbold@robocracy.org> | 2021-11-02 18:13:14 -0700 |
---|---|---|
committer | Bryan Newbold <bnewbold@robocracy.org> | 2021-11-02 18:13:14 -0700 |
commit | cdfd6b85b386b7bbf9d5a5179ef26970b6e5a4e7 (patch) | |
tree | 5e4034027b51f3ee4d2a488bb2cbb7a75c3bd0d8 /python/tests/import_grobid_metadata.py | |
parent | 78f08280edea4ff65ca613ad30005c45cc48dea6 (diff) | |
download | fatcat-cdfd6b85b386b7bbf9d5a5179ef26970b6e5a4e7.tar.gz fatcat-cdfd6b85b386b7bbf9d5a5179ef26970b6e5a4e7.zip |
fmt (black): tests/
Diffstat (limited to 'python/tests/import_grobid_metadata.py')
-rw-r--r-- | python/tests/import_grobid_metadata.py | 52 |
1 files changed, 29 insertions, 23 deletions
diff --git a/python/tests/import_grobid_metadata.py b/python/tests/import_grobid_metadata.py index 862b310f..a06e7aa4 100644 --- a/python/tests/import_grobid_metadata.py +++ b/python/tests/import_grobid_metadata.py @@ -1,4 +1,3 @@ - import base64 import json import os @@ -13,17 +12,21 @@ WARNING: these tests are currently very fragile because they have database side-effects. Should probably be disabled or re-written. """ + @pytest.fixture(scope="function") def grobid_metadata_importer(api): yield GrobidMetadataImporter(api) def test_grobid_metadata_parse(grobid_metadata_importer): - with open('tests/files/example_grobid_metadata_lines.tsv', 'r') as f: - raw = json.loads(f.readline().split('\t')[4]) + with open("tests/files/example_grobid_metadata_lines.tsv", "r") as f: + raw = json.loads(f.readline().split("\t")[4]) re = grobid_metadata_importer.parse_grobid_json(raw) assert re - assert re.title == "PEMBELAJARAN FISIKA DENGAN PENDEKATAN KETERAMPILAN PROSES MELALUI METODE DEMONSTRASI MENGGUNAKAN MEDIA RIIL DAN MEDIA VIRTUIL DITINJAU DARI MOTIVASI DAN GAYA BERFIKIR SISWA" + assert ( + re.title + == "PEMBELAJARAN FISIKA DENGAN PENDEKATAN KETERAMPILAN PROSES MELALUI METODE DEMONSTRASI MENGGUNAKAN MEDIA RIIL DAN MEDIA VIRTUIL DITINJAU DARI MOTIVASI DAN GAYA BERFIKIR SISWA" + ) assert len(re.contribs) == 5 print(re.contribs) assert re.contribs[0].raw_name == "Wahyu Ary" @@ -31,19 +34,21 @@ def test_grobid_metadata_parse(grobid_metadata_importer): assert re.contribs[0].surname == "Ary" assert re.publisher is None if re.extra: - assert re.extra.get('container_name') is None + assert re.extra.get("container_name") is None assert len(re.refs) == 27 + def test_file_metadata_parse(grobid_metadata_importer): - with open('tests/files/example_grobid_metadata_lines.tsv', 'r') as f: + with open("tests/files/example_grobid_metadata_lines.tsv", "r") as f: f.readline() - raw = f.readline().split('\t') + raw = f.readline().split("\t") # randomize sha1 so tests are repeatable - random_sha1 = "sha1:{}".format(base64.b32encode(os.urandom(20)).decode('utf-8').upper()) + random_sha1 = "sha1:{}".format(base64.b32encode(os.urandom(20)).decode("utf-8").upper()) fe = grobid_metadata_importer.parse_file_metadata( - random_sha1, json.loads(raw[1]), raw[2], int(raw[3])) + random_sha1, json.loads(raw[1]), raw[2], int(raw[3]) + ) assert fe - #assert fe.sha1 == "d4a841744719518bf8bdd5d91576ccedc55efbb5" # "sha1:2SUEC5CHDFIYX6F52XMRK5WM5XCV565V" + # assert fe.sha1 == "d4a841744719518bf8bdd5d91576ccedc55efbb5" # "sha1:2SUEC5CHDFIYX6F52XMRK5WM5XCV565V" assert fe.md5 is None assert fe.mimetype == "application/pdf" assert fe.size == 142710 @@ -53,29 +58,30 @@ def test_file_metadata_parse(grobid_metadata_importer): assert fe.urls[0].rel == "webarchive" assert len(fe.release_ids) == 0 + 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: + with open("tests/files/example_grobid_metadata_lines.tsv", "r") as f: grobid_metadata_importer.bezerk_mode = True counts = LinePusher(grobid_metadata_importer, f).run() - assert counts['insert'] == 10 - assert counts['inserted.release'] == 10 - assert counts['exists'] == 0 - assert counts['skip'] == 0 + 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) + change = grobid_metadata_importer.api.get_changelog_entry(index=last_index + 1) eg = change.editgroup assert eg.description assert "grobid" in eg.description.lower() - assert eg.extra['git_rev'] - assert "fatcat_tools.GrobidMetadataImporter" in eg.extra['agent'] + 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: + 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 + assert counts["insert"] == 0 + assert counts["inserted.release"] == 0 + assert counts["exists"] == 10 + assert counts["skip"] == 0 |