From 689da76d1c759d6368d760b4a1fa942e16095a40 Mon Sep 17 00:00:00 2001 From: Bryan Newbold Date: Wed, 15 Jan 2020 14:13:34 -0800 Subject: ingest: improve tests, support old ingest results --- python/tests/import_ingest.py | 16 ++++++++++++++++ 1 file changed, 16 insertions(+) (limited to 'python/tests/import_ingest.py') diff --git a/python/tests/import_ingest.py b/python/tests/import_ingest.py index 7c0a85cd..5089d99e 100644 --- a/python/tests/import_ingest.py +++ b/python/tests/import_ingest.py @@ -56,3 +56,19 @@ def test_ingest_dict_parse(ingest_importer): if u.rel == "webarchive": assert u.url.startswith("https://web.archive.org/") assert len(f.release_ids) == 1 + +def test_ingest_dict_parse_old(ingest_importer): + with open('tests/files/example_ingest.old.json', 'r') as f: + raw = json.loads(f.readline()) + f = ingest_importer.parse_record(raw) + assert f.sha1 == "00242a192acc258bdfdb151943419437f440c313" + assert f.md5 == "f4de91152c7ab9fdc2a128f962faebff" + assert f.mimetype == "application/pdf" + assert f.size == 255629 + assert len(f.urls) == 2 + for u in f.urls: + if u.rel == "web": + assert u.url.startswith("http://journals.plos.org") + if u.rel == "webarchive": + assert u.url.startswith("https://web.archive.org/") + assert len(f.release_ids) == 1 -- cgit v1.2.3