summaryrefslogtreecommitdiffstats
path: root/python/tests/matched_importer_test.py
diff options
context:
space:
mode:
authorBryan Newbold <bnewbold@robocracy.org>2018-11-12 23:37:28 -0800
committerBryan Newbold <bnewbold@robocracy.org>2018-11-12 23:37:28 -0800
commit7ebda2e051b51e49544ab75673b19ec5f27d9d45 (patch)
tree671fd995b2fdb0b9ec69cfcb0e7771f43c13d32d /python/tests/matched_importer_test.py
parent889606ffac5096610e6123134cb89423c15ef464 (diff)
downloadfatcat-7ebda2e051b51e49544ab75673b19ec5f27d9d45.tar.gz
fatcat-7ebda2e051b51e49544ab75673b19ec5f27d9d45.zip
more python module refactoring
Diffstat (limited to 'python/tests/matched_importer_test.py')
-rw-r--r--python/tests/matched_importer_test.py32
1 files changed, 32 insertions, 0 deletions
diff --git a/python/tests/matched_importer_test.py b/python/tests/matched_importer_test.py
new file mode 100644
index 00000000..4042eabb
--- /dev/null
+++ b/python/tests/matched_importer_test.py
@@ -0,0 +1,32 @@
+
+import json
+import pytest
+from fatcat_tools.matched_importer import FatcatMatchedImporter
+
+
+@pytest.fixture(scope="function")
+def matched_importer():
+ yield FatcatMatchedImporter("http://localhost:9411/v0")
+
+# TODO: use API to check that entities actually created...
+def test_matched_importer_batch(matched_importer):
+ with open('tests/files/example_matched.json', 'r') as f:
+ matched_importer.process_batch(f)
+
+def test_matched_importer(matched_importer):
+ with open('tests/files/example_matched.json', 'r') as f:
+ matched_importer.process_source(f)
+
+def test_matched_dict_parse(matched_importer):
+ with open('tests/files/example_matched.json', 'r') as f:
+ raw = json.loads(f.readline())
+ f = matched_importer.parse_matched_dict(raw)
+ assert f.sha1 == "00242a192acc258bdfdb151943419437f440c313"
+ assert f.md5 == "f4de91152c7ab9fdc2a128f962faebff"
+ assert f.mimetype == "application/pdf"
+ assert f.size == 255629
+ assert f.urls[1].url.startswith("http://journals.plos.org")
+ assert f.urls[1].rel == "web"
+ assert f.urls[0].url.startswith("https://web.archive.org/")
+ assert f.urls[0].rel == "webarchive"
+ assert len(f.releases) == 1