summaryrefslogtreecommitdiffstats
path: root/tests/test_grobid2json.py
diff options
context:
space:
mode:
authorBryan Newbold <bnewbold@archive.org>2021-10-21 13:31:27 -0700
committerBryan Newbold <bnewbold@archive.org>2021-10-21 13:31:27 -0700
commit2f3fd33c33f12b4426072a0279de85ff797611fe (patch)
tree089d0eb973ae78c9750ee57239950fe88feba927 /tests/test_grobid2json.py
downloadgrobid_tei_xml-2f3fd33c33f12b4426072a0279de85ff797611fe.tar.gz
grobid_tei_xml-2f3fd33c33f12b4426072a0279de85ff797611fe.zip
init project from existing code and tests in sandcrawler, fatcat-scholar, fuzzycat, etc
Diffstat (limited to 'tests/test_grobid2json.py')
-rw-r--r--tests/test_grobid2json.py22
1 files changed, 22 insertions, 0 deletions
diff --git a/tests/test_grobid2json.py b/tests/test_grobid2json.py
new file mode 100644
index 0000000..8497b10
--- /dev/null
+++ b/tests/test_grobid2json.py
@@ -0,0 +1,22 @@
+
+import xml
+import json
+import pytest
+from grobid2json import *
+
+
+def test_small_xml():
+
+ with open('tests/files/small.xml', 'r') as f:
+ tei_xml = f.read()
+ with open('tests/files/small.json', 'r') as f:
+ json_form = json.loads(f.read())
+
+ assert teixml2json(tei_xml) == json_form
+
+def test_invalid_xml():
+
+ with pytest.raises(xml.etree.ElementTree.ParseError):
+ teixml2json("this is not XML")
+ with pytest.raises(ValueError):
+ teixml2json("<xml></xml>")