refactor old python hadoop code into new directory

1 files changed, 22 insertions, 0 deletions

diff --git a/python_hadoop/tests/test_grobid2json.py b/python_hadoop/tests/test_grobid2json.py
new file mode 100644
index 0000000..8497b10
--- /dev/null
+++ b/python_hadoop/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>")