From 2f3fd33c33f12b4426072a0279de85ff797611fe Mon Sep 17 00:00:00 2001 From: Bryan Newbold Date: Thu, 21 Oct 2021 13:31:27 -0700 Subject: init project from existing code and tests in sandcrawler, fatcat-scholar, fuzzycat, etc --- tests/test_grobid2json.py | 22 ++++++++++++++++++++++ 1 file changed, 22 insertions(+) create mode 100644 tests/test_grobid2json.py (limited to 'tests/test_grobid2json.py') 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("") -- cgit v1.2.3