From 132d257befc2088a98b3fe10aa71713338d15673 Mon Sep 17 00:00:00 2001 From: Bryan Newbold Date: Wed, 3 Nov 2021 20:30:02 -0700 Subject: add test for XML parse exception behavior --- tests/files/citation_list/parse_error.tei.xml | 64 +++++++++++++++++++++++++++ tests/test_errors.py | 23 ++++++++++ 2 files changed, 87 insertions(+) create mode 100644 tests/files/citation_list/parse_error.tei.xml create mode 100644 tests/test_errors.py (limited to 'tests') diff --git a/tests/files/citation_list/parse_error.tei.xml b/tests/files/citation_list/parse_error.tei.xml new file mode 100644 index 0000000..fc7b4fd --- /dev/null +++ b/tests/files/citation_list/parse_error.tei.xml @@ -0,0 +1,64 @@ + + + + + + +
+ + + + + Design and Analysis of an Optimi-zation Model by using Scheduling Algorithm for Electric Power Cycles + + SMaheswari + + + CVijayalakshmi + + + + Proceedings of the National Conference On Applied Ma-thematics + + BSAbdur Rahman University + Chennai + + the National Conference On Applied Ma-thematics + + NCAM 2010. January 2010 + + + + Maheswari, S., Vijayalakshmi, C.: Design and Analysis of an Optimi-zation Model by using Scheduling Algorithm for Electric Power Cycles. In: Proceedings of the National Conference On Applied Ma-thematics (NCAM 2010), B.S. Abdur Rahman University Chennai, pp. 160–163 (January 2010) + + + + + Optimization Model for Electricity Distribution System Control using Communication System by La-grangian Relaxation Technique + + SMaheswari + + + CVijayalakshmi + + + + CiiT International Journal of Wireless Communication + 0974 - 9756 & Online: 0974 - 9640 + + 3 + 3 + + 2011 + + + Print: + Maheswari, S., Vijayalakshmi, C.: Optimization Model for Electricity Distribution System Control using Communication System by La-grangian Relaxation Technique. CiiT International Journal of Wireless Communication 3(3), 183–187 (2011) (Print: ISSN 0974 – 9756 & Online: ISSN 0974 – 9640) + + + +
+
+
+
diff --git a/tests/test_errors.py b/tests/test_errors.py new file mode 100644 index 0000000..8537551 --- /dev/null +++ b/tests/test_errors.py @@ -0,0 +1,23 @@ +import xml.etree.ElementTree + +import pytest + +from grobid_tei_xml import parse_citation_list_xml + + +def test_doc_parse_error() -> None: + """ + This XML document has a bare '&' (should be '&') and results in a parse + error. + + See also: https://github.com/kermitt2/grobid/issues/848 + + The intent of this test is to ensure that the exception raised is the one + expected, especially if that behavior changes in the future. + """ + + with open("tests/files/citation_list/parse_error.tei.xml", "r") as f: + tei_xml = f.read() + + with pytest.raises(xml.etree.ElementTree.ParseError): + parse_citation_list_xml(tei_xml) -- cgit v1.2.3