diff options
author | Bryan Newbold <bnewbold@archive.org> | 2020-10-30 17:20:22 -0700 |
---|---|---|
committer | Bryan Newbold <bnewbold@archive.org> | 2020-10-30 17:20:22 -0700 |
commit | e61d6e8cc3b6824816a83dff56ffbdbbb6329e57 (patch) | |
tree | 1a59b220e4f984b2a805b23961e4bd8a150c0f47 /python/tests/test_html_ingest.py | |
parent | a8387ac21bf6f9693cef24f9ef39482b9337f3af (diff) | |
download | sandcrawler-e61d6e8cc3b6824816a83dff56ffbdbbb6329e57.tar.gz sandcrawler-e61d6e8cc3b6824816a83dff56ffbdbbb6329e57.zip |
html: work around firstmonday DOCTYPE issue
Diffstat (limited to 'python/tests/test_html_ingest.py')
-rw-r--r-- | python/tests/test_html_ingest.py | 14 |
1 files changed, 14 insertions, 0 deletions
diff --git a/python/tests/test_html_ingest.py b/python/tests/test_html_ingest.py new file mode 100644 index 0000000..96b3883 --- /dev/null +++ b/python/tests/test_html_ingest.py @@ -0,0 +1,14 @@ + +import datetime +import pytest + +from sandcrawler.html_ingest import * + + +def test_html_extract_ojs3() -> None: + + with open('tests/files/first_monday_ojs3_fulltext.html', 'rb') as f: + ojs3_html = f.read() + + fulltext = html_extract_fulltext_teixml(ojs3_html) + assert fulltext['status'] == 'success' |