diff options
Diffstat (limited to 'python/tests/test_html_ingest.py')
-rw-r--r-- | python/tests/test_html_ingest.py | 10 |
1 files changed, 10 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..ba4acf1 --- /dev/null +++ b/python/tests/test_html_ingest.py @@ -0,0 +1,10 @@ +from sandcrawler.ingest_html 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_body_teixml(ojs3_html) + assert fulltext["status"] == "success" |