diff options
| author | Bryan Newbold <bnewbold@archive.org> | 2020-11-06 18:32:35 -0800 |
|---|---|---|
| committer | Bryan Newbold <bnewbold@archive.org> | 2020-11-06 18:32:35 -0800 |
| commit | 175019c96fced3e21d0f60ea1a4a37da6b8872ac (patch) | |
| tree | f42fbbe9c8ac06ae9eb06373ab9eec96d2b3a177 /python/tests/test_pdfextract.py | |
| parent | b0b66c20c6ffb9d8acc626068964d7dfd5d3bcdc (diff) | |
| parent | 47ca1a273912c8836630b0930b71a4e66fd2c85b (diff) | |
| download | sandcrawler-175019c96fced3e21d0f60ea1a4a37da6b8872ac.tar.gz sandcrawler-175019c96fced3e21d0f60ea1a4a37da6b8872ac.zip | |
Merge branch 'bnewbold-html-ingest'
Diffstat (limited to 'python/tests/test_pdfextract.py')
| -rw-r--r-- | python/tests/test_pdfextract.py | 2 |
1 files changed, 2 insertions, 0 deletions
diff --git a/python/tests/test_pdfextract.py b/python/tests/test_pdfextract.py index ed93341..255e3fb 100644 --- a/python/tests/test_pdfextract.py +++ b/python/tests/test_pdfextract.py @@ -2,6 +2,7 @@ import pytest import struct import responses +import poppler from sandcrawler import PdfExtractWorker, PdfExtractBlobWorker, CdxLinePusher, BlackholeSink, WaybackClient from sandcrawler.pdfextract import process_pdf @@ -20,6 +21,7 @@ def test_process_fake_pdf(): resp = process_pdf(pdf_bytes) assert resp.status == 'not-pdf' +@pytest.mark.skipif(poppler.version_string() == '0.71.0', reason="unsupported version of poppler") def test_process_dummy_pdf(): with open('tests/files/dummy.pdf', 'rb') as f: pdf_bytes = f.read() |