diff options
| author | Bryan Newbold <bnewbold@archive.org> | 2021-10-26 16:59:32 -0700 |
|---|---|---|
| committer | Bryan Newbold <bnewbold@archive.org> | 2021-10-26 16:59:32 -0700 |
| commit | 4a46f166f8514b5620d2bcb13a5c5f3e6cee66c8 (patch) | |
| tree | a15635b11ca66d5cdbbc1c3f6eaa73fd5fe35801 /python/sandcrawler/ingest_file.py | |
| parent | f08bbeb7981fd692ffc9277d15d282883a408051 (diff) | |
| download | sandcrawler-4a46f166f8514b5620d2bcb13a5c5f3e6cee66c8.tar.gz sandcrawler-4a46f166f8514b5620d2bcb13a5c5f3e6cee66c8.zip | |
more progress on type annotations and linting
Diffstat (limited to 'python/sandcrawler/ingest_file.py')
| -rw-r--r-- | python/sandcrawler/ingest_file.py | 2 |
1 files changed, 2 insertions, 0 deletions
diff --git a/python/sandcrawler/ingest_file.py b/python/sandcrawler/ingest_file.py index bc8643b..281c6d3 100644 --- a/python/sandcrawler/ingest_file.py +++ b/python/sandcrawler/ingest_file.py @@ -364,6 +364,8 @@ class IngestFileWorker(SandcrawlerWorker): # Need to actually processes result = process_pdf(resource.body) + assert result.sha1hex == file_meta['sha1hex'] + assert result.file_meta is not None assert result.file_meta['sha1hex'] == file_meta['sha1hex'] if self.thumbnail_sink and result.page0_thumbnail is not None: self.thumbnail_sink.push_record(result.page0_thumbnail, key=result.sha1hex) |