diff options
author | Bryan Newbold <bnewbold@archive.org> | 2019-12-26 21:32:56 -0800 |
---|---|---|
committer | Bryan Newbold <bnewbold@archive.org> | 2020-01-02 18:12:58 -0800 |
commit | d4f40c4401e479f6db366ea104687938b9d2345e (patch) | |
tree | 34b21bab26150fc61d6f9fd1dde3c628b9a09eb5 /python/sandcrawler | |
parent | d6c744002607d8789927cc98fe6c8a6a76da24bd (diff) | |
download | sandcrawler-d4f40c4401e479f6db366ea104687938b9d2345e.tar.gz sandcrawler-d4f40c4401e479f6db366ea104687938b9d2345e.zip |
fix sandcrawler persist workers
Diffstat (limited to 'python/sandcrawler')
-rw-r--r-- | python/sandcrawler/__init__.py | 1 |
1 files changed, 1 insertions, 0 deletions
diff --git a/python/sandcrawler/__init__.py b/python/sandcrawler/__init__.py index c9cc0c9..236570e 100644 --- a/python/sandcrawler/__init__.py +++ b/python/sandcrawler/__init__.py @@ -4,4 +4,5 @@ from .misc import gen_file_metadata, b32_hex, parse_cdx_line, parse_cdx_datetime from .workers import KafkaSink, KafkaGrobidSink, JsonLinePusher, CdxLinePusher, CdxLinePusher, KafkaJsonPusher, BlackholeSink, ZipfilePusher, MultiprocessWrapper from .ia import WaybackClient, WaybackError, CdxApiClient, CdxApiError, SavePageNowClient, SavePageNowError, SavePageNowRemoteError from .ingest import IngestFileWorker +from .persist import PersistCdxWorker, PersistIngestFileResultWorker, PersistGrobidWorker, PersistGrobidDiskWorker |