diff options
author | Bryan Newbold <bnewbold@archive.org> | 2021-09-30 15:09:42 -0700 |
---|---|---|
committer | Bryan Newbold <bnewbold@archive.org> | 2021-10-15 18:15:04 -0700 |
commit | 86107e39b761e5b799562af662219fda04ade1be (patch) | |
tree | 9a9801fc1d8a583a792457c2b8468ce6dbea22f1 /python/sandcrawler/__init__.py | |
parent | 7430ddbcdec76091220de474060b968f0ef1bb70 (diff) | |
download | sandcrawler-86107e39b761e5b799562af662219fda04ade1be.tar.gz sandcrawler-86107e39b761e5b799562af662219fda04ade1be.zip |
refactoring; progress on filesets
Diffstat (limited to 'python/sandcrawler/__init__.py')
-rw-r--r-- | python/sandcrawler/__init__.py | 3 |
1 files changed, 2 insertions, 1 deletions
diff --git a/python/sandcrawler/__init__.py b/python/sandcrawler/__init__.py index e461462..724a39c 100644 --- a/python/sandcrawler/__init__.py +++ b/python/sandcrawler/__init__.py @@ -4,7 +4,8 @@ from .pdftrio import PdfTrioClient, PdfTrioWorker, PdfTrioBlobWorker from .misc import gen_file_metadata, b32_hex, parse_cdx_line, parse_cdx_datetime, clean_url from .workers import KafkaSink, KafkaCompressSink, JsonLinePusher, CdxLinePusher, CdxLinePusher, KafkaJsonPusher, BlackholeSink, ZipfilePusher, MultiprocessWrapper from .ia import WaybackClient, WaybackError, WaybackContentError, CdxApiClient, CdxApiError, SavePageNowClient, SavePageNowError, PetaboxError, ResourceResult, WarcResource, CdxPartial, CdxRow -from .ingest import IngestFileWorker +from .ingest_file import IngestFileWorker +from .ingest_fileset import IngestFilesetWorker from .persist import PersistCdxWorker, PersistIngestFileResultWorker, PersistGrobidWorker, PersistGrobidDiskWorker, PersistPdfTrioWorker, PersistIngestRequestWorker, PersistPdfTextWorker, PersistThumbnailWorker from .db import SandcrawlerPostgrestClient, SandcrawlerPostgresClient from .pdfextract import PdfExtractWorker, PdfExtractBlobWorker |