diff options
author | Bryan Newbold <bnewbold@archive.org> | 2020-06-03 23:16:07 -0700 |
---|---|---|
committer | Bryan Newbold <bnewbold@archive.org> | 2020-06-03 23:16:07 -0700 |
commit | 62fde540ef9c38d403ea89a6fb1db51bfba23da8 (patch) | |
tree | 2142c739349237b0ef137a57000850db1b2ff39d /fatcat_scholar/sim_pipeline.py | |
parent | 74ef1c6f4e0f08dd6a6b3e6eacc4d780c990eb3f (diff) | |
download | fatcat-scholar-62fde540ef9c38d403ea89a6fb1db51bfba23da8.tar.gz fatcat-scholar-62fde540ef9c38d403ea89a6fb1db51bfba23da8.zip |
flake8 fixes (partial)
Diffstat (limited to 'fatcat_scholar/sim_pipeline.py')
-rw-r--r-- | fatcat_scholar/sim_pipeline.py | 17 |
1 files changed, 4 insertions, 13 deletions
diff --git a/fatcat_scholar/sim_pipeline.py b/fatcat_scholar/sim_pipeline.py index b84ac47..cfc197f 100644 --- a/fatcat_scholar/sim_pipeline.py +++ b/fatcat_scholar/sim_pipeline.py @@ -1,24 +1,15 @@ -import os import io import sys import sqlite3 import argparse +from typing import List, Dict, Optional, Any + import requests -from pydantic import BaseModel, validator -from typing import List, Dict, Tuple, Optional, Any, Sequence -from fatcat_openapi_client import ReleaseEntity, FileEntity import internetarchive -from fatcat_scholar.api_entities import * from fatcat_scholar.djvu import djvu_extract_leaf_texts -from fatcat_scholar.sandcrawler import ( - SandcrawlerPostgrestClient, - SandcrawlerMinioClient, -) -from fatcat_scholar.issue_db import IssueDB, SimIssueRow +from fatcat_scholar.issue_db import IssueDB from fatcat_scholar.schema import ( - es_biblio_from_release, - es_release_from_release, DocType, IntermediateBundle, ) @@ -57,7 +48,7 @@ class SimPipeline: def fetch_sim_issue(self, issue_db_row: Any) -> Optional[Any]: """ - issue_item + issue_item pages: str page_texts: list raw_text |