mirror of
https://github.com/ArchiveBox/ArchiveBox
synced 2024-11-21 19:53:06 +00:00
29 lines
690 B
Python
29 lines
690 B
Python
__package__ = 'archivebox.parsers'
|
|
|
|
import json
|
|
from typing import IO, Iterable
|
|
|
|
from archivebox.misc.util import enforce_types
|
|
|
|
from ..index.schema import Link
|
|
from .generic_json import jsonObjectToLink
|
|
|
|
def parse_line(line: str):
|
|
if line.strip() != "":
|
|
return json.loads(line)
|
|
|
|
@enforce_types
|
|
def parse_generic_jsonl_export(json_file: IO[str], **_kwargs) -> Iterable[Link]:
|
|
"""Parse JSONL format bookmarks export files"""
|
|
|
|
json_file.seek(0)
|
|
|
|
links = [ parse_line(line) for line in json_file ]
|
|
|
|
for link in links:
|
|
if link:
|
|
yield jsonObjectToLink(link,json_file.name)
|
|
|
|
KEY = 'jsonl'
|
|
NAME = 'Generic JSONL'
|
|
PARSER = parse_generic_jsonl_export
|