mirrors/ArchiveBox
1
0
Fork 0
mirror of https://github.com/ArchiveBox/ArchiveBox.git synced 2025-05-16 16:14:28 -04:00
Code Issues Projects Releases Packages Wiki Activity Actions

refactor: All parsers return snapshot instead of link

Browse source
This commit is contained in:
Cristian 2021-01-04 09:31:14 -05:00
parent 15d88be229
commit b8efaa5b6a
11 changed files with 72 additions and 53 deletions
Download patch file Download diff file Expand all files Collapse all files

12
archivebox/parsers/pocket_html.py
View file

@ -6,7 +6,8 @@ import re
from typing import IO, Iterable
from datetime import datetime
from ..index.schema import Link
from django.db.models import Model
from ..util import (
htmldecode,
enforce_types,
@ -14,8 +15,9 @@ from ..util import (
@enforce_types
def parse_pocket_html_export(html_file: IO[str], **_kwargs) -> Iterable[Link]:
def parse_pocket_html_export(html_file: IO[str], **_kwargs) -> Iterable[Model]:
"""Parse Pocket-format bookmarks export files (produced by getpocket.com/export/)"""
from core.models import Snapshot
html_file.seek(0)
pattern = re.compile("^\\s*<li><a href=\"(.+)\" time_added=\"(\\d+)\" tags=\"(.*)\">(.+)</a></li>", re.UNICODE)
@ -29,10 +31,10 @@ def parse_pocket_html_export(html_file: IO[str], **_kwargs) -> Iterable[Link]:
tags = match.group(3)
title = match.group(4).replace(' — Readability', '').replace('http://www.readability.com/read?url=', '')
yield Link(
yield Snapshot(
url=htmldecode(url),
timestamp=str(time.timestamp()),
title=htmldecode(title) or None,
tags=tags or '',
sources=[html_file.name],
#tags=tags or '',
#sources=[html_file.name],
)