browser: parse browser history using browserexport (#216)

* browser: parse browser history using browserexport

from seanbreckenridge/HPI module:
1fba8ccf2f/my/browser/export.py
This commit is contained in:
seanbreckenridge 2022-02-13 15:56:05 -08:00 committed by GitHub
parent 059c4ae791
commit 9e5cd60ff2
No known key found for this signature in database
GPG key ID: 4AEE18F83AFDEB23
7 changed files with 198 additions and 24 deletions

50
my/browser/export.py Normal file
View file

@ -0,0 +1,50 @@
"""
Parses browser history using [[http://github.com/seanbreckenridge/browserexport][browserexport]]
"""
REQUIRES = ["browserexport"]
from my.config import browser as user_config
from my.core import Paths, dataclass
@dataclass
class config(user_config.export):
# path[s]/glob to your backed up browser history sqlite files
export_path: Paths
from pathlib import Path
from typing import Iterator, Sequence, List
from my.core import Stats, get_files, LazyLogger
from my.core.common import mcachew
from browserexport.merge import read_and_merge, Visit
from .common import _patch_browserexport_logs
logger = LazyLogger(__name__, level="warning")
_patch_browserexport_logs()
# all of my backed up databases
def inputs() -> Sequence[Path]:
return get_files(config.export_path)
def _cachew_depends_on() -> List[str]:
return [str(f) for f in inputs()]
@mcachew(depends_on=_cachew_depends_on, logger=logger)
def history() -> Iterator[Visit]:
yield from read_and_merge(inputs())
def stats() -> Stats:
from my.core import stat
return {**stat(history)}