mirrors/ArchiveBox
1
0
Fork 0
mirror of https://github.com/ArchiveBox/ArchiveBox.git synced 2025-05-23 03:06:55 -04:00
Code Issues Projects Releases Packages Wiki Activity Actions

refactor: status command is functional

Browse source
This commit is contained in:
Cristian 2020-12-30 12:53:20 -05:00
parent d92083b928
commit 973f8b6abc
5 changed files with 33 additions and 26 deletions
Download patch file Download diff file Expand all files Collapse all files

6
archivebox/index/__init__.py
View file

@ -42,7 +42,7 @@ from .html import (
write_html_snapshot_details,
)
from .json import (
load_json_snapshot_details,
load_json_snapshot,
parse_json_snapshot_details,
write_json_snapshot_details,
)
@ -441,7 +441,7 @@ def get_duplicate_folders(snapshots, out_dir: Path=OUTPUT_DIR) -> Dict[str, Opti
path = path.snapshot_dir
try:
snapshot = parse_json_snapshot_details(path)
snapshot = load_json_snapshot_details(path)
except Exception:
pass
@ -530,7 +530,7 @@ def is_valid(snapshot: Model) -> bool:
if dir_exists and index_exists:
try:
# TODO: review if the `guess` was necessary here
parsed_snapshot = parse_json_snapshot_details(snapshot.snapshot_dir)
parsed_snapshot = load_json_snapshot(snapshot.snapshot_dir)
return snapshot.url == parsed_snapshot.url
except Exception:
pass

11
archivebox/index/json.py
View file

@ -91,17 +91,18 @@ def write_json_snapshot_details(snapshot: Model, out_dir: Optional[str]=None) ->
@enforce_types
def load_json_snapshot_details(out_dir: Path) -> Optional[Model]:
def load_json_snapshot(out_dir: Path) -> Optional[Model]:
"""
Loads the detail from the local json index
"""
from core.models import Snapshot
existing_index = Path(out_dir) / JSON_INDEX_FILENAME
if existing_index.exists():
with open(existing_index, 'r', encoding='utf-8') as f:
try:
output = pyjson.load(f)
if "history" not in output.keys():
output["history"] = {}
output = Snapshot.from_json(output)
return output
except pyjson.JSONDecodeError:
pass
@ -110,13 +111,13 @@ def load_json_snapshot_details(out_dir: Path) -> Optional[Model]:
@enforce_types
def parse_json_snapshot_details(out_dir: Union[Path, str]) -> Iterator[dict]:
"""read through all the archive data folders and return the parsed links"""
"""read through all the archive data folders and return the parsed snapshots"""
for entry in os.scandir(Path(out_dir)):
if entry.is_dir(follow_symlinks=True):
if (Path(entry.path) / 'index.json').exists():
try:
snapshot_details = load_snapshot_details(entry.path)
snapshot_details = load_json_snapshot_details(entry.path)
except KeyError:
snapshot_details = None
if snapshot_details: