Files
linkding/bookmarks/services/assets.py
Sascha Ißbrücker 2d81ea6f6e Add REST endpoint for uploading snapshots from the Singlefile extension (#996)
* Extract asset logic

* Allow disabling HTML snapshot when creating bookmark

* Add endpoint for uploading singlefile snapshots

* Add URL parameter to disable HTML snapshots

* Allow using asset list in base Docker image

* Expose app version through profile
2025-02-23 22:58:14 +01:00

129 lines
4.3 KiB
Python

import gzip
import logging
import os
import shutil
from django.conf import settings
from django.core.files.uploadedfile import UploadedFile
from django.utils import timezone, formats
from bookmarks.models import Bookmark, BookmarkAsset
from bookmarks.services import singlefile
MAX_ASSET_FILENAME_LENGTH = 192
logger = logging.getLogger(__name__)
def create_snapshot_asset(bookmark: Bookmark) -> BookmarkAsset:
date_created = timezone.now()
timestamp = formats.date_format(date_created, "SHORT_DATE_FORMAT")
asset = BookmarkAsset(
bookmark=bookmark,
asset_type=BookmarkAsset.TYPE_SNAPSHOT,
date_created=date_created,
content_type=BookmarkAsset.CONTENT_TYPE_HTML,
display_name=f"HTML snapshot from {timestamp}",
status=BookmarkAsset.STATUS_PENDING,
)
return asset
def create_snapshot(asset: BookmarkAsset):
try:
# Create snapshot into temporary file
temp_filename = _generate_asset_filename(asset, asset.bookmark.url, "tmp")
temp_filepath = os.path.join(settings.LD_ASSET_FOLDER, temp_filename)
singlefile.create_snapshot(asset.bookmark.url, temp_filepath)
# Store as gzip in asset folder
filename = _generate_asset_filename(asset, asset.bookmark.url, "html.gz")
filepath = os.path.join(settings.LD_ASSET_FOLDER, filename)
with open(temp_filepath, "rb") as temp_file, gzip.open(
filepath, "wb"
) as gz_file:
shutil.copyfileobj(temp_file, gz_file)
# Remove temporary file
os.remove(temp_filepath)
asset.status = BookmarkAsset.STATUS_COMPLETE
asset.file = filename
asset.gzip = True
asset.save()
except Exception as error:
asset.status = BookmarkAsset.STATUS_FAILURE
asset.save()
raise error
def upload_snapshot(bookmark: Bookmark, html: bytes):
asset = create_snapshot_asset(bookmark)
filename = _generate_asset_filename(asset, asset.bookmark.url, "html.gz")
filepath = os.path.join(settings.LD_ASSET_FOLDER, filename)
with gzip.open(filepath, "wb") as gz_file:
gz_file.write(html)
# Only save the asset if the file was written successfully
asset.status = BookmarkAsset.STATUS_COMPLETE
asset.file = filename
asset.gzip = True
asset.save()
return asset
def upload_asset(bookmark: Bookmark, upload_file: UploadedFile):
try:
asset = BookmarkAsset(
bookmark=bookmark,
asset_type=BookmarkAsset.TYPE_UPLOAD,
date_created=timezone.now(),
content_type=upload_file.content_type,
display_name=upload_file.name,
status=BookmarkAsset.STATUS_COMPLETE,
gzip=False,
)
name, extension = os.path.splitext(upload_file.name)
filename = _generate_asset_filename(asset, name, extension.lstrip("."))
filepath = os.path.join(settings.LD_ASSET_FOLDER, filename)
with open(filepath, "wb") as f:
for chunk in upload_file.chunks():
f.write(chunk)
asset.file = filename
asset.file_size = upload_file.size
asset.save()
logger.info(
f"Successfully uploaded asset file. bookmark={bookmark} file={upload_file.name}"
)
return asset
except Exception as e:
logger.error(
f"Failed to upload asset file. bookmark={bookmark} file={upload_file.name}",
exc_info=e,
)
raise e
def _generate_asset_filename(
asset: BookmarkAsset, filename: str, extension: str
) -> str:
def sanitize_char(char):
if char.isalnum() or char in ("-", "_", "."):
return char
else:
return "_"
formatted_datetime = asset.date_created.strftime("%Y-%m-%d_%H%M%S")
sanitized_filename = "".join(sanitize_char(char) for char in filename)
# Calculate the length of fixed parts of the final filename
non_filename_length = len(f"{asset.asset_type}_{formatted_datetime}_.{extension}")
# Calculate the maximum length for the dynamic part of the filename
max_filename_length = MAX_ASSET_FILENAME_LENGTH - non_filename_length
# Truncate the filename if necessary
sanitized_filename = sanitized_filename[:max_filename_length]
return f"{asset.asset_type}_{formatted_datetime}_{sanitized_filename}.{extension}"