mirror of
https://github.com/bellingcat/auto-archiver.git
synced 2026-06-12 13:18:28 +03:00
Merge main
This commit is contained in:
@@ -1 +1 @@
|
||||
from .local_storage import LocalStorage
|
||||
from .local_storage import LocalStorage
|
||||
|
||||
@@ -13,11 +13,15 @@
|
||||
},
|
||||
"filename_generator": {
|
||||
"default": "static",
|
||||
"help": "how to name stored files: 'random' creates a random string; 'static' uses a replicable strategy such as a hash.",
|
||||
"help": "how to name stored files: 'random' creates a random string; 'static' uses a hash, with the settings of the 'hash_enricher' module (defaults to SHA256 if not enabled)",
|
||||
"choices": ["random", "static"],
|
||||
},
|
||||
"save_to": {"default": "./local_archive", "help": "folder where to save archived content"},
|
||||
"save_absolute": {"default": True, "help": "whether the path to the stored file is absolute or relative in the output result inc. formatters (Warning: saving an absolute path will show your computer's file structure)"},
|
||||
"save_absolute": {
|
||||
"default": False,
|
||||
"type": "bool",
|
||||
"help": "whether the path to the stored file is absolute or relative in the output result inc. formatters (Warning: saving an absolute path will show your computer's file structure)",
|
||||
},
|
||||
},
|
||||
"description": """
|
||||
LocalStorage: A storage module for saving archived content locally on the filesystem.
|
||||
@@ -31,5 +35,5 @@
|
||||
### Notes
|
||||
- Default storage folder is `./archived`, but this can be changed via the `save_to` configuration.
|
||||
- The `save_absolute` option can reveal the file structure in output formats; use with caution.
|
||||
"""
|
||||
""",
|
||||
}
|
||||
|
||||
@@ -1,4 +1,3 @@
|
||||
|
||||
import shutil
|
||||
from typing import IO
|
||||
import os
|
||||
@@ -6,25 +5,43 @@ from loguru import logger
|
||||
|
||||
from auto_archiver.core import Media
|
||||
from auto_archiver.core import Storage
|
||||
from auto_archiver.core.consts import SetupError
|
||||
|
||||
|
||||
class LocalStorage(Storage):
|
||||
def setup(self) -> None:
|
||||
if len(self.save_to) > 200:
|
||||
raise SetupError(
|
||||
"Your save_to path is too long, this will cause issues saving files on your computer. Please use a shorter path."
|
||||
)
|
||||
|
||||
def get_cdn_url(self, media: Media) -> str:
|
||||
# TODO: is this viable with Storage.configs on path/filename?
|
||||
dest = os.path.join(self.save_to, media.key)
|
||||
dest = media.key
|
||||
|
||||
if self.save_absolute:
|
||||
dest = os.path.abspath(dest)
|
||||
return dest
|
||||
|
||||
def set_key(self, media, url, metadata):
|
||||
# clarify we want to save the file to the save_to folder
|
||||
|
||||
old_folder = metadata.get("folder", "")
|
||||
metadata.set_context("folder", os.path.join(self.save_to, metadata.get("folder", "")))
|
||||
super().set_key(media, url, metadata)
|
||||
# don't impact other storages that might want a different 'folder' set
|
||||
metadata.set_context("folder", old_folder)
|
||||
|
||||
def upload(self, media: Media, **kwargs) -> bool:
|
||||
# override parent so that we can use shutil.copy2 and keep metadata
|
||||
dest = os.path.join(self.save_to, media.key)
|
||||
dest = media.key
|
||||
|
||||
os.makedirs(os.path.dirname(dest), exist_ok=True)
|
||||
logger.debug(f'[{self.__class__.__name__}] storing file {media.filename} with key {media.key} to {dest}')
|
||||
logger.debug(f"[{self.__class__.__name__}] storing file {media.filename} with key {media.key} to {dest}")
|
||||
|
||||
res = shutil.copy2(media.filename, dest)
|
||||
logger.info(res)
|
||||
return True
|
||||
|
||||
# must be implemented even if unused
|
||||
def uploadf(self, file: IO[bytes], key: str, **kwargs: dict) -> bool: pass
|
||||
def uploadf(self, file: IO[bytes], key: str, **kwargs: dict) -> bool:
|
||||
pass
|
||||
|
||||
Reference in New Issue
Block a user