This commit is contained in:
msramalho
2022-06-07 18:41:58 +02:00
parent d46b8e1157
commit f87acb6d1d
13 changed files with 187 additions and 157 deletions

View File

@@ -1,24 +1,23 @@
import os, time
from loguru import logger
from .base_storage import Storage
from dataclasses import dataclass
from googleapiclient.discovery import build
from googleapiclient.http import MediaFileUpload
from google.oauth2 import service_account
import time
@dataclass
class GDConfig:
root_folder_id: str
default_folder: str = "default"
folder: str = "default"
service_account: str = "service_account.json"
class GDStorage(Storage):
def __init__(self, config: GDConfig):
self.default_folder = config.default_folder
self.folder = config.folder
self.root_folder_id = config.root_folder_id
creds = service_account.Credentials.from_service_account_file(
config.service_account, scopes=['https://www.googleapis.com/auth/drive'])
@@ -29,77 +28,73 @@ class GDStorage(Storage):
only support files saved in a folder for GD
S3 supports folder and all stored in the root
"""
self.subfolder = self._clean_path(self.subfolder, self.default_folder, False)
filename = key
logger.debug(f'Looking for {self.subfolder} and filename: {filename} on GD')
folder_id = self._get_id_from_parent_and_name(self.root_folder_id, self.subfolder, 5, 10)
# check for sub folder in file youtube_dl_abcde/index.html, needed for thumbnails
# a='youtube_dl_abcde', b='index.html'
a, _, b = filename.partition('/')
if b != '':
logger.debug(f'get_cdn_url: Found a subfolder so need to split on: {a=} and {b=}')
folder_id = self._get_id_from_parent_and_name(folder_id, a, use_mime_type=True)
filename = b
full_name = os.path.join(self.folder, key)
parent_id, folder_id = self.root_folder_id, None
path_parts = full_name.split(os.path.sep)
filename = path_parts[-1]
logger.info(f"looking for folders for {path_parts=} before uploading {filename=}")
for folder in path_parts[0:-1]:
folder_id = self._get_id_from_parent_and_name(parent_id, folder, use_mime_type=True, raise_on_missing=True)
parent_id = folder_id
# get id of file inside folder (or sub folder)
file_id = self._get_id_from_parent_and_name(folder_id, filename)
return f"https://drive.google.com/file/d/{file_id}/view?usp=sharing"
def exists(self, _key):
# TODO: How to check for google drive, as it accepts different names?
return False
def exists(self, key):
try:
self.get_cdn_url(key)
return True
except: return False
def uploadf(self, file, key, **_kwargs):
def uploadf(self, file: str, key: str, **_kwargs):
"""
1. check if subfolder exists or create it
2. check if key contains sub-subfolder, check if exists or create it
3. upload file to root_id/subfolder[/sub-subfolder]/filename
1. for each sub-folder in the path check if exists or create
2. upload file to root_id/other_paths.../filename
"""
self.subfolder = self._clean_path(self.subfolder, GDStorage.DEFAULT_UPLOAD_FOLDER_NAME, False)
filename = key
# get id of subfolder or create if it does not exist
folder_id_to_upload_to = self._get_id_from_parent_and_name(self.root_folder_id, self.subfolder, use_mime_type=True, raise_on_missing=False)
if folder_id_to_upload_to is None:
folder_id_to_upload_to = self._mkdir(self.subfolder, self.root_folder_id)
# check for sub folder in file youtube_dl_abcde/index.html, needed for thumbnails
# a='youtube_dl_abcde', b='index.html'
a, _, b = filename.partition('/')
if b != '':
logger.debug(f'uploadf: Found a subfolder so need to split on: {a=} and {b=}')
# get id of subfolder or create if it does not exist
sub_folder_id_to_upload_to = self._get_id_from_parent_and_name(folder_id_to_upload_to, a, use_mime_type=True, raise_on_missing=False)
if sub_folder_id_to_upload_to is None:
sub_folder_id_to_upload_to = self._mkdir(a, folder_id_to_upload_to)
filename = b
folder_id_to_upload_to = sub_folder_id_to_upload_to
full_name = os.path.join(self.folder, key)
parent_id, upload_to = self.root_folder_id, None
path_parts = full_name.split(os.path.sep)
filename = path_parts[-1]
logger.info(f"checking folders {path_parts[0:-1]} exist (or creating) before uploading {filename=}")
for folder in path_parts[0:-1]:
upload_to = self._get_id_from_parent_and_name(parent_id, folder, use_mime_type=True, raise_on_missing=False)
if upload_to is None:
upload_to = self._mkdir(folder, parent_id)
parent_id = upload_to
# upload file to gd
logger.debug(f'uploading {filename=} to folder id {upload_to}')
file_metadata = {
'name': [filename],
'parents': [folder_id_to_upload_to]
'parents': [upload_to]
}
media = MediaFileUpload(file, resumable=True)
gd_file = self.service.files().create(body=file_metadata, media_body=media, fields='id').execute()
logger.debug(f'uploadf: uploaded file {gd_file["id"]} succesfully in folder={folder_id_to_upload_to}')
logger.debug(f'uploadf: uploaded file {gd_file["id"]} succesfully in folder={upload_to}')
def upload(self, filename: str, key: str, **kwargs):
# GD only requires the filename not a file reader
logger.debug(f'[{self.__class__.__name__}] uploading file {filename} with key {key}')
self.uploadf(filename, key, **kwargs)
def _get_id_from_parent_and_name(self, parent_id: str, name: str, retries: int = 1, sleep_seconds: int = 10, use_mime_type: bool = False, raise_on_missing: bool = True):
def _get_id_from_parent_and_name(self, parent_id: str, name: str, retries: int = 1, sleep_seconds: int = 10, use_mime_type: bool = False, raise_on_missing: bool = True, use_cache=True):
"""
Retrieves the id of a folder or file from its @name and the @parent_id folder
Optionally does multiple @retries and sleeps @sleep_seconds between them
If @use_mime_type will restrict search to "mimeType='application/vnd.google-apps.folder'"
If @raise_on_missing will throw error when not found, or returns None
Will remember previous calls to avoid duplication if @use_cache
Returns the id of the file or folder from its name as a string
"""
# cache logic
if use_cache:
self.api_cache = getattr(self, "api_cache", {})
cache_key = f"{parent_id}_{name}_{use_mime_type}"
if cache_key in self.api_cache:
logger.debug(f"cache hit for {cache_key=}")
return self.api_cache[cache_key]
# API logic
debug_header: str = f"[searching {name=} in {parent_id=}]"
query_string = f"'{parent_id}' in parents and name = '{name}' "
if use_mime_type:
@@ -115,10 +110,14 @@ class GDStorage(Storage):
if len(items) > 0:
logger.debug(f"{debug_header} found {len(items)} matches, returning last of {','.join([i['id'] for i in items])}")
return items[-1]['id']
_id = items[-1]['id']
if use_cache: self.api_cache[cache_key] = _id
return _id
else:
logger.debug(f'{debug_header} not found, attempt {attempt+1}/{retries}. sleeping for {sleep_seconds} second(s)')
if attempt < retries - 1: time.sleep(sleep_seconds)
logger.debug(f'{debug_header} not found, attempt {attempt+1}/{retries}.')
if attempt < retries - 1:
logger.debug(f'sleeping for {sleep_seconds} second(s)')
time.sleep(sleep_seconds)
if raise_on_missing:
raise ValueError(f'{debug_header} not found after {retries} attempt(s)')
@@ -129,7 +128,7 @@ class GDStorage(Storage):
Creates a new GDrive folder @name inside folder @parent_id
Returns id of the created folder
"""
logger.debug(f'[_mkdir] Creating new folder with {name=} inside {parent_id=}')
logger.debug(f'Creating new folder with {name=} inside {parent_id=}')
file_metadata = {
'name': [name],
'mimeType': 'application/vnd.google-apps.folder',