Files
auto-archiver/example.orchestration.yaml
msramalho d6dbdec6ac example
2023-02-09 12:32:55 +00:00

81 lines
2.0 KiB
YAML

steps:
# only 1 feeder allowed
# feeder: cli_feeder # default feeder
feeder: gsheet_feeder # default -> only expects URL from CLI
archivers: # order matters
# - vk_archiver
# - telethon_archiver
# - telegram_archiver
# - twitter_archiver
# - twitter_api_archiver
# # - instagram_archiver
# - tiktok_archiver
- youtubedl_enricher
# - wayback_archiver_enricher
enrichers:
- hash_enricher
- screenshot_enricher
- thumbnail_enricher
# - wayback_archiver_enricher
# - wacz_enricher
formatter: html_formatter # defaults to mute_formatter
storages:
- local_storage
# - s3_storage
# - gdrive_storage
databases:
# - console_db
# - csv_db
- gsheet_db
# - mongo_db
configurations:
global: #TODO: implement this logic, does nothing ATM
- save_logs: false
gsheet_feeder:
sheet: auto-archiver-test
header: 2 # defaults to 1 in GSheetsFeeder
service_account: "secrets/service_account.json"
use_sheet_names_in_stored_paths: false
columns:
url: link
status: archive status
folder: destination folder
archive: archive location
date: archive date
thumbnail: thumbnail
thumbnail_index: thumbnail index
timestamp: upload timestamp
title: upload title
text: textual content
duration: duration
screenshot: screenshot
hash: hash
wacz: wacz
replaywebpage: replaywebpage
screenshot_enricher:
width: 1280
height: 2300
wayback_archiver_enricher:
timeout: 10
key: ""
secret: ""
hash_enricher:
algorithm: "SHA3-512"
# wacz:
# profile: secrets/profile.tar.gz
local_storage:
save_to: "./local_archive"
save_absolute: true
filename_generator: static
path_generator: flat
gdrive_storage:
path_generator: url
filename_generator: random
root_folder_id: TODO
oauth_token: secrets/gd-token.json
service_account: "secrets/service_account.json"