2022-11-11 02:08:48 +00:00
|
|
|
steps:
|
|
|
|
|
# only 1 feeder allowed
|
2023-02-09 12:32:55 +00:00
|
|
|
# feeder: cli_feeder # default feeder
|
2023-01-04 16:37:36 +00:00
|
|
|
feeder: gsheet_feeder # default -> only expects URL from CLI
|
2022-11-11 02:08:48 +00:00
|
|
|
archivers: # order matters
|
2023-02-09 12:32:55 +00:00
|
|
|
# - vk_archiver
|
|
|
|
|
# - telethon_archiver
|
|
|
|
|
# - telegram_archiver
|
|
|
|
|
# - twitter_archiver
|
|
|
|
|
# - twitter_api_archiver
|
2023-02-17 16:15:25 +00:00
|
|
|
# - instagram_archiver
|
|
|
|
|
# - instagram_tbot_archiver
|
2023-02-09 12:32:55 +00:00
|
|
|
# - tiktok_archiver
|
2023-02-17 15:45:35 +00:00
|
|
|
- youtubedl_archiver
|
2023-02-09 12:32:55 +00:00
|
|
|
# - wayback_archiver_enricher
|
|
|
|
|
enrichers:
|
|
|
|
|
- hash_enricher
|
|
|
|
|
- screenshot_enricher
|
|
|
|
|
- thumbnail_enricher
|
|
|
|
|
# - wayback_archiver_enricher
|
|
|
|
|
# - wacz_enricher
|
|
|
|
|
|
|
|
|
|
formatter: html_formatter # defaults to mute_formatter
|
2022-11-11 02:08:48 +00:00
|
|
|
storages:
|
|
|
|
|
- local_storage
|
2023-02-09 12:32:55 +00:00
|
|
|
# - s3_storage
|
|
|
|
|
# - gdrive_storage
|
2022-11-11 02:08:48 +00:00
|
|
|
databases:
|
2023-02-09 12:32:55 +00:00
|
|
|
# - console_db
|
|
|
|
|
# - csv_db
|
|
|
|
|
- gsheet_db
|
|
|
|
|
# - mongo_db
|
2022-11-11 02:08:48 +00:00
|
|
|
|
|
|
|
|
configurations:
|
2023-01-04 16:37:36 +00:00
|
|
|
gsheet_feeder:
|
2023-02-09 12:32:55 +00:00
|
|
|
sheet: auto-archiver-test
|
2022-11-24 15:44:25 +00:00
|
|
|
header: 2 # defaults to 1 in GSheetsFeeder
|
|
|
|
|
service_account: "secrets/service_account.json"
|
2023-02-09 12:32:55 +00:00
|
|
|
use_sheet_names_in_stored_paths: false
|
2022-11-24 15:44:25 +00:00
|
|
|
columns:
|
2023-02-09 12:32:55 +00:00
|
|
|
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
|
2022-12-14 14:01:39 +00:00
|
|
|
|
2023-02-09 12:32:55 +00:00
|
|
|
screenshot_enricher:
|
2022-11-11 02:08:48 +00:00
|
|
|
width: 1280
|
2023-02-09 12:32:55 +00:00
|
|
|
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
|
2022-11-11 02:08:48 +00:00
|
|
|
|
2023-02-09 12:32:55 +00:00
|
|
|
gdrive_storage:
|
|
|
|
|
path_generator: url
|
|
|
|
|
filename_generator: random
|
|
|
|
|
root_folder_id: TODO
|
|
|
|
|
oauth_token: secrets/gd-token.json
|
|
|
|
|
service_account: "secrets/service_account.json"
|
