Feature Request: Allow local text or TSV files instead of Google Spreadsheets

[kkarhan]

Hi,
as I asked on the fediverse, there's like a not-so insignificant need to allow self-hosting, which admittedly it doesn't do as of now.

• Ideally one could just specify i.e. a --local [filename] flag (similar to the batch option of youtube-dlp and curl) and just use that.
• I have built something like that in the past out of need to circumvent extra-obnoxious anti-botting firewalls and similar attempts at preventing people from archiving contents.
• It is not always practical to or possible to invoke or access Google Services which may or may not be blocked by networks, ISPs and upstreams.

I sincerely hope this will help your project going forward and if needed I'll gladly provide samples of sites that one may want to archive.

Yours faithfully,
Kevin Karhan

[GalenReich]

Hi kkarhan, thanks for opening the issue - this is something we may look at and would welcome pull requests to add a TSV feeder.

Currently we do support using a command line feeder (cli_feeder) if you wanted to bypass the Google dependency in the interim - you can set this in your orchestration.yaml.

We are planning on working on the documentation of the auto-archiver so hopefully that will help with correctly configuring for different workflows.

[msramalho]

Hey @kkarhan thanks for the clear issue and suggestion.

Adding to Galen's answer: for now we only implemented 2 main feeders: GoogleSheets and CommandLine. Internally, that covers all our needs, so this is not something we will not be worked on by us atm (adding wontfix label).

Still, we'll leave this issue open for a while in case you or others find it a valuable addition and want to contribute it to the project.

[kkarhan]

Thanks so far for the feedback and keeping the issue open.

Is there any conclusive documentation re: cli_feeder ?

• Cuz at worst I could just make like a downloads.list file [similar to the batching functionality of youtube-dlp] and just iterate over it's contents in i.e. a bash-loop & array as I did in the past elsewhere.

Cuz if similar to wget & curl I could just iterate over things that way...

[msramalho]

No good documentation on it unfortunately.

If you look at the code

auto-archiver/src/auto_archiver/feeders/cli_feeder.py

Lines 17 to 24 in b166d57

	def configs() -> dict:
	return {
	"urls": {
	"default": None,
	"help": "URL(s) to archive, either a single URL or a list of urls, should not come from config.yaml",
	"cli_set": lambda cli_val, cur_val: list(set(cli_val.split(",")))
	},
	}

it is essentially expecting a --cli_feeder.urls parameter, an example call would be:
python -m src.auto_archiver --config secrets/orchestration.yaml --cli_feeder.urls="https://example.com,https://example2.wow"

What I'd suggest is you either create a new, very similar feeder, that accepts a filename instead of a csv of hardcoded urls OR actually modify the cli_feeder to have another parameter just for filenames and force at least one of them to be present.

This should not be hard to achieve assuming you've been able to run/test the auto-archiver locally on your development environment.

[msramalho]

*this would be preferable to piping giving the current sofware architecture of the library.