Repository navigation
Normalizing twarc2 commands #608
Description
I'm finding that some command line commands are starting to diverge, and become inconsistent. This issue is just for investigating this and coordinating some changes i think we should make. Maybe some breaking changes? It would be great to align all the commands so that they have consistent commonalities and documentation and examples.
To begin with:
- Add the singular user command, analogous to the single tweet command #599
name-or-idpattern is way better than guessing, it should be the same across the others. - Timelines and mentions should only take user IDs, not usernames #587 should use the
name or idpattern maybe - Timeline command is missing --max-results #596
- Add an extra progress bar for users in timelines command #561
- accept usernames with @ in twarc2 timelines #568 maybe but this is more of a data / validation thing
- The users CLI command can't handle quoted strings like the timelines command can #542 as above, more validation maybe
- Command completion #444 is super important now that there's a bunch more commands
- Option to use full archive search by default #650 changing defaults
- Use
name-or-idpattern for supporting commands specifying user IDs and usernames
Activity
SamHames commentedon Mar 14, 2022
I'm in favour of this, especially if it means we stop guessing or inferring what was intended.
This is probably enough of a breaking change that we'd want to have at least a major version bump, and ideally some nice managing of errors that we can tell people about the new syntax. This would also let us standardise the handling of usernames, so hopefully we don't have to deal with other instances like #568 and #542.
igorbrigadir commentedon Mar 17, 2022
Maybe all commands that do bulk operations on an input text file vs a single argument should be like
twarc2 bulk followers input.txt output.txt? Just another idea.SamHames commentedon Mar 21, 2022
I like the ID of a bulk subcommand, I can definitely see how that would contribute to consistency: anything that takes an input file is a
bulkcommand, everything that takes an argument on the command line is at the top level.That also gives us a nicer migration path because we can decouple creating new (more consistent) commands under bulk from removing the old commands, allowing for some deprecation warnings or similar.