Bitbucket: Add support for queries and sorting

[alexfikl]

Make better use of the bitbucket API to sort tags:

• use q="my_query" to narrow down the results. This does not support fancy regex queries, just contains and comparisons.
• use sort=field to sort the results by the given field.
• use fields=values.name,next to only return relevant data.

[lilydjwg]

I didn't know bitbucket has query and sort support on this. I'd like to utilize this instead of paging, like the github source does.

[alexfikl]

I didn't know bitbucket has query and sort support on this. I'd like to utilize this instead of paging, like the github source does.

Not super familiar with the bitbucket API, but it seems like the results are always paginated, even when sorted or filtered.

I can definitely use the sorted results to get a latest tag like the github plugin. Is that what you meant? EDIT: Gave it a go in e6012b8. Let me know what you think!

[lilydjwg]

I want to expose the q and sort parameters from the API to the configuration, and we only return results on the first page. Hopefully the desired tag can be queried and sorted to be on the first page, and any outliners can be filtered using the "list options" mechanism.

We may add three new options: use_sorted_tags (not latest since we can sort differently) means to use this new behavior and optional query and sort can be provided, and we pass those directly to bitbucket's API.

(However the old behavior needs to be kept for backward compatibility.)

[alexfikl]

I want to expose the q and sort parameters from the API to the configuration, and we only return results on the first page. Hopefully the desired tag can be queried and sorted to be on the first page, and any outliners can be filtered using the "list options" mechanism.

We may add three new options: use_sorted_tags (not latest since we can sort differently) means to use this new behavior and optional query and sort can be provided, and we pass those directly to bitbucket's API.

(However the old behavior needs to be kept for backward compatibility.)

Tried do implement something like that in a8228bb. It seems to work for the simple tests I played with. What do you think?

[alexfikl]

There are a few other things we can add to the query:

• fields=values.name tells bitbucket to only return the tag names (as that's all we use anyway). This could be always added?
• pagelen=NUMBER can return more results (default is just 10). This could be a config option like max_page?

[lilydjwg]

• Yes, the smaller the response is, the faster it works.
• I don't feel like we need to add pagelen. max_page is for repos with a lot of unwanted tags, but since we can sort and filter, it should be unlikely we need a large pagelen to get what we want. Users can open an issue if it's actually useful.

[alexfikl]

• Yes, the smaller the response is, the faster it works.

Added support for this in the latest commit and fixed the mypy issues (at least locally, fingers crossed!).