Change repo uniqueness to be based on repo_src_id not url

[collectoss-issue-migrator]

Note

Migrated from augurlabs/augur#3056
Originally opened by @cdolfi on 2025-03-14

---

When repositories move like https://github.com/openai/triton to https://github.com/triton-lang/triton, both can be added and cause neither to fully complete collection. If the check before adding to the repository was based on repo_src_id I believe this problem could be prevented

[collectoss-issue-migrator]

Note

Migrated comment from original
Originally posted by @sgoggins on 2025-03-14

---

This is an issue on instances that started before May, 2024. This is addressed/fixable, using the scripts here: https://github.com/chaoss/augur-utilities/tree/main/more_cowbell

[collectoss-issue-migrator]

Note

Migrated comment from original
Originally posted by @sgoggins on 2025-03-14

---

@ABrain7710 : I think this is possibly not fixed by our Augur patch to the 100% level. @cdolfi is reporting that a new repository that was a duplicate got added in December. I really thought this was patched, and I know it was tested. So, perhaps there is an edge case missed?

https://github.com/openai/triton to https://github.com/triton-lang/triton,

[collectoss-issue-migrator]

Note

Migrated comment from original
Originally posted by @sgoggins on 2025-03-14

---

I verified this occurred on Padres in January of this year:

select * from repo where repo_git like '%triton'; 

[collectoss-issue-migrator]

Note

Migrated comment from original
Originally posted by @ABrain7710 on 2025-07-08

---

@sgoggins I haven't heard about this in awhile. Where are we at on this?

[collectoss-issue-migrator]

Note

Migrated comment from original
Originally posted by @MoralCode on 2025-10-31

---

repo source id applied to frontend repo additions in https://github.com/chaoss/augur/pull/2929

This issue should be resolved as a result

[collectoss-issue-migrator]

Note

Migrated comment from original
Originally posted by @MoralCode on 2025-10-31

---

Still an issue with CLI additions

per @sgoggins analysis:

augur/tasks/frontend.py has the method called first. (add_new_github_repos)
augur/application/cli/db.py seems to handle the insertion of repos at the command line (add_repos)

A brief examination of the code, [...] suggests to me that you possibly added repositories using the command line tools? I say that because it "looks like" we actually already have the GH ID in the table, and we are checking for duplicates in that interface.

We should compare the add_new_github_repos and add_repos functions to see how they differ and refactor them to use the same methods to add repositories

[MoralCode]

Using an AI tool to recheck this assertion and it seems as though the suspicion that the CLI repo addition path is not correctly deduplicating by source ID is valid.