Code Metrics Star/Fork counts are reporting incorrectly for some

[ravenac95]

Which area(s) are affected? (leave empty if unsure)

No response

To Reproduce

This was from a user in the discord (thanks daehyun from discord!)

This query:

SELECT *
FROM `oso_production.code_metrics_by_project_v1`
WHERE project_name = 'sonobe-privacy-scaling-explorations'

Returns the following data

• star_count: 644
• fork_count: 152

But the numbers that we have in both the raw data from github (the cloudquery github resolver) and also the data on the UI in github shows that those star / fork counts are 162 and 38 respectively.

Describe the Bug

See above

Expected Behavior

The numbers should match the github data and add up correctly based on all the repos in a project.

[ravenac95]

The issue seems to be somewhere in int_repo_metrics_by_project:

SELECT * 
FROM `opensource-observer.oso.int_repo_metrics_by_project` 
WHERE project_id = 'YIX6II_0b1iDDkzlII1s4Nek-cFYoFCweXAnbjrN62w='

This returns 4 copies of the project. I imagine that gets aggregated somewhere and that causes this issue.

The project_id above is the project id of the sonobe-privacy-scaling-explorations project.

[ravenac95]

This doesn't seem to be happening on all things. If you search for opensource-observer's project (Erx9J64anc8oSeN-wDKm0sojJf8ONrFVYbQ7GFnqSyc=) it's just fine.

[baumstern]

zk-eigentrust has 110 stars and 10 forks but the query to the oso_production in BigQuery returns 220 stars and 20 forks count. It's project_name is zk-eigentrust-privacy-scaling-explorations