Rossi

@flooie it wasn't assigned to anyone, but I am assigning you in case you can review it or assing it to someone else I have just pushed an update for...

Thanks for the review @quevon24 I updated the PR with your comments. Since it's a big one do you want to do another review @flooie ?

> Did we have an answer to this as well @grossir ? @flooie Yes, you can see it here https://github.com/freelawproject/courtlistener/blob/adfcb7eacc4c0222afe24855c169440d8f118c11/cl/scrapers/management/commands/merge_opinion_versions.py#L324-L356 I fixed the conflicts, too

@flooie CL PRs now need a explicit approval to merge @mlissner this PR has migrations; so I think it should actually be merged by Chaco?

If you copy the new ORM code into a Django shell and print the query, you get the query below, which are very close to the queries profiled in this...

Ran ``` ./manage.py merge_opinion_versions download_url --url_template 'https://www.supremecourt.ohio.gov/rod/docs/pdf/' --verbosity 3 INFO {'seen_urls': 732, 'success': 636, 'different dockets': 51, 'text too different': 108}) ``` I have saved the logs; will analyze them...

From the logs, some that need review [ [ohio-versions-logs.txt](https://github.com/user-attachments/files/21415795/ohio-versions-logs.txt) ](url) These around harvard fields definetly need a check ``` WARNING Unexpected difference in filepath_json_harvard: 'law.free.cap.ohio-st-3d.144/1213.4043319.json' 'law.free.cap.ohio-st-3d.143/1295.4353884.json'. opinioncluster: 2842992, 2841898 WARNING...

New run with loose text similarity ``` {'seen_urls': 638, 'different dockets': 32, 'loose text similarity': 71, 'success': 71, 'text too different': 34, 'merging error': 9} ```

In this [issue](https://github.com/freelawproject/courtlistener/issues/3992), Rebeca correctly points to missing neutral citations in older years. We have data from the Harvard import, that is missing the neutral citations. See, for example: https://www.courtlistener.com/opinion/10680819/state-ex-rel-albright-v-indus-comm/pdf/...

We have observed this issue, and have been drafting a way to [solve](https://github.com/freelawproject/juriscraper/issues/858#issuecomment-2240043029) it, for [all courts](https://github.com/freelawproject/juriscraper/issues/858#issuecomment-2248873504) including `scotus`. Do you want me to work on this next @flooie ?...