BUG: loc setitem with duplicate columns and new columns corrupts Data…

[roeimed0]

…Frame (#58317)

• closes BUG: loc __setitem__ has incorrect behavior when assigned a DataFrame and new columns and duplicated columns are added. #58317 (Replace xxxx with the GitHub issue number)
• Tests added and passed if fixing a bug or adding a new feature
• All code checks passed.
• Added type annotations to new arguments/methods/functions.
• Added an entry in the latest doc/source/whatsnew/vX.X.X.rst file if fixing a bug or adding a new feature.
• I have reviewed and followed all the contribution guidelines
• If I used AI to develop this pull request, I prompted it to follow AGENTS.md.

Summary

fixing stale PR #64079
When using df.loc to assign a DataFrame with duplicate column names and new columns ,unrelated columns were corrupted.

The indexer in _ensure_listlike_indexer assumed the expanded columns mapped 1-to-1
to the original columns in order, which broke when union inserted duplicates in the
middle. Fixed by using get_indexer to correctly map each column position.

AI was used to explore the code path and trace the root cause.

[afurm]

The test creates a DataFrame with columns ["D", "B", "C", "A"] and assigns with ["B", "E", "B"]. When there's a mismatch between the column counts and the assignment target, the get_indexer approach needs to guarantee it produces the same indexer array length as len(keys) — otherwise reindex_indexer may misalign columns silently. Can you confirm the indexer length matches len(keys) in all paths?

[roeimed0]

I can confirm the indexer length always matches len(keys). get_indexer is guaranteed to return one value per element in its input — either the position of that key in the existing columns, or -1 for new ones. So the length is always correct.

The problem with the old code was that it assumed new columns always end up at the tail of keys, which isn't true when there are duplicate columns.

get_indexer handles duplicates and any ordering correctly without making that assumption.

[jbrockmendel]

thanks @roeimed0