Module stats 

Counting a repo, and counting a review.

Two tallies over the same crate::language table: the working tree’s lines per language, which needs a read per file, and the diff’s churn per language, which the model already holds. Both are pure functions of what they are handed, so the caller decides which thread pays.

Structs

LanguageChurn

One language’s share of a review.

LanguageCount

How much of one language a repo holds.

RepoStats

The whole scan: one entry per language, heaviest first, plus what it skipped.

Functions

review_mix

What the review is written in, busiest language first. Files in no known language are left out: naming them would take a row from the languages the reader can act on.

scan

Count every path under root, grouped by language and ordered by code lines. Reads each file once; a path that is binary, oversized or unreadable is counted as skipped and never parsed.

scan_repo

Scan a checkout from its root, opening the backend here so the whole job, the index read included, happens on the caller’s thread. extra carries what git does not track yet, the status screen’s untracked files.