#!/usr/bin/env python3
"""ext-census: extension census of the Troupe standup repo.

Reads the curated file census `repo_files.tsv` (3,344 data rows) in the
troupe repo, applies a stated rule, and emits an ordered table of
extension -> file_count.

Rule
----
basename = final path component.
  . dotfile (leading dot, e.g. .gitignore) -> "(none)"
  . no dot in basename                     -> "(none)"
  . otherwise                             -> substring after LAST dot, lowercased

Paths default to the curated census at the repo root and to the ledger
artifact, and are overridable with EXT_CENSUS_SRC / EXT_CENSUS_OUT so a
reader can regenerate into scratch instead of clobbering the committed
table. (Q101: four string literals had a real line break where "\n"
belonged, so this file did not compile; repaired 2026-10-27.)
"""
import collections
import os
import sys

SRC = os.environ.get("EXT_CENSUS_SRC", "/root/troupe-standup/repo_files.tsv")
OUT = os.environ.get(
    "EXT_CENSUS_OUT", "/root/.hermes/wizard-outputs/ledger/ext-census.tsv"
)

def ext_of(basename):
    if basename.startswith(".") or "." not in basename:
        return "(none)"
    return basename.rsplit(".", 1)[1].lower()

def parse_census(path):
    rows = []
    with open(path) as fh:
        for line in fh:
            line = line.rstrip("\n")
            if not line.strip():
                continue
            if line.startswith("rank\tpath"):
                continue
            key, _, val = line.partition("\t")
            rows.append((key.strip(), val.strip()))
    return rows

def main():
    rows = parse_census(SRC)
    cnt = collections.Counter(ext_of(os.path.basename(p)) for _, p in rows)
    total = sum(cnt.values())
    if total != len(rows):
        sys.stderr.write("WARNING: parsed rows=%d counts=%d\n" % (len(rows), total))
    with open(OUT, "w") as fh:
        fh.write("rank\textension\tfile_count\n")
        for i, (ext, n) in enumerate(cnt.most_common(), 1):
            fh.write("%d\t%s\t%d\n" % (i, ext, n))
    print("rows=%d total=%d -> %s" % (len(rows), total, OUT))

if __name__ == "__main__":
    main()