#!/usr/bin/env python3
"""Extract the troupe repo's per-day commit counts as of HEAD (Extract lane).

Source: the git repository /root/troupe-standup (HEAD at capture time).
Produces: a TSV table (date + count + TOTAL), a CSV, and a markdown landing
page. Verified by wizard-work/ledger/CHECK, which re-derives the table from
`git log HEAD` independently of this script.
"""
import csv
import subprocess
from collections import Counter

REPO = "/root/troupe-standup"
OUT_DIR = "/root/.hermes/wizard-outputs/ledger"
REPO_OUT = "/root/troupe-standup/wizard-work/ledger"

MD_HEADER = """# Troupe commit census — as of HEAD
"""


def git(*args):
    return subprocess.run(
        ["git", "-C", REPO] + list(args), check=True, capture_output=True, text=True
    ).stdout


def main():
    # Independent per-day counts straight from git HEAD.
    raw = git("log", "--date=short", "--pretty=format:%ad").splitlines()
    counts = Counter(line for line in raw if line)
    rows = sorted(counts.items())
    total = sum(counts.values())

    # Shared writer for the machine-readable tables.
    def write_tsv(path):
        with open(path, "w", newline="") as f:
            w = csv.writer(f, delimiter="\t", lineterminator="\n")
            w.writerow(["date", "count"])
            for d, c in rows:
                w.writerow([d, c])
            w.writerow(["TOTAL", total])

    def write_csv(path):
        with open(path, "w", newline="") as f:
            w = csv.writer(f, lineterminator="\n")
            w.writerow(["date", "count"])
            for d, c in rows:
                w.writerow([d, c])
            w.writerow(["TOTAL", total])

    # Markdown landing page.
    head = git("rev-parse", "HEAD").strip()
    top = ", ".join("%s:%d" % (d, c) for d, c in rows[-5:])  # last 5 days
    lines = [
        MD_HEADER,
        "Source: git history of " + REPO + " (HEAD " + head[:12] + "), one `git log` pass.",
        "Extraction: `git log --date=short --pretty=format:%ad | awk 'NF' | sort | uniq -c`.",
        "Rows: per-calendar-day commit counts for the full history to HEAD.",
        "Provenance: `git rev-parse HEAD` = " + head,
        "Last 5 days of the table: " + top + ".",
        "Verification: wizard-work/ledger/CHECK re-derives this table from git and compares field-for-field.",
        "",
        "| date | commits |",
        "| --- | ---: |",
    ]
    for d, c in rows:
        lines.append("| " + d + " | " + str(c) + " |")
    lines += [
        "",
        "| TOTAL | " + str(total) + " |",
        "",
        "## Notes",
        "",
        "- Each row counts commits whose author date (committed by that day, local) starts with that date. A merge commit counts once, under its own author-date.",
        "- `TOTAL` is the sum of the per-day rows and equals `git rev-list --count HEAD`.",
        "- History spans " + str(len(rows)) + " distinct calendar days.",
        "",
    ]
    md = "\n".join(lines)

    for path, fn in [
        (OUT_DIR + "/" + "2026-10-27-commit-census.tsv", write_tsv),
        (REPO_OUT + "/" + "2026-10-27-commit-census.tsv", write_tsv),
        (OUT_DIR + "/" + "2026-10-27-commit-census.csv", write_csv),
        (REPO_OUT + "/" + "2026-10-27-commit-census.csv", write_csv),
        (OUT_DIR + "/" + "2026-10-27-commit-census.md", lambda p: open(p, "w").write(md)),
        (REPO_OUT + "/" + "2026-10-27-commit-census.md", lambda p: open(p, "w").write(md)),
    ]:
        fn(path)

    print("rows=%d total=%d days=%d head=%s" % (len(rows), total, len(rows), head[:12]))


if __name__ == "__main__":
    main()