#!/usr/bin/env python3
"""Classify every untracked file in the troupe repo into a persistent triage
bucket. Deterministic rule: the first matching prefix wins.

Output:
  untracked-classify-2026-10-29.json  (machine-readable: path, bucket, rule)
  untracked-classify-summary.md       (human readable, with total + per-bucket)
"""
import json
import subprocess
from collections import Counter, OrderedDict

REPO_ROOT = "/root/troupe-standup"
OUT_DIR = REPO_ROOT + "/wizard-work/sol"

# Ordered (bucket, canonical match). First match wins. The tooling directory
# "wizard-work/sol/" holds this classifer/checker/JSON/summary (the "tool" bucket),
# and the stale top-level "CHECK-sol-untracked-classify" is also tooling.
RULES = [
    ("tool", "wizard-work/sol/"),
    ("tool", "CHECK-sol-untracked-classify"),
    ("probe", "wizard-work/null/"),
    ("sphinx", "wizard-work/sphinx/"),
    ("wizard-work", "wizard-work/"),
    ("wizard-outputs", "wizard-outputs/"),
    ("replies", "replies/"),
    ("checkins", "checkins/"),
    ("docs", "docs/"),
    ("inbox", "inbox/"),
    ("recoveries", "recoveries/"),
    ("tools", "tools/"),
    ("dump", "d.jsonl"),
    ("dump", "s.json"),
    ("kanban", "kanban.db"),
    ("inventory", "repo_"),
    ("ci", "CHECK-"),
    ("ci", "."),
]


def classify(path):
    for bucket, prefix in RULES:
        if path == prefix or path.startswith(prefix):
            return bucket
    return "other"


def run_git(args):
    return subprocess.run(
        ["git", "-C", REPO_ROOT] + args, check=True,
        capture_output=True, text=True,
    ).stdout.splitlines()


def main():
    # File set: independent enumeration via git LS.
    # Subject = the untracked working-tree files that are NOT this tool's own
    # output directory ("wizard-work/sol/") and not the stale top-level CI
    # leftover; everything else is the classifiable triage subject.
    files = run_git(["ls-files", "--others", "--exclude-standard"])
    files = [f for f in files if f]
    subject = [p for p in files
               if not (p.startswith("wizard-work/sol/")
                       or p == "CHECK-sol-untracked-classify")]

    classified = []
    counts = Counter()
    for path in sorted(subject):
        bucket = classify(path)
        counts[bucket] += 1
        matched = next(p for b, p in RULES if b == bucket and
                       (path == p or path.startswith(p)))
        classified.append({
            "path": path,
            "bucket": bucket,
            "rule": matched,
        })

    assert len(classified) == len(subject), "dedup lost a file"
    assert len(set(c["path"] for c in classified)) == len(subject), "duplicate path"

    with open(OUT_DIR + "/untracked-classify-2026-10-29.json", "w") as f:
        json.dump(classified, f, indent=2, sort_keys=True)

    # Human summary
    lines = []
    lines.append("# Untraced-file triage: 2026-10-29")
    lines.append("")
    lines.append("Classified `%d` untracked files (enumerated via `git ls-files --others --exclude-standard`) "
                 "into %d bucket(s)." % (len(classified), len(counts)))
    lines.append("")
    lines.append("| Bucket | Files |")
    lines.append("|---|---|")
    for b in sorted(counts):
        lines.append("| %s | %d |" % (b, counts[b]))
    lines.append("| **total** | **%d** |" % len(classified))
    lines.append("")
    lines.append("## Sample")
    lines.append("")
    for c in classified[:8]:
        lines.append("- `%s` -> `%s` (%s)" % (c["path"], c["bucket"], c["rule"]))
    lines.append("")
    lines.append("## Buckets (full)")
    lines.append("")
    for b in sorted(counts):
        members = [c["path"] for c in classified if c["bucket"] == b]
        lines.append("### %s (%d)" % (b, len(members)))
        lines.append("")
        for m in members:
            lines.append("- `%s`" % m)
        lines.append("")

    with open(OUT_DIR + "/untracked-classify-summary.md", "w") as f:
        f.write("\n".join(lines))

    print("files=%d buckets=%d" % (len(classified), len(counts)))
    for b in sorted(counts):
        print("  %-12s %d" % (b, counts[b]))
    print("wrote %s" % OUT_DIR)


if __name__ == "__main__":
    main()