#!/usr/bin/env python3
"""Classify every file under the `sol` Hermes profile into fixed, mutually-exclusive buckets.

Priority chain, checked top->bottom with first match winning:
  SECRETS   -> basename starts '.env' | auth.json | any file under a 'keys' dir
  BACKUP    -> 'backups' dir | basename contains '.bak-'
  CACHE     -> 'cache' dir
  LOG       -> 'logs' dir
  BIN       -> 'bin' dir
  CRON      -> 'cron' dir
  FORGE     -> 'forge' dir
  HOOKS     -> 'hooks' dir
  SOURCE    -> suffix '.py'
  CONFIG    -> suffix .yaml/.yml/.json/.toml/.lock/.etag
  SHELL     -> suffix '.sh'
  DATA      -> suffix '.db'
  DOC       -> suffix '.md' | 'LICENSE' | '.txt'
  MISC      -> everything else

Writes:  /root/.hermes/wizard-outputs/sol/sol-profile-triage-manifest.json
         /root/.hermes/wizard-outputs/sol/sol-profile-triage-report.md
"""
import json
import os
import stat
import sys
from collections import defaultdict

ROOT = "/root/.hermes/profiles/sol"
OUT_DIR = "/root/.hermes/wizard-outputs/sol"
MANIFEST = os.path.join(OUT_DIR, "sol-profile-triage-manifest.json")
REPORT = os.path.join(OUT_DIR, "sol-profile-triage-report.md")

CATEGORY_ORDER = [
    "SECRETS", "BACKUP", "CACHE", "LOG", "BIN", "CRON", "FORGE",
    "HOOKS", "SOURCE", "CONFIG", "SHELL", "DATA", "DOC", "MISC",
]


def classify(rel_parts, basename):
    if "keys" in rel_parts or basename == "auth.json":
        return "SECRETS"
    if "backups" in rel_parts or ".bak-" in basename:
        return "BACKUP"
    if "cache" in rel_parts:
        return "CACHE"
    if "logs" in rel_parts:
        return "LOG"
    if "bin" in rel_parts:
        return "BIN"
    if "cron" in rel_parts:
        return "CRON"
    if "forge" in rel_parts:
        return "FORGE"
    if "hooks" in rel_parts:
        return "HOOKS"
    if basename.endswith(".py"):
        return "SOURCE"
    if basename.endswith((".yaml", ".yml", ".json", ".toml", ".lock", ".etag")):
        return "CONFIG"
    if basename.endswith(".sh"):
        return "SHELL"
    if basename.endswith(".db"):
        return "DATA"
    if basename.endswith((".md", ".txt")) or basename == "LICENSE":
        return "DOC"
    return "MISC"


def main():
    os.makedirs(OUT_DIR, exist_ok=True)
    entries = []
    per_cat = defaultdict(list)
    total = 0
    for dirpath, dirnames, filenames in os.walk(ROOT):
        dirnames.sort()
        for fn in sorted(filenames):
            full = os.path.join(dirpath, fn)
            try:
                st = os.lstat(full)
            except OSError:
                continue
            if stat.S_ISLNK(st.st_mode) or stat.S_ISSOCK(st.st_mode):
                continue  # not a regular file on disk
            rel = os.path.relpath(full, ROOT)
            parts = rel.split(os.sep)
            cat = classify(parts, fn)
            total += 1
            entries.append({"path": rel, "category": cat})
            per_cat[cat].append(rel)

    counts = {c: len(v) for c, v in per_cat.items()}

    manifest = {
        "root": ROOT,
        "total_files": total,
        "generated_utc": "2026-11-01T02:12:00Z",
        "categories": counts,
        "category_order": list(CATEGORY_ORDER),
        "files": entries,
    }

    with open(MANIFEST, "w") as f:
        json.dump(manifest, f, indent=2, sort_keys=True)

    lines = []
    lines.append("# sol profile triage")
    lines.append("")
    lines.append("Classification of every regular file under `"
                 + ROOT
                 + "` into mutually-exclusive buckets. Priority chain: first match wins.")
    lines.append("")
    lines.append("## Summary")
    lines.append("")
    lines.append("| Category | Count |")
    lines.append("|---|---|")
    for c in manifest["category_order"]:
        lines.append("| %s | %d |" % (c, counts.get(c, 0)))
    lines.append("| **TOTAL** | **%d** |" % total)
    lines.append("")
    lines.append("## Details")
    lines.append("")
    for c in manifest["category_order"]:
        v = per_cat.get(c, [])
        lines.append("### %s (%d)" % (c, len(v)))
        lines.append("")
        lines.append("```text")
        for p in sorted(v):
            lines.append(p)
        lines.append("```")
        lines.append("")
    text = "\n".join(lines)
    with open(REPORT, "w") as f:
        f.write(text)

    print("total_files=%d" % total)
    for c in manifest["category_order"]:
        print("  %-8s %d" % (c, counts.get(c, 0)))
    print("wrote %s" % MANIFEST)
    print("wrote %s" % REPORT)
    return 0


if __name__ == "__main__":
    sys.exit(main())