#!/usr/bin/env python3
"""Verify the sol-profile triage artifact against real disk state.

Recomputes the classification from the active profile directory
/root/.hermes/profiles/sol/ and diffs it against the recorded artifact.
Exit code 0 = artifact matches disk. Nonzero = mismatch.

Rules (must mirror the generator's EXACT map / ordering):
  identity: .env, .env.bak-20261014T165715Z
  soul:       SOUL.md
  config:     .env.lock, config.yaml, config.yaml.bak-20261014T165715Z
  state_db:   state.db, state.db-shm, state.db-wal,
              state.db.auto-maintenance.lock, state.db.fts_rebuild.lock,
              state.db.quarantine.lock, auth.json, auth.lock,
              .skills_prompt_snapshot.json, models_dev_cache.json,
              models_dev_cache.etag
  cache:      cache, audio_cache, image_cache, context_length_cache.yaml
  sessions:   sessions
  skills:     skills
  plugins:    plugins, plugin-update-checks
  logs:       logs
  backups:    backups
  vault:      vault
  extension_dir: any other directory
  script:     name ends with .py
  config_data: name ends with .yaml or .json
  other:      none of the above
"""
import json
import os
import sys

PROFILE = "/root/.hermes/profiles/sol"
ARTIFACT = "/root/.hermes/wizard-outputs/sol/sol-profile-triage-2026-10-28.json"

EXACT = {
    "identity": {".env", ".env.bak-20261014T165715Z"},
    "soul": {"SOUL.md"},
    "config": {".env.lock", "config.yaml", "config.yaml.bak-20261014T165715Z"},
    "state_db": {
        "state.db", "state.db-shm", "state.db-wal",
        "state.db.auto-maintenance.lock", "state.db.fts_rebuild.lock",
        "state.db.quarantine.lock", "auth.json", "auth.lock",
        ".skills_prompt_snapshot.json", "models_dev_cache.json",
        "models_dev_cache.etag",
    },
    "cache": {"cache", "audio_cache", "image_cache",
              "context_length_cache.yaml"},
    "sessions": {"sessions"},
    "skills": {"skills"},
    "plugins": {"plugins", "plugin-update-checks"},
    "logs": {"logs"},
    "backups": {"backups"},
    "vault": {"vault"},
}
EXTS = {"bin", "cron", "forge", "home", "hooks", "keys", "lsp", "memories",
        "pairings", "plans", "platforms", "runtime", "sandboxes", "skins",
        "source-checks", "wizard-outputs", "workspace"}


def expected_category(name, is_dir):
    if name in EXACT["identity"]:
        return "identity"
    if name in EXACT["soul"]:
        return "soul"
    if name in EXACT["config"]:
        return "config"
    if name in EXACT["state_db"]:
        return "state_db"
    if name in EXACT["cache"]:
        return "cache"
    if name in EXACT["sessions"]:
        return "sessions"
    if name in EXACT["skills"]:
        return "skills"
    if name in EXACT["plugins"]:
        return "plugins"
    if name in EXACT["logs"]:
        return "logs"
    if name in EXACT["backups"]:
        return "backups"
    if name in EXACT["vault"]:
        return "vault"
    if is_dir:
        return "extension_dir"
    if name.endswith(".py"):
        return "script"
    if name.endswith(".yaml") or name.endswith(".json"):
        return "config_data"
    return "other"


def main():
    with open(ARTIFACT, encoding="utf-8") as fh:
        art = json.load(fh)

    disk = sorted(os.listdir(PROFILE))
    expected = {}
    for name in disk:
        path = os.path.join(PROFILE, name)
        expected[name] = expected_category(name, os.path.isdir(path))

    recorded = {e["name"]: e.get("category") for e in art.get("entries", [])}

    problems = []
    if art.get("total_entries") != len(disk):
        problems.append("total_entries %r != disk %d" % (art.get("total_entries"), len(disk)))
    if set(recorded.keys()) != set(disk):
        problems.append("entry names differ: missing=%s extra=%s" % (
            sorted(set(disk) - set(recorded)), sorted(set(recorded) - set(disk))))
    for n in sorted(set(disk) & set(recorded)):
        if recorded[n] != expected[n]:
            problems.append("category mismatch for %r: recorded=%r expected=%r" % (n, recorded[n], expected[n]))
    expected_counts = {}
    for n, c in expected.items():
        expected_counts[c] = expected_counts.get(c, 0) + 1
    if art.get("categories") != expected_counts:
        problems.append("category counts %r != expected %r" % (art.get("categories"), expected_counts))

    if problems:
        print("FAIL")
        for p in problems:
            print("  -", p)
        sys.exit(1)

    print("PASS: artifact matches disk (%d entries, %d bytes)" % (len(disk), os.path.getsize(ARTIFACT)))
    print("categories=%s" % json.dumps(art.get("categories")))
    sys.exit(0)


if __name__ == "__main__":
    main()