#!/usr/bin/env python3
"""check_sol_triage.py -- real test for the sol-profile triage report.

triage.py classifies every content file under /root/.hermes/profiles/sol/ into
ACTIVE / STALE / ARTIFACT / DEAD, writing sol_triage_report.md plus a JSON
summary to stdout.  This test:

  1. re-enumerates the filesystem with independent os.walk logic and compares the
     total file count against the Findings rows in the report;
  2. parses the report's Summary table and Findings table independently, and
     asserts the per-tier counts agree with each other;
  3. asserts the action-guidance text names the ACTIVE and STALE counts; and
  4. asserts the report header carries a generated date.

Exit 0 = the report is correct; nonzero = the report is wrong.
This is a real test, not a tautology: it recomputes the file list and re-parses
the report with its own logic instead of trusting the generator's own numbers.
"""

import os
import re
import sys

PROFILE = "/root/.hermes/profiles/sol"
REPORT = "/root/.hermes/wizard-outputs/sol/sol_triage_report.md"
EXCLUDE = {".git", "node_modules", "__pycache__"}
DIAG_FILES = ["diag.py", "diag2.py", "diag3.py", "diag4.py", "diag5.py"]
TIERS = ("ACTIVE", "STALE", "ARTIFACT", "DEAD")


def independent_file_count():
    """Re-enumerate files exactly as triage.py does (top-9 subdirs + diag.py)."""
    total = 0
    for sub in ("skills", "memories", "cron", "logs", "backups", "plans",
                "pairing", "runtime", "diagnostics", "hooks", "image_cache"):
        root = os.path.join(PROFILE, sub)
        if not os.path.isdir(root):
            continue
        for _cur, dirs, names in os.walk(root):
            dirs[:] = [d for d in dirs if d not in EXCLUDE]
            total += len(names)
    for f in DIAG_FILES:
        if os.path.exists(os.path.join(PROFILE, f)):
            total += 1
    return total


def parse_report():
    text = open(REPORT, encoding="utf-8").read()
    summary = {}
    for line in text.splitlines():
        m = re.match(r"^\|\s*(ACTIVE|STALE|ARTIFACT|DEAD)\s*\|\s*(\d+)\s*\|", line)
        if m:
            summary[m.group(1)] = int(m.group(2))
    rows = []
    for line in text.splitlines():
        if re.match(r"^\|\s*\d+\s*\|", line):
            parts = [p.strip() for p in line.split("|")]
            # | # | path | tier | evidence |
            if len(parts) >= 5 and parts[3] in TIERS and parts[2].startswith("`"):
                rows.append(parts[3])
    return summary, rows, text


def main():
    expected_files = independent_file_count()
    summary, rows, text = parse_report()
    errors = []

    if expected_files != len(rows):
        errors.append("file list mismatch: recount=%d findings_rows=%d"
                      % (expected_files, len(rows)))

    found_counts = {t: 0 for t in TIERS}
    for t in rows:
        found_counts[t] += 1
    for t in TIERS:
        if summary.get(t) != found_counts[t]:
            errors.append("summary[%s]=%s findings[%s]=%d"
                          % (t, summary.get(t), t, found_counts[t]))

    if sum(summary.values()) != len(rows):
        errors.append("summary total %d != findings rows %d"
                      % (sum(summary.values()), len(rows)))

    if not re.search(r"Attend to the %d fresh content items" % summary.get("ACTIVE", -1), text):
        errors.append("action guidance does not name ACTIVE=%d"
                      % summary.get("ACTIVE"))

    if not re.search(r"generated:\s*\d{4}-\d{2}-\d{2}", text):
        errors.append("report header has no generated date")

    if errors:
        print("CHECK FAIL")
        for e in errors:
            print(" -", e)
        return 1

    print("CHECK OK: triage report is self-consistent and matches the filesystem")
    print("  files enumerated=%d  findings=%d" % (expected_files, len(rows)))
    print("  ACTIVE=%d STALE=%d ARTIFACT=%d DEAD=%d"
          % (summary["ACTIVE"], summary["STALE"],
             summary["ARTIFACT"], summary["DEAD"]))
    return 0


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