#!/usr/bin/env python3
"""Independent verifier for sol-l030-classify-messages.json.

Exits 0 iff the committed artifact re-derives EVERY inbox message by the
triage rule — re-read from the raw inputs, without importing the generator —
and the aggregate counts match.

Run:  python3 verify_sol_l030_classify.py
"""

import json
import sys
from pathlib import Path

REPO = Path("/root/troupe-standup")
INBOX = REPO / "wizard-work" / "sol" / "inbox"
ART = REPO / "wizard-work" / "sol" / "sol-l030-classify-messages.json"

# The exact rule the artifact is supposed to implement (copied for
# independent re-derivation, not imported).
LANE_RULES = [
    ("ops", ["cron", "deploy", "disk", "backup", "server", "fail", "uptime"]),
    ("finance", ["invoice", "payment", "expense", "salary", "bitcoin", "fee"]),
    ("person", ["alexander", "sol", "wizard", "troupe", "meeting", "alex"]),
]
LANES = ["ops", "finance", "person", "other"]


def rederive():
    """Re-derive the classification of every inbox message straight from the
    raw files, using only the rule above."""
    items = []
    for path in sorted(INBOX.glob("*.txt")):
        subject, body = "", ""
        for raw in path.read_text(encoding="utf-8").splitlines():
            if raw.startswith("subject:"):
                subject = raw.split(":", 1)[1].strip()
            elif raw.startswith("body:"):
                body = raw.split(":", 1)[1].strip()
        text = (subject + " " + body).lower()
        lane = "other"
        for name, kws in LANE_RULES:
            if any(k in text for k in kws):
                lane = name
                break
        items.append({"file": path.name, "lane": lane})
    return items


def main():
    problems = []

    if not INBOX.is_dir():
        problems.append("inbox dir missing: " + str(INBOX))
    if not ART.is_file():
        problems.append("artifact missing: " + str(ART))
    if problems:
        _fail(problems)
        return 1

    expected = rederive()
    expected_names = {e["file"] for e in expected}
    art = json.loads(ART.read_text(encoding="utf-8"))

    # 1. every input re-derives
    for e in expected:
        a = next((i for i in art.get("items", []) if i.get("file") == e["file"]),
                 None)
        if a is None:
            problems.append("no report item for " + e["file"])
        elif a.get("lane") != e["lane"]:
            problems.append("lane mismatch for %r: got %r want %r" % (
                e["file"], a.get("lane"), e["lane"]))

    # 2. every report item came from an input
    for a in art.get("items", []):
        if a.get("file") not in expected_names:
            problems.append("report item not in inbox: " + a.get("file", "?"))

    # 3. counts
    exp_counts = {l: 0 for l in LANES}
    for e in expected:
        exp_counts[e["lane"]] += 1
    if art.get("counts") != exp_counts:
        problems.append("counts %r != expected %r" % (art.get("counts"), exp_counts))
    if art.get("total") != len(expected):
        problems.append("total %r != expected %d" % (art.get("total"), len(expected)))

    if problems:
        _fail(problems)
        return 1

    print("PASS: sol-l030-classify-messages.json re-derives %d/%d inbox messages "
          "by the triage rule; counts=%s" % (len(expected), len(expected), art.get("counts")))
    return 0


def _fail(problems):
    print("FAIL")
    for p in problems:
        print("  -", p)


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