#!/usr/bin/env python3
"""Remove from a registry fragment every sensor that the validator reported as FAIL (and, with
--drop-warn, WARN). Works on the YAML text line by line so the compact flow style is preserved.

Usage: python3 scripts/prune-fragment.py <fragment.yaml> <validator-report.json> [--drop-warn] [--keep-empty]
Sources left without any sensor are removed unless --keep-empty (they may still be useful with
discover: flags — decide case by case).
"""
import json
import re
import sys

import yaml


def main() -> None:
    frag, report = sys.argv[1], sys.argv[2]
    drop_warn = "--drop-warn" in sys.argv
    keep_empty = "--keep-empty" in sys.argv
    rep = json.load(open(report))
    bad = {r["url"] for r in rep["results"] if r["status"] == "FAIL" or (drop_warn and r["status"] == "WARN")}
    lines = open(frag).read().split("\n")
    out = []
    removed = 0
    for ln in lines:
        m = re.search(r'url:\s*"([^"]+)"', ln) or re.search(r"url:\s*'([^']+)'", ln) or re.search(r"url:\s*(\S+?)[,}]", ln)
        if ln.lstrip().startswith("- {") and m and m.group(1) in bad:
            removed += 1
            continue
        out.append(ln)
    text = "\n".join(out)
    # Drop sources with no sensors left (flow-style sensors → check the parsed structure).
    if not keep_empty:
        data = yaml.safe_load(text)
        empty = {s["id"] for s in data["sources"] if not s.get("sensors") and not s.get("discover")}
        if empty:
            blocks = re.split(r"(?m)^(?=  - id: )", text)
            kept = [b for b in blocks if not any(b.startswith(f"  - id: {e}\n") for e in empty)]
            text = "".join(kept)
            print(f"removed {len(empty)} empty sources: {', '.join(sorted(empty))}", file=sys.stderr)
    open(frag, "w").write(text)
    print(f"removed {removed} sensors from {frag}", file=sys.stderr)


if __name__ == "__main__":
    main()