"""只读请求商务接口，记录耗时、完整数据摘要及 SQL；敏感内容必须存放仓库外。"""

import argparse
import hashlib
import json
import re
import statistics
import time
import urllib.parse
import urllib.request
from pathlib import Path


def log_offsets(directory):
    return {p: p.stat().st_size for p in directory.rglob("*.log")}


def read_queries(directory, offsets):
    queries = []
    files = list(directory.rglob("*.log"))
    files = [file for file in files if file.parent.name == "sql"] or files
    for file in files:
        with file.open("rb") as stream:
            stream.seek(offsets.get(file, 0))
            for line in stream:
                try:
                    row = json.loads(line)
                except (ValueError, UnicodeDecodeError):
                    continue
                if row.get("gorm_logger") and "sql" in row:
                    queries.append(row)
    return queries


def main():
    parser = argparse.ArgumentParser(description=__doc__)
    parser.add_argument("--runtime", type=Path, required=True)
    parser.add_argument("--base-url", default="http://127.0.0.1:19004")
    parser.add_argument("--phase", required=True)
    parser.add_argument("--repeat", type=int, default=2)
    parser.add_argument("--cases", type=Path)
    args = parser.parse_args()
    if args.runtime.resolve().is_relative_to(Path(__file__).resolve().parents[1]):
        raise ValueError("完整接口响应和 SQL 必须留在仓库外")
    output = args.runtime / args.phase
    output.mkdir(exist_ok=True)
    cases = json.loads((args.cases or args.runtime / "cases.json").read_text(encoding="utf-8"))
    token = (args.runtime / "new-admin.token").read_text(encoding="utf-8").strip()
    report = []
    for case in cases:
        elapsed = []
        for round_number in range(args.repeat):
            offsets = log_offsets(args.runtime / "sql-logs")
            url = args.base_url + case["path"] + "?" + urllib.parse.urlencode(case["params"], doseq=True)
            request = urllib.request.Request(url, headers={"x-token": token, "X-Request-Id": f"perf-{args.phase}-{case['name']}-{round_number}"})
            started = time.monotonic()
            with urllib.request.urlopen(request, timeout=300) as response:
                body = json.load(response)
            elapsed.append(round((time.monotonic() - started) * 1000, 2))
            if body.get("code") != 0:
                raise RuntimeError("请求失败：" + case["name"])
            queries = read_queries(args.runtime / "sql-logs", offsets)
            data = body["data"]
            capture = {"case": case, "data": data, "elapsedMs": elapsed[-1], "queries": queries}
            (output / f"{case['name']}-{round_number}.json").write_text(json.dumps(capture, ensure_ascii=False, indent=2), encoding="utf-8")
            digest = hashlib.sha256(json.dumps(data, sort_keys=True, ensure_ascii=False).encode()).hexdigest()
            slow = sorted(queries, key=lambda row: row.get("elapsed_ms", 0), reverse=True)[:5]
            item = {"case": case["name"], "round": round_number, "elapsedMs": elapsed[-1], "dataSha256": digest, "queryCount": len(queries), "slowQueries": [{"elapsedMs": q.get("elapsed_ms"), "rows": q.get("rows"), "tables": list(dict.fromkeys(re.findall(r"(?:FROM|JOIN)\s+`?([\w.]+)", q["sql"], re.I)))} for q in slow]}
            report.append(item)
            (output / "report.json").write_text(json.dumps(report, indent=2) + "\n", encoding="utf-8")
            print(json.dumps(item), flush=True)
        print(json.dumps({"case": case["name"], "medianMs": statistics.median(elapsed)}), flush=True)


if __name__ == "__main__":
    main()