stocks-query/scripts/yahoo-institutional-browserless.py

#!/usr/bin/env python3
"""
用 Browserless 抓 Yahoo 三大法人頁，輸出精簡摘要（代號可變）

用法：
  python3 stocks/yahoo-institutional-browserless.py 3481
  python3 stocks/yahoo-institutional-browserless.py 2330
"""

import json
import os
from pathlib import Path
import re
import sys
import subprocess



def load_dotenv() -> None:
    """載入 .env（優先目前目錄，其次 workspace 根目錄）。"""
    candidates = [
        Path.cwd() / ".env",
        Path(__file__).resolve().parents[3] / ".env",
    ]
    for env_path in candidates:
        if not env_path.exists():
            continue
        for raw in env_path.read_text(encoding="utf-8").splitlines():
            line = raw.strip()
            if not line or line.startswith("#") or "=" not in line:
                continue
            k, v = line.split("=", 1)
            k = k.strip()
            v = v.strip().strip('"').strip("'")
            if k and k not in os.environ:
                os.environ[k] = v
        break


load_dotenv()
ENDPOINT = os.getenv("BROWSERLESS_ENDPOINT")
TOKEN = os.getenv("BROWSERLESS_TOKEN")
if not ENDPOINT or not TOKEN:
    raise RuntimeError("缺少必要環境變數：BROWSERLESS_ENDPOINT / BROWSERLESS_TOKEN")
STOCK_NO = sys.argv[1] if len(sys.argv) > 1 else "3481"


def fetch_html(stock_no: str) -> str:
    url = f"https://tw.stock.yahoo.com/quote/{stock_no}.TW/institutional-trading"
    api = f"{ENDPOINT}/content?token={TOKEN}"
    payload = json.dumps(
        {
            "url": url,
            "gotoOptions": {"waitUntil": "domcontentloaded", "timeout": 60000},
        },
        ensure_ascii=False,
    )

    cmd = [
        "curl",
        "-sS",
        "--max-time",
        "90",
        "-X",
        "POST",
        api,
        "-H",
        "Content-Type: application/json",
        "-d",
        payload,
    ]
    try:
        text = subprocess.check_output(cmd, text=True, timeout=95)
    except subprocess.CalledProcessError as e:
        raise RuntimeError(f"Browserless 呼叫失敗: {e}") from e
    except subprocess.TimeoutExpired as e:
        raise RuntimeError("Browserless 呼叫逾時") from e

    if "TimeoutError" in text:
        raise RuntimeError("Browserless 導航逾時（TimeoutError）")
    return text


def pick(html: str, pattern: str, default: str = "-") -> str:
    m = re.search(pattern, html, flags=re.S)
    return m.group(1) if m else default


def parse_inst(html: str, stock_no: str) -> dict:
    name = pick(html, r'property="og:title" content="([^"(]+)\([0-9]+\.TW\)', "").strip()
    if not name:
        name = pick(html, r"<title>([^<(]+)\([0-9]+\.TW\)", "-").strip()

    # 抓 summary data（避免大範圍 regex 回溯，先縮小搜尋範圍）
    anchor = html.find('institutionBuySellSummaryData')
    chunk = html[anchor: anchor + 250000] if anchor != -1 else html

    rows = []
    obj_pat = re.compile(r'\{[^{}]*"formattedDate":"[^"]+"[^{}]*\}')
    for om in obj_pat.finditer(chunk):
        s = om.group(0)
        foreign_m = re.search(r'"foreignDiffVolK":(-?\d+)', s)
        trust_m = re.search(r'"investmentTrustDiffVolK":(-?\d+)', s)
        dealer_m = re.search(r'"dealerDiffVolK":(-?\d+)', s)
        total_m = re.search(r'"totalDiffVolK":(-?\d+)', s)
        date_m = re.search(r'"formattedDate":"([^"]+)"', s)
        if foreign_m and trust_m and dealer_m and total_m and date_m:
            rows.append((foreign_m.group(1), trust_m.group(1), dealer_m.group(1), total_m.group(1), date_m.group(1)))

    if not rows:
        raise RuntimeError("抓不到三大法人資料（頁面結構可能變更）")

    def normalize_date(ds: str) -> str:
        return ds.replace("\\u002F", "/")

    def date_key(ds: str):
        s = normalize_date(ds)
        try:
            y, m, d = s.split("/")
            return int(y), int(m), int(d)
        except Exception:
            return (0, 0, 0)

    foreign, trust, dealer, total, date = max(rows, key=lambda r: date_key(r[4]))
    date = normalize_date(date)

    return {
        "name": name,
        "stock_no": stock_no,
        "date": date,
        "foreign": int(foreign),
        "trust": int(trust),
        "dealer": int(dealer),
        "total": int(total),
    }


def sign(v: int) -> str:
    return f"+{v:,}" if v >= 0 else f"{v:,}"


def print_summary(d: dict) -> None:
    print(f"{d['name']}（{d['stock_no']}）｜法人買賣超摘要")
    print(f"• 日期：{d['date']}")
    print(f"• 外資：{sign(d['foreign'])} 張")
    print(f"• 投信：{sign(d['trust'])} 張")
    print(f"• 自營商：{sign(d['dealer'])} 張")
    print(f"• 三大法人合計：{sign(d['total'])} 張")


if __name__ == "__main__":
    try:
        html = fetch_html(STOCK_NO)
        data = parse_inst(html, STOCK_NO)
        print_summary(data)
    except Exception as e:
        print(f"取得失敗：{e}", file=sys.stderr)
        sys.exit(1)