"""
Google Trends "search interest" — a free retail-attention signal.

Uses Google's undocumented Trends JSON endpoints directly (no pytrends/pandas
dependency). Google frequently rate-limits/blocks datacenter IPs, so this is
best-effort: on ANY failure it returns the last cached value (or {}), and the
pipeline continues unaffected. Results are cached for ~1h to avoid rate limits.
"""

import json
import time
import logging
import requests
from urllib.parse import quote

logger = logging.getLogger(__name__)

_EXPLORE = "https://trends.google.com/trends/api/explore"
_MULTILINE = "https://trends.google.com/trends/api/widgetdata/multiline"
_HEADERS = {"User-Agent": "Mozilla/5.0 (Windows NT 10.0; Win64; x64) "
                          "AppleWebKit/537.36 (KHTML, like Gecko) Chrome/120 Safari/537.36",
            "Accept-Language": "en-US,en;q=0.9"}

_cache: dict = {}
_last_fetch = 0.0
_MIN_INTERVAL = 3600            # 1h


def _strip(txt: str):
    # Google prefixes JSON with )]}' — drop the first line.
    i = txt.find("{")
    return json.loads(txt[i:]) if i >= 0 else None


def _interest(keyword: str, timeframe="now 7-d", geo="") -> int | None:
    """Latest search-interest value (0-100) for one keyword, or None."""
    s = requests.Session(); s.headers.update(_HEADERS)
    req = {"comparisonItem": [{"keyword": keyword, "geo": geo, "time": timeframe}],
           "category": 0, "property": ""}
    r = s.get(_EXPLORE, params={"hl": "en-US", "tz": "0", "req": json.dumps(req)}, timeout=12)
    if r.status_code != 200:
        raise RuntimeError(f"explore {r.status_code}")
    widgets = _strip(r.text).get("widgets", [])
    w = next((w for w in widgets if w.get("id") == "TIMESERIES"), None)
    if not w:
        raise RuntimeError("no timeseries widget")
    r2 = s.get(_MULTILINE, params={"hl": "en-US", "tz": "0",
               "req": json.dumps(w["request"]), "token": w["token"]}, timeout=12)
    if r2.status_code != 200:
        raise RuntimeError(f"multiline {r2.status_code}")
    vals = [pt["value"][0] for pt in _strip(r2.text).get("default", {}).get("timelineData", [])
            if pt.get("value")]
    return vals[-1] if vals else None


def fetch_trends(keywords: list[str]) -> dict:
    """{keyword: interest 0-100}. Cached ~1h; best-effort (never raises)."""
    global _last_fetch
    now = time.time()
    if _cache and now - _last_fetch < _MIN_INTERVAL:
        return _cache
    out = {}
    try:
        for kw in keywords[:5]:                       # keep it light
            try:
                v = _interest(kw)
                if v is not None:
                    out[kw] = v
                time.sleep(1.0)                        # be gentle
            except Exception as exc:
                logger.info(f"[Trends] '{kw}' failed: {exc}")
        if out:
            _cache.clear(); _cache.update(out); _last_fetch = now
    except Exception as exc:
        logger.info(f"[Trends] fetch failed: {exc}")
    return _cache or out