product_index.py

"""Trek katalog XML'ini bir kez parse edip hash index halinde tutar.
Tum lookup'lar O(1) — re-parse yok. Background refresh ve fetch lock'u var."""
from __future__ import annotations

import asyncio
import logging
import re
import threading
import time

import requests
import urllib3

from config import TREK_XML_URL, TREK_XML_TIMEOUT, CACHE_TTL_TREK_XML

urllib3.disable_warnings(urllib3.exceptions.InsecureRequestWarning)
logger = logging.getLogger(__name__)


# Turkce -> ASCII normalizasyon
_TR_MAP = {
    "İ": "i", "I": "i", "ı": "i",
    "Ğ": "g", "ğ": "g",
    "Ü": "u", "ü": "u",
    "Ş": "s", "ş": "s",
    "Ö": "o", "ö": "o",
    "Ç": "c", "ç": "c",
}


def normalize(s: str) -> str:
    if not s:
        return ""
    for tr, en in _TR_MAP.items():
        s = s.replace(tr, en)
    return s.lower()


# Pre-compiled regex'ler (modul yuklemesinde bir kez)
_ITEM_RE = re.compile(r"<item>(.*?)</item>", re.DOTALL)
_VARIANT_LABEL_SEP = re.compile(r"\s*[-/]\s*")
_TOKEN_RE = re.compile(r"[a-z0-9]+")
_SIZE_PAT = re.compile(
    r"^(?:XX?S|XS|S|M|L|XL|XXL|XXXL|\d{2}(?:\.\d)?(?:\s*CM)?)$", re.I
)

_FIELD_RES = {
    "rootlabel": re.compile(r"<rootlabel><!\[CDATA\[(.*?)\]\]></rootlabel>"),
    "label": re.compile(r"<label><!\[CDATA\[(.*?)\]\]></label>"),
    "productLink": re.compile(r"<productLink><!\[CDATA\[(.*?)\]\]></productLink>"),
    "stockCode": re.compile(r"<stockCode><!\[CDATA\[(.*?)\]\]></stockCode>"),
    "isOptionOfAProduct": re.compile(r"<isOptionOfAProduct>(\d+)</isOptionOfAProduct>"),
    "rootProductStockCode": re.compile(
        r"<rootProductStockCode><!\[CDATA\[(.*?)\]\]></rootProductStockCode>"
    ),
    "priceTaxWithCur": re.compile(r"<priceTaxWithCur>([\d.]+)</priceTaxWithCur>"),
}
def _parse_item(it: str) -> dict:
    """Tek bir <item> blogunu dict'e cevir + arama icin token cache'i ekle.
    Resim alanlari (picture1..8) parse edilmiyor — browser sayfayi kendisi gosteriyor."""
    def grab(name: str) -> str:
        m = _FIELD_RES[name].search(it)
        return m.group(1).strip() if m else ""

    rootlabel = grab("rootlabel")
    var_label = grab("label")
    link = grab("productLink")
    sku = grab("stockCode")
    iv_m = _FIELD_RES["isOptionOfAProduct"].search(it)
    is_variant = bool(iv_m and iv_m.group(1) == "1")
    root_sku_raw = grab("rootProductStockCode")
    root_sku = root_sku_raw if root_sku_raw and root_sku_raw != "0" else None

    price_str = grab("priceTaxWithCur")
    price: float | None = None
    if price_str:
        try:
            p = float(price_str)
            if p > 0:
                price = p
        except (TypeError, ValueError):
            pass

    color: str | None = None
    size: str | None = None
    if is_variant and var_label:
        parts = [p.strip() for p in _VARIANT_LABEL_SEP.split(var_label) if p.strip()]
        for p in parts:
            if _SIZE_PAT.match(p) and not size:
                size = p.upper()
            elif not color:
                color = p.upper()

    label_norm = normalize(rootlabel)
    tokens = [t for t in _TOKEN_RE.findall(label_norm) if len(t) >= 1]

    return {
        "name": rootlabel,
        "link": link,
        "sku": sku,
        "color": color,
        "size": size,
        "price": price,
        "is_variant": is_variant,
        "root_sku": root_sku,
        "_tokens": tokens,
        "_label_norm": label_norm,
    }


def public_view(p: dict | None) -> dict | None:
    """Internal field'lari (_tokens, _label_norm) cikar — client'a gonderilebilir."""
    if not p:
        return None
    return {k: v for k, v in p.items() if not k.startswith("_")}


class ProductIndex:
    """Thread-safe parse-once index. Re-parse sadece XML degisirse."""

    def __init__(self) -> None:
        self._lock = threading.Lock()
        self._fetch_lock = threading.Lock()
        self._xml_data: bytes | None = None
        self._xml_time: float = 0
        self._xml_id: int | None = None
        self.products: list[dict] = []
        self.by_link: dict[str, dict] = {}
        self.by_sku: dict[str, dict] = {}
        self.variants_by_root: dict[str, list[dict]] = {}
        self.main_count: int = 0

    # ---------- XML fetch (lock'lu, thundering herd onleyici) ----------
    def _fetch_xml(self) -> bytes | None:
        with self._fetch_lock:
            now = time.time()
            # Lock alindiginda baska bir thread fetch yapmis olabilir
            if self._xml_data and (now - self._xml_time < CACHE_TTL_TREK_XML):
                return self._xml_data
            try:
                r = requests.get(TREK_XML_URL, verify=False, timeout=TREK_XML_TIMEOUT)
                if r.status_code == 200 and r.content:
                    self._xml_data = r.content
                    self._xml_time = now
                    logger.info(f"[index] fetched Trek XML ({len(r.content)} bytes)")
                    return r.content
            except Exception:
                logger.exception("[index] Trek XML fetch hatasi")
            # Eski (stale) data varsa onu kullan
            return self._xml_data

    # ---------- Index build ----------
    def _build(self, xml_bytes: bytes) -> None:
        text = xml_bytes.decode("utf-8", errors="replace")
        products: list[dict] = []
        by_link: dict[str, dict] = {}
        by_sku: dict[str, dict] = {}
        variants_by_root: dict[str, list[dict]] = {}
        main_count = 0

        for it in _ITEM_RE.findall(text):
            p = _parse_item(it)
            products.append(p)
            if p["link"]:
                by_link[p["link"]] = p
            if p["sku"]:
                by_sku[p["sku"]] = p
            if not p["is_variant"]:
                main_count += 1
            else:
                if p["root_sku"]:
                    variants_by_root.setdefault(p["root_sku"], []).append(p)

        with self._lock:
            self.products = products
            self.by_link = by_link
            self.by_sku = by_sku
            self.variants_by_root = variants_by_root
            self.main_count = main_count
            self._xml_id = id(xml_bytes)

        logger.info(
            f"[index] built: {len(products)} items, {main_count} main, "
            f"{sum(len(v) for v in variants_by_root.values())} variants"
        )

    def ensure(self) -> bool:
        """XML cache'ini guncelle ve index'i (gerekirse) yeniden build et.
        True donerse data hazir. Sync, thread-safe."""
        xml = self._fetch_xml()
        if not xml:
            return False
        if id(xml) != self._xml_id:
            self._build(xml)
        return True

    # ---------- Public lookups (sync, hizli) ----------
    def find_by_link(self, link: str) -> dict | None:
        """productLink ile eslesen urun. Varyant ise ana urune cikar."""
        if not link:
            return None
        with self._lock:
            p = self.by_link.get(link.strip())
            if not p:
                return None
            if p["is_variant"] and p["root_sku"]:
                main = self.by_sku.get(p["root_sku"])
                if main and not main["is_variant"]:
                    return public_view(main)
            return public_view(p)

    def variants_of(self, main_link: str) -> list[dict]:
        """Verilen ana urunun tum varyantlari."""
        if not main_link:
            return []
        with self._lock:
            main = self.by_link.get(main_link.strip())
            if not main or main["is_variant"] or not main["sku"]:
                return []
            return [public_view(v) for v in self.variants_by_root.get(main["sku"], [])]

    def snapshot(self) -> list[dict]:
        """Tum urunlerin internal-field'li listesi (matcher icin)."""
        with self._lock:
            return list(self.products)


# Singleton
_index = ProductIndex()


def get_index() -> ProductIndex:
    return _index


async def background_refresh_loop(interval: int):
    """Periodic XML refresh in thread executor."""
    while True:
        try:
            await asyncio.to_thread(_index.ensure)
        except Exception:
            logger.exception("background_refresh_loop hatasi")
        await asyncio.sleep(interval)