rakuten-api/app/scraping/parsers/rakuma/shop.py

"""ラクマ 店铺页 HTML → 卖家详情与卖家商品列表

C2C 集市里的「商家」就是个人卖家，页面在 fril.jp/shop/{hash}：
- 店铺页本身：卖家资料 + 该卖家的商品分页列表（含已售出）
- /review 子页：评价明细与好评/普通/差评分档计数

商品卡片与搜索页共用 `.item-box` 结构，直接复用 search 里的解析。
"""
from __future__ import annotations

import json
import re

from selectolax.parser import HTMLParser

from app.scraping.core import rakuma_site as site
from app.shared.errors import ScrapeParseError
from app.scraping.models.scrape import (
    RakumaRatingBreakdown,
    RakumaReview,
    RakumaShopDetailData,
    RakumaShopItemsData,
)
from app.scraping.parsers.rakuma.base import (
    attr,
    event_payload,
    image_url,
    node_text,
    parse_float,
    parse_int,
    parse_total_count,
)
from app.scraping.parsers.rakuma.search import parse_item_cards

# 评价条目标题左侧的图标 class → 评价档位
_RATING_ICONS = {
    "icon_review_sun": "good",  # よい
    "icon_review_cloud": "normal",  # ふつう
    "icon_review_rain": "bad",  # わるい
}

# /review 页上三组分档计数的容器 id 前缀
_ALL_RATINGS_PREFIX = "all"
_SELLER_RATINGS_PREFIX = "seller"

_LD_JSON_RE = re.compile(
    r'<script[^>]*type="application/ld\+json"[^>]*>(.*?)</script>', re.S
)


def _require_shop_page(html: str) -> HTMLParser:
    tree = HTMLParser(html)
    if tree.css_first(f".{site.SHOP_PAGE_MARKER}") is None:
        raise ScrapeParseError("页面不是店铺页（缺少店铺资料区块）")
    return tree


def _parse_rating_breakdown(tree: HTMLParser, prefix: str) -> RakumaRatingBreakdown:
    """解析一组好评/普通/差评计数

    页面用 `<ul class="nav-pills">` 里的三个链接展示，锚点形如
    `#all-good` / `#seller-normal`，按锚点前缀区分「全部」与「出品」两组。
    """
    counts = {"good": 0, "normal": 0, "bad": 0}
    for link in tree.css("ul.nav-pills a"):
        href = attr(link, "href")
        for rating in counts:
            if href == f"#{prefix}-{rating}":
                counts[rating] = parse_int(node_text(link))
    return RakumaRatingBreakdown(**counts)


def _parse_reviews(tree: HTMLParser) -> list[RakumaReview]:
    """解析评价列表

    站点在「すべての評価」标签页里最多展示最新 100 条，且三个标签页
    （全部/出品/购入）的条目在 DOM 里重复出现，这里只取第一个激活面板。
    """
    panel = tree.css_first("#all-all") or tree.css_first(".tab-pane.active")
    if panel is None:
        return []

    reviews: list[RakumaReview] = []
    for article in panel.css("article.review-item"):
        title_node = article.css_first(".review-item-title")
        icon = title_node.css_first("i") if title_node else None
        classes = attr(icon, "class").split() if icon else []
        rating = next((_RATING_ICONS[name] for name in classes if name in _RATING_ICONS), "")

        reviews.append(
            RakumaReview(
                rating=rating,
                title=node_text(title_node),
                comment=node_text(article.css_first(".review-item-text")),
                reviewer_name=node_text(article.css_first(".review-item-name")),
                reviewed_at=node_text(article.css_first(".review-item-date")),
            )
        )
    return reviews


def _parse_store_rating(html: str) -> tuple[float, int]:
    """从店铺页的 ld+json Store 节点取评分与评价数

    评价数只有这里给得出来——可见 DOM 上只有星级和分数，没有条数。
    """
    for match in _LD_JSON_RE.finditer(html):
        try:
            data = json.loads(match.group(1))
        except ValueError:
            continue
        if not isinstance(data, dict) or data.get("@type") != "Store":
            continue
        rating = data.get("aggregateRating")
        if isinstance(rating, dict):
            return parse_float(rating.get("ratingValue")), parse_int(rating.get("ratingCount"))
    return 0.0, 0


def parse_shop_detail(
    html: str, *, shop_id: str, shop_url: str, review_html: str | None = None
) -> RakumaShopDetailData:
    """把店铺页 HTML 解析为卖家详情

    Args:
        review_html: /review 子页的 HTML；给出时才填充评价明细与分档计数

    Raises:
        ScrapeParseError: 页面不是店铺页
    """
    tree = _require_shop_page(html)

    total_count, _, _ = parse_total_count(node_text(tree.css_first(".page-count")))
    badge = tree.css_first(".badge-status")
    verification_label = node_text(badge)

    # 简介在侧栏「プロフィール」区块；卖家未填写时站点会写一句占位文案
    introduction = node_text(tree.css_first("[data-test=profile-text-top]"))
    if "設定されていません" in introduction:
        introduction = ""

    score, review_count = _parse_store_rating(html)
    detail = RakumaShopDetailData(
        shop_id=shop_id,
        shop_name=node_text(tree.css_first(".profile-area__shop-name")),
        user_name=node_text(tree.css_first("[data-test=profile_user_name], .profile-area__user-name")),
        shop_url=shop_url,
        icon_url=image_url(tree.css_first(".profile-area__user-icon img")),
        cover_url=_cover_url(tree),
        introduction=introduction,
        review_score=score or parse_float(node_text(tree.css_first(".shop_score__score"))),
        review_count=review_count,
        is_verified="未完了" not in verification_label and bool(verification_label),
        verification_label=verification_label,
        item_count=total_count,
    )

    # 用户数值 ID：优先取商品卡片埋点里的 seller_user_id（店铺页所有商品都属于
    # 该卖家），卖家未设头像时头像地址是站点默认图，取不到 ID。
    detail.user_id = _seller_user_id(tree) or _user_id_from_icon(detail.icon_url)

    if review_html is not None:
        review_tree = HTMLParser(review_html)
        detail.rating_breakdown = _parse_rating_breakdown(review_tree, _ALL_RATINGS_PREFIX)
        detail.seller_rating_breakdown = _parse_rating_breakdown(review_tree, _SELLER_RATINGS_PREFIX)
        detail.reviews = _parse_reviews(review_tree)

    return detail


def _cover_url(tree: HTMLParser) -> str:
    """封面图挂在 inline style 的 background url() 里"""
    cover = tree.css_first(".profile-area__shop-cover")
    style = attr(cover, "style")
    start = style.find("url(")
    if start < 0:
        return ""
    end = style.find(")", start)
    return style[start + 4 : end].strip("'\" ") if end > start else ""


def _seller_user_id(tree: HTMLParser) -> str:
    """从店铺页商品卡片的埋点里取卖家数值 ID"""
    for node in tree.css("[data-gtm-click], [onclick]"):
        payload = event_payload(node)
        user_id = payload.get("seller_user_id")
        if user_id:
            return str(user_id)
    return ""


def _user_id_from_icon(icon_url: str) -> str:
    """从头像地址 https://img.fril.jp/user/{id}/s/{id}.jpg 里取用户数值 ID"""
    marker = "/user/"
    start = icon_url.find(marker)
    if start < 0:
        return ""
    rest = icon_url[start + len(marker) :]
    user_id = rest.split("/", 1)[0]
    return user_id if user_id.isdigit() else ""


def parse_shop_items(
    html: str, *, shop_id: str, request_url: str, page: int
) -> RakumaShopItemsData:
    """把店铺页 HTML 解析为该卖家的商品列表

    Raises:
        ScrapeParseError: 页面不是店铺页
    """
    tree = _require_shop_page(html)
    items = parse_item_cards(tree)
    total_count, _, end = parse_total_count(node_text(tree.css_first(".page-count")))

    return RakumaShopItemsData(
        shop_id=shop_id,
        shop_name=node_text(tree.css_first(".profile-area__shop-name")),
        page=page,
        total_count=total_count,
        has_more=bool(items) and bool(end) and end < total_count,
        request_url=request_url,
        items=items,
    )