Files
surugaya/app/utils/parse_util.py
T
2026-06-09 14:55:00 +08:00

186 lines
6.2 KiB
Python

"""骏河屋页面解析工具函数"""
import re
from urllib.parse import parse_qs, urlparse
def extract_low_price(text: str) -> int | None:
"""从价格范围文本中提取最低价格
例如 "¥4,700 ~ ¥6,200" 会返回 4700
"""
prices = re.findall(r'(\d[0-9,.]*)', text)
if prices:
low_price_str = prices[0].replace(',', '')
return int(float(low_price_str))
return None
def format_price(price_text: str) -> int | None:
"""格式化商品价格文字为整数(日元)
支持的格式:
- 单价:¥4,700 → 4700
- 范围:¥4,700 ~ ¥6,200 → 4700(取最低价)
- 円后缀:4700円 → 4700
"""
if not price_text:
return None
# 价格范围,取最低价
if "" in price_text:
return extract_low_price(price_text)
match = re.search(r"[¥¥]\s*([0-9][0-9,]*)", price_text)
if not match:
match = re.search(r"([0-9][0-9,]*)\s*円", price_text)
if not match:
return None
try:
return int(match.group(1).replace(",", ""))
except ValueError:
return None
def normalize_goods_id(goods_id: str) -> str:
"""
高效处理商品ID:如果包含字母则转小写。
Python 的 lower() 会处理整个字符串,但在全是数字的情况下开销极小。
"""
# 如果 ID 已经是全数字,直接返回原对象,避免额外的处理
if goods_id.isdigit():
return goods_id
return goods_id.lower()
def surugaya_photo_url_to_cdn_m(photo_url: str | None = None, item_id: str | None = None) -> str:
"""将骏河屋商品照片 URL 转换为 CDN URL(webp 格式)
示例:
输入: https://www.suruga-ya.jp/database/photo.php?shinaban=220035970&size=m
输出: https://cdn.suruga-ya.jp/pics_webp/boxart_m/220035970m.jpg.webp
当 photo_url 无法解析时,回退使用 item_id 拼接 CDN 地址;
两者都为空时返回空字符串。
"""
# 如果无法从 photo_url 解析出商品编号,则用 item_id 直接拼接
if not photo_url:
if item_id:
return f"https://cdn.suruga-ya.jp/pics_webp/boxart_m/{normalize_goods_id(item_id)}m.jpg.webp"
return ""
parsed = urlparse(photo_url)
qs = parse_qs(parsed.query)
shinaban = (qs.get("shinaban") or [None])[0]
if not shinaban:
# photo_url 中没有 shinaban 参数,回退到 item_id
if item_id:
return f"https://cdn.suruga-ya.jp/pics_webp/boxart_m/{normalize_goods_id(item_id)}m.jpg.webp"
return photo_url
size = (qs.get("size") or ["m"])[0] or "m"
size = str(size).strip().lower()
folder = f"boxart_{size}"
filename = f"{normalize_goods_id(shinaban)}{size}"
return f"https://cdn.suruga-ya.jp/pics_webp/{folder}/{filename}.jpg.webp"
def surugaya_photo_url_to_cdn(photo_url: str | None = None, item_id: str | None = None) -> str:
"""将骏河屋商品照片 URL 转换为 CDN URL(webp 格式)
示例:
输入: https://www.suruga-ya.jp/database/photo.php?shinaban=220035970&size=m
输出: https://cdn.suruga-ya.jp/database/pics_webp/game/220035970.jpg.webp
优先使用 item_id 拼接 CDN 地址;
两者都为空时返回空字符串。
"""
if item_id:
return f"https://cdn.suruga-ya.jp/database/pics_webp/game/{normalize_goods_id(item_id)}.jpg.webp"
if photo_url:
parsed = urlparse(photo_url)
qs = parse_qs(parsed.query)
shinaban = (qs.get("shinaban") or [None])[0]
if not shinaban:
return photo_url
return f"https://cdn.suruga-ya.jp/database/pics_webp/game/{normalize_goods_id(shinaban)}.jpg.webp"
return ""
def extract_product_id(url: str | None = None) -> str:
"""从各种商品URL或路径中精准提取商品ID"""
if not url:
return ""
# 核心逻辑:匹配 /detail/ 后面紧跟的非斜杠、非问号、非点号的连续字符
match = re.search(r"/detail/([^/?.]+)", url)
if match:
return match.group(1)
return ""
# --- 官方店铺日本境内运费规则 ---
# amount < OFFICIAL_SHIPPING_FEE_LOW_THRESHOLD: OFFICIAL_SHIPPING_FEE_LOW
# OFFICIAL_SHIPPING_FEE_LOW_THRESHOLD <= amount < OFFICIAL_SHIPPING_FEE_HIGH_THRESHOLD: OFFICIAL_SHIPPING_FEE_MID
# amount >= OFFICIAL_SHIPPING_FEE_HIGH_THRESHOLD: 免运费(0)
OFFICIAL_SHIPPING_FEE_LOW_THRESHOLD = 1000
OFFICIAL_SHIPPING_FEE_HIGH_THRESHOLD = 1500
OFFICIAL_SHIPPING_FEE_LOW = 440
OFFICIAL_SHIPPING_FEE_MID = 385
# --- 加盟店日本境内运费规则 ---
# 每 1 家加盟店,无论订单数量与金额,统一收 FRANCHISE_SHIPPING_FEE_PER_STORE 日元
FRANCHISE_SHIPPING_FEE_PER_STORE = 800
# --- 通贩手续费规则 ---
# amount < HANDLE_FEE_FREE_THRESHOLD: HANDLE_FEE_BELOW
# amount >= HANDLE_FEE_FREE_THRESHOLD: 免手续费(0)
HANDLE_FEE_FREE_THRESHOLD = 5000
HANDLE_FEE_BELOW = 240
def get_shipping_fee(amount: int) -> int:
"""计算官方店铺日本境内运费(本州/四国/九州统一价)。
规则:
- amount < 1000 日元: 440 日元
- 1000 <= amount < 1500 日元: 385 日元
- amount >= 1500 日元: 0(免运费)
备注: 北海道/冲绳/离岛可能需额外支付;此处按本州口径计算。
加盟店运费另由 get_franchise_shipping_fee 计算。
"""
if amount >= OFFICIAL_SHIPPING_FEE_HIGH_THRESHOLD:
return 0
if amount >= OFFICIAL_SHIPPING_FEE_LOW_THRESHOLD:
return OFFICIAL_SHIPPING_FEE_MID
return OFFICIAL_SHIPPING_FEE_LOW
def get_franchise_shipping_fee(franchise_store_count: int) -> int:
"""计算加盟店日本境内运费。
每 1 家加盟店统一收 800 日元,与订单数量/金额无关。
franchise_store_count <= 0 时返回 0。
"""
if franchise_store_count <= 0:
return 0
return franchise_store_count * FRANCHISE_SHIPPING_FEE_PER_STORE
def get_handle_fee(amount: int) -> int:
"""计算通贩手续费(本州/四国/九州口径)。
规则:
- amount < 5000 日元: 240 日元
- amount >= 5000 日元: 0(免手续费)
参考: https://www.suruga-ya.jp/man/qa/hanbai_qa/shiharai.html#shiharai_1_2
"""
if amount >= HANDLE_FEE_FREE_THRESHOLD:
return 0
return HANDLE_FEE_BELOW