spider_work/qiandao_spider/qd_toy_spider/hmc256_demo.py

# -*- coding: utf-8 -*-
# Author : Charley
# Python : 3.10.8
# Date   : 2025/7/3 16:36
import base64
import re
import urllib.parse
from collections import OrderedDict
from typing import List, Dict


# ========================
# Step 1: 模拟 JS 中的 $u 映射表
# ========================
def get_string(index):
    lu_list = [
        "dat", "htt", "ueO", "1669242vsNCrp", "hash", "2079476BHgCTx", "MzA", "2651740OMCnHx", "UKA", "development",
        "lYm", "linjieapp.com", "337909LsZqJc", "hanfu086.com", "some", "olv", "path", "10BvkOVt", "url",
        "563436jgaQxo",
        "0123456789abcdef", "ine", "hBd", "que", "fromCharCode", "1Y2", "T9f", "cke", "charAt", "l0b", "JlY", "Lio",
        "3NT", "UNH", "qiandao.com", "yxt", "red", "location", "WZh", "per", "462588MQJvSP", "sor", "HMA",
        "1647704YNAAwy",
        "length", "wM2", "charCodeAt", "972075UMDbSu", "20AarWCs", "def", "VjY", "iYW", "Q1M", "ZhY", "eUL", "rZB",
        "N2Y", "val", "161519TCUroc", "yNj", "isA", "8eV", "qiandao.cn", "pi.", "gRI", "JmZ", "concat", "ogr",
        "502096DJfVUj", "web", "0123456789ABCDEF", "nda", "DxB", "-Id", "136664rALOgK", "OTY", "6AoEOPC", "9JilTKK",
        "4977ANskNQ", "ZaC", "query", "host", "shift", "push", "equ", "str", "53FmkAxN", "spl", "341952KmUHty", "Tim",
        "forEach", "echoing.tech", "rra", "X-R", "fT6", "ipr", "184MSesMH", "test", "Pro", "NWR", "ues", "n/n",
        "735588HRQEfZ", "O1N", "1345113mxYnWn", "2460130PiswuR", "protocol", "production", "14037hrVEJe",
        "986934SQjjnp",
        "8bUOEvE", "HA2", "echo.tech", "localhost", "par", "1195uBtgZp", "res", "est", "//a", "3647kNKLxG",
        "12463moFXsz",
        "o.c", "28yNQNBZ", "abs", "28503cIknPz", "wXW", "min", "get", "req", "WM1", "354926IYrsjA", "44wJFZkm",
        "5868HCgLkL", "2274EbjwLA", "NTB", "qia", "9150940OsTyJa", "key", "ps:"
    ]
    return lu_list[index - 469]


# ========================
# Step 2: 实现 au() - UTF-8 编码字符串
# ========================
def au(s: str) -> bytes:
    result = []
    for c in s:
        if ord(c) <= 0x7F:
            result.append(ord(c))
        elif ord(c) <= 0x7FF:
            result.extend([
                0xC0 | (ord(c) >> 6),
                0x80 | (ord(c) & 0x3F)
            ])
        elif ord(c) <= 0xFFFF:
            result.extend([
                0xE0 | (ord(c) >> 12),
                0x80 | ((ord(c) >> 6) & 0x3F),
                0x80 | (ord(c) & 0x3F)
            ])
        elif ord(c) <= 0x1FFFFF:
            result.extend([
                0xF0 | (ord(c) >> 18),
                0x80 | ((ord(c) >> 12) & 0x3F),
                0x80 | ((ord(c) >> 6) & 0x3F),
                0x80 | (ord(c) & 0x3F)
            ])
    return bytes(result)


# ========================
# Step 3: 实现 su() - 字符串转整型数组
# ========================
def su(data: bytes) -> List[int]:
    out = [0] * (len(data) // 4)
    for i in range(len(out)):
        for j in range(4):
            if i * 4 + j < len(data):
                out[i] |= data[i * 4 + j] << (24 - j * 8)
    return out


# ========================
# Step 4: 实现 pu() - 自定义哈希函数（类似 SHA-256）
# ========================
def cu(x, n):
    return (x >> n) | ((x & ((1 << n) - 1)) << (32 - n))


def hu(a, b):
    lower = (a & 0xFFFF) + (b & 0xFFFF)
    upper = (a >> 16) + (b >> 16) + (lower >> 16)
    return (upper << 16) | (lower & 0xFFFF)


# fu = [
#     1116352408, 1899447441, -1245643825, -373957723, 961987163, 1508970993, -1841331548, -1424204075,
#     -670586216, 310598401, 607225278, 1426881987, 1925078388, -2132889090, -1680079193, -1046744716
# ]

fu = [1116352408, 1899447441, -1245643825, -373957723, 961987163, 1508970993, -1841331548, -1424204075, -670586216,
      310598401, 607225278, 1426881987, 1925078388, -2132889090, -1680079193, -1046744716, -459576895, -272742522,
      264347078, 604807628, 770255983, 1249150122, 1555081692, 1996064986, -1740746414, -1473132947, -1341970488,
      -1084653625, -958395405, -710438585, 113926993, 338241895, 666307205, 773529912, 1294757372, 1396182291,
      1695183700, 1986661051, -2117940946, -1838011259, -1564481375, -1474664885, -1035236496, -949202525, -778901479,
      -694614492, -200395387, 275423344, 430227734, 506948616, 659060556, 883997877, 958139571, 1322822218, 1537002063,
      1747873779, 1955562222, 2024104815, -2067236844, -1933114872, -1866530822, -1538233109, -1090935817, -965641998]

# print(len(fu))


def pu(data: List[int], length: int) -> List[int]:
    # 计算需要的最小 msg 长度（每项为 32 位整数）
    needed_blocks = (length + 64) // (32 * 4) + 1
    needed_length = needed_blocks * 16  # 每块 16 个整数
    msg = data[:]
    while len(msg) < needed_length:
        msg.append(0)

    # 填充长度信息
    msg[length >> 5] |= 0x80 << (24 - length % 32)
    msg[15 + ((length + 64) >> 9 << 4)] = length

    h = [1779033703, -1150833019, 1013904242, -1521486534, 1359893119, -1694144372, 528734635, 1541459225]

    for i in range(0, len(msg), 16):
        # 确保 w 至少有 64 个元素
        w = [0] * 64
        for x in range(min(len(msg) - i, 16)):
            w[x] = msg[i + x]

        for j in range(16, 64):
            s0 = cu(w[j - 15], 7) ^ cu(w[j - 15], 18) ^ (w[j - 15] >> 3)
            s1 = cu(w[j - 2], 17) ^ cu(w[j - 2], 19) ^ (w[j - 2] >> 10)
            w[j] = (w[j - 16] + s0 + w[j - 7] + s1) & 0xFFFFFFFF

        a, b, c, d, e, f, g, h_ = h

        for j in range(64):
            s1 = cu(e, 6) ^ cu(e, 11) ^ cu(e, 25)
            ch = (e & f) ^ (~e & g)
            temp = (s1 + ch) & 0xFFFFFFFF
            s0 = cu(a, 2) ^ cu(a, 13) ^ cu(a, 22)
            maj = (a & b) ^ (a & c) ^ (b & c)
            t2 = (s0 + maj) & 0xFFFFFFFF
            t1 = (h_ + temp + fu[j] + w[j]) & 0xFFFFFFFF

            h_ = g
            g = f
            f = e
            e = (d + t1) & 0xFFFFFFFF
            d = c
            c = b
            b = a
            a = (t1 + t2) & 0xFFFFFFFF

        h = [(h[k] + [a, b, c, d, e, f, g, h_][k]) & 0xFFFFFFFF for k in range(8)]

    return h


# ========================
# Step 5: 实现 iu() - Base64 编码
# ========================
def iu(data: bytes, url_safe: bool = False) -> str:
    encoded = base64.b64encode(data).decode()
    if url_safe:
        encoded = encoded.replace('+', '-').replace('/', '_').rstrip('=')
    return encoded


# ========================
# Step 6: 构建签名数据 w
# ========================
def parse_url(url: str) -> dict:
    pattern = re.compile(r'^((?:https?|ftp):)?//([^/?#]+)([^?#]*)(\?[^#]*)?(#.*)?$')
    match = pattern.match(url)
    keys = ['protocol', 'host', 'path', 'query', 'hash']
    return {k: (match.group(i + 1) if match and match.group(i + 1) else '') for i, k in enumerate(keys)}


def parse_query(query: str) -> Dict[str, List[str]]:
    params = {}
    if query.startswith('?'):
        query = query[1:]
    for part in query.split('&'):
        if '=' in part:
            k, v = part.split('=', 1)
            k, v = urllib.parse.unquote(k), urllib.parse.unquote(v)
            params.setdefault(k, []).append(v)
    return params


def sort_and_normalize(params: Dict[str, list]) -> Dict[str, list]:
    sorted_params = OrderedDict()
    for key in sorted(params.keys()):
        values = params[key]
        if isinstance(values, list) and len(values) == 1:
            sorted_params[key] = values[0]
        else:
            sorted_params[key] = values
    return sorted_params


def build_sign_data(url: str) -> str:
    parsed = parse_url(url)
    query_str = parsed['query']
    params = parse_query(query_str)
    sorted_params = sort_and_normalize(params)

    # 拼接参数部分
    param_str = ''
    for k, v in sorted_params.items():
        if isinstance(v, list):
            for val in v:
                param_str += f"{urllib.parse.quote(k)}={urllib.parse.quote(val)}&"
        else:
            param_str += f"{urllib.parse.quote(k)}={urllib.parse.quote(v)}&"
    param_str = param_str.rstrip('&')

    # 拼接签名字符串
    sign_data = parsed['path'] + parsed['host'] + param_str
    return sign_data


# ========================
# Step 7: 最终签名函数
# ========================
def generate_sign_(m: str, url: str) -> str:
    w = build_sign_data(url)
    x = au(m)
    S = au(w)

    # Step 1: su(x)
    r = su(x)
    if len(r) > 16:
        r = pu(r, len(x))

    # 确保 r 长度至少为 16
    while len(r) < 16:
        r.append(0)

    # Step 2: 异或混淆
    i = [909522486 ^ r[a] for a in range(16)]
    o = [1549556828 ^ r[a] for a in range(16)]

    # Step 3: su(S)
    e_bytes = su(S)
    s = pu(i + e_bytes, 512 + 8 * len(S))
    final_hash = pu(o + s, 768)

    # Step 4: 转换为十六进制字符串
    hash_hex = ''.join(f"{(word & 0xFFFFFFFF):08x}" for word in final_hash)

    # Step 5: 将 hex 字符串转为 bytes 并进行 Base64 编码
    hash_bytes = hash_hex.encode('utf-8')
    return base64.b64encode(hash_bytes).decode('utf-8')

import hmac
from hashlib import sha256


def generate_sign(m: str, url: str) -> str:
    w = build_sign_data(url)

    # 使用 HMAC-SHA256 计算签名
    key = m.encode('utf-8')
    msg = w.encode('utf-8')
    digest = hmac.new(key, msg, sha256).digest()

    # 转换为十六进制字符串
    hash_hex = digest.hex()

    # Base64 编码
    return base64.b64encode(hash_hex.encode()).decode()


# ========================
# Step 8: 示例使用
# ========================
if __name__ == '__main__':
    m = "8eVqsqhBdEKxeULQ1M72yrZB5HET9fx0"
    # url = 'https://api.qiandao.com/shelf-web/list-home-recommender?limit=10&offset=280&cmd=b2c_homepage_feed&project=channel&name=manghe&type%5B%5D=BLIND_BOX_MACHINE&type%5B%5D=KUJI&type%5B%5D=NEW_LUCKY_BAG&type%5B%5D=B2C&typeIds%5B%5D=15&channelId=397228442736842784&screenName=%E6%8E%A8%E8%8D%90'
    # url = 'https://api.qiandao.com/shelf-web/list-home-recommender?limit=10&offset=300&cmd=b2c_homepage_feed&project=channel&name=manghe&type%5B%5D=BLIND_BOX_MACHINE&type%5B%5D=KUJI&type%5B%5D=NEW_LUCKY_BAG&type%5B%5D=B2C&typeIds%5B%5D=15&channelId=397228442736842784&screenName=%E6%8E%A8%E8%8D%90'
    url = "/shelf-web/list-home-recommendertypeIds%5B%5D=15&type%5B%5D=NEW_LUCKY_BAG&type%5B%5D=KUJI&type%5B%5D=BLIND_BOX_MACHINE&type%5B%5D=B2C&screenName=%E6%8E%A8%E8%8D%90&project=channel&offset=310&name=manghe&limit=10&cmd=b2c_homepage_feed&channelId=3972284427368427841751530694089"

    t ="https://api.qiandao.com/shelf-web/list-home-recommender?limit=10&offset=310&cmd=b2c_homepage_feed&project=channel&name=manghe&type%5B%5D=BLIND_BOX_MACHINE&type%5B%5D=KUJI&type%5B%5D=NEW_LUCKY_BAG&type%5B%5D=B2C&typeIds%5B%5D=15&channelId=397228442736842784&screenName=%E6%8E%A8%E8%8D%90"
    timestr = '1751530647285'
    signature = generate_sign(m, url)
    print("Signature:", signature)

    # 原始签名
    # OTMxYmRiMDhhYjYxM2IyMDBiN2U0ZGFlNDUzNGU4YzhiZWYxOTk4MjNjYTVkODg4OWYzZDJkOTUxMjU2Zjc2Yw==

    # YzNjOTNiY2NlN2UzYTcwNmJlNzQ0YTJiOTJjMTkzYTAzNmRiZGQyNjZiNGZjMzM5NDNiMGUxNjQ4MDQwYzAzYw==

    "OTYwYjQ1NWI2ZmYyMGNjNzQ1NDJhNmFhY2I0OGJlMTZjZGM3NzVjZjJhZTRiMzIxZTQxYTg0NTZmZjVkODgwZg=="