fix(bv2av): negative av number

原转换代码针对 2023 年 3 月 28 日 19:45:02 (av99999999) 之后的新视频出现了转换为负数 av 号的 bug，此处使用新的算法来实现：
https://github.com/SocialSisterYi/bilibili-API-collect/blob/master/docs/misc/bvid_desc.md

Author: otomad

.gitignore

@@ -12,4 +12,5 @@ desktop.ini
 217/217bak/
 hotkid/tel/old_ver2
 wwwroot/aspnet_client
-generator/pyfmt/c/pyfmt.exe
+generator/pyfmt/c/pyfmt.exe
+__pycache__/

math/bv2av/BV 号跋扈 ～ Who done it!.mjs

@@ -1,69 +1,65 @@
-// 改写自 https://www.zhihu.com/question/381784377/answer/1099438784，并加上一些适当的处理
-// 我这人虽然是写 JS 的，但是看 Python 不是问题
+// 原转换代码针对 2023 年 3 月 28 日 19:45:02 (av99999999) 之后的新视频出现了转换为负数 av 号的 bug，此处使用新的算法来实现：
+// https://github.com/SocialSisterYi/bilibili-API-collect/blob/master/docs/misc/bvid_desc.md
 
 // @ts-check
-"use strict";
 
-const table = [..."fZodR9XQDSUm21yCkr6zBqiveYah8bt4xsWpHnJE7jL5VG3guMTKNPAwcF"];
-const slots = [11, 10, 3, 8, 4, 6];
-const xor = 177451812;
-const add = 8728348608;
+const XOR_CODE = 23442827791579n;
+const MASK_CODE = 2251799813685247n;
+const MAX_AID = 1n << 51n;
+const BASE = 58n;
 const ERROR = "¿你在想桃子？";
 
-export const av2bv = (/** @type {string | number | bigint} */ av) => {
-	/** @type {number | undefined} */
-	let num;
+const data = 'FcwAPNKTMug3GV5Lj7EJnHpWsx4tb8haYeviqBz6rkCy12mUSDQX9RdoZf';
+
+export function av2bv(/** @type {bigint | number | string} */ aid) {
 	try {
-		if (typeof av === "number") {
-			num = Math.trunc(av);
-		} else if (typeof av === "string") {
-			num = Number(av.replace(/[^0-9]/gu, ""));
-		} else if (typeof av === "bigint") {
-			num = Number(av);
-		};
-		if (num === undefined || !Number.isFinite(num) || num <= 0) {
-			throw new Error();
-		};
+		if (typeof aid === "bigint") { }
+		else if (typeof aid === "number") {
+			aid = BigInt(Math.trunc(aid));
+		} else if (typeof aid === "string") {
+			aid = BigInt(aid.replace(/[^0-9]/gu, ""));
+		} else throw new Error();
+		if (aid === undefined || aid <= 0) throw new Error();
 	} catch (cause) {
 		console.error(new Error(ERROR, { cause }))
 		return ERROR;
 	}
 
-	num = (num ^ xor) + add;
-	let result = [..."BV1  4 1 7  "];
-	let i = 0;
-	while (i < 6) {
-		// 这里改写差点犯了运算符优先级的坑
-		// 果然 Python 也不是特别熟练
-		// 说起来 ** 按照传统语法应该写成 Math.pow()，但是我个人更喜欢 ** 一些
-		result[slots[i]] = table[Math.floor(num / 58 ** i) % 58];
-		i++;
-	};
-	return result.join("");
-};
+	const bytes = [..."BV1000000000"];
+	let bvIndex = bytes.length - 1;
+	let tmp = (MAX_AID | BigInt(aid)) ^ XOR_CODE;
+	while (tmp > 0) {
+		bytes[bvIndex] = data[Number(tmp % BigInt(BASE))];
+		tmp = tmp / BASE;
+		bvIndex -= 1;
+	}
+	[bytes[3], bytes[9]] = [bytes[9], bytes[3]];
+	[bytes[4], bytes[7]] = [bytes[7], bytes[4]];
+	return bytes.join('');
+}
 
-export const bv2av = (/** @type {string} */ bv) => {
-	let str = "";
-	if (bv.length === 12) {
-		str = bv;
-	} else if (bv.length === 10) {
-		str = `BV${bv}`;
-		// 根据官方 API，BV 号开头的 BV1 其实可以省略
-		// 不过单独省略个 B 又不行（
-	} else if (bv.length === 9) {
-		str = `BV1${bv}`;
+export function bv2av(/** @type {`BV1${string}`} */ bvid) {
+	if (bvid.length === 12) { }
+	else if (bvid.length === 10) {
+		bvid = /** @type {never} */ (`BV${bvid}`);
+		// 根据官方 API，BV 号开头的 BV1 其实可以省略，不过单独省略个 B 又不行。
+	} else if (bvid.length === 9) {
+		bvid = `BV1${bvid}`;
 	} else {
 		return ERROR;
 	};
-	if (!str.match(/[Bb][Vv][fZodR9XQDSUm21yCkr6zBqiveYah8bt4xsWpHnJE7jL5VG3guMTKNPAwcF]{10}/gu)) {
+	if (!bvid.match(/[Bb][Vv][fZodR9XQDSUm21yCkr6zBqiveYah8bt4xsWpHnJE7jL5VG3guMTKNPAwcF]{10}/gu)) {
 		return ERROR;
 	};
 
-	let result = 0;
-	let i = 0;
-	while (i < 6) {
-		result += table.indexOf(str[slots[i]]) * 58 ** i;
-		i += 1;
-	};
-	return `av${result - add ^ xor}`;
-};
+	const bvidArr = Array.from(bvid);
+	[bvidArr[3], bvidArr[9]] = [bvidArr[9], bvidArr[3]];
+	[bvidArr[4], bvidArr[7]] = [bvidArr[7], bvidArr[4]];
+	bvidArr.splice(0, 3);
+	const tmp = bvidArr.reduce((pre, bvidChar) => pre * BASE + BigInt(data.indexOf(bvidChar)), 0n);
+	const aid = (tmp & MASK_CODE) ^ XOR_CODE;
+	return `av${aid}`;
+}
+
+// console.log(av2bv(111298867365120));
+// console.log(bv2av('BV1L9Uoa9EUx'));

math/bv2av/python/bv2av.py

@@ -1,45 +1,56 @@
 #!/usr/bin/env python3
 import sys
 
-table = 'fZodR9XQDSUm21yCkr6zBqiveYah8bt4xsWpHnJE7jL5VG3guMTKNPAwcF'
-transform: dict[str, int] = {}
-for i in range(58):
-	transform[table[i]] = i
-slots = [11, 10, 3, 8, 4, 6]
-xor = 177451812
-add = 8728348608
-error = '¿你在想桃子？'
-
-def decode(input: str) -> int:
-	try:
-		result: int = 0
-		for i in range(6):
-			result += transform[input[slots[i]]] * 58 ** i
-		return (result - add) ^ xor
-	except:
-		print(error)
-		exit()
+# 原转换代码针对 2023 年 3 月 28 日 19:45:02 (av99999999) 之后的新视频出现了转换为负数 av 号的 bug，此处使用新的算法来实现：
+# https://github.com/SocialSisterYi/bilibili-API-collect/blob/master/docs/misc/bvid_desc.md
+
+XOR_CODE = 23442827791579
+MASK_CODE = 2251799813685247
+MAX_AID = 1 << 51
+ALPHABET = "FcwAPNKTMug3GV5Lj7EJnHpWsx4tb8haYeviqBz6rkCy12mUSDQX9RdoZf"
+ENCODE_MAP = 8, 7, 0, 5, 1, 3, 2, 4, 6
+DECODE_MAP = tuple(reversed(ENCODE_MAP))
+
+BASE = len(ALPHABET)
+PREFIX = "BV1"
+PREFIX_LEN = len(PREFIX)
+CODE_LEN = len(ENCODE_MAP)
+ERROR = "¿你在想桃子？"
 
-def encode(input: int) -> str:
-	input = (input ^ xor) + add
-	result = list('BV1  4 1 7  ')
-	for i in range(6):
-		result[slots[i]] = table[input // 58 ** i % 58]
-	return ''.join(result)
+def av2bv(aid: int) -> str:
+	bvid = [""] * 9
+	tmp = (MAX_AID | aid) ^ XOR_CODE
+	for i in range(CODE_LEN):
+		bvid[ENCODE_MAP[i]] = ALPHABET[tmp % BASE]
+		tmp //= BASE
+	return PREFIX + "".join(bvid)
 
-if __name__ == '__main__':
+def bv2av(bvid: str) -> int:
+	assert bvid[:3] == PREFIX
+
+	bvid = bvid[3:]
+	tmp = 0
+	for i in range(CODE_LEN):
+		idx = ALPHABET.index(bvid[DECODE_MAP[i]])
+		tmp = tmp * BASE + idx
+	return (tmp & MASK_CODE) ^ XOR_CODE
+
+if __name__ == "__main__":
 	if len(sys.argv) < 2:
-		print('Please enter av/bv number!')
+		print("Please input av/BV number!")
 		exit()
 	original = sys.argv[1].strip()
-	result = error
-	if original[:2].lower() == 'bv':
-		result = 'av' + str(decode(original))
-	elif original[:2].lower() == 'av':
-		result = encode(int(original[2:]))
+	result = ERROR
+	if original[:2].lower() == "bv":
+		result = "av" + str(bv2av(original))
+	elif original[:2].lower() == "av":
+		result = av2bv(int(original[2:]))
 	elif original.isdigit():
-		result = encode(int(original))
+		result = av2bv(int(original))
 	else:
-		print('Please enter a legal av/bv number!')
+		print("Please input a valid av/BV number!")
 		exit()
 	print(result)
+
+# assert av2bv(111298867365120) == "BV1L9Uoa9EUx"
+# assert bv2av("BV1L9Uoa9EUx") == 111298867365120