@traceryn/tiktok-sdk 1.0.0
This diff represents the content of publicly available package versions that have been released to one of the supported registries. The information contained in this diff is provided for informational purposes only and reflects changes between package versions as they appear in their respective public registries.
- package/LICENSE +21 -0
- package/README.md +276 -0
- package/engineRequirements.js +7 -0
- package/lib/ProxiflyProvider.d.ts +24 -0
- package/lib/ProxiflyProvider.d.ts.map +1 -0
- package/lib/ProxiflyProvider.js +112 -0
- package/lib/ProxiflyProvider.js.map +1 -0
- package/lib/client.d.ts +42 -0
- package/lib/client.d.ts.map +1 -0
- package/lib/client.js +377 -0
- package/lib/client.js.map +1 -0
- package/lib/core/CacheManager.d.ts +13 -0
- package/lib/core/CacheManager.d.ts.map +1 -0
- package/lib/core/CacheManager.js +49 -0
- package/lib/core/CacheManager.js.map +1 -0
- package/lib/core/CircuitBreaker.d.ts +21 -0
- package/lib/core/CircuitBreaker.d.ts.map +1 -0
- package/lib/core/CircuitBreaker.js +64 -0
- package/lib/core/CircuitBreaker.js.map +1 -0
- package/lib/core/Proxymanager.d.ts +32 -0
- package/lib/core/Proxymanager.d.ts.map +1 -0
- package/lib/core/Proxymanager.js +129 -0
- package/lib/core/Proxymanager.js.map +1 -0
- package/lib/core/RateLimiter.d.ts +17 -0
- package/lib/core/RateLimiter.d.ts.map +1 -0
- package/lib/core/RateLimiter.js +53 -0
- package/lib/core/RateLimiter.js.map +1 -0
- package/lib/core/RequestQueue.d.ts +17 -0
- package/lib/core/RequestQueue.d.ts.map +1 -0
- package/lib/core/RequestQueue.js +51 -0
- package/lib/core/RequestQueue.js.map +1 -0
- package/lib/core/RetryHandler.d.ts +13 -0
- package/lib/core/RetryHandler.d.ts.map +1 -0
- package/lib/core/RetryHandler.js +37 -0
- package/lib/core/RetryHandler.js.map +1 -0
- package/lib/core/SessionManager.d.ts +17 -0
- package/lib/core/SessionManager.d.ts.map +1 -0
- package/lib/core/SessionManager.js +57 -0
- package/lib/core/SessionManager.js.map +1 -0
- package/lib/core/Signer.d.ts +3 -0
- package/lib/core/Signer.d.ts.map +1 -0
- package/lib/core/Signer.js +26 -0
- package/lib/core/Signer.js.map +1 -0
- package/lib/index.d.ts +22 -0
- package/lib/index.d.ts.map +1 -0
- package/lib/index.js +21 -0
- package/lib/index.js.map +1 -0
- package/lib/playwright.d.ts +46 -0
- package/lib/playwright.d.ts.map +1 -0
- package/lib/playwright.js +330 -0
- package/lib/playwright.js.map +1 -0
- package/lib/scrapers/CommentScraper.d.ts +6 -0
- package/lib/scrapers/CommentScraper.d.ts.map +1 -0
- package/lib/scrapers/CommentScraper.js +107 -0
- package/lib/scrapers/CommentScraper.js.map +1 -0
- package/lib/scrapers/ExtraScraper.d.ts +15 -0
- package/lib/scrapers/ExtraScraper.d.ts.map +1 -0
- package/lib/scrapers/ExtraScraper.js +244 -0
- package/lib/scrapers/ExtraScraper.js.map +1 -0
- package/lib/scrapers/FormatExtractor.d.ts +14 -0
- package/lib/scrapers/FormatExtractor.d.ts.map +1 -0
- package/lib/scrapers/FormatExtractor.js +110 -0
- package/lib/scrapers/FormatExtractor.js.map +1 -0
- package/lib/scrapers/ParserEngine.d.ts +17 -0
- package/lib/scrapers/ParserEngine.d.ts.map +1 -0
- package/lib/scrapers/ParserEngine.js +201 -0
- package/lib/scrapers/ParserEngine.js.map +1 -0
- package/lib/scrapers/SearchScraper.d.ts +5 -0
- package/lib/scrapers/SearchScraper.d.ts.map +1 -0
- package/lib/scrapers/SearchScraper.js +149 -0
- package/lib/scrapers/SearchScraper.js.map +1 -0
- package/lib/scrapers/UrlExtractor.d.ts +9 -0
- package/lib/scrapers/UrlExtractor.d.ts.map +1 -0
- package/lib/scrapers/UrlExtractor.js +34 -0
- package/lib/scrapers/UrlExtractor.js.map +1 -0
- package/lib/scrapers/UserScraper.d.ts +3 -0
- package/lib/scrapers/UserScraper.d.ts.map +1 -0
- package/lib/scrapers/UserScraper.js +137 -0
- package/lib/scrapers/UserScraper.js.map +1 -0
- package/lib/scrapers/VideoScraper.d.ts +3 -0
- package/lib/scrapers/VideoScraper.d.ts.map +1 -0
- package/lib/scrapers/VideoScraper.js +65 -0
- package/lib/scrapers/VideoScraper.js.map +1 -0
- package/lib/types.d.ts +483 -0
- package/lib/types.d.ts.map +1 -0
- package/lib/types.js +2 -0
- package/lib/types.js.map +1 -0
- package/lib/utils/CookieJar.d.ts +8 -0
- package/lib/utils/CookieJar.d.ts.map +1 -0
- package/lib/utils/CookieJar.js +51 -0
- package/lib/utils/CookieJar.js.map +1 -0
- package/lib/utils/HttpClient.d.ts +12 -0
- package/lib/utils/HttpClient.d.ts.map +1 -0
- package/lib/utils/HttpClient.js +39 -0
- package/lib/utils/HttpClient.js.map +1 -0
- package/lib/utils/constants.d.ts +41 -0
- package/lib/utils/constants.d.ts.map +1 -0
- package/lib/utils/constants.js +48 -0
- package/lib/utils/constants.js.map +1 -0
- package/lib/utils/errors.d.ts +31 -0
- package/lib/utils/errors.d.ts.map +1 -0
- package/lib/utils/errors.js +63 -0
- package/lib/utils/errors.js.map +1 -0
- package/lib/utils/helpers.d.ts +9 -0
- package/lib/utils/helpers.d.ts.map +1 -0
- package/lib/utils/helpers.js +41 -0
- package/lib/utils/helpers.js.map +1 -0
- package/lib/utils/xbogus.cjs +565 -0
- package/package.json +88 -0
|
@@ -0,0 +1,149 @@
|
|
|
1
|
+
import { ofetch } from 'ofetch';
|
|
2
|
+
import { HEADERS } from '../utils/constants.js';
|
|
3
|
+
import { CookieJar } from '../utils/CookieJar.js';
|
|
4
|
+
import { signUrl } from '../core/Signer.js';
|
|
5
|
+
import { TikTokFetchError } from '../utils/errors.js';
|
|
6
|
+
const SEARCH_USER_URI = 'https://www.tiktok.com/api/search/user/full/';
|
|
7
|
+
const USER_PAGE = 'https://www.tiktok.com/@tiktok';
|
|
8
|
+
const WEB_SEARCH_CODE = JSON.stringify({
|
|
9
|
+
tiktok: {
|
|
10
|
+
client_params_x: {
|
|
11
|
+
search_engine: {
|
|
12
|
+
ies_mt_user_live_video_card_use_libra: 1,
|
|
13
|
+
mt_search_general_user_live_card: 1,
|
|
14
|
+
},
|
|
15
|
+
},
|
|
16
|
+
search_server: {},
|
|
17
|
+
},
|
|
18
|
+
});
|
|
19
|
+
export function buildAvatarInfo(url) {
|
|
20
|
+
const m = url.match(/cropcenter:(\d+):(\d+)/);
|
|
21
|
+
const w = m && m[1] ? parseInt(m[1], 10) : 0;
|
|
22
|
+
const h = m && m[2] ? parseInt(m[2], 10) : 0;
|
|
23
|
+
let a = w, b = h;
|
|
24
|
+
while (b) {
|
|
25
|
+
const t = b;
|
|
26
|
+
b = a % b;
|
|
27
|
+
a = t;
|
|
28
|
+
}
|
|
29
|
+
const gcd = a || 1;
|
|
30
|
+
const maxDim = Math.max(w, h);
|
|
31
|
+
const fmtM = url.match(/\.(\w+)(?:\?|$)/);
|
|
32
|
+
let fmt = 'jpeg';
|
|
33
|
+
if (fmtM && fmtM[1]) {
|
|
34
|
+
const ext = fmtM[1].toLowerCase();
|
|
35
|
+
fmt = ext === 'jpg' ? 'jpeg' : ext;
|
|
36
|
+
}
|
|
37
|
+
return {
|
|
38
|
+
url,
|
|
39
|
+
width: w,
|
|
40
|
+
height: h,
|
|
41
|
+
ratio: w && h ? `${w / gcd}:${h / gcd}` : '',
|
|
42
|
+
quality: maxDim > 0 ? `${maxDim}p` : 'unknown',
|
|
43
|
+
format: fmt,
|
|
44
|
+
size: 0,
|
|
45
|
+
};
|
|
46
|
+
}
|
|
47
|
+
export function parseSearchUser(data) {
|
|
48
|
+
const u = data.user_info ?? data;
|
|
49
|
+
return {
|
|
50
|
+
by: 'Traceryn',
|
|
51
|
+
id: String(u.uid ?? u.id ?? ''),
|
|
52
|
+
uniqueId: String(u.uniqueId ?? u.unique_id ?? ''),
|
|
53
|
+
nickname: String(u.nickname ?? ''),
|
|
54
|
+
signature: String(u.signature ?? ''),
|
|
55
|
+
verified: Boolean(u.verified ?? false),
|
|
56
|
+
secUid: String(u.secUid ?? ''),
|
|
57
|
+
avatar: {
|
|
58
|
+
thumb: buildAvatarInfo(String(u.avatarThumb ?? u.avatar_thumb?.url_list?.[0] ?? '')),
|
|
59
|
+
medium: buildAvatarInfo(String(u.avatarMedium ?? u.avatar_medium?.url_list?.[0] ?? '')),
|
|
60
|
+
larger: buildAvatarInfo(String(u.avatarLarger ?? u.avatar_larger?.url_list?.[0] ?? '')),
|
|
61
|
+
},
|
|
62
|
+
followerCount: Number(u.followerCount ?? u.follower_count ?? 0),
|
|
63
|
+
followingCount: Number(u.followingCount ?? u.following_count ?? 0),
|
|
64
|
+
heartCount: Number(u.heartCount ?? u.heart ?? 0),
|
|
65
|
+
videoCount: Number(u.videoCount ?? u.video_count ?? 0),
|
|
66
|
+
commerceUserInfo: u.commerceUserInfo
|
|
67
|
+
? { commerceUser: Boolean(u.commerceUserInfo.commerceUser), category: u.commerceUserInfo.category }
|
|
68
|
+
: undefined,
|
|
69
|
+
privateAccount: Boolean(u.privateAccount ?? u.private_account ?? false),
|
|
70
|
+
raw: data,
|
|
71
|
+
};
|
|
72
|
+
}
|
|
73
|
+
async function fetchCookies(proxy) {
|
|
74
|
+
const jar = new CookieJar();
|
|
75
|
+
await ofetch(USER_PAGE, {
|
|
76
|
+
headers: HEADERS.desktop,
|
|
77
|
+
...{ proxy },
|
|
78
|
+
parseResponse: (t) => t,
|
|
79
|
+
onResponse(_ctx) {
|
|
80
|
+
const resp = _ctx.response;
|
|
81
|
+
if (resp?.headers)
|
|
82
|
+
jar.setFromHeaders(resp.headers);
|
|
83
|
+
},
|
|
84
|
+
});
|
|
85
|
+
const blocked = new Set(['tt_chain_token', 'msToken']);
|
|
86
|
+
return Object.entries(jar.all)
|
|
87
|
+
.filter(([k]) => !blocked.has(k))
|
|
88
|
+
.map(([k, v]) => `${k}=${v}`)
|
|
89
|
+
.join('; ');
|
|
90
|
+
}
|
|
91
|
+
export async function searchUsers(query, proxy, cursor = 0, count = 20, session) {
|
|
92
|
+
if (session?.isReady) {
|
|
93
|
+
try {
|
|
94
|
+
const code = JSON.stringify({ tiktok: { client_params_x: { search_engine: { ies_mt_user_live_video_card_use_libra: 1, mt_search_general_user_live_card: 1 } }, search_server: {} } });
|
|
95
|
+
const res = await session.request('/api/search/user/full/', {
|
|
96
|
+
keyword: query, cursor: String(cursor), count: String(count), from_page: 'search', web_search_code: code,
|
|
97
|
+
});
|
|
98
|
+
const rawUsers = res.user_list ?? [];
|
|
99
|
+
const users = rawUsers.map(parseSearchUser);
|
|
100
|
+
return { users, cursor: Number(res.cursor ?? cursor), hasMore: Boolean(res.has_more ?? false), raw: res };
|
|
101
|
+
}
|
|
102
|
+
catch {
|
|
103
|
+
// if the session path fails, use plain HTTP instead
|
|
104
|
+
}
|
|
105
|
+
}
|
|
106
|
+
const cookieStr = await fetchCookies(proxy);
|
|
107
|
+
const params = new URLSearchParams({
|
|
108
|
+
keyword: query,
|
|
109
|
+
cursor: String(cursor),
|
|
110
|
+
count: String(count),
|
|
111
|
+
from_page: 'search',
|
|
112
|
+
web_search_code: WEB_SEARCH_CODE,
|
|
113
|
+
});
|
|
114
|
+
const ua = HEADERS.desktop['User-Agent'];
|
|
115
|
+
const baseUrl = `${SEARCH_USER_URI}?${params.toString()}`;
|
|
116
|
+
const signedUrl = signUrl(baseUrl, ua);
|
|
117
|
+
let res;
|
|
118
|
+
try {
|
|
119
|
+
res = await ofetch(signedUrl, {
|
|
120
|
+
headers: {
|
|
121
|
+
...HEADERS.api,
|
|
122
|
+
Referer: `https://www.tiktok.com/search/user?q=${encodeURIComponent(query)}`,
|
|
123
|
+
Cookie: cookieStr,
|
|
124
|
+
},
|
|
125
|
+
...{ proxy },
|
|
126
|
+
});
|
|
127
|
+
}
|
|
128
|
+
catch (err) {
|
|
129
|
+
const ofetchErr = err;
|
|
130
|
+
throw new TikTokFetchError(`Search API failed: ${err.message}`, ofetchErr.status);
|
|
131
|
+
}
|
|
132
|
+
if (!res || typeof res !== 'object') {
|
|
133
|
+
throw new TikTokFetchError('TikTok API returned empty response for searchUsers. ' +
|
|
134
|
+
'Try using a PlaywrightSession — install with: npm install playwright', 0);
|
|
135
|
+
}
|
|
136
|
+
const resObj = res;
|
|
137
|
+
if (resObj.status_code && resObj.status_code !== 0) {
|
|
138
|
+
throw new TikTokFetchError(`Search API returned error: ${resObj.status_msg ?? resObj.status_code}`, resObj.status_code);
|
|
139
|
+
}
|
|
140
|
+
const rawUsers = resObj.user_list ?? [];
|
|
141
|
+
const users = rawUsers.map(parseSearchUser);
|
|
142
|
+
return {
|
|
143
|
+
users,
|
|
144
|
+
cursor: Number(resObj.cursor ?? cursor),
|
|
145
|
+
hasMore: Boolean(resObj.has_more ?? false),
|
|
146
|
+
raw: resObj,
|
|
147
|
+
};
|
|
148
|
+
}
|
|
149
|
+
//# sourceMappingURL=SearchScraper.js.map
|
|
@@ -0,0 +1 @@
|
|
|
1
|
+
{"version":3,"file":"SearchScraper.js","sourceRoot":"","sources":["../../src/scrapers/SearchScraper.ts"],"names":[],"mappings":"AAAA,OAAO,EAAE,MAAM,EAAE,MAAM,QAAQ,CAAC;AAEhC,OAAO,EAAE,OAAO,EAAE,MAAM,uBAAuB,CAAC;AAChD,OAAO,EAAE,SAAS,EAAE,MAAM,uBAAuB,CAAC;AAClD,OAAO,EAAE,OAAO,EAAE,MAAM,mBAAmB,CAAC;AAC5C,OAAO,EAAE,gBAAgB,EAAE,MAAM,oBAAoB,CAAC;AAEtD,MAAM,eAAe,GAAG,8CAA8C,CAAC;AACvE,MAAM,SAAS,GAAG,gCAAgC,CAAC;AAEnD,MAAM,eAAe,GAAG,IAAI,CAAC,SAAS,CAAC;IACrC,MAAM,EAAE;QACN,eAAe,EAAE;YACf,aAAa,EAAE;gBACb,qCAAqC,EAAE,CAAC;gBACxC,gCAAgC,EAAE,CAAC;aACpC;SACF;QACD,aAAa,EAAE,EAAE;KAClB;CACF,CAAC,CAAC;AAEH,MAAM,UAAU,eAAe,CAAC,GAAW;IACzC,MAAM,CAAC,GAAG,GAAG,CAAC,KAAK,CAAC,wBAAwB,CAAC,CAAC;IAC9C,MAAM,CAAC,GAAG,CAAC,IAAI,CAAC,CAAC,CAAC,CAAC,CAAC,CAAC,CAAC,QAAQ,CAAC,CAAC,CAAC,CAAC,CAAC,EAAE,EAAE,CAAC,CAAC,CAAC,CAAC,CAAC,CAAC;IAC7C,MAAM,CAAC,GAAG,CAAC,IAAI,CAAC,CAAC,CAAC,CAAC,CAAC,CAAC,CAAC,QAAQ,CAAC,CAAC,CAAC,CAAC,CAAC,EAAE,EAAE,CAAC,CAAC,CAAC,CAAC,CAAC,CAAC;IAC7C,IAAI,CAAC,GAAG,CAAC,EAAE,CAAC,GAAG,CAAC,CAAC;IACjB,OAAO,CAAC,EAAE,CAAC;QAAC,MAAM,CAAC,GAAG,CAAC,CAAC;QAAC,CAAC,GAAG,CAAC,GAAG,CAAC,CAAC;QAAC,CAAC,GAAG,CAAC,CAAC;IAAC,CAAC;IAC5C,MAAM,GAAG,GAAG,CAAC,IAAI,CAAC,CAAC;IACnB,MAAM,MAAM,GAAG,IAAI,CAAC,GAAG,CAAC,CAAC,EAAE,CAAC,CAAC,CAAC;IAC9B,MAAM,IAAI,GAAG,GAAG,CAAC,KAAK,CAAC,iBAAiB,CAAC,CAAC;IAC1C,IAAI,GAAG,GAAG,MAAM,CAAC;IACjB,IAAI,IAAI,IAAI,IAAI,CAAC,CAAC,CAAC,EAAE,CAAC;QACpB,MAAM,GAAG,GAAG,IAAI,CAAC,CAAC,CAAC,CAAC,WAAW,EAAE,CAAC;QAClC,GAAG,GAAG,GAAG,KAAK,KAAK,CAAC,CAAC,CAAC,MAAM,CAAC,CAAC,CAAC,GAAG,CAAC;IACrC,CAAC;IACD,OAAO;QACL,GAAG;QACH,KAAK,EAAE,CAAC;QACR,MAAM,EAAE,CAAC;QACT,KAAK,EAAE,CAAC,IAAI,CAAC,CAAC,CAAC,CAAC,GAAG,CAAC,GAAG,GAAG,IAAI,CAAC,GAAG,GAAG,EAAE,CAAC,CAAC,CAAC,EAAE;QAC5C,OAAO,EAAE,MAAM,GAAG,CAAC,CAAC,CAAC,CAAC,GAAG,MAAM,GAAG,CAAC,CAAC,CAAC,SAAS;QAC9C,MAAM,EAAE,GAAG;QACX,IAAI,EAAE,CAAC;KACR,CAAC;AACJ,CAAC;AAED,MAAM,UAAU,eAAe,CAAC,IAAyB;IACvD,MAAM,CAAC,GAAG,IAAI,CAAC,SAAS,IAAI,IAAI,CAAC;IACjC,OAAO;QACL,EAAE,EAAE,UAAU;QACd,EAAE,EAAE,MAAM,CAAC,CAAC,CAAC,GAAG,IAAI,CAAC,CAAC,EAAE,IAAI,EAAE,CAAC;QAC/B,QAAQ,EAAE,MAAM,CAAC,CAAC,CAAC,QAAQ,IAAI,CAAC,CAAC,SAAS,IAAI,EAAE,CAAC;QACjD,QAAQ,EAAE,MAAM,CAAC,CAAC,CAAC,QAAQ,IAAI,EAAE,CAAC;QAClC,SAAS,EAAE,MAAM,CAAC,CAAC,CAAC,SAAS,IAAI,EAAE,CAAC;QACpC,QAAQ,EAAE,OAAO,CAAC,CAAC,CAAC,QAAQ,IAAI,KAAK,CAAC;QACtC,MAAM,EAAE,MAAM,CAAC,CAAC,CAAC,MAAM,IAAI,EAAE,CAAC;QAC9B,MAAM,EAAE;YACN,KAAK,EAAE,eAAe,CAAC,MAAM,CAAC,CAAC,CAAC,WAAW,IAAI,CAAC,CAAC,YAAY,EAAE,QAAQ,EAAE,CAAC,CAAC,CAAC,IAAI,EAAE,CAAC,CAAC;YACpF,MAAM,EAAE,eAAe,CAAC,MAAM,CAAC,CAAC,CAAC,YAAY,IAAI,CAAC,CAAC,aAAa,EAAE,QAAQ,EAAE,CAAC,CAAC,CAAC,IAAI,EAAE,CAAC,CAAC;YACvF,MAAM,EAAE,eAAe,CAAC,MAAM,CAAC,CAAC,CAAC,YAAY,IAAI,CAAC,CAAC,aAAa,EAAE,QAAQ,EAAE,CAAC,CAAC,CAAC,IAAI,EAAE,CAAC,CAAC;SACxF;QACD,aAAa,EAAE,MAAM,CAAC,CAAC,CAAC,aAAa,IAAI,CAAC,CAAC,cAAc,IAAI,CAAC,CAAC;QAC/D,cAAc,EAAE,MAAM,CAAC,CAAC,CAAC,cAAc,IAAI,CAAC,CAAC,eAAe,IAAI,CAAC,CAAC;QAClE,UAAU,EAAE,MAAM,CAAC,CAAC,CAAC,UAAU,IAAI,CAAC,CAAC,KAAK,IAAI,CAAC,CAAC;QAChD,UAAU,EAAE,MAAM,CAAC,CAAC,CAAC,UAAU,IAAI,CAAC,CAAC,WAAW,IAAI,CAAC,CAAC;QACtD,gBAAgB,EAAE,CAAC,CAAC,gBAAgB;YAClC,CAAC,CAAC,EAAE,YAAY,EAAE,OAAO,CAAC,CAAC,CAAC,gBAAgB,CAAC,YAAY,CAAC,EAAE,QAAQ,EAAE,CAAC,CAAC,gBAAgB,CAAC,QAAQ,EAAE;YACnG,CAAC,CAAC,SAAS;QACb,cAAc,EAAE,OAAO,CAAC,CAAC,CAAC,cAAc,IAAI,CAAC,CAAC,eAAe,IAAI,KAAK,CAAC;QACvE,GAAG,EAAE,IAAI;KACV,CAAC;AACJ,CAAC;AAED,KAAK,UAAU,YAAY,CAAC,KAAa;IACvC,MAAM,GAAG,GAAG,IAAI,SAAS,EAAE,CAAC;IAC5B,MAAM,MAAM,CAAS,SAAS,EAAE;QAC9B,OAAO,EAAE,OAAO,CAAC,OAAO;QACxB,GAAI,EAAE,KAAK,EAAU;QACrB,aAAa,EAAE,CAAC,CAAS,EAAE,EAAE,CAAC,CAAC;QAC/B,UAAU,CAAC,IAAI;YACb,MAAM,IAAI,GAAG,IAAI,CAAC,QAAQ,CAAC;YAC3B,IAAI,IAAI,EAAE,OAAO;gBAAE,GAAG,CAAC,cAAc,CAAC,IAAI,CAAC,OAA6B,CAAC,CAAC;QAC5E,CAAC;KACF,CAAC,CAAC;IACH,MAAM,OAAO,GAAG,IAAI,GAAG,CAAC,CAAC,gBAAgB,EAAE,SAAS,CAAC,CAAC,CAAC;IACvD,OAAO,MAAM,CAAC,OAAO,CAAC,GAAG,CAAC,GAAG,CAAC;SAC3B,MAAM,CAAC,CAAC,CAAC,CAAC,CAAC,EAAE,EAAE,CAAC,CAAC,OAAO,CAAC,GAAG,CAAC,CAAC,CAAC,CAAC;SAChC,GAAG,CAAC,CAAC,CAAC,CAAC,EAAE,CAAC,CAAC,EAAE,EAAE,CAAC,GAAG,CAAC,IAAI,CAAC,EAAE,CAAC;SAC5B,IAAI,CAAC,IAAI,CAAC,CAAC;AAChB,CAAC;AAED,MAAM,CAAC,KAAK,UAAU,WAAW,CAC/B,KAAa,EACb,KAAa,EACb,MAAM,GAAG,CAAC,EACV,KAAK,GAAG,EAAE,EACV,OAAiB;IAEjB,IAAI,OAAO,EAAE,OAAO,EAAE,CAAC;QACrB,IAAI,CAAC;YACH,MAAM,IAAI,GAAG,IAAI,CAAC,SAAS,CAAC,EAAE,MAAM,EAAE,EAAE,eAAe,EAAE,EAAE,aAAa,EAAE,EAAE,qCAAqC,EAAE,CAAC,EAAE,gCAAgC,EAAE,CAAC,EAAE,EAAE,EAAE,aAAa,EAAE,EAAE,EAAE,EAAE,CAAC,CAAC;YACtL,MAAM,GAAG,GAAG,MAAM,OAAO,CAAC,OAAO,CAAsB,wBAAwB,EAAE;gBAC/E,OAAO,EAAE,KAAK,EAAE,MAAM,EAAE,MAAM,CAAC,MAAM,CAAC,EAAE,KAAK,EAAE,MAAM,CAAC,KAAK,CAAC,EAAE,SAAS,EAAE,QAAQ,EAAE,eAAe,EAAE,IAAI;aACzG,CAAC,CAAC;YACH,MAAM,QAAQ,GAA+B,GAAG,CAAC,SAAS,IAAI,EAAE,CAAC;YACjE,MAAM,KAAK,GAAG,QAAQ,CAAC,GAAG,CAAC,eAAe,CAAC,CAAC;YAC5C,OAAO,EAAE,KAAK,EAAE,MAAM,EAAE,MAAM,CAAC,GAAG,CAAC,MAAM,IAAI,MAAM,CAAC,EAAE,OAAO,EAAE,OAAO,CAAC,GAAG,CAAC,QAAQ,IAAI,KAAK,CAAC,EAAE,GAAG,EAAE,GAAG,EAAE,CAAC;QAC5G,CAAC;QAAC,MAAM,CAAC;YACP,oDAAoD;QACtD,CAAC;IACH,CAAC;IAED,MAAM,SAAS,GAAG,MAAM,YAAY,CAAC,KAAK,CAAC,CAAC;IAE5C,MAAM,MAAM,GAAG,IAAI,eAAe,CAAC;QACjC,OAAO,EAAE,KAAK;QACd,MAAM,EAAE,MAAM,CAAC,MAAM,CAAC;QACtB,KAAK,EAAE,MAAM,CAAC,KAAK,CAAC;QACpB,SAAS,EAAE,QAAQ;QACnB,eAAe,EAAE,eAAe;KACjC,CAAC,CAAC;IAEH,MAAM,EAAE,GAAG,OAAO,CAAC,OAAO,CAAC,YAAY,CAAC,CAAC;IACzC,MAAM,OAAO,GAAG,GAAG,eAAe,IAAI,MAAM,CAAC,QAAQ,EAAE,EAAE,CAAC;IAC1D,MAAM,SAAS,GAAG,OAAO,CAAC,OAAO,EAAE,EAAE,CAAC,CAAC;IAEvC,IAAI,GAAY,CAAC;IACjB,IAAI,CAAC;QACH,GAAG,GAAG,MAAM,MAAM,CAAU,SAAS,EAAE;YACrC,OAAO,EAAE;gBACP,GAAG,OAAO,CAAC,GAAG;gBACd,OAAO,EAAE,wCAAwC,kBAAkB,CAAC,KAAK,CAAC,EAAE;gBAC5E,MAAM,EAAE,SAAS;aAClB;YACD,GAAI,EAAE,KAAK,EAAU;SACtB,CAAC,CAAC;IACL,CAAC;IAAC,OAAO,GAAY,EAAE,CAAC;QACtB,MAAM,SAAS,GAAG,GAA0B,CAAC;QAC7C,MAAM,IAAI,gBAAgB,CACxB,sBAAuB,GAAa,CAAC,OAAO,EAAE,EAC9C,SAAS,CAAC,MAAM,CACjB,CAAC;IACJ,CAAC;IAED,IAAI,CAAC,GAAG,IAAI,OAAO,GAAG,KAAK,QAAQ,EAAE,CAAC;QACpC,MAAM,IAAI,gBAAgB,CACxB,sDAAsD;YACtD,sEAAsE,EACtE,CAAC,CACF,CAAC;IACJ,CAAC;IACD,MAAM,MAAM,GAAG,GAA0B,CAAC;IAE1C,IAAI,MAAM,CAAC,WAAW,IAAI,MAAM,CAAC,WAAW,KAAK,CAAC,EAAE,CAAC;QACnD,MAAM,IAAI,gBAAgB,CACxB,8BAA8B,MAAM,CAAC,UAAU,IAAI,MAAM,CAAC,WAAW,EAAE,EACvE,MAAM,CAAC,WAAW,CACnB,CAAC;IACJ,CAAC;IAED,MAAM,QAAQ,GAA+B,MAAM,CAAC,SAAS,IAAI,EAAE,CAAC;IACpE,MAAM,KAAK,GAAG,QAAQ,CAAC,GAAG,CAAC,eAAe,CAAC,CAAC;IAE5C,OAAO;QACL,KAAK;QACL,MAAM,EAAE,MAAM,CAAC,MAAM,CAAC,MAAM,IAAI,MAAM,CAAC;QACvC,OAAO,EAAE,OAAO,CAAC,MAAM,CAAC,QAAQ,IAAI,KAAK,CAAC;QAC1C,GAAG,EAAE,MAAM;KACZ,CAAC;AACJ,CAAC"}
|
|
@@ -0,0 +1,9 @@
|
|
|
1
|
+
export declare function extractVideoId(urlOrId: string): string;
|
|
2
|
+
export declare function normalizePhotoUrl(url: string): string;
|
|
3
|
+
export declare function buildVideoUrl(uniqueId: string, videoId: string): string;
|
|
4
|
+
export declare function buildUserUrl(username: string): string;
|
|
5
|
+
export declare function buildFeedUrl(secUid: string, cursor: number): string;
|
|
6
|
+
export declare function buildWebpageUrl(videoUrl: string): string;
|
|
7
|
+
export declare function isValidTikTokUrl(url: string): boolean;
|
|
8
|
+
export declare function isPhotoUrl(url: string): boolean;
|
|
9
|
+
//# sourceMappingURL=UrlExtractor.d.ts.map
|
|
@@ -0,0 +1 @@
|
|
|
1
|
+
{"version":3,"file":"UrlExtractor.d.ts","sourceRoot":"","sources":["../../src/scrapers/UrlExtractor.ts"],"names":[],"mappings":"AAEA,wBAAgB,cAAc,CAAC,OAAO,EAAE,MAAM,GAAG,MAAM,CAGtD;AAED,wBAAgB,iBAAiB,CAAC,GAAG,EAAE,MAAM,GAAG,MAAM,CAErD;AAED,wBAAgB,aAAa,CAAC,QAAQ,EAAE,MAAM,EAAE,OAAO,EAAE,MAAM,GAAG,MAAM,CAEvE;AAED,wBAAgB,YAAY,CAAC,QAAQ,EAAE,MAAM,GAAG,MAAM,CAGrD;AAED,wBAAgB,YAAY,CAAC,MAAM,EAAE,MAAM,EAAE,MAAM,EAAE,MAAM,GAAG,MAAM,CAOnE;AAED,wBAAgB,eAAe,CAAC,QAAQ,EAAE,MAAM,GAAG,MAAM,CAGxD;AAED,wBAAgB,gBAAgB,CAAC,GAAG,EAAE,MAAM,GAAG,OAAO,CAErD;AAED,wBAAgB,UAAU,CAAC,GAAG,EAAE,MAAM,GAAG,OAAO,CAE/C"}
|
|
@@ -0,0 +1,34 @@
|
|
|
1
|
+
import { TIKTOK_DOMAINS } from '../utils/constants.js';
|
|
2
|
+
export function extractVideoId(urlOrId) {
|
|
3
|
+
const m = urlOrId.match(/\/(?:video|photo)\/(\d+)/);
|
|
4
|
+
return m?.[1] ?? urlOrId;
|
|
5
|
+
}
|
|
6
|
+
export function normalizePhotoUrl(url) {
|
|
7
|
+
return url.replace('/photo/', '/video/');
|
|
8
|
+
}
|
|
9
|
+
export function buildVideoUrl(uniqueId, videoId) {
|
|
10
|
+
return `${TIKTOK_DOMAINS.www}/@${uniqueId}/video/${videoId}`;
|
|
11
|
+
}
|
|
12
|
+
export function buildUserUrl(username) {
|
|
13
|
+
const name = username.replace('@', '');
|
|
14
|
+
return `${TIKTOK_DOMAINS.www}/@${name}`;
|
|
15
|
+
}
|
|
16
|
+
export function buildFeedUrl(secUid, cursor) {
|
|
17
|
+
const params = new URLSearchParams({
|
|
18
|
+
secUid,
|
|
19
|
+
count: '30',
|
|
20
|
+
cursor: String(cursor),
|
|
21
|
+
});
|
|
22
|
+
return `${TIKTOK_DOMAINS.api}/post/item_list/?${params}`;
|
|
23
|
+
}
|
|
24
|
+
export function buildWebpageUrl(videoUrl) {
|
|
25
|
+
const id = extractVideoId(videoUrl);
|
|
26
|
+
return `${TIKTOK_DOMAINS.www}/video/${id}`;
|
|
27
|
+
}
|
|
28
|
+
export function isValidTikTokUrl(url) {
|
|
29
|
+
return /https?:\/\/(?:www\.)?tiktok\.com\/(?:@[\w.-]+\/(?:video|photo)\/\d+|@[\w.-]+)/.test(url);
|
|
30
|
+
}
|
|
31
|
+
export function isPhotoUrl(url) {
|
|
32
|
+
return url.includes('/photo/');
|
|
33
|
+
}
|
|
34
|
+
//# sourceMappingURL=UrlExtractor.js.map
|
|
@@ -0,0 +1 @@
|
|
|
1
|
+
{"version":3,"file":"UrlExtractor.js","sourceRoot":"","sources":["../../src/scrapers/UrlExtractor.ts"],"names":[],"mappings":"AAAA,OAAO,EAAE,cAAc,EAAE,MAAM,uBAAuB,CAAC;AAEvD,MAAM,UAAU,cAAc,CAAC,OAAe;IAC5C,MAAM,CAAC,GAAG,OAAO,CAAC,KAAK,CAAC,0BAA0B,CAAC,CAAC;IACpD,OAAO,CAAC,EAAE,CAAC,CAAC,CAAC,IAAI,OAAO,CAAC;AAC3B,CAAC;AAED,MAAM,UAAU,iBAAiB,CAAC,GAAW;IAC3C,OAAO,GAAG,CAAC,OAAO,CAAC,SAAS,EAAE,SAAS,CAAC,CAAC;AAC3C,CAAC;AAED,MAAM,UAAU,aAAa,CAAC,QAAgB,EAAE,OAAe;IAC7D,OAAO,GAAG,cAAc,CAAC,GAAG,KAAK,QAAQ,UAAU,OAAO,EAAE,CAAC;AAC/D,CAAC;AAED,MAAM,UAAU,YAAY,CAAC,QAAgB;IAC3C,MAAM,IAAI,GAAG,QAAQ,CAAC,OAAO,CAAC,GAAG,EAAE,EAAE,CAAC,CAAC;IACvC,OAAO,GAAG,cAAc,CAAC,GAAG,KAAK,IAAI,EAAE,CAAC;AAC1C,CAAC;AAED,MAAM,UAAU,YAAY,CAAC,MAAc,EAAE,MAAc;IACzD,MAAM,MAAM,GAAG,IAAI,eAAe,CAAC;QACjC,MAAM;QACN,KAAK,EAAE,IAAI;QACX,MAAM,EAAE,MAAM,CAAC,MAAM,CAAC;KACvB,CAAC,CAAC;IACH,OAAO,GAAG,cAAc,CAAC,GAAG,oBAAoB,MAAM,EAAE,CAAC;AAC3D,CAAC;AAED,MAAM,UAAU,eAAe,CAAC,QAAgB;IAC9C,MAAM,EAAE,GAAG,cAAc,CAAC,QAAQ,CAAC,CAAC;IACpC,OAAO,GAAG,cAAc,CAAC,GAAG,UAAU,EAAE,EAAE,CAAC;AAC7C,CAAC;AAED,MAAM,UAAU,gBAAgB,CAAC,GAAW;IAC1C,OAAO,+EAA+E,CAAC,IAAI,CAAC,GAAG,CAAC,CAAC;AACnG,CAAC;AAED,MAAM,UAAU,UAAU,CAAC,GAAW;IACpC,OAAO,GAAG,CAAC,QAAQ,CAAC,SAAS,CAAC,CAAC;AACjC,CAAC"}
|
|
@@ -0,0 +1 @@
|
|
|
1
|
+
{"version":3,"file":"UserScraper.d.ts","sourceRoot":"","sources":["../../src/scrapers/UserScraper.ts"],"names":[],"mappings":"AACA,OAAO,KAAK,EAAE,UAAU,EAAc,MAAM,aAAa,CAAC;AAiC1D,wBAAsB,UAAU,CAAC,QAAQ,EAAE,MAAM,EAAE,KAAK,EAAE,MAAM,GAAG,OAAO,CAAC,UAAU,CAAC,CA0GrF"}
|
|
@@ -0,0 +1,137 @@
|
|
|
1
|
+
import { ofetch } from 'ofetch';
|
|
2
|
+
import { CookieJar, fetchTikTokCookies } from '../utils/CookieJar.js';
|
|
3
|
+
import { HEADERS, DESKTOP_UA } from '../utils/constants.js';
|
|
4
|
+
import { extractUserPageData } from './ParserEngine.js';
|
|
5
|
+
import { buildUserUrl } from './UrlExtractor.js';
|
|
6
|
+
import { TikTokFetchError, TikTokParseError } from '../utils/errors.js';
|
|
7
|
+
import { parseCookieString, formatISODate } from '../utils/helpers.js';
|
|
8
|
+
function buildAvatarInfo(url) {
|
|
9
|
+
const m = url.match(/cropcenter:(\d+):(\d+)/);
|
|
10
|
+
const w = m && m[1] ? parseInt(m[1], 10) : 0;
|
|
11
|
+
const h = m && m[2] ? parseInt(m[2], 10) : 0;
|
|
12
|
+
let a = w, b = h;
|
|
13
|
+
while (b) {
|
|
14
|
+
const t = b;
|
|
15
|
+
b = a % b;
|
|
16
|
+
a = t;
|
|
17
|
+
}
|
|
18
|
+
const gcd = a || 1;
|
|
19
|
+
const maxDim = Math.max(w, h);
|
|
20
|
+
const fmtM = url.match(/\.(\w+)(?:\?|$)/);
|
|
21
|
+
let fmt = 'jpeg';
|
|
22
|
+
if (fmtM && fmtM[1]) {
|
|
23
|
+
const ext = fmtM[1].toLowerCase();
|
|
24
|
+
fmt = ext === 'jpg' ? 'jpeg' : ext;
|
|
25
|
+
}
|
|
26
|
+
return {
|
|
27
|
+
url,
|
|
28
|
+
width: w,
|
|
29
|
+
height: h,
|
|
30
|
+
ratio: w && h ? `${w / gcd}:${h / gcd}` : '',
|
|
31
|
+
quality: maxDim > 0 ? `${maxDim}p` : 'unknown',
|
|
32
|
+
format: fmt,
|
|
33
|
+
size: 0,
|
|
34
|
+
};
|
|
35
|
+
}
|
|
36
|
+
export async function scrapeUser(username, proxy) {
|
|
37
|
+
const name = username.replace('@', '');
|
|
38
|
+
const url = buildUserUrl(name);
|
|
39
|
+
const jar = new CookieJar();
|
|
40
|
+
let html;
|
|
41
|
+
try {
|
|
42
|
+
html = await ofetch(url, {
|
|
43
|
+
headers: HEADERS.desktop,
|
|
44
|
+
...{ proxy },
|
|
45
|
+
parseResponse: (txt) => txt,
|
|
46
|
+
onResponse(_ctx) {
|
|
47
|
+
const resp = _ctx.response;
|
|
48
|
+
if (resp?.headers) {
|
|
49
|
+
jar.setFromHeaders(resp.headers);
|
|
50
|
+
}
|
|
51
|
+
},
|
|
52
|
+
});
|
|
53
|
+
}
|
|
54
|
+
catch (err) {
|
|
55
|
+
const ofetchErr = err;
|
|
56
|
+
throw new TikTokFetchError(`Failed to fetch user page: ${err.message}`, ofetchErr.status);
|
|
57
|
+
}
|
|
58
|
+
let pageData = extractUserPageData(html);
|
|
59
|
+
if (!pageData) {
|
|
60
|
+
const cookieStr = await fetchTikTokCookies();
|
|
61
|
+
if (cookieStr) {
|
|
62
|
+
try {
|
|
63
|
+
html = await ofetch(url, {
|
|
64
|
+
headers: { ...HEADERS.desktop, Cookie: cookieStr },
|
|
65
|
+
...{ proxy },
|
|
66
|
+
parseResponse: (txt) => txt,
|
|
67
|
+
onResponse(_ctx) {
|
|
68
|
+
const resp = _ctx.response;
|
|
69
|
+
if (resp?.headers)
|
|
70
|
+
jar.setFromHeaders(resp.headers);
|
|
71
|
+
},
|
|
72
|
+
});
|
|
73
|
+
pageData = extractUserPageData(html);
|
|
74
|
+
}
|
|
75
|
+
catch { }
|
|
76
|
+
}
|
|
77
|
+
}
|
|
78
|
+
if (!pageData) {
|
|
79
|
+
throw new TikTokParseError('Could not extract user data from profile page');
|
|
80
|
+
}
|
|
81
|
+
const u = pageData.user;
|
|
82
|
+
const s = pageData.stats;
|
|
83
|
+
const createTimestamp = Number(u.createTime ?? 0);
|
|
84
|
+
const httpHeaders = {
|
|
85
|
+
...HEADERS.desktop,
|
|
86
|
+
Referer: url,
|
|
87
|
+
};
|
|
88
|
+
const cu = u.commerceUserInfo;
|
|
89
|
+
return {
|
|
90
|
+
by: 'Traceryn',
|
|
91
|
+
raw: pageData.userInfo,
|
|
92
|
+
id: String(u.id ?? ''),
|
|
93
|
+
uniqueId: String(u.uniqueId ?? ''),
|
|
94
|
+
nickname: String(u.nickname ?? ''),
|
|
95
|
+
signature: String(u.signature ?? ''),
|
|
96
|
+
verified: Boolean(u.verified),
|
|
97
|
+
secUid: String(u.secUid ?? ''),
|
|
98
|
+
avatar: {
|
|
99
|
+
thumb: buildAvatarInfo(String(u.avatarThumb ?? '')),
|
|
100
|
+
medium: buildAvatarInfo(String(u.avatarMedium ?? '')),
|
|
101
|
+
larger: buildAvatarInfo(String(u.avatarLarger ?? '')),
|
|
102
|
+
},
|
|
103
|
+
stats: {
|
|
104
|
+
followerCount: Number(s.followerCount ?? 0),
|
|
105
|
+
followingCount: Number(s.followingCount ?? 0),
|
|
106
|
+
heartCount: Number(s.heartCount ?? 0),
|
|
107
|
+
videoCount: Number(s.videoCount ?? 0),
|
|
108
|
+
diggCount: Number(s.diggCount ?? 0),
|
|
109
|
+
friendCount: Number(s.friendCount ?? 0),
|
|
110
|
+
},
|
|
111
|
+
createTime: formatISODate(createTimestamp),
|
|
112
|
+
createTimestamp,
|
|
113
|
+
language: String(u.language ?? ''),
|
|
114
|
+
region: String(u.region ?? ''),
|
|
115
|
+
bioLink: u.bioLink?.link,
|
|
116
|
+
privateAccount: Boolean(u.privateAccount),
|
|
117
|
+
secret: Boolean(u.secret),
|
|
118
|
+
ftc: Boolean(u.ftc),
|
|
119
|
+
isOrganization: Boolean(u.isOrganization),
|
|
120
|
+
ttSeller: Boolean(u.ttSeller),
|
|
121
|
+
openFavorite: Boolean(u.openFavorite),
|
|
122
|
+
isADVirtual: Boolean(u.isADVirtual),
|
|
123
|
+
commentSetting: Number(u.commentSetting ?? 0),
|
|
124
|
+
duetSetting: Number(u.duetSetting ?? 0),
|
|
125
|
+
stitchSetting: Number(u.stitchSetting ?? 0),
|
|
126
|
+
downloadSetting: Number(u.downloadSetting ?? 0),
|
|
127
|
+
followingVisibility: Number(u.followingVisibility ?? 0),
|
|
128
|
+
profileEmbedPermission: Number(u.profileEmbedPermission ?? 0),
|
|
129
|
+
commerceUserInfo: cu?.commerceUser ? {
|
|
130
|
+
commerceUser: true,
|
|
131
|
+
category: cu.category,
|
|
132
|
+
} : undefined,
|
|
133
|
+
httpHeaders,
|
|
134
|
+
cookies: parseCookieString(jar.all),
|
|
135
|
+
};
|
|
136
|
+
}
|
|
137
|
+
//# sourceMappingURL=UserScraper.js.map
|
|
@@ -0,0 +1 @@
|
|
|
1
|
+
{"version":3,"file":"UserScraper.js","sourceRoot":"","sources":["../../src/scrapers/UserScraper.ts"],"names":[],"mappings":"AAAA,OAAO,EAAE,MAAM,EAAE,MAAM,QAAQ,CAAC;AAEhC,OAAO,EAAE,SAAS,EAAE,kBAAkB,EAAE,MAAM,uBAAuB,CAAC;AACtE,OAAO,EAAE,OAAO,EAAE,UAAU,EAAE,MAAM,uBAAuB,CAAC;AAC5D,OAAO,EAAE,mBAAmB,EAAE,MAAM,mBAAmB,CAAC;AACxD,OAAO,EAAE,YAAY,EAAE,MAAM,mBAAmB,CAAC;AACjD,OAAO,EAAE,gBAAgB,EAAE,gBAAgB,EAAE,MAAM,oBAAoB,CAAC;AACxE,OAAO,EAAE,iBAAiB,EAAE,aAAa,EAAE,MAAM,qBAAqB,CAAC;AAEvE,SAAS,eAAe,CAAC,GAAW;IAClC,MAAM,CAAC,GAAG,GAAG,CAAC,KAAK,CAAC,wBAAwB,CAAC,CAAC;IAC9C,MAAM,CAAC,GAAG,CAAC,IAAI,CAAC,CAAC,CAAC,CAAC,CAAC,CAAC,CAAC,QAAQ,CAAC,CAAC,CAAC,CAAC,CAAC,EAAE,EAAE,CAAC,CAAC,CAAC,CAAC,CAAC,CAAC;IAC7C,MAAM,CAAC,GAAG,CAAC,IAAI,CAAC,CAAC,CAAC,CAAC,CAAC,CAAC,CAAC,QAAQ,CAAC,CAAC,CAAC,CAAC,CAAC,EAAE,EAAE,CAAC,CAAC,CAAC,CAAC,CAAC,CAAC;IAC7C,IAAI,CAAC,GAAG,CAAC,EAAE,CAAC,GAAG,CAAC,CAAC;IACjB,OAAO,CAAC,EAAE,CAAC;QAAC,MAAM,CAAC,GAAG,CAAC,CAAC;QAAC,CAAC,GAAG,CAAC,GAAG,CAAC,CAAC;QAAC,CAAC,GAAG,CAAC,CAAC;IAAC,CAAC;IAC5C,MAAM,GAAG,GAAG,CAAC,IAAI,CAAC,CAAC;IACnB,MAAM,MAAM,GAAG,IAAI,CAAC,GAAG,CAAC,CAAC,EAAE,CAAC,CAAC,CAAC;IAC9B,MAAM,IAAI,GAAG,GAAG,CAAC,KAAK,CAAC,iBAAiB,CAAC,CAAC;IAC1C,IAAI,GAAG,GAAG,MAAM,CAAC;IACjB,IAAI,IAAI,IAAI,IAAI,CAAC,CAAC,CAAC,EAAE,CAAC;QACpB,MAAM,GAAG,GAAG,IAAI,CAAC,CAAC,CAAC,CAAC,WAAW,EAAE,CAAC;QAClC,GAAG,GAAG,GAAG,KAAK,KAAK,CAAC,CAAC,CAAC,MAAM,CAAC,CAAC,CAAC,GAAG,CAAC;IACrC,CAAC;IACD,OAAO;QACL,GAAG;QACH,KAAK,EAAE,CAAC;QACR,MAAM,EAAE,CAAC;QACT,KAAK,EAAE,CAAC,IAAI,CAAC,CAAC,CAAC,CAAC,GAAG,CAAC,GAAG,GAAG,IAAI,CAAC,GAAG,GAAG,EAAE,CAAC,CAAC,CAAC,EAAE;QAC5C,OAAO,EAAE,MAAM,GAAG,CAAC,CAAC,CAAC,CAAC,GAAG,MAAM,GAAG,CAAC,CAAC,CAAC,SAAS;QAC9C,MAAM,EAAE,GAAG;QACX,IAAI,EAAE,CAAC;KACR,CAAC;AACJ,CAAC;AAED,MAAM,CAAC,KAAK,UAAU,UAAU,CAAC,QAAgB,EAAE,KAAa;IAC9D,MAAM,IAAI,GAAG,QAAQ,CAAC,OAAO,CAAC,GAAG,EAAE,EAAE,CAAC,CAAC;IACvC,MAAM,GAAG,GAAG,YAAY,CAAC,IAAI,CAAC,CAAC;IAE/B,MAAM,GAAG,GAAG,IAAI,SAAS,EAAE,CAAC;IAC5B,IAAI,IAAY,CAAC;IACjB,IAAI,CAAC;QACH,IAAI,GAAG,MAAM,MAAM,CAAS,GAAG,EAAE;YAC/B,OAAO,EAAE,OAAO,CAAC,OAAO;YACxB,GAAI,EAAE,KAAK,EAAU;YACrB,aAAa,EAAE,CAAC,GAAW,EAAE,EAAE,CAAC,GAAG;YACnC,UAAU,CAAC,IAAI;gBACb,MAAM,IAAI,GAAG,IAAI,CAAC,QAAQ,CAAC;gBAC3B,IAAI,IAAI,EAAE,OAAO,EAAE,CAAC;oBAClB,GAAG,CAAC,cAAc,CAAC,IAAI,CAAC,OAA6B,CAAC,CAAC;gBACzD,CAAC;YACH,CAAC;SACF,CAAC,CAAC;IACL,CAAC;IAAC,OAAO,GAAY,EAAE,CAAC;QACtB,MAAM,SAAS,GAAG,GAA0B,CAAC;QAC7C,MAAM,IAAI,gBAAgB,CACxB,8BAA+B,GAAa,CAAC,OAAO,EAAE,EACtD,SAAS,CAAC,MAAM,CACjB,CAAC;IACJ,CAAC;IAED,IAAI,QAAQ,GAAG,mBAAmB,CAAC,IAAI,CAAC,CAAC;IACzC,IAAI,CAAC,QAAQ,EAAE,CAAC;QACd,MAAM,SAAS,GAAG,MAAM,kBAAkB,EAAE,CAAC;QAC7C,IAAI,SAAS,EAAE,CAAC;YACd,IAAI,CAAC;gBACH,IAAI,GAAG,MAAM,MAAM,CAAS,GAAG,EAAE;oBAC/B,OAAO,EAAE,EAAE,GAAG,OAAO,CAAC,OAAO,EAAE,MAAM,EAAE,SAAS,EAAE;oBAClD,GAAI,EAAE,KAAK,EAAU;oBACrB,aAAa,EAAE,CAAC,GAAW,EAAE,EAAE,CAAC,GAAG;oBACnC,UAAU,CAAC,IAAI;wBACb,MAAM,IAAI,GAAG,IAAI,CAAC,QAAQ,CAAC;wBAC3B,IAAI,IAAI,EAAE,OAAO;4BAAE,GAAG,CAAC,cAAc,CAAC,IAAI,CAAC,OAA6B,CAAC,CAAC;oBAC5E,CAAC;iBACF,CAAC,CAAC;gBACH,QAAQ,GAAG,mBAAmB,CAAC,IAAI,CAAC,CAAC;YACvC,CAAC;YAAC,MAAM,CAAC,CAAA,CAAC;QACZ,CAAC;IACH,CAAC;IACD,IAAI,CAAC,QAAQ,EAAE,CAAC;QACd,MAAM,IAAI,gBAAgB,CAAC,+CAA+C,CAAC,CAAC;IAC9E,CAAC;IAED,MAAM,CAAC,GAAG,QAAQ,CAAC,IAAI,CAAC;IACxB,MAAM,CAAC,GAAG,QAAQ,CAAC,KAAK,CAAC;IACzB,MAAM,eAAe,GAAG,MAAM,CAAC,CAAC,CAAC,UAAU,IAAI,CAAC,CAAC,CAAC;IAElD,MAAM,WAAW,GAA2B;QAC1C,GAAG,OAAO,CAAC,OAAO;QAClB,OAAO,EAAE,GAAG;KACb,CAAC;IAEF,MAAM,EAAE,GAAG,CAAC,CAAC,gBAA6E,CAAC;IAE3F,OAAO;QACL,EAAE,EAAE,UAAU;QACd,GAAG,EAAE,QAAQ,CAAC,QAA+C;QAC7D,EAAE,EAAE,MAAM,CAAC,CAAC,CAAC,EAAE,IAAI,EAAE,CAAC;QACtB,QAAQ,EAAE,MAAM,CAAC,CAAC,CAAC,QAAQ,IAAI,EAAE,CAAC;QAClC,QAAQ,EAAE,MAAM,CAAC,CAAC,CAAC,QAAQ,IAAI,EAAE,CAAC;QAClC,SAAS,EAAE,MAAM,CAAC,CAAC,CAAC,SAAS,IAAI,EAAE,CAAC;QACpC,QAAQ,EAAE,OAAO,CAAC,CAAC,CAAC,QAAQ,CAAC;QAC7B,MAAM,EAAE,MAAM,CAAC,CAAC,CAAC,MAAM,IAAI,EAAE,CAAC;QAC9B,MAAM,EAAE;YACN,KAAK,EAAE,eAAe,CAAC,MAAM,CAAC,CAAC,CAAC,WAAW,IAAI,EAAE,CAAC,CAAC;YACnD,MAAM,EAAE,eAAe,CAAC,MAAM,CAAC,CAAC,CAAC,YAAY,IAAI,EAAE,CAAC,CAAC;YACrD,MAAM,EAAE,eAAe,CAAC,MAAM,CAAC,CAAC,CAAC,YAAY,IAAI,EAAE,CAAC,CAAC;SACtD;QACD,KAAK,EAAE;YACL,aAAa,EAAE,MAAM,CAAC,CAAC,CAAC,aAAa,IAAI,CAAC,CAAC;YAC3C,cAAc,EAAE,MAAM,CAAC,CAAC,CAAC,cAAc,IAAI,CAAC,CAAC;YAC7C,UAAU,EAAE,MAAM,CAAC,CAAC,CAAC,UAAU,IAAI,CAAC,CAAC;YACrC,UAAU,EAAE,MAAM,CAAC,CAAC,CAAC,UAAU,IAAI,CAAC,CAAC;YACrC,SAAS,EAAE,MAAM,CAAC,CAAC,CAAC,SAAS,IAAI,CAAC,CAAC;YACnC,WAAW,EAAE,MAAM,CAAC,CAAC,CAAC,WAAW,IAAI,CAAC,CAAC;SACxC;QACD,UAAU,EAAE,aAAa,CAAC,eAAe,CAAC;QAC1C,eAAe;QACf,QAAQ,EAAE,MAAM,CAAC,CAAC,CAAC,QAAQ,IAAI,EAAE,CAAC;QAClC,MAAM,EAAE,MAAM,CAAC,CAAC,CAAC,MAAM,IAAI,EAAE,CAAC;QAC9B,OAAO,EAAG,CAAC,CAAC,OAAyC,EAAE,IAAI;QAC3D,cAAc,EAAE,OAAO,CAAC,CAAC,CAAC,cAAc,CAAC;QACzC,MAAM,EAAE,OAAO,CAAC,CAAC,CAAC,MAAM,CAAC;QACzB,GAAG,EAAE,OAAO,CAAC,CAAC,CAAC,GAAG,CAAC;QACnB,cAAc,EAAE,OAAO,CAAC,CAAC,CAAC,cAAc,CAAC;QACzC,QAAQ,EAAE,OAAO,CAAC,CAAC,CAAC,QAAQ,CAAC;QAC7B,YAAY,EAAE,OAAO,CAAC,CAAC,CAAC,YAAY,CAAC;QACrC,WAAW,EAAE,OAAO,CAAC,CAAC,CAAC,WAAW,CAAC;QACnC,cAAc,EAAE,MAAM,CAAC,CAAC,CAAC,cAAc,IAAI,CAAC,CAAC;QAC7C,WAAW,EAAE,MAAM,CAAC,CAAC,CAAC,WAAW,IAAI,CAAC,CAAC;QACvC,aAAa,EAAE,MAAM,CAAC,CAAC,CAAC,aAAa,IAAI,CAAC,CAAC;QAC3C,eAAe,EAAE,MAAM,CAAC,CAAC,CAAC,eAAe,IAAI,CAAC,CAAC;QAC/C,mBAAmB,EAAE,MAAM,CAAC,CAAC,CAAC,mBAAmB,IAAI,CAAC,CAAC;QACvD,sBAAsB,EAAE,MAAM,CAAC,CAAC,CAAC,sBAAsB,IAAI,CAAC,CAAC;QAC7D,gBAAgB,EAAE,EAAE,EAAE,YAAY,CAAC,CAAC,CAAC;YACnC,YAAY,EAAE,IAAI;YAClB,QAAQ,EAAE,EAAE,CAAC,QAAQ;SACtB,CAAC,CAAC,CAAC,SAAS;QACb,WAAW;QACX,OAAO,EAAE,iBAAiB,CAAC,GAAG,CAAC,GAAG,CAAC;KACpC,CAAC;AACJ,CAAC"}
|
|
@@ -0,0 +1 @@
|
|
|
1
|
+
{"version":3,"file":"VideoScraper.d.ts","sourceRoot":"","sources":["../../src/scrapers/VideoScraper.ts"],"names":[],"mappings":"AACA,OAAO,KAAK,EAAE,OAAO,EAAE,YAAY,EAAuB,MAAM,aAAa,CAAC;AAoC9E,wBAAsB,WAAW,CAC/B,GAAG,EAAE,MAAM,EACX,OAAO,EAAE,OAAO,GAAG,SAAS,EAC5B,KAAK,EAAE,MAAM,GACZ,OAAO,CAAC,YAAY,CAAC,CA8BvB"}
|
|
@@ -0,0 +1,65 @@
|
|
|
1
|
+
import { ofetch } from 'ofetch';
|
|
2
|
+
import { CookieJar } from '../utils/CookieJar.js';
|
|
3
|
+
import { HEADERS } from '../utils/constants.js';
|
|
4
|
+
import { normalizePhotoUrl, extractVideoId } from './UrlExtractor.js';
|
|
5
|
+
import { ParserEngine } from './ParserEngine.js';
|
|
6
|
+
import { TikTokFetchError, TikTokWafError } from '../utils/errors.js';
|
|
7
|
+
let parserEngine = null;
|
|
8
|
+
function getParser() {
|
|
9
|
+
if (!parserEngine)
|
|
10
|
+
parserEngine = new ParserEngine();
|
|
11
|
+
return parserEngine;
|
|
12
|
+
}
|
|
13
|
+
async function scrapeVideoHTML(url, proxy) {
|
|
14
|
+
const jar = new CookieJar();
|
|
15
|
+
const fetchUrl = normalizePhotoUrl(url);
|
|
16
|
+
const html = await ofetch(fetchUrl, {
|
|
17
|
+
headers: { ...HEADERS.desktop, Referer: url },
|
|
18
|
+
parseResponse: (txt) => txt,
|
|
19
|
+
...{ proxy },
|
|
20
|
+
onResponse(_ctx) {
|
|
21
|
+
const resp = _ctx.response;
|
|
22
|
+
if (resp?.headers)
|
|
23
|
+
jar.setFromHeaders(resp.headers);
|
|
24
|
+
},
|
|
25
|
+
});
|
|
26
|
+
// Detect WAF: empty or tiny HTML means blocked
|
|
27
|
+
if (!html || html.length < 500) {
|
|
28
|
+
throw new TikTokWafError(`Video HTML too short (${html?.length ?? 0} bytes)`, proxy);
|
|
29
|
+
}
|
|
30
|
+
const { itemStruct } = getParser().parse(html);
|
|
31
|
+
return { html, itemStruct, cookies: jar.all };
|
|
32
|
+
}
|
|
33
|
+
export async function scrapeVideo(url, session, proxy) {
|
|
34
|
+
// Strategy 1: Session API path
|
|
35
|
+
if (session?.isReady) {
|
|
36
|
+
try {
|
|
37
|
+
const videoId = extractVideoId(url);
|
|
38
|
+
const res = await session.request('/api/item/detail/', {
|
|
39
|
+
item_id: videoId,
|
|
40
|
+
});
|
|
41
|
+
const itemStruct = res?.itemInfo?.itemStruct;
|
|
42
|
+
if (itemStruct)
|
|
43
|
+
return { html: '', itemStruct, cookies: {} };
|
|
44
|
+
}
|
|
45
|
+
catch {
|
|
46
|
+
// fall through
|
|
47
|
+
}
|
|
48
|
+
}
|
|
49
|
+
// 2) render the page if the session path misses
|
|
50
|
+
if (session?.render) {
|
|
51
|
+
try {
|
|
52
|
+
const renderedHtml = await session.render(url);
|
|
53
|
+
if (renderedHtml && renderedHtml.includes('itemStruct')) {
|
|
54
|
+
const { itemStruct } = getParser().parse(renderedHtml);
|
|
55
|
+
return { html: renderedHtml, itemStruct, cookies: {} };
|
|
56
|
+
}
|
|
57
|
+
}
|
|
58
|
+
catch {
|
|
59
|
+
// fall through to HTML scrape
|
|
60
|
+
}
|
|
61
|
+
}
|
|
62
|
+
// 3) raw HTML scrape as the last shot
|
|
63
|
+
return scrapeVideoHTML(url, proxy);
|
|
64
|
+
}
|
|
65
|
+
//# sourceMappingURL=VideoScraper.js.map
|
|
@@ -0,0 +1 @@
|
|
|
1
|
+
{"version":3,"file":"VideoScraper.js","sourceRoot":"","sources":["../../src/scrapers/VideoScraper.ts"],"names":[],"mappings":"AAAA,OAAO,EAAE,MAAM,EAAE,MAAM,QAAQ,CAAC;AAEhC,OAAO,EAAE,SAAS,EAAE,MAAM,uBAAuB,CAAC;AAClD,OAAO,EAAE,OAAO,EAAE,MAAM,uBAAuB,CAAC;AAChD,OAAO,EAAE,iBAAiB,EAAE,cAAc,EAAE,MAAM,mBAAmB,CAAC;AACtE,OAAO,EAAE,YAAY,EAAE,MAAM,mBAAmB,CAAC;AACjD,OAAO,EAAE,gBAAgB,EAAE,cAAc,EAAE,MAAM,oBAAoB,CAAC;AAEtE,IAAI,YAAY,GAAwB,IAAI,CAAC;AAC7C,SAAS,SAAS;IAChB,IAAI,CAAC,YAAY;QAAE,YAAY,GAAG,IAAI,YAAY,EAAE,CAAC;IACrD,OAAO,YAAY,CAAC;AACtB,CAAC;AAED,KAAK,UAAU,eAAe,CAAC,GAAW,EAAE,KAAa;IACvD,MAAM,GAAG,GAAG,IAAI,SAAS,EAAE,CAAC;IAC5B,MAAM,QAAQ,GAAG,iBAAiB,CAAC,GAAG,CAAC,CAAC;IAExC,MAAM,IAAI,GAAG,MAAM,MAAM,CAAS,QAAQ,EAAE;QAC1C,OAAO,EAAE,EAAE,GAAG,OAAO,CAAC,OAAO,EAAE,OAAO,EAAE,GAAG,EAAE;QAC7C,aAAa,EAAE,CAAC,GAAW,EAAE,EAAE,CAAC,GAAG;QACnC,GAAI,EAAE,KAAK,EAAU;QACrB,UAAU,CAAC,IAAI;YACb,MAAM,IAAI,GAAG,IAAI,CAAC,QAAQ,CAAC;YAC3B,IAAI,IAAI,EAAE,OAAO;gBAAE,GAAG,CAAC,cAAc,CAAC,IAAI,CAAC,OAA6B,CAAC,CAAC;QAC5E,CAAC;KACF,CAAC,CAAC;IAEH,+CAA+C;IAC/C,IAAI,CAAC,IAAI,IAAI,IAAI,CAAC,MAAM,GAAG,GAAG,EAAE,CAAC;QAC/B,MAAM,IAAI,cAAc,CAAC,yBAAyB,IAAI,EAAE,MAAM,IAAI,CAAC,SAAS,EAAE,KAAK,CAAC,CAAC;IACvF,CAAC;IAED,MAAM,EAAE,UAAU,EAAE,GAAG,SAAS,EAAE,CAAC,KAAK,CAAC,IAAI,CAAC,CAAC;IAC/C,OAAO,EAAE,IAAI,EAAE,UAAU,EAAE,OAAO,EAAE,GAAG,CAAC,GAAG,EAAE,CAAC;AAChD,CAAC;AAED,MAAM,CAAC,KAAK,UAAU,WAAW,CAC/B,GAAW,EACX,OAA4B,EAC5B,KAAa;IAEb,+BAA+B;IAC/B,IAAI,OAAO,EAAE,OAAO,EAAE,CAAC;QACrB,IAAI,CAAC;YACH,MAAM,OAAO,GAAG,cAAc,CAAC,GAAG,CAAC,CAAC;YACpC,MAAM,GAAG,GAAG,MAAM,OAAO,CAAC,OAAO,CAAsB,mBAAmB,EAAE;gBAC1E,OAAO,EAAE,OAAO;aACjB,CAAC,CAAC;YACH,MAAM,UAAU,GAAG,GAAG,EAAE,QAAQ,EAAE,UAAU,CAAC;YAC7C,IAAI,UAAU;gBAAE,OAAO,EAAE,IAAI,EAAE,EAAE,EAAE,UAAU,EAAE,OAAO,EAAE,EAAE,EAAE,CAAC;QAC/D,CAAC;QAAC,MAAM,CAAC;YACP,eAAe;QACjB,CAAC;IACH,CAAC;IAED,gDAAgD;IAChD,IAAI,OAAO,EAAE,MAAM,EAAE,CAAC;QACpB,IAAI,CAAC;YACH,MAAM,YAAY,GAAG,MAAM,OAAO,CAAC,MAAM,CAAC,GAAG,CAAC,CAAC;YAC/C,IAAI,YAAY,IAAI,YAAY,CAAC,QAAQ,CAAC,YAAY,CAAC,EAAE,CAAC;gBACxD,MAAM,EAAE,UAAU,EAAE,GAAG,SAAS,EAAE,CAAC,KAAK,CAAC,YAAY,CAAC,CAAC;gBACvD,OAAO,EAAE,IAAI,EAAE,YAAY,EAAE,UAAU,EAAE,OAAO,EAAE,EAAE,EAAE,CAAC;YACzD,CAAC;QACH,CAAC;QAAC,MAAM,CAAC;YACP,8BAA8B;QAChC,CAAC;IACH,CAAC;IAED,sCAAsC;IACtC,OAAO,eAAe,CAAC,GAAG,EAAE,KAAK,CAAC,CAAC;AACrC,CAAC"}
|