tt-help-cli-ycl 1.3.62 → 1.3.64
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/package.json +1 -1
- package/scripts/run-explore.bat +1 -1
- package/src/cli/explore.js +8 -4
- package/src/cli/refresh.js +437 -163
- package/src/lib/args.js +54 -8
- package/src/lib/browser/cdp.js +11 -5
- package/src/lib/browser/page.js +41 -0
- package/src/lib/constants.js +31 -9
- package/src/main.js +3 -0
- package/src/npm-main.js +3 -0
- package/src/watch/data-store.js +145 -9
- package/src/watch/public/app.js +296 -42
- package/src/watch/public/index.html +63 -0
- package/src/watch/public/style.css +80 -3
- package/src/watch/server.js +91 -2
- package/src/scraper/refresh-core.js +0 -213
package/src/watch/server.js
CHANGED
|
@@ -356,7 +356,8 @@ export function startWatchServer(dataAnchor, port = 3000, existingStore) {
|
|
|
356
356
|
|
|
357
357
|
if (req.method === "GET" && routePath === "/api/redo-job") {
|
|
358
358
|
const userId = params.userId || "";
|
|
359
|
-
const
|
|
359
|
+
const maxAge = parseInt(params.maxAge) || 43200;
|
|
360
|
+
const job = store.getNextRedoJob(userId, maxAge);
|
|
360
361
|
if (job) {
|
|
361
362
|
logJob("REDO-CLAIM", { user: job.uniqueId, clientId: userId });
|
|
362
363
|
sendJSON(res, 200, { hasJob: true, user: job });
|
|
@@ -525,6 +526,66 @@ export function startWatchServer(dataAnchor, port = 3000, existingStore) {
|
|
|
525
526
|
return;
|
|
526
527
|
}
|
|
527
528
|
|
|
529
|
+
if (
|
|
530
|
+
req.method === "GET" &&
|
|
531
|
+
routePath === "/api/target-users-by-country"
|
|
532
|
+
) {
|
|
533
|
+
const result = store.getTargetUsersByCountry(DEFAULT_TARGET_LOCATIONS);
|
|
534
|
+
if (req.headers["accept"]?.includes("text/csv")) {
|
|
535
|
+
const columns = [
|
|
536
|
+
"uniqueId",
|
|
537
|
+
"nickname",
|
|
538
|
+
"followerCount",
|
|
539
|
+
"videoCount",
|
|
540
|
+
"locationCreated",
|
|
541
|
+
"latestVideoTime",
|
|
542
|
+
"refreshTime",
|
|
543
|
+
"status",
|
|
544
|
+
"sources",
|
|
545
|
+
];
|
|
546
|
+
const allUsers = [];
|
|
547
|
+
for (const country of result.countries) {
|
|
548
|
+
for (const u of country.users) {
|
|
549
|
+
allUsers.push({
|
|
550
|
+
uniqueId: u.uniqueId,
|
|
551
|
+
nickname: u.nickname || "",
|
|
552
|
+
followerCount: u.followerCount ?? 0,
|
|
553
|
+
videoCount: u.videoCount ?? 0,
|
|
554
|
+
locationCreated: u.locationCreated || "",
|
|
555
|
+
latestVideoTime: u.latestVideoTime
|
|
556
|
+
? new Date(u.latestVideoTime * 1000)
|
|
557
|
+
.toISOString()
|
|
558
|
+
.slice(0, 19)
|
|
559
|
+
.replace("T", " ")
|
|
560
|
+
: "",
|
|
561
|
+
refreshTime: u.refreshTime
|
|
562
|
+
? new Date(u.refreshTime)
|
|
563
|
+
.toISOString()
|
|
564
|
+
.slice(0, 19)
|
|
565
|
+
.replace("T", " ")
|
|
566
|
+
: "",
|
|
567
|
+
status: u.status || "",
|
|
568
|
+
sources: (u.sources || []).join(";"),
|
|
569
|
+
});
|
|
570
|
+
}
|
|
571
|
+
}
|
|
572
|
+
res.writeHead(200, {
|
|
573
|
+
"Content-Type": "text/csv; charset=utf-8",
|
|
574
|
+
"Content-Disposition":
|
|
575
|
+
'attachment; filename="target-users-by-country.csv"',
|
|
576
|
+
});
|
|
577
|
+
const BOM = "\uFEFF";
|
|
578
|
+
const header = columns.join(",");
|
|
579
|
+
const lines = allUsers.map((r) =>
|
|
580
|
+
columns.map((c) => csvEscape(r[c])).join(","),
|
|
581
|
+
);
|
|
582
|
+
res.end(BOM + [header, ...lines].join("\r\n"));
|
|
583
|
+
} else {
|
|
584
|
+
sendJSON(res, 200, result);
|
|
585
|
+
}
|
|
586
|
+
return;
|
|
587
|
+
}
|
|
588
|
+
|
|
528
589
|
if (req.method === "GET" && routePath === "/api/client-errors") {
|
|
529
590
|
sendJSON(res, 200, { clients: store.getClientErrors() });
|
|
530
591
|
return;
|
|
@@ -563,6 +624,8 @@ export function startWatchServer(dataAnchor, port = 3000, existingStore) {
|
|
|
563
624
|
location: params.location,
|
|
564
625
|
limit: params.limit,
|
|
565
626
|
offset: params.offset,
|
|
627
|
+
hasVideo: params.hasVideo,
|
|
628
|
+
hasFollower: params.hasFollower,
|
|
566
629
|
});
|
|
567
630
|
sendJSON(res, 200, result);
|
|
568
631
|
return;
|
|
@@ -589,10 +652,17 @@ export function startWatchServer(dataAnchor, port = 3000, existingStore) {
|
|
|
589
652
|
let result;
|
|
590
653
|
if (body.uniqueId) {
|
|
591
654
|
result = store.restoreRawJobById(body.uniqueId);
|
|
592
|
-
} else if (
|
|
655
|
+
} else if (
|
|
656
|
+
body.search ||
|
|
657
|
+
body.location ||
|
|
658
|
+
body.hasVideo ||
|
|
659
|
+
body.hasFollower
|
|
660
|
+
) {
|
|
593
661
|
result = store.restoreRawJobsByFilter({
|
|
594
662
|
search: body.search || "",
|
|
595
663
|
location: body.location || "",
|
|
664
|
+
hasVideo: body.hasVideo,
|
|
665
|
+
hasFollower: body.hasFollower,
|
|
596
666
|
});
|
|
597
667
|
} else if (body.country) {
|
|
598
668
|
result = store.restoreRawJobsByCountry(body.country);
|
|
@@ -628,6 +698,24 @@ export function startWatchServer(dataAnchor, port = 3000, existingStore) {
|
|
|
628
698
|
return;
|
|
629
699
|
}
|
|
630
700
|
|
|
701
|
+
if (
|
|
702
|
+
req.method === "POST" &&
|
|
703
|
+
routePath === "/api/pending-by-country/reset"
|
|
704
|
+
) {
|
|
705
|
+
try {
|
|
706
|
+
const body = await readBody(req);
|
|
707
|
+
const result = store.resetPendingByCountry(body.country);
|
|
708
|
+
if (result.error) {
|
|
709
|
+
sendJSON(res, 400, result);
|
|
710
|
+
return;
|
|
711
|
+
}
|
|
712
|
+
sendJSON(res, 200, result);
|
|
713
|
+
} catch (e) {
|
|
714
|
+
sendJSON(res, 400, { error: e.message });
|
|
715
|
+
}
|
|
716
|
+
return;
|
|
717
|
+
}
|
|
718
|
+
|
|
631
719
|
if (
|
|
632
720
|
req.method === "DELETE" &&
|
|
633
721
|
routePath.startsWith("/api/client-error/")
|
|
@@ -683,6 +771,7 @@ export function startWatchServer(dataAnchor, port = 3000, existingStore) {
|
|
|
683
771
|
followerCount: u.followerCount,
|
|
684
772
|
locationCreated: u.locationCreated,
|
|
685
773
|
latestVideoTime: u.latestVideoTime,
|
|
774
|
+
refreshTime: u.refreshTime,
|
|
686
775
|
guessedLocation: u.guessedLocation,
|
|
687
776
|
pinned: u.pinned,
|
|
688
777
|
processedAt: u.processedAt,
|
|
@@ -1,213 +0,0 @@
|
|
|
1
|
-
import {
|
|
2
|
-
delay,
|
|
3
|
-
retryWithBackoff,
|
|
4
|
-
detectPageError,
|
|
5
|
-
assertPageUrl,
|
|
6
|
-
} from "./modules/page-helpers.js";
|
|
7
|
-
import { detectCaptcha } from "./modules/captcha-handler.js";
|
|
8
|
-
import { getUserInfo, collectVideos } from "../videos/core.js";
|
|
9
|
-
import { extractFollowAndFollowers } from "./modules/follow-extractor.js";
|
|
10
|
-
import { processExplore } from "./explore-core.js";
|
|
11
|
-
import { DEFAULT_TARGET_LOCATIONS_CSV } from "../lib/target-locations.js";
|
|
12
|
-
|
|
13
|
-
export async function processRefresh(page, username, serverUrl, options, log) {
|
|
14
|
-
const { maxFollowing = 100, maxFollowers = 100, maxVideos = 100 } = options;
|
|
15
|
-
|
|
16
|
-
const result = {
|
|
17
|
-
userInfo: null,
|
|
18
|
-
discoveredVideoAuthors: [],
|
|
19
|
-
discoveredFollowing: [],
|
|
20
|
-
discoveredFollowers: [],
|
|
21
|
-
newUsersAdded: 0,
|
|
22
|
-
collectedVideos: 0,
|
|
23
|
-
error: null,
|
|
24
|
-
};
|
|
25
|
-
|
|
26
|
-
try {
|
|
27
|
-
log(` 访问 @${username} 主页...`);
|
|
28
|
-
const homeUrl = `https://www.tiktok.com/@${username}`;
|
|
29
|
-
await retryWithBackoff(
|
|
30
|
-
async () => {
|
|
31
|
-
await page.goto(homeUrl, {
|
|
32
|
-
waitUntil: "domcontentloaded",
|
|
33
|
-
timeout: 30000,
|
|
34
|
-
});
|
|
35
|
-
assertPageUrl(page, `@${username}`);
|
|
36
|
-
},
|
|
37
|
-
{ log },
|
|
38
|
-
);
|
|
39
|
-
await page
|
|
40
|
-
.waitForSelector('[class*="DivVideoList"]', { timeout: 10000 })
|
|
41
|
-
.catch(() => {});
|
|
42
|
-
await delay(1000, 2000);
|
|
43
|
-
|
|
44
|
-
log(" 获取用户信息...");
|
|
45
|
-
const info = await getUserInfo(page);
|
|
46
|
-
if (info) {
|
|
47
|
-
result.userInfo = info;
|
|
48
|
-
log(
|
|
49
|
-
` 用户: ${info.nickname || username} | 粉丝: ${info.followerCount || "-"} | 视频: ${info.videoCount || "-"}`,
|
|
50
|
-
);
|
|
51
|
-
}
|
|
52
|
-
|
|
53
|
-
const captcha = await detectCaptcha(page);
|
|
54
|
-
if (captcha && captcha.visible) {
|
|
55
|
-
log(`[验证码] @${username} 页面出现验证码`);
|
|
56
|
-
result.captchaDetected = true;
|
|
57
|
-
result.captchaStage = result.captchaStage || "video-page";
|
|
58
|
-
result.captchaMessage = result.captchaMessage || "视频页出现验证码";
|
|
59
|
-
}
|
|
60
|
-
|
|
61
|
-
// 采集视频
|
|
62
|
-
log(` 采集视频 (最多 ${maxVideos} 个)...`);
|
|
63
|
-
const videoList = await collectVideos(page, username, maxVideos, log);
|
|
64
|
-
const videoArray = videoList ? [...videoList.values()] : [];
|
|
65
|
-
result.collectedVideos = videoArray.length;
|
|
66
|
-
result.discoveredVideoAuthors = videoArray.map((v) => v.author);
|
|
67
|
-
|
|
68
|
-
if (videoArray.length <= 0) {
|
|
69
|
-
result.noVideo = true;
|
|
70
|
-
const pageError = await detectPageError(page);
|
|
71
|
-
if (pageError) {
|
|
72
|
-
result.restricted = true;
|
|
73
|
-
log(` @${username} 页面受限(${pageError}),标记跳过`);
|
|
74
|
-
}
|
|
75
|
-
return result;
|
|
76
|
-
}
|
|
77
|
-
|
|
78
|
-
// 采集关注和粉丝
|
|
79
|
-
log(` 采集关注 (最多 ${maxFollowing}) + 粉丝 (最多 ${maxFollowers})...`);
|
|
80
|
-
try {
|
|
81
|
-
const followResult = await extractFollowAndFollowers(page, {
|
|
82
|
-
maxFollowing,
|
|
83
|
-
maxFollowers,
|
|
84
|
-
});
|
|
85
|
-
result.discoveredFollowing = followResult.following || [];
|
|
86
|
-
result.discoveredFollowers = followResult.followers || [];
|
|
87
|
-
log(
|
|
88
|
-
` 关注: ${result.discoveredFollowing.length}, 粉丝: ${result.discoveredFollowers.length}`,
|
|
89
|
-
);
|
|
90
|
-
} catch (e) {
|
|
91
|
-
log(` [关注/粉丝采集失败] ${e.message}`);
|
|
92
|
-
result.discoveredFollowing = [];
|
|
93
|
-
result.discoveredFollowers = [];
|
|
94
|
-
}
|
|
95
|
-
|
|
96
|
-
// 处理新发现的用户(关注 + 粉丝),循环执行完整 explore
|
|
97
|
-
// follow-extractor 返回 [handle, displayName] 数组
|
|
98
|
-
const allDiscovered = [
|
|
99
|
-
...result.discoveredFollowing.map((h) => ({
|
|
100
|
-
handle: Array.isArray(h) ? h[0] : h,
|
|
101
|
-
source: "refresh-following",
|
|
102
|
-
})),
|
|
103
|
-
...result.discoveredFollowers.map((h) => ({
|
|
104
|
-
handle: Array.isArray(h) ? h[0] : h,
|
|
105
|
-
source: "refresh-follower",
|
|
106
|
-
})),
|
|
107
|
-
];
|
|
108
|
-
|
|
109
|
-
for (const { handle, source } of allDiscovered) {
|
|
110
|
-
const uniqueId = handle.replace("@", "");
|
|
111
|
-
|
|
112
|
-
// 检查用户是否已存在
|
|
113
|
-
const existsResp = await fetch(
|
|
114
|
-
`${serverUrl}/api/user-exists/${encodeURIComponent(uniqueId)}`,
|
|
115
|
-
);
|
|
116
|
-
const existsData = await existsResp.json();
|
|
117
|
-
|
|
118
|
-
if (existsData.exists) {
|
|
119
|
-
continue;
|
|
120
|
-
}
|
|
121
|
-
|
|
122
|
-
log(` [新用户] @${uniqueId} 不存在,开始探索 (来源: ${source})...`);
|
|
123
|
-
await delay(1000, 2000);
|
|
124
|
-
|
|
125
|
-
// 对新用户做完整 explore(与 explore 命令逻辑一致)
|
|
126
|
-
const exploreResult = await processExplore(
|
|
127
|
-
page,
|
|
128
|
-
uniqueId,
|
|
129
|
-
{
|
|
130
|
-
maxComments: 10,
|
|
131
|
-
maxGuess: 0,
|
|
132
|
-
enableFollow: true,
|
|
133
|
-
maxFollowing: 5,
|
|
134
|
-
maxFollowers: 5,
|
|
135
|
-
location: DEFAULT_TARGET_LOCATIONS_CSV,
|
|
136
|
-
},
|
|
137
|
-
log,
|
|
138
|
-
);
|
|
139
|
-
|
|
140
|
-
// 提交 explore 结果到服务端(和 explore 命令的 commitJob 一致)
|
|
141
|
-
if (exploreResult.userInfo) {
|
|
142
|
-
const guessedLocation = exploreResult.locationCreated || null;
|
|
143
|
-
|
|
144
|
-
const payload = {
|
|
145
|
-
userInfo: exploreResult.userInfo || {},
|
|
146
|
-
discoveredVideoAuthors: (
|
|
147
|
-
exploreResult.discoveredVideoAuthors || []
|
|
148
|
-
).map((item) =>
|
|
149
|
-
typeof item === "object" ? { ...item, guessedLocation } : item,
|
|
150
|
-
),
|
|
151
|
-
discoveredCommentAuthors: (
|
|
152
|
-
exploreResult.discoveredCommentAuthors || []
|
|
153
|
-
).map((author) => ({ author, guessedLocation })),
|
|
154
|
-
discoveredGuessAuthors: (
|
|
155
|
-
exploreResult.discoveredGuessAuthors || []
|
|
156
|
-
).map((author) => ({ author, guessedLocation })),
|
|
157
|
-
discoveredFollowing: (exploreResult.discoveredFollowing || []).map(
|
|
158
|
-
(f) => ({
|
|
159
|
-
handle: Array.isArray(f) ? f[0] : f,
|
|
160
|
-
displayName: Array.isArray(f) ? f[1] : null,
|
|
161
|
-
guessedLocation,
|
|
162
|
-
}),
|
|
163
|
-
),
|
|
164
|
-
discoveredFollowers: (exploreResult.discoveredFollowers || []).map(
|
|
165
|
-
(f) => ({
|
|
166
|
-
handle: Array.isArray(f) ? f[0] : f,
|
|
167
|
-
displayName: Array.isArray(f) ? f[1] : null,
|
|
168
|
-
guessedLocation,
|
|
169
|
-
}),
|
|
170
|
-
),
|
|
171
|
-
processed: exploreResult.processed,
|
|
172
|
-
hasFollowData: exploreResult.hasFollowData,
|
|
173
|
-
keepFollow: exploreResult.keepFollow,
|
|
174
|
-
locationCreated: exploreResult.locationCreated,
|
|
175
|
-
noVideo: exploreResult.noVideo,
|
|
176
|
-
restricted: exploreResult.restricted,
|
|
177
|
-
error: exploreResult.error,
|
|
178
|
-
};
|
|
179
|
-
|
|
180
|
-
const addResp = await fetch(
|
|
181
|
-
`${serverUrl}/api/explore-new/${uniqueId}`,
|
|
182
|
-
{
|
|
183
|
-
method: "POST",
|
|
184
|
-
headers: { "Content-Type": "application/json" },
|
|
185
|
-
body: JSON.stringify(payload),
|
|
186
|
-
},
|
|
187
|
-
);
|
|
188
|
-
const addResult = await addResp.json();
|
|
189
|
-
|
|
190
|
-
if (!addResult.saved) {
|
|
191
|
-
log(` [跳过] @${uniqueId} 提交失败`);
|
|
192
|
-
continue;
|
|
193
|
-
}
|
|
194
|
-
|
|
195
|
-
result.newUsersAdded++;
|
|
196
|
-
if (exploreResult.captchaDetected) {
|
|
197
|
-
result.captchaDetected = true;
|
|
198
|
-
}
|
|
199
|
-
log(
|
|
200
|
-
` [已提交] @${uniqueId} ${addResult.created ? "(新用户)" : "(已存在)"} | 发现: ${addResult.newUsers?.length || 0} 个`,
|
|
201
|
-
);
|
|
202
|
-
}
|
|
203
|
-
|
|
204
|
-
await delay(2000, 4000);
|
|
205
|
-
}
|
|
206
|
-
} catch (e) {
|
|
207
|
-
log(` [错误] ${e.message}`);
|
|
208
|
-
result.error = e.message;
|
|
209
|
-
result.errorStack = e.stack || "";
|
|
210
|
-
}
|
|
211
|
-
|
|
212
|
-
return result;
|
|
213
|
-
}
|