1import type { Adapter, AdapterLoginInfo } from "../types";
2import { exportCookies, restoreCookies, type CookieSidecarConfig } from "../../browser/cookie-sidecar";
3import { detectInteractionGate } from "../../browser/interaction-gates";
4import type { ExtractedDocument } from "../../extract/document";
5import { collectMediaFromDocument } from "../../media/markdown-media";
6import { extractArticleDocumentFromPayload } from "./article";
7import { buildNeedsLoginResult, detectXLogin } from "./login";
8import { extractStatusId, isXHost } from "./match";
9import { collectXJsonPayloads, waitForInitialXPayload } from "./payloads";
10import { extractSingleTweetDocumentFromPayload } from "./single";
11import { extractThreadDocumentFromPayloads } from "./thread";
12import { loadFullXThread } from "./thread-loader";
13 
14const cookieConfig: CookieSidecarConfig = {
15  urls: ["https://x.com/", "https://twitter.com/"],
16  filename: "x-session-cookies.json",
17  requiredCookieNames: ["auth_token", "ct0"],
18  filterCookie: (c) => {
19    const d = c.domain ?? "";
20    return d.endsWith("x.com") || d.endsWith("twitter.com");
21  },
22};
23 
24function extractDocumentFromPayloads(
25  payloads: unknown[],
26  statusId: string,
27  pageUrl: string,
28): ExtractedDocument | null {
29  for (const payload of payloads) {
30    const articleDocument = extractArticleDocumentFromPayload(payload, statusId, pageUrl, payloads);
31    if (articleDocument) {
32      return articleDocument;
33    }
34  }
35 
36  const threadDocument = extractThreadDocumentFromPayloads(payloads, statusId, pageUrl);
37  if (threadDocument) {
38    return threadDocument;
39  }
40 
41  for (const payload of payloads) {
42    const singleDocument = extractSingleTweetDocumentFromPayload(payload, statusId, pageUrl);
43    if (singleDocument) {
44      return singleDocument;
45    }
46  }
47 
48  return null;
49}
50 
51async function ensureXLoginState(context: Parameters<Adapter["process"]>[0]): Promise<AdapterLoginInfo> {
52  return detectXLogin(context);
53}
54 
55export const xAdapter: Adapter = {
56  name: "x",
57  match(input) {
58    return isXHost(input.url.hostname);
59  },
60  async checkLogin(context) {
61    return detectXLogin(context);
62  },
63  async exportCookies(context, profileDir) {
64    return exportCookies(context.browser.targetSession, cookieConfig, profileDir);
65  },
66  async restoreCookies(context, profileDir) {
67    return restoreCookies(context.browser.targetSession, cookieConfig, profileDir);
68  },
69  async process(context) {
70    const statusId = extractStatusId(context.input.url);
71    if (!statusId) {
72      return {
73        status: "no_document",
74      };
75    }
76 
77    context.log.info(`Loading ${context.input.url.toString()} with x adapter`);
78    await context.browser.goto(context.input.url.toString(), context.timeoutMs);
79 
80    const interaction = await detectInteractionGate(context.browser);
81    if (interaction) {
82      return {
83        status: "needs_interaction",
84        interaction,
85      };
86    }
87 
88    let login = await ensureXLoginState(context);
89    if (login.state === "logged_out") {
90      return buildNeedsLoginResult(login);
91    }
92 
93    await waitForInitialXPayload(context);
94    await loadFullXThread(context, statusId);
95 
96    const pageUrl = await context.browser.getURL();
97    const postLoadInteraction = await detectInteractionGate(context.browser);
98    if (postLoadInteraction) {
99      return {
100        status: "needs_interaction",
101        interaction: postLoadInteraction,
102        login,
103      };
104    }
105 
106    login = await ensureXLoginState(context).catch(() => login);
107    if (login.state === "logged_out") {
108      return buildNeedsLoginResult(login);
109    }
110 
111    const payloads = await collectXJsonPayloads(context);
112    if (payloads.length === 0) {
113      return {
114        status: "no_document",
115        login,
116      };
117    }
118 
119    const document = extractDocumentFromPayloads(payloads, statusId, pageUrl);
120    if (document) {
121      return {
122        status: "ok",
123        document,
124        media: collectMediaFromDocument(document),
125        login,
126      };
127    }
128 
129    return {
130      status: "no_document",
131      login,
132    };
133  },
134};
135