1import { fetchXArticle } from "./graphql.js";
2import type { ArticleEntity } from "./types.js";
3 
4function coerceArticleEntity(value: unknown): ArticleEntity | null {
5  if (!value || typeof value !== "object") return null;
6  const candidate = value as ArticleEntity;
7  if (
8    typeof candidate.title === "string" ||
9    typeof candidate.plain_text === "string" ||
10    typeof candidate.preview_text === "string" ||
11    candidate.content_state
12  ) {
13    return candidate;
14  }
15  return null;
16}
17 
18function hasArticleContent(article: ArticleEntity): boolean {
19  const blocks = article.content_state?.blocks;
20  if (Array.isArray(blocks) && blocks.length > 0) {
21    return true;
22  }
23  if (typeof article.plain_text === "string" && article.plain_text.trim()) {
24    return true;
25  }
26  if (typeof article.preview_text === "string" && article.preview_text.trim()) {
27    return true;
28  }
29  return false;
30}
31 
32function parseArticleIdFromUrl(raw: string | undefined): string | null {
33  if (!raw) return null;
34  try {
35    const parsed = new URL(raw);
36    const match = parsed.pathname.match(/\/(?:i\/)?article\/(\d+)/);
37    if (match?.[1]) return match[1];
38  } catch {
39    return null;
40  }
41  return null;
42}
43 
44function extractArticleIdFromUrls(urls: any[] | undefined): string | null {
45  if (!Array.isArray(urls)) return null;
46  for (const url of urls) {
47    const candidate =
48      url?.expanded_url ?? url?.url ?? (url?.display_url ? `https://${url.display_url}` : undefined);
49    const id = parseArticleIdFromUrl(candidate);
50    if (id) return id;
51  }
52  return null;
53}
54 
55export function extractArticleEntityFromTweet(tweet: any): unknown | null {
56  return (
57    tweet?.article?.article_results?.result ??
58    tweet?.article?.result ??
59    tweet?.legacy?.article?.article_results?.result ??
60    tweet?.legacy?.article?.result ??
61    tweet?.article_results?.result ??
62    null
63  );
64}
65 
66export function extractArticleIdFromTweet(tweet: any): string | null {
67  const embedded = extractArticleEntityFromTweet(tweet);
68  const embeddedArticle = embedded as { rest_id?: string } | null;
69  if (embeddedArticle?.rest_id) {
70    return embeddedArticle.rest_id;
71  }
72 
73  const noteUrls = tweet?.note_tweet?.note_tweet_results?.result?.entity_set?.urls;
74  const legacyUrls = tweet?.legacy?.entities?.urls;
75  return extractArticleIdFromUrls(noteUrls) ?? extractArticleIdFromUrls(legacyUrls);
76}
77 
78export async function resolveArticleEntityFromTweet(
79  tweet: any,
80  cookieMap: Record<string, string>
81): Promise<unknown | null> {
82  if (!tweet) return null;
83  const embedded = extractArticleEntityFromTweet(tweet);
84  const embeddedArticle = coerceArticleEntity(embedded);
85  if (embeddedArticle && hasArticleContent(embeddedArticle)) {
86    return embedded;
87  }
88 
89  const articleId = extractArticleIdFromTweet(tweet);
90  if (!articleId) {
91    return embedded ?? null;
92  }
93 
94  const fetched = await fetchXArticle(articleId, cookieMap, false);
95  return fetched ?? embedded ?? null;
96}
97