1import type { ExtractedDocument, ContentBlock } from "../../extract/document";
2import { findTweetNode, formatMediaList, formatTweetAuthor, getTweetAuthorMetadata, normalizeTitle, toXTweet } from "./shared";
3 
4export function extractSingleTweetDocumentFromPayload(
5  payload: unknown,
6  statusId: string,
7  pageUrl: string,
8): ExtractedDocument | null {
9  const tweet = findTweetNode(payload, statusId);
10  if (!tweet) {
11    return null;
12  }
13 
14  const xTweet = toXTweet(tweet, pageUrl);
15  const content: ContentBlock[] = [];
16 
17  if (xTweet.text) {
18    content.push({ type: "paragraph", text: xTweet.text });
19  }
20 
21  for (const mediaLine of formatMediaList(xTweet.media)) {
22    if (mediaLine.startsWith("photo: ")) {
23      content.push({
24        type: "image",
25        url: mediaLine.slice("photo: ".length),
26      });
27    } else {
28      content.push({
29        type: "list",
30        ordered: false,
31        items: [mediaLine],
32      });
33    }
34  }
35 
36  if (xTweet.quotedTweet) {
37    const quotedLines: string[] = [];
38    const quotedAuthor =
39      xTweet.quotedTweet.author && xTweet.quotedTweet.authorName
40        ? `${xTweet.quotedTweet.authorName} (@${xTweet.quotedTweet.author})`
41        : xTweet.quotedTweet.author
42          ? `@${xTweet.quotedTweet.author}`
43          : xTweet.quotedTweet.authorName;
44 
45    if (quotedAuthor) {
46      quotedLines.push(quotedAuthor);
47    }
48    if (xTweet.quotedTweet.text) {
49      quotedLines.push(xTweet.quotedTweet.text);
50    }
51    quotedLines.push(...formatMediaList(xTweet.quotedTweet.media));
52 
53    if (quotedLines.length > 0) {
54      content.push({ type: "heading", depth: 2, text: "Quoted Tweet" });
55      content.push({ type: "quote", text: quotedLines.join("\n\n") });
56    }
57  }
58 
59  return {
60    url: pageUrl,
61    canonicalUrl: xTweet.url,
62    title: normalizeTitle(
63      xTweet.author ? `@${xTweet.author}: ${xTweet.text}` : xTweet.text,
64      "Tweet",
65    ),
66    author: formatTweetAuthor(xTweet),
67    siteName: "X",
68    publishedAt: xTweet.createdAt,
69    summary: xTweet.text.slice(0, 200) || undefined,
70    adapter: "x",
71    metadata: {
72      kind: "x/post",
73      tweetId: xTweet.id,
74      ...getTweetAuthorMetadata(xTweet),
75      conversationId:
76        typeof tweet.legacy === "object" &&
77        tweet.legacy !== null &&
78        typeof (tweet.legacy as Record<string, unknown>).conversation_id_str === "string"
79          ? (tweet.legacy as Record<string, unknown>).conversation_id_str
80          : undefined,
81      favoriteCount: xTweet.likes,
82      replyCount: xTweet.replies,
83      retweetCount: xTweet.retweets,
84    },
85    content,
86  };
87}
88