fix(baoyu-fetch): extract X video variants

2026-07-12 13:59:47 +08:00 · 2026-04-21 13:39:01 -05:00
parent d7f623158c
commit fe26c2b639
4 changed files with 261 additions and 22 deletions
@@ -339,4 +339,105 @@ describe("x article extraction", () => {
    expect(content.markdown).toContain("https://example.com/report");
    expect(content.markdown).not.toContain("https://t.co/example");
  });
+
+  test("renders article video media as the highest bitrate mp4 link", () => {
+    const payload = {
+      data: {
+        tweetResult: {
+          result: {
+            rest_id: "2046628728210350366",
+            legacy: {
+              full_text: "Fallback text",
+              favorite_count: 12,
+              retweet_count: 3,
+              reply_count: 1,
+              created_at: "Tue Apr 21 16:34:47 +0000 2026",
+            },
+            core: {
+              user_results: {
+                result: {
+                  legacy: {
+                    name: "Google AI Studio",
+                    screen_name: "GoogleAIStudio",
+                  },
+                },
+              },
+            },
+            article: {
+              article_results: {
+                result: {
+                  title: "Article with video",
+                  media_entities: [
+                    {
+                      media_id: "2046627051822530560",
+                      media_info: {
+                        __typename: "ApiVideo",
+                        variants: [
+                          {
+                            bit_rate: 2176000,
+                            content_type: "video/mp4",
+                            url: "https://video.twimg.com/amplify_video/2046627051822530560/vid/avc1/1280x720/medium.mp4",
+                          },
+                          {
+                            content_type: "application/x-mpegURL",
+                            url: "https://video.twimg.com/amplify_video/2046627051822530560/pl/playlist.m3u8",
+                          },
+                          {
+                            bit_rate: 10368000,
+                            content_type: "video/mp4",
+                            url: "https://video.twimg.com/amplify_video/2046627051822530560/vid/avc1/1920x1080/high.mp4",
+                          },
+                        ],
+                      },
+                    },
+                  ],
+                  content_state: {
+                    blocks: [
+                      {
+                        type: "atomic",
+                        text: " ",
+                        data: {},
+                        entityRanges: [{ key: 0, length: 1, offset: 0 }],
+                        inlineStyleRanges: [],
+                      },
+                    ],
+                    entityMap: [
+                      {
+                        key: "0",
+                        value: {
+                          type: "MEDIA",
+                          mutability: "Immutable",
+                          data: {
+                            mediaItems: [{ mediaId: "2046627051822530560" }],
+                          },
+                        },
+                      },
+                    ],
+                  },
+                },
+              },
+            },
+          },
+        },
+      },
+    };
+
+    const document = extractArticleDocumentFromPayload(
+      payload,
+      "2046628728210350366",
+      "https://x.com/GoogleAIStudio/status/2046628728210350366",
+    );
+
+    expect(document).not.toBeNull();
+
+    const content = document?.content[0];
+    expect(content?.type).toBe("markdown");
+    if (!content || content.type !== "markdown") {
+      throw new Error("Expected markdown content");
+    }
+
+    expect(content.markdown).toBe(
+      "[video](https://video.twimg.com/amplify_video/2046627051822530560/vid/avc1/1920x1080/high.mp4)",
+    );
+  });
 });
@@ -184,4 +184,77 @@ describe("x single tweet extraction", () => {
        "Quoted Author (@quoted_author)\n\nQuoted tweet text\n\nphoto: https://pbs.twimg.com/media/quoted?format=jpg&name=4096x4096",
    });
  });
+
+  test("uses the highest bitrate mp4 variant for tweet video media", () => {
+    const payload = {
+      data: {
+        tweetResult: {
+          result: {
+            rest_id: "2046628728210350366",
+            legacy: {
+              full_text: "Video post https://t.co/video",
+              favorite_count: 12,
+              retweet_count: 3,
+              reply_count: 1,
+              created_at: "Tue Apr 21 16:34:47 +0000 2026",
+              extended_entities: {
+                media: [
+                  {
+                    type: "video",
+                    media_url_https: "https://pbs.twimg.com/amplify_video_thumb/2046627051822530560/img/poster.jpg",
+                    url: "https://t.co/video",
+                    video_info: {
+                      variants: [
+                        {
+                          content_type: "application/x-mpegURL",
+                          url: "https://video.twimg.com/amplify_video/2046627051822530560/pl/playlist.m3u8",
+                        },
+                        {
+                          bitrate: 256000,
+                          content_type: "video/mp4",
+                          url: "https://video.twimg.com/amplify_video/2046627051822530560/vid/avc1/480x270/low.mp4",
+                        },
+                        {
+                          bitrate: 10368000,
+                          content_type: "video/mp4",
+                          url: "https://video.twimg.com/amplify_video/2046627051822530560/vid/avc1/1920x1080/high.mp4",
+                        },
+                      ],
+                    },
+                  },
+                ],
+              },
+            },
+            core: {
+              user_results: {
+                result: {
+                  legacy: {
+                    name: "Google AI Studio",
+                    screen_name: "GoogleAIStudio",
+                  },
+                },
+              },
+            },
+          },
+        },
+      },
+    };
+
+    const document = extractSingleTweetDocumentFromPayload(
+      payload,
+      "2046628728210350366",
+      "https://x.com/GoogleAIStudio/status/2046628728210350366",
+    );
+
+    expect(document).not.toBeNull();
+
+    const listBlock = document?.content.find((block) => block.type === "list");
+    expect(listBlock).toEqual({
+      type: "list",
+      ordered: false,
+      items: [
+        "video: https://video.twimg.com/amplify_video/2046627051822530560/vid/avc1/1920x1080/high.mp4",
+      ],
+    });
+  });
 });
@@ -9,18 +9,43 @@ import {
  getUser,
  isRecord,
  normalizeTitle,
+  resolveBestXVideoVariantUrl,
  toHighResXImageUrl,
  toXTweet,
 } from "./shared";
 import type { JsonObject } from "./types";

-function resolveArticleMediaUrl(mediaInfo: JsonObject): string {
+interface ArticleMedia {
+  kind: "image" | "video";
+  url: string;
+}
+
+function resolveArticleMedia(mediaInfo: JsonObject): ArticleMedia | null {
+  const videoUrl = resolveBestXVideoVariantUrl(mediaInfo);
+  if (videoUrl) {
+    return {
+      kind: "video",
+      url: videoUrl,
+    };
+  }
+
  const rawUrl =
    (typeof mediaInfo.original_img_url === "string" && mediaInfo.original_img_url) ||
    (typeof mediaInfo.url === "string" && mediaInfo.url) ||
    "";

-  return rawUrl ? toHighResXImageUrl(rawUrl) : "";
+  if (!rawUrl) {
+    return null;
+  }
+
+  return {
+    kind: "image",
+    url: toHighResXImageUrl(rawUrl),
+  };
+}
+
+function resolveArticleMediaUrl(mediaInfo: JsonObject): string {
+  return resolveArticleMedia(mediaInfo)?.url ?? "";
 }

 function normalizeEntityMap(entityMap: unknown): Map<string, JsonObject> {
@@ -139,8 +164,8 @@ function getTweetId(entityMap: Map<string, JsonObject>, entityKey: unknown): str
  return data.tweetId;
 }

-function buildMediaUrlMap(articleResult: JsonObject): Map<string, string> {
-  const mediaMap = new Map<string, string>();
+function buildMediaMap(articleResult: JsonObject): Map<string, ArticleMedia> {
+  const mediaMap = new Map<string, ArticleMedia>();
  const mediaEntities = Array.isArray(articleResult.media_entities) ? articleResult.media_entities : [];

  for (const entity of mediaEntities) {
@@ -148,25 +173,28 @@ function buildMediaUrlMap(articleResult: JsonObject): Map<string, string> {
      continue;
    }

-    const mediaInfo = entity.media_info;
-    const url = resolveArticleMediaUrl(mediaInfo);
-    if (url) {
-      mediaMap.set(entity.media_id, url);
+    const media = resolveArticleMedia(entity.media_info);
+    if (media) {
+      mediaMap.set(entity.media_id, media);
    }
  }

  const coverMedia = isRecord(articleResult.cover_media) ? articleResult.cover_media : null;
  if (coverMedia && typeof coverMedia.media_id === "string" && isRecord(coverMedia.media_info)) {
-    const url = resolveArticleMediaUrl(coverMedia.media_info);
-    if (url) {
-      mediaMap.set(coverMedia.media_id, url);
+    const media = resolveArticleMedia(coverMedia.media_info);
+    if (media) {
+      mediaMap.set(coverMedia.media_id, media);
    }
  }

  return mediaMap;
 }

-function getMediaMarkdown(entityMap: Map<string, JsonObject>, entityKey: unknown, mediaMap: Map<string, string>): string[] {
+function getMediaMarkdown(
+  entityMap: Map<string, JsonObject>,
+  entityKey: unknown,
+  mediaMap: Map<string, ArticleMedia>,
+): string[] {
  const key =
    typeof entityKey === "string" || typeof entityKey === "number"
      ? String(entityKey)
@@ -182,19 +210,19 @@ function getMediaMarkdown(entityMap: Map<string, JsonObject>, entityKey: unknown

  const data = isRecord(entity.data) ? entity.data : {};
  const mediaItems = Array.isArray(data.mediaItems) ? data.mediaItems : [];
-  const urls: string[] = [];
+  const media: ArticleMedia[] = [];

  for (const item of mediaItems) {
    if (!isRecord(item) || typeof item.mediaId !== "string") {
      continue;
    }
-    const url = mediaMap.get(item.mediaId);
-    if (url && !urls.includes(url)) {
-      urls.push(url);
+    const mediaItem = mediaMap.get(item.mediaId);
+    if (mediaItem && !media.some((value) => value.url === mediaItem.url)) {
+      media.push(mediaItem);
    }
  }

-  return urls.map((url) => `![](${url})`);
+  return media.map((item) => item.kind === "image" ? `![](${item.url})` : `[video](${item.url})`);
 }

 function resolveTweetMarkdown(payloads: unknown[], tweetId: string, pageUrl: string): string | null {
@@ -250,7 +278,7 @@ function replaceLinkEntities(text: string, block: JsonObject, entityMap: Map<str
 function renderAtomicBlock(
  block: JsonObject,
  entityMap: Map<string, JsonObject>,
-  mediaMap: Map<string, string>,
+  mediaMap: Map<string, ArticleMedia>,
  payloads: unknown[],
  pageUrl: string,
 ): string | null {
@@ -293,7 +321,7 @@ function renderAtomicBlock(
 function renderArticleBlocks(
  blocks: unknown[],
  entityMap: Map<string, JsonObject>,
-  mediaMap: Map<string, string>,
+  mediaMap: Map<string, ArticleMedia>,
  payloads: unknown[],
  pageUrl: string,
 ): string {
@@ -396,7 +424,7 @@ export function extractArticleDocumentFromPayload(
  const contentState = isRecord(articleResult.content_state) ? articleResult.content_state : {};
  const blocks = Array.isArray(contentState.blocks) ? contentState.blocks : [];
  const entityMap = normalizeEntityMap(contentState.entityMap);
-  const mediaMap = buildMediaUrlMap(articleResult);
+  const mediaMap = buildMediaMap(articleResult);
  const richMarkdown = renderArticleBlocks(blocks, entityMap, mediaMap, payloads, pageUrl);
  const plainText = typeof articleResult.plain_text === "string" ? articleResult.plain_text.trim() : "";
  const markdown = richMarkdown || plainText || getTweetText(tweet);
@@ -251,6 +251,39 @@ export function toHighResXImageUrl(rawUrl: string): string {
  }
 }

+function getVideoVariantBitrate(variant: JsonObject): number {
+  const value = variant.bitrate ?? variant.bit_rate;
+  return typeof value === "number" && Number.isFinite(value) ? value : 0;
+}
+
+function getVideoVariantContentType(variant: JsonObject): string {
+  const value = variant.content_type ?? variant.contentType;
+  return typeof value === "string" ? value.toLowerCase() : "";
+}
+
+export function resolveBestXVideoVariantUrl(mediaInfo: unknown): string | undefined {
+  if (!isRecord(mediaInfo)) {
+    return undefined;
+  }
+
+  const variantsSource =
+    Array.isArray(mediaInfo.variants)
+      ? mediaInfo.variants
+      : isRecord(mediaInfo.video_info) && Array.isArray(mediaInfo.video_info.variants)
+        ? mediaInfo.video_info.variants
+        : [];
+
+  const variants = variantsSource
+    .filter(
+      (variant): variant is JsonObject =>
+        isRecord(variant) && typeof variant.url === "string" && variant.url.length > 0,
+    )
+    .filter((variant) => getVideoVariantContentType(variant) === "video/mp4")
+    .sort((left, right) => getVideoVariantBitrate(right) - getVideoVariantBitrate(left));
+
+  return typeof variants[0]?.url === "string" ? variants[0].url : undefined;
+}
+
 export function getTweetMedia(tweet: JsonObject): XMedia[] {
  const legacy = getLegacy(tweet);
  const extendedEntities = isRecord(legacy.extended_entities) ? legacy.extended_entities : emptyObject();
@@ -268,10 +301,14 @@ export function getTweetMedia(tweet: JsonObject): XMedia[] {
          alt: typeof value.ext_alt_text === "string" ? value.ext_alt_text : undefined,
        };
      }
-      if ((value.type === "video" || value.type === "animated_gif") && typeof value.media_url_https === "string") {
+      if (value.type === "video" || value.type === "animated_gif") {
+        const videoUrl = resolveBestXVideoVariantUrl(value);
+        if (!videoUrl) {
+          return null;
+        }
        return {
          type: value.type,
-          url: value.media_url_https,
+          url: videoUrl,
        };
      }
      return null;