api/tiktok: add transcript-only downloads

api/README.md

@@ -31,7 +31,7 @@ if the desired service isn't supported yet, feel free to create an appropriate i
 | snapchat          | ✅            | ✅         | ✅         | ❌              | ➖         | ➖              |
 | soundcloud        | ➖            | ✅         | ➖         | ❌              | ✅         | ✅              |
 | streamable        | ✅            | ✅         | ✅         | ❌              | ➖         | ➖              |
-| tiktok            | ✅            | ✅         | ✅         | ❌              | ❌         | ❌              |
+| tiktok            | ✅            | ✅         | ✅         | ✅              | ❌         | ❌              |
 | tumblr            | ✅            | ✅         | ✅         | ❌              | ➖         | ➖              |
 | twitch clips      | ✅            | ✅         | ✅         | ❌              | ✅         | ✅              |
 | twitter/x         | ✅            | ✅         | ✅         | ❌              | ➖         | ➖              |

api/src/processing/match.js

@@ -167,6 +167,9 @@ export default async function({ host, patternMatch, params, authType }) {
                     h265: params.allowH265,
                     alwaysProxy: params.alwaysProxy,
                     subtitleLang,
+                    isCaptionOnly,
+                    captionLanguage: params.captionLanguage || subtitleLang,
+                    captionFormat: params.captionFormat,
                 });
                 break;
 

api/src/processing/service-config.js

@@ -1,7 +1,7 @@
 import UrlPattern from "url-pattern";
 
 export const audioIgnore = new Set(["vk", "ok", "loom"]);
-export const captionOnlyServices = new Set(["loom", "rutube", "vimeo", "vk", "youtube"]);
+export const captionOnlyServices = new Set(["loom", "rutube", "tiktok", "vimeo", "vk", "youtube"]);
 export const hlsExceptions = new Set(["dailymotion", "vimeo", "rutube", "bsky", "youtube"]);
 
 export const services = {

api/src/processing/services/tiktok.js

@@ -5,6 +5,10 @@ import { genericUserAgent } from "../../config.js";
 import { updateCookie } from "../cookie/manager.js";
 import { createStream } from "../../stream/manage.js";
 import { convertLanguageCode } from "../../misc/language-codes.js";
+import {
+    createCaptionResponse,
+    matchesCaptionLanguage,
+} from "../helpers/captions.js";
 
 const shortDomain = "https://vt.tiktok.com/";
 
@@ -72,6 +76,33 @@ export default async function(obj) {
         return { error: "fetch.empty" };
     }
 
+    if (obj.isCaptionOnly) {
+        const requestedLanguage = convertLanguageCode(obj.captionLanguage)
+            || obj.captionLanguage;
+        const subtitle = detail.video?.subtitleInfos?.find(subtitle =>
+            subtitle.Format === "webvtt"
+            && matchesCaptionLanguage(subtitle.LanguageCodeName, requestedLanguage)
+        );
+        if (!subtitle) return { error: "fetch.empty" };
+
+        const language = convertLanguageCode(subtitle.LanguageCodeName)
+            || subtitle.LanguageCodeName;
+        return createCaptionResponse({
+            url: subtitle.Url,
+            format: obj.captionFormat,
+            language,
+            service: "tiktok",
+            id: postId,
+            title: detail.desc?.trim(),
+            author: detail.author.uniqueId,
+            source: `https://www.tiktok.com/@${detail.author.uniqueId}/video/${postId}`,
+            headers: {
+                cookie: cookie.toString(),
+                "user-agent": genericUserAgent,
+            },
+        });
+    }
+
     let video, videoFilename, audioFilename, audio, images,
         filenameBase = `tiktok_${detail.author?.uniqueId}_${postId}`,
         bestAudio; // will get defaulted to m4a later on in match-action

api/src/util/tests/tiktok.json

@@ -8,6 +8,19 @@
             "status": "tunnel"
         }
     },
+    {
+        "name": "standalone English transcript",
+        "url": "https://www.tiktok.com/@tedtoks/video/7168241231892319531",
+        "params": {
+            "downloadMode": "captions",
+            "captionFormat": "txt",
+            "captionLanguage": "en"
+        },
+        "expected": {
+            "code": 200,
+            "status": "tunnel"
+        }
+    },
     {
         "name": "images",
         "url": "https://www.tiktok.com/@matryoshk4/video/7231234675476532526",
@@ -44,4 +57,4 @@
             "status": "error"
         }
     }
-]
+]