@mux/ai 0.5.0 → 0.5.2

package/dist/{types-KcVfWtUl.d.ts → types-BhVuLeSp.d.ts}

@@ -71,12 +71,6 @@ declare function decryptFromWorkflow<T>(payload: EncryptedPayload, key: Uint8Arr
 interface MuxAIOptions {
     /** Optional timeout (ms) for helper utilities that support request limits. */
     timeout?: number;
-    /**
-     * Optional cancellation signal passed through to underlying AI SDK calls.
-     * When aborted, in-flight model requests will be
-     * cancelled where supported.
-     */
-    abortSignal?: AbortSignal;
     /**
      * Optional credentials for workflow execution.
      * Use encryptForWorkflow when running in Workflow Dev Kit environments.

package/dist/workflows/index.d.ts

@@ -1,7 +1,7 @@
-export { J as AudioTranslationOptions, I as AudioTranslationResult, e as BurnedInCaptionsAnalysis, c as BurnedInCaptionsOptions, b as BurnedInCaptionsPromptOverrides, a as BurnedInCaptionsPromptSections, B as BurnedInCaptionsResult, C as Chapter, o as ChapterSystemPromptSections, n as ChaptersOptions, m as ChaptersPromptOverrides, l as ChaptersPromptSections, k as ChaptersResult, j as ChaptersType, E as EmbeddingsOptions, q as EmbeddingsResult, u as HiveModerationOutput, H as HiveModerationSource, v as ModerationOptions, t as ModerationProvider, M as ModerationResult, S as SUMMARY_KEYWORD_LIMIT, F as SummarizationOptions, D as SummarizationPromptOverrides, A as SummarizationPromptSections, z as SummaryAndTagsResult, y as SummaryType, T as ThumbnailModerationScore, N as TranslationOptions, P as TranslationPayload, L as TranslationResult, d as burnedInCaptionsSchema, f as chapterSchema, g as chaptersSchema, p as generateChapters, r as generateEmbeddings, s as generateVideoEmbeddings, w as getModerationScores, G as getSummaryAndTags, h as hasBurnedInCaptions, x as summarySchema, K as translateAudio, Q as translateCaptions, O as translationSchema } from '../index-BdpEVCbj.js';
+export { J as AudioTranslationOptions, I as AudioTranslationResult, e as BurnedInCaptionsAnalysis, c as BurnedInCaptionsOptions, b as BurnedInCaptionsPromptOverrides, a as BurnedInCaptionsPromptSections, B as BurnedInCaptionsResult, C as Chapter, o as ChapterSystemPromptSections, n as ChaptersOptions, m as ChaptersPromptOverrides, l as ChaptersPromptSections, k as ChaptersResult, j as ChaptersType, E as EmbeddingsOptions, q as EmbeddingsResult, u as HiveModerationOutput, H as HiveModerationSource, v as ModerationOptions, t as ModerationProvider, M as ModerationResult, S as SUMMARY_KEYWORD_LIMIT, F as SummarizationOptions, D as SummarizationPromptOverrides, A as SummarizationPromptSections, z as SummaryAndTagsResult, y as SummaryType, T as ThumbnailModerationScore, N as TranslationOptions, P as TranslationPayload, L as TranslationResult, d as burnedInCaptionsSchema, f as chapterSchema, g as chaptersSchema, p as generateChapters, r as generateEmbeddings, s as generateVideoEmbeddings, w as getModerationScores, G as getSummaryAndTags, h as hasBurnedInCaptions, x as summarySchema, K as translateAudio, Q as translateCaptions, O as translationSchema } from '../index-DgCfxP4T.js';
 import 'zod';
 import '@ai-sdk/anthropic';
 import '@ai-sdk/google';
 import '@ai-sdk/openai';
-import '../types-KcVfWtUl.js';
+import '../types-BhVuLeSp.js';
 import '@mux/mux-node';

package/dist/workflows/index.js

@@ -28018,7 +28018,7 @@ var init_dist10 = __esm({
 });
 
 // src/workflows/burned-in-captions.ts
-import { generateObject } from "ai";
+import { generateText, Output } from "ai";
 import dedent from "dedent";
 import { z as z21 } from "zod";
 
@@ -28669,7 +28669,7 @@ async function getStoryboardUrl(playbackId, width = DEFAULT_STORYBOARD_WIDTH, sh
 // src/workflows/burned-in-captions.ts
 var burnedInCaptionsSchema = z21.object({
   hasBurnedInCaptions: z21.boolean(),
-  confidence: z21.number().min(0).max(1),
+  confidence: z21.number(),
   detectedLanguage: z21.string().nullable()
 });
 var SYSTEM_PROMPT = dedent`
@@ -28770,9 +28770,9 @@ async function analyzeStoryboard({
 }) {
   "use step";
   const model = await createLanguageModelFromConfig(provider, modelId, credentials);
-  const response = await generateObject({
+  const response = await generateText({
     model,
-    schema: burnedInCaptionsSchema,
+    output: Output.object({ schema: burnedInCaptionsSchema }),
     experimental_telemetry: { isEnabled: true },
     messages: [
       {
@@ -28789,7 +28789,10 @@ async function analyzeStoryboard({
     ]
   });
   return {
-    result: response.object,
+    result: {
+      ...response.output,
+      confidence: Math.min(1, Math.max(0, response.output.confidence))
+    },
     usage: {
       inputTokens: response.usage.inputTokens,
       outputTokens: response.usage.outputTokens,
@@ -28853,7 +28856,7 @@ async function hasBurnedInCaptions(assetId, options = {}) {
 }
 
 // src/workflows/chapters.ts
-import { generateObject as generateObject2 } from "ai";
+import { generateText as generateText2, Output as Output2 } from "ai";
 import dedent2 from "dedent";
 import { z as z22 } from "zod";
 
@@ -29082,9 +29085,9 @@ async function generateChaptersWithAI({
   "use step";
   const model = await createLanguageModelFromConfig(provider, modelId, credentials);
   const response = await withRetry(
-    () => generateObject2({
+    () => generateText2({
       model,
-      schema: chaptersSchema,
+      output: Output2.object({ schema: chaptersSchema }),
       messages: [
         {
           role: "system",
@@ -29098,7 +29101,7 @@ async function generateChaptersWithAI({
     })
   );
   return {
-    chapters: response.object,
+    chapters: response.output,
     usage: {
       inputTokens: response.usage.inputTokens,
       outputTokens: response.usage.outputTokens,
@@ -29532,8 +29535,8 @@ async function generateVideoEmbeddings(assetId, options = {}) {
 // src/primitives/thumbnails.ts
 async function getThumbnailUrls(playbackId, duration, options = {}) {
   "use step";
-  const { interval = 10, width = 640, shouldSign = false, credentials } = options;
-  const timestamps = [];
+  const { interval = 10, width = 640, shouldSign = false, maxSamples, credentials } = options;
+  let timestamps = [];
   if (duration <= 50) {
     const spacing = duration / 6;
     for (let i = 1; i <= 5; i++) {
@@ -29544,6 +29547,18 @@ async function getThumbnailUrls(playbackId, duration, options = {}) {
       timestamps.push(time);
     }
   }
+  if (maxSamples !== void 0 && timestamps.length > maxSamples) {
+    const newTimestamps = [];
+    newTimestamps.push(0);
+    if (maxSamples >= 2) {
+      const spacing = duration / (maxSamples - 1);
+      for (let i = 1; i < maxSamples - 1; i++) {
+        newTimestamps.push(spacing * i);
+      }
+      newTimestamps.push(duration);
+    }
+    timestamps = newTimestamps;
+  }
   const baseUrl = `https://image.mux.com/${playbackId}/thumbnail.png`;
   const urlPromises = timestamps.map(async (time) => {
     if (shouldSign) {
@@ -29809,6 +29824,7 @@ async function getModerationScores(assetId, options = {}) {
     thresholds = DEFAULT_THRESHOLDS,
     thumbnailInterval = 10,
     thumbnailWidth = 640,
+    maxSamples,
     maxConcurrent = 5,
     imageSubmissionMode = "url",
     imageDownloadOptions,
@@ -29860,6 +29876,7 @@ async function getModerationScores(assetId, options = {}) {
       interval: thumbnailInterval,
       width: thumbnailWidth,
       shouldSign: policy === "signed",
+      maxSamples,
       credentials
     });
     if (provider === "openai") {
@@ -29901,7 +29918,7 @@ async function getModerationScores(assetId, options = {}) {
 }
 
 // src/workflows/summarization.ts
-import { generateObject as generateObject3 } from "ai";
+import { generateText as generateText3, Output as Output3 } from "ai";
 import dedent3 from "dedent";
 import { z as z23 } from "zod";
 var SUMMARY_KEYWORD_LIMIT = 10;
@@ -30131,9 +30148,9 @@ function buildUserPrompt3({
 async function analyzeStoryboard2(imageDataUrl, provider, modelId, userPrompt, systemPrompt, credentials) {
   "use step";
   const model = await createLanguageModelFromConfig(provider, modelId, credentials);
-  const response = await generateObject3({
+  const response = await generateText3({
     model,
-    schema: summarySchema,
+    output: Output3.object({ schema: summarySchema }),
     messages: [
       {
         role: "system",
@@ -30149,7 +30166,7 @@ async function analyzeStoryboard2(imageDataUrl, provider, modelId, userPrompt, s
     ]
   });
   return {
-    result: response.object,
+    result: response.output,
     usage: {
       inputTokens: response.usage.inputTokens,
       outputTokens: response.usage.outputTokens,
@@ -30162,9 +30179,9 @@ async function analyzeStoryboard2(imageDataUrl, provider, modelId, userPrompt, s
 async function analyzeAudioOnly(provider, modelId, userPrompt, systemPrompt, credentials) {
   "use step";
   const model = await createLanguageModelFromConfig(provider, modelId, credentials);
-  const response = await generateObject3({
+  const response = await generateText3({
     model,
-    schema: summarySchema,
+    output: Output3.object({ schema: summarySchema }),
     messages: [
       {
         role: "system",
@@ -30177,7 +30194,7 @@ async function analyzeAudioOnly(provider, modelId, userPrompt, systemPrompt, cre
     ]
   });
   return {
-    result: response.object,
+    result: response.output,
     usage: {
       inputTokens: response.usage.inputTokens,
       outputTokens: response.usage.outputTokens,
@@ -30220,7 +30237,6 @@ async function getSummaryAndTags(assetId, options) {
     cleanTranscript = true,
     imageSubmissionMode = "url",
     imageDownloadOptions,
-    abortSignal: _abortSignal,
     promptOverrides,
     credentials
   } = options ?? {};
@@ -30881,7 +30897,7 @@ async function translateAudio(assetId, toLanguageCode, options = {}) {
 
 // src/workflows/translate-captions.ts
 import Mux4 from "@mux/mux-node";
-import { generateObject as generateObject4 } from "ai";
+import { generateText as generateText4, Output as Output4 } from "ai";
 import { z as z24 } from "zod";
 var translationSchema = z24.object({
   translation: z24.string()
@@ -30900,15 +30916,13 @@ async function translateVttWithAI({
   toLanguageCode,
   provider,
   modelId,
-  abortSignal,
   credentials
 }) {
   "use step";
-  const languageModel = await createLanguageModelFromConfig(provider, modelId, credentials);
-  const response = await generateObject4({
-    model: languageModel,
-    schema: translationSchema,
-    abortSignal,
+  const model = await createLanguageModelFromConfig(provider, modelId, credentials);
+  const response = await generateText4({
+    model,
+    output: Output4.object({ schema: translationSchema }),
     messages: [
       {
         role: "user",
@@ -30919,7 +30933,7 @@ ${vttContent}`
     ]
   });
   return {
-    translatedVtt: response.object.translation,
+    translatedVtt: response.output.translation,
     usage: {
       inputTokens: response.usage.inputTokens,
       outputTokens: response.usage.outputTokens,
@@ -31066,7 +31080,6 @@ async function translateCaptions(assetId, fromLanguageCode, toLanguageCode, opti
       toLanguageCode,
       provider: modelConfig.provider,
       modelId: modelConfig.modelId,
-      abortSignal: options.abortSignal,
       credentials
     });
     translatedVtt = result.translatedVtt;