@vibeframe/cli 0.27.0 → 0.30.0

package/src/commands/ai-narrate.ts

@@ -1,451 +0,0 @@
-/**
- * @module ai-narrate
- *
- * Auto-narration pipeline and provider listing.
- *
- * CLI commands: narrate, providers
- *
- * Execute function:
- *   autoNarrate - Analyze video -> generate script -> TTS voiceover
- *
- * Extracted from ai.ts as part of modularisation.
- * ai.ts calls registerNarrateCommands(aiCommand).
- *
- * @dependencies Gemini (video analysis), Claude/OpenAI (script generation),
- *              ElevenLabs (TTS), FFmpeg (duration probe)
- */
-
-import { type Command } from "commander";
-import { readFile, writeFile, mkdir } from "node:fs/promises";
-import { resolve, dirname, basename } from "node:path";
-import { existsSync } from "node:fs";
-import chalk from "chalk";
-import ora from "ora";
-import {
-  providerRegistry,
-  whisperProvider,
-  geminiProvider,
-  openaiProvider,
-  claudeProvider,
-  elevenLabsProvider,
-  openaiImageProvider,
-  runwayProvider,
-  klingProvider,
-  replicateProvider,
-  GeminiProvider,
-  OpenAIProvider,
-  ClaudeProvider,
-  ElevenLabsProvider,
-} from "@vibeframe/ai-providers";
-import { Project, type ProjectFile } from "../engine/index.js";
-import { getApiKey } from "../utils/api-key.js";
-import { ffprobeDuration } from "../utils/exec-safe.js";
-import { getAudioDuration } from "../utils/audio.js";
-import { formatTime } from "./ai-helpers.js";
-
-// ==========================================
-// Auto-Narrate Feature Types and Functions
-// ==========================================
-
-/** Options for {@link autoNarrate}. */
-export interface AutoNarrateOptions {
-  /** Path to video file */
-  videoPath: string;
-  /** Duration of the video in seconds */
-  duration: number;
-  /** Output directory for generated files */
-  outputDir: string;
-  /** ElevenLabs voice name or ID (default: "rachel") */
-  voice?: string;
-  /** Narration style */
-  style?: "informative" | "energetic" | "calm" | "dramatic";
-  /** Language for narration (default: "en") */
-  language?: string;
-  /** LLM provider for script generation: "claude" (default) or "openai" */
-  scriptProvider?: "claude" | "openai";
-}
-
-/** Result from {@link autoNarrate}. */
-export interface AutoNarrateResult {
-  success: boolean;
-  /** Path to generated audio file */
-  audioPath?: string;
-  /** Generated narration script */
-  script?: string;
-  /** Transcript segments for timeline sync */
-  segments?: Array<{
-    startTime: number;
-    endTime: number;
-    text: string;
-  }>;
-  /** Error message if failed */
-  error?: string;
-}
-
-/**
- * Generate narration for a video that doesn't have one.
- *
- * Pipeline:
- * 1. Analyze video with Gemini Video Understanding
- * 2. Generate narration script with Claude (fallback to OpenAI on 529)
- * 3. Convert to speech with ElevenLabs TTS
- *
- * Saves both the audio file and script text to the output directory.
- *
- * @param options - Auto-narrate configuration
- * @returns Result with audio path, script text, and timed segments
- */
-export async function autoNarrate(options: AutoNarrateOptions): Promise<AutoNarrateResult> {
-  const {
-    videoPath,
-    duration,
-    outputDir,
-    voice = "rachel",
-    style = "informative",
-    language = "en",
-    scriptProvider = "claude",
-  } = options;
-
-  // Validate API keys
-  const geminiApiKey = await getApiKey("GOOGLE_API_KEY", "Google");
-  if (!geminiApiKey) {
-    return { success: false, error: "GOOGLE_API_KEY required for video analysis. Run 'vibe setup' or set GOOGLE_API_KEY in .env" };
-  }
-
-  let claudeApiKey: string | null = null;
-  let openaiScriptApiKey: string | null = null;
-  if (scriptProvider === "openai") {
-    openaiScriptApiKey = await getApiKey("OPENAI_API_KEY", "OpenAI");
-    if (!openaiScriptApiKey) {
-      return { success: false, error: "OPENAI_API_KEY required for script generation. Run 'vibe setup' or set OPENAI_API_KEY in .env" };
-    }
-  } else {
-    claudeApiKey = await getApiKey("ANTHROPIC_API_KEY", "Anthropic");
-    if (!claudeApiKey) {
-      return { success: false, error: "ANTHROPIC_API_KEY required for script generation. Run 'vibe setup' or set ANTHROPIC_API_KEY in .env" };
-    }
-  }
-
-  const elevenlabsApiKey = await getApiKey("ELEVENLABS_API_KEY", "ElevenLabs");
-  if (!elevenlabsApiKey) {
-    return { success: false, error: "ELEVENLABS_API_KEY required for TTS. Run 'vibe setup' or set ELEVENLABS_API_KEY in .env" };
-  }
-
-  try {
-    // Step 1: Analyze video with Gemini
-    const gemini = new GeminiProvider();
-    await gemini.initialize({ apiKey: geminiApiKey });
-
-    const videoBuffer = await readFile(videoPath);
-
-    const analysisPrompt = `Analyze this video in detail for narration purposes. Describe:
-1. What is happening visually (actions, movements, subjects)
-2. The setting and environment
-3. Any text or graphics visible
-4. The mood and tone of the content
-5. Key moments and their approximate timestamps
-
-Provide a detailed description that could be used to write a voiceover narration.
-Focus on what viewers need to know to understand and appreciate the video.`;
-
-    const analysisResult = await gemini.analyzeVideo(videoBuffer, analysisPrompt, {
-      fps: 0.5, // Lower FPS for cost optimization
-      lowResolution: duration > 60, // Use low res for longer videos
-    });
-
-    if (!analysisResult.success || !analysisResult.response) {
-      return { success: false, error: `Video analysis failed: ${analysisResult.error}` };
-    }
-
-    // Step 2: Generate narration script with Claude or OpenAI
-    let scriptResult: { success: boolean; script?: string; segments?: Array<{ startTime: number; endTime: number; text: string }>; error?: string };
-
-    if (scriptProvider === "openai") {
-      const gpt = new OpenAIProvider();
-      await gpt.initialize({ apiKey: openaiScriptApiKey! });
-      scriptResult = await gpt.generateNarrationScript(
-        analysisResult.response,
-        duration,
-        style,
-        language
-      );
-    } else {
-      const claude = new ClaudeProvider();
-      await claude.initialize({ apiKey: claudeApiKey! });
-      scriptResult = await claude.generateNarrationScript(
-        analysisResult.response,
-        duration,
-        style,
-        language
-      );
-
-      // Auto-fallback to OpenAI on Claude overload (529)
-      if (!scriptResult.success && scriptResult.error?.includes("529")) {
-        const fallbackKey = await getApiKey("OPENAI_API_KEY", "OpenAI");
-        if (fallbackKey) {
-          console.error("⚠️  Claude overloaded, falling back to OpenAI...");
-          const gpt = new OpenAIProvider();
-          await gpt.initialize({ apiKey: fallbackKey });
-          scriptResult = await gpt.generateNarrationScript(
-            analysisResult.response,
-            duration,
-            style,
-            language
-          );
-        }
-      }
-    }
-
-    if (!scriptResult.success || !scriptResult.script) {
-      return { success: false, error: `Script generation failed: ${scriptResult.error}` };
-    }
-
-    // Step 3: Convert to speech with ElevenLabs
-    const elevenlabs = new ElevenLabsProvider();
-    await elevenlabs.initialize({ apiKey: elevenlabsApiKey });
-
-    const ttsResult = await elevenlabs.textToSpeech(scriptResult.script, {
-      voiceId: voice,
-    });
-
-    if (!ttsResult.success || !ttsResult.audioBuffer) {
-      return { success: false, error: `TTS generation failed: ${ttsResult.error}` };
-    }
-
-    // Ensure output directory exists
-    if (!existsSync(outputDir)) {
-      await mkdir(outputDir, { recursive: true });
-    }
-
-    // Save audio file
-    const audioPath = resolve(outputDir, "auto-narration.mp3");
-    await writeFile(audioPath, ttsResult.audioBuffer);
-
-    // Save script for reference
-    const scriptPath = resolve(outputDir, "narration-script.txt");
-    await writeFile(scriptPath, scriptResult.script, "utf-8");
-
-    return {
-      success: true,
-      audioPath,
-      script: scriptResult.script,
-      segments: scriptResult.segments,
-    };
-  } catch (error) {
-    return {
-      success: false,
-      error: error instanceof Error ? error.message : "Unknown error in autoNarrate",
-    };
-  }
-}
-
-// ==========================================
-// CLI Command Registration
-// ==========================================
-
-export function registerNarrateCommands(ai: Command): void {
-
-ai
-  .command("providers")
-  .description("List available AI providers")
-  .action(async () => {
-    // Register default providers
-    providerRegistry.register(whisperProvider);
-    providerRegistry.register(geminiProvider);
-    providerRegistry.register(openaiProvider);
-    providerRegistry.register(claudeProvider);
-    providerRegistry.register(elevenLabsProvider);
-    providerRegistry.register(openaiImageProvider);
-    providerRegistry.register(runwayProvider);
-    providerRegistry.register(klingProvider);
-    providerRegistry.register(replicateProvider);
-
-    console.log();
-    console.log(chalk.bold.cyan("Available AI Providers"));
-    console.log(chalk.dim("─".repeat(60)));
-
-    const providers = providerRegistry.getAll();
-    for (const provider of providers) {
-      const status = provider.isAvailable ? chalk.green("●") : chalk.red("○");
-      console.log();
-      console.log(`${status} ${chalk.bold(provider.name)} ${chalk.dim(`(${provider.id})`)}`);
-      console.log(`  ${provider.description}`);
-      console.log(`  ${chalk.dim("Capabilities:")} ${provider.capabilities.join(", ")}`);
-    }
-
-    console.log();
-  });
-
-// Auto-Narrate command
-ai
-  .command("narrate")
-  .description("Generate AI narration for a video file or project")
-  .argument("<input>", "Video file or project file (.vibe.json)")
-  .option("-o, --output <dir>", "Output directory for generated files", ".")
-  .option("-v, --voice <name>", "ElevenLabs voice name (rachel, adam, josh, etc.)", "rachel")
-  .option("-s, --style <style>", "Narration style: informative, energetic, calm, dramatic", "informative")
-  .option("-l, --language <lang>", "Language code (e.g., en, ko)", "en")
-  .option("-p, --provider <name>", "LLM for script generation: claude (default), openai", "claude")
-  .option("--add-to-project", "Add narration to project (only for .vibe.json input)")
-  .action(async (inputPath: string, options) => {
-    try {
-      const absPath = resolve(process.cwd(), inputPath);
-      if (!existsSync(absPath)) {
-        console.error(chalk.red(`File not found: ${absPath}`));
-        process.exit(1);
-      }
-
-      console.log();
-      console.log(chalk.bold.cyan("🎙️ Auto-Narrate Pipeline"));
-      console.log(chalk.dim("─".repeat(60)));
-      console.log();
-
-      const isProject = inputPath.endsWith(".vibe.json");
-      let videoPath: string;
-      let project: Project | null = null;
-      let outputDir = resolve(process.cwd(), options.output);
-
-      if (isProject) {
-        // Load project to find video source
-        const content = await readFile(absPath, "utf-8");
-        const data: ProjectFile = JSON.parse(content);
-        project = Project.fromJSON(data);
-        const sources = project.getSources();
-        const videoSource = sources.find((s) => s.type === "video");
-
-        if (!videoSource) {
-          console.error(chalk.red("No video source found in project"));
-          process.exit(1);
-        }
-
-        videoPath = resolve(dirname(absPath), videoSource.url);
-        if (!existsSync(videoPath)) {
-          console.error(chalk.red(`Video file not found: ${videoPath}`));
-          process.exit(1);
-        }
-
-        // Use project directory as output if not specified
-        if (options.output === ".") {
-          outputDir = dirname(absPath);
-        }
-
-        console.log(`📁 Project: ${chalk.bold(project.getMeta().name)}`);
-      } else {
-        videoPath = absPath;
-        console.log(`🎬 Video: ${chalk.bold(basename(videoPath))}`);
-      }
-
-      // Get video duration
-      const durationSpinner = ora("📊 Analyzing video...").start();
-      let duration: number;
-      try {
-        duration = await ffprobeDuration(videoPath);
-        durationSpinner.succeed(chalk.green(`Duration: ${formatTime(duration)}`));
-      } catch {
-        durationSpinner.fail(chalk.red("Failed to get video duration"));
-        process.exit(1);
-      }
-
-      // Validate style option
-      const validStyles = ["informative", "energetic", "calm", "dramatic"];
-      if (!validStyles.includes(options.style)) {
-        console.error(chalk.red(`Invalid style: ${options.style}`));
-        console.error(chalk.dim(`Valid styles: ${validStyles.join(", ")}`));
-        process.exit(1);
-      }
-
-      // Generate narration
-      const generateSpinner = ora("🤖 Generating narration...").start();
-
-      generateSpinner.text = "📹 Analyzing video with Gemini...";
-      const result = await autoNarrate({
-        videoPath,
-        duration,
-        outputDir,
-        voice: options.voice,
-        style: options.style as "informative" | "energetic" | "calm" | "dramatic",
-        language: options.language,
-        scriptProvider: options.provider as "claude" | "openai",
-      });
-
-      if (!result.success) {
-        generateSpinner.fail(chalk.red(`Failed: ${result.error}`));
-        process.exit(1);
-      }
-
-      generateSpinner.succeed(chalk.green("Narration generated successfully"));
-
-      // Display result
-      console.log();
-      console.log(chalk.bold.cyan("Generated Files"));
-      console.log(chalk.dim("─".repeat(60)));
-      console.log(`  🎵 Audio: ${chalk.green(result.audioPath)}`);
-      console.log(`  📝 Script: ${chalk.green(resolve(outputDir, "narration-script.txt"))}`);
-
-      if (result.segments && result.segments.length > 0) {
-        console.log();
-        console.log(chalk.bold.cyan("Narration Segments"));
-        console.log(chalk.dim("─".repeat(60)));
-        for (const seg of result.segments.slice(0, 5)) {
-          console.log(`  [${formatTime(seg.startTime)} - ${formatTime(seg.endTime)}] ${chalk.dim(seg.text.substring(0, 50))}${seg.text.length > 50 ? "..." : ""}`);
-        }
-        if (result.segments.length > 5) {
-          console.log(chalk.dim(`  ... and ${result.segments.length - 5} more segments`));
-        }
-      }
-
-      // Add to project if requested
-      if (options.addToProject && project && isProject) {
-        const addSpinner = ora("Adding narration to project...").start();
-
-        // Get audio duration
-        let audioDuration: number;
-        try {
-          audioDuration = await getAudioDuration(result.audioPath!);
-        } catch {
-          audioDuration = duration; // Fallback to video duration
-        }
-
-        // Add audio source
-        const audioSource = project.addSource({
-          name: "Auto-generated narration",
-          url: basename(result.audioPath!),
-          type: "audio",
-          duration: audioDuration,
-        });
-
-        // Add audio clip to audio track
-        const audioTrack = project.getTracks().find((t) => t.type === "audio");
-        if (audioTrack) {
-          project.addClip({
-            sourceId: audioSource.id,
-            trackId: audioTrack.id,
-            startTime: 0,
-            duration: Math.min(audioDuration, duration),
-            sourceStartOffset: 0,
-            sourceEndOffset: Math.min(audioDuration, duration),
-          });
-        }
-
-        // Save updated project
-        await writeFile(absPath, JSON.stringify(project.toJSON(), null, 2), "utf-8");
-        addSpinner.succeed(chalk.green("Narration added to project"));
-      }
-
-      console.log();
-      console.log(chalk.bold.green("✅ Auto-narrate complete!"));
-
-      if (!options.addToProject && isProject) {
-        console.log();
-        console.log(chalk.dim("Tip: Use --add-to-project to automatically add the narration to your project"));
-      }
-
-      console.log();
-    } catch (error) {
-      console.error(chalk.red("Auto-narrate failed"));
-      console.error(error);
-      process.exit(1);
-    }
-  });
-
-} // end registerNarrateCommands