feat: ✨ (api) transcribeYoutubeVideo を実装

apps/api/.gitignore

@@ -4,5 +4,9 @@ schema.gql
 # auto generated folder
 generated
 
+# tmp files
+tmp/*
+!tmp/.gitkeep
+
 # certificates
 *.crt

apps/api/package.json

@@ -21,6 +21,7 @@
   },
   "dependencies": {
     "@as-integrations/fastify": "2.1.1",
+    "@distube/ytdl-core": "4.13.3",
     "@fastify/compress": "7.0.0",
     "@fastify/cors": "9.0.1",
     "@fastify/helmet": "11.1.1",
@@ -33,7 +34,9 @@
     "@nestjs/core": "10.3.3",
     "@nestjs/platform-fastify": "10.3.3",
     "@nestjs/testing": "10.3.3",
+    "fluent-ffmpeg": "2.1.2",
     "langchain": "0.1.27",
+    "openai": "4.29.1",
     "ts-pattern": "5.0.8",
     "uuid": "9.0.1",
     "zod": "3.22.4"
@@ -42,6 +45,7 @@
     "@hanjaemeo-api/tsconfig": "workspace:*",
     "@hanjaemeo-api/type": "workspace:*",
     "@nestjs/schematics": "10.1.1",
+    "@types/fluent-ffmpeg": "2.1.24",
     "@types/uuid": "9.0.8"
   }
 }

apps/api/src/common/service/env/env.service.ts

@@ -6,7 +6,6 @@ export class EnvService {
   private readonly logger = new Logger(EnvService.name);
 
   constructor(@Inject(ConfigService) private readonly configService: ConfigService) {
-    this.logger.debug(`${EnvService.name} constructed`);
     this.logger.log(`NODE_ENV: ${this.NodeEnv}`);
   }
 

apps/api/src/infra/langchain/langchain.controller.ts

@@ -0,0 +1,27 @@
+import { Controller, Get, Inject, Logger, Param } from '@nestjs/common';
+import { OpenaiService } from '#api/infra/openai/openai.service';
+import { YoutubeService } from '../youtube/youtube.service';
+
+@Controller()
+export class LangchainController {
+  private readonly logger = new Logger(LangchainController.name);
+
+  constructor(
+    @Inject(OpenaiService)
+    private readonly openaiService: OpenaiService,
+    @Inject(YoutubeService)
+    private readonly youtubeService: YoutubeService,
+  ) {}
+
+  @Get('/predict/:id')
+  async transcribeYoutubeVideo(@Param('id') id: string): Promise<string> {
+    this.logger.log(`${this.transcribeYoutubeVideo.name} called`);
+    this.logger.debug(`Transcribing video: https://www.youtube.com/watch?v=${id}`);
+
+    const outputFilePath = await this.youtubeService.saveVideoStreamToFile(`https://www.youtube.com/watch?v=${id}`);
+
+    const transcription = await this.openaiService.transcribeFromFile(outputFilePath);
+
+    return transcription;
+  }
+}

apps/api/src/infra/langchain/langchain.module.ts

@@ -1,8 +1,13 @@
 import { Global, Module } from '@nestjs/common';
+import { OpenaiModule } from '#api/infra/openai/openai.module';
+import { YoutubeModule } from '#api/infra/youtube/youtube.module';
+import { LangchainController } from './langchain.controller';
 import { LangchainService } from './langchain.service';
 
 @Global()
 @Module({
+  imports: [OpenaiModule, YoutubeModule],
+  controllers: [LangchainController],
   providers: [LangchainService],
   exports: [LangchainService],
 })

apps/api/src/infra/langchain/langchain.service.ts

@@ -30,9 +30,7 @@ export class LangchainService {
 
   private readonly logger = new Logger(LangchainService.name);
 
-  constructor(@Inject(EnvService) private readonly envService: EnvService) {
-    this.logger.debug(`${LangchainService.name} constructed`);
-  }
+  constructor(@Inject(EnvService) private readonly envService: EnvService) {}
 
   async embedding(text: string) {
     if (!this.embeddingModel) {

apps/api/src/infra/openai/openai.module.ts

@@ -0,0 +1,8 @@
+import { Module } from '@nestjs/common';
+import { OpenaiService } from './openai.service';
+
+@Module({
+  providers: [OpenaiService],
+  exports: [OpenaiService],
+})
+export class OpenaiModule {}

apps/api/src/infra/openai/openai.service.ts

@@ -0,0 +1,25 @@
+import fs from 'node:fs';
+import { Inject, Injectable } from '@nestjs/common';
+import { OpenAI } from 'openai';
+import { EnvService } from '#api/common/service/env/env.service';
+
+@Injectable()
+export class OpenaiService {
+  private readonly openai: OpenAI;
+
+  constructor(@Inject(EnvService) private readonly envService: EnvService) {
+    this.openai = new OpenAI({ apiKey: this.envService.OpenaiApiKey });
+  }
+
+  async transcribeFromFile(path: string): Promise<string> {
+    const res = await this.openai.audio.transcriptions.create({
+      model: 'whisper-1',
+      language: 'ko',
+      timestamp_granularities: ['segment'],
+      response_format: 'verbose_json',
+      file: fs.createReadStream(path),
+    });
+
+    return res.text;
+  }
+}

apps/api/src/infra/youtube/youtube.module.ts

@@ -0,0 +1,8 @@
+import { Module } from '@nestjs/common';
+import { YoutubeService } from './youtube.service';
+
+@Module({
+  providers: [YoutubeService],
+  exports: [YoutubeService],
+})
+export class YoutubeModule {}

apps/api/src/infra/youtube/youtube.service.ts

@@ -0,0 +1,74 @@
+import fs from 'node:fs';
+import ytdl from '@distube/ytdl-core';
+import { Injectable, Logger } from '@nestjs/common';
+import ffmpeg from 'fluent-ffmpeg';
+
+@Injectable()
+export class YoutubeService {
+  private readonly logger = new Logger(YoutubeService.name);
+
+  async saveVideoStreamToFile(url: string): Promise<string> {
+    const videoInfo = await this.getVideoInfo(url);
+
+    const downloadedPath = this.isAlreadyDownloaded(videoInfo.videoDetails.videoId)
+      ? `./tmp/${videoInfo.videoDetails.videoId}.mp4`
+      : await this.getSoundOnlyVideoStream(videoInfo);
+
+    const audioFilePath = this.isAlreadyConverted(videoInfo.videoDetails.videoId)
+      ? `./tmp/${videoInfo.videoDetails.videoId}.mp3`
+      : await this.convertToMp3(downloadedPath, videoInfo.videoDetails.videoId);
+
+    return audioFilePath;
+  }
+
+  private async getVideoInfo(url: string) {
+    const videoInfo = await ytdl.getInfo(url);
+
+    return videoInfo;
+  }
+
+  private isAlreadyDownloaded(id: string): boolean {
+    const downloaded = fs.existsSync(`./tmp/${id}.mp4`);
+
+    return downloaded;
+  }
+
+  private async getSoundOnlyVideoStream(videoInfo: ytdl.videoInfo): Promise<string> {
+    const stream = await ytdl
+      .downloadFromInfo(videoInfo, {
+        filter: 'audioonly',
+        requestOptions: {
+          reset: true,
+        },
+      })
+      .pipe(fs.createWriteStream(`./tmp/${videoInfo.videoDetails.videoId}.mp4`));
+
+    this.logger.debug(`Downloaded: ${videoInfo.videoDetails.videoId}`);
+
+    return stream.path as string;
+  }
+
+  private isAlreadyConverted(id: string): boolean {
+    const downloaded = fs.existsSync(`./tmp/${id}.mp3`);
+
+    return downloaded;
+  }
+
+  private async convertToMp3(inputPath: string, id: string): Promise<string> {
+    await new Promise<void>((resolve, reject) => {
+      ffmpeg(inputPath)
+        .outputFormat('mp3')
+        .on('end', () => {
+          resolve(void undefined);
+        })
+        .on('error', error => {
+          reject(error);
+        })
+        .save(`./tmp/${id}.mp3`);
+    });
+
+    this.logger.debug(`Converted to mp3: ${id}`);
+
+    return `./tmp/${id}.mp3`;
+  }
+}

docker/api/Dockerfile.production

@@ -12,9 +12,15 @@ RUN bun install --frozen-lockfile --ignore-scripts
 # hadolint ignore=DL3059
 RUN bun turbo --filter='@hanjaemeo-api/api' build
 
-FROM oven/bun:1.0.30-alpine AS runner
+FROM oven/bun:1.0.30-slim AS runner
 ENV NODE_ENV=production
 
+# hadolint ignore=DL3008
+RUN apt-get update \
+  && apt-get --no-install-recommends -y install ffmpeg \
+  && apt-get clean \
+  && rm -rf /var/lib/apt/lists
+
 WORKDIR /hanjaemeo-api/
 
 COPY --from=builder /hanjaemeo-api/node_modules/ ./node_modules/