Merge pull request #302 from boostcampwm-2024/feat/#92/api-summary

[Feat] 오디오 텍스트로 변환 및 요약 로직 구현

apps/api/src/entity/summary.entity.ts

@@ -15,13 +15,10 @@ export class Summary {
   id: number;
 
   @Column('varchar')
-  name: string;
+  audioUrl: string;
 
-  @Column('varchar')
-  source: string;
-
-  @Column('varchar')
-  summary: string;
+  @Column('json')
+  summaryText: string[];
 
   @CreateDateColumn({ type: 'timestamp', name: 'created_at' })
   createdAt: Date;

apps/api/src/stream/dto/createSummaryDto.ts

@@ -0,0 +1,4 @@
+export class CreateSummaryDto {
+  ticleId: number;
+  audioUrl: string;
+}

apps/api/src/stream/stream.controller.ts

@@ -1,4 +1,25 @@
-import { Controller } from '@nestjs/common';
+import { Body, Controller, Get, Param, Post } from '@nestjs/common';
+
+import { CreateSummaryDto } from './dto/createSummaryDto';
+import { StreamService } from './stream.service';
 
 @Controller('stream')
-export class StreamController {}
+export class StreamController {
+  constructor(private readonly streamService: StreamService) {}
+
+  @Post('audio')
+  async getAudioFileUrl(@Body() createSummaryDto: CreateSummaryDto) {
+    const summary = await this.streamService.createSummary(createSummaryDto);
+
+    const fulltext = await this.streamService.transcribeAudio(createSummaryDto.audioUrl);
+    const summaryResult = await this.streamService.summaryAudio(fulltext);
+
+    await this.streamService.updateSummaryText(summary, summaryResult);
+  }
+
+  @Get('summary/:ticleId')
+  async getSummaryByTicleId(@Param('ticleId') ticleId: number) {
+    const text = await this.streamService.getSummaryText(ticleId);
+    return { summary: text };
+  }
+}

apps/api/src/stream/stream.module.ts

@@ -1,8 +1,15 @@
 import { Module } from '@nestjs/common';
+import { TypeOrmModule } from '@nestjs/typeorm';
+
+import { Summary } from '@/entity/summary.entity';
+import { Ticle } from '@/entity/ticle.entity';
 
 import { StreamController } from './stream.controller';
+import { StreamService } from './stream.service';
 
 @Module({
+  imports: [TypeOrmModule.forFeature([Ticle, Summary])],
   controllers: [StreamController],
+  providers: [StreamService],
 })
 export class StreamModule {}

apps/api/src/stream/stream.service.ts

@@ -0,0 +1,117 @@
+import { Injectable, InternalServerErrorException, NotFoundException } from '@nestjs/common';
+import { ConfigService } from '@nestjs/config';
+import { InjectRepository } from '@nestjs/typeorm';
+import { Repository } from 'typeorm';
+import { ErrorMessage } from '@repo/types';
+
+import { Summary } from '@/entity/summary.entity';
+import { Ticle } from '@/entity/ticle.entity';
+
+import { CreateSummaryDto } from './dto/createSummaryDto';
+
+@Injectable()
+export class StreamService {
+  constructor(
+    @InjectRepository(Summary)
+    private summaryRepository: Repository<Summary>,
+    @InjectRepository(Ticle)
+    private ticleRepository: Repository<Ticle>,
+    private configService: ConfigService
+  ) {}
+
+  async createSummary(createSummaryDto: CreateSummaryDto) {
+    const { ticleId, audioUrl } = createSummaryDto;
+    const ticle = await this.ticleRepository.findOne({
+      where: { id: ticleId },
+    });
+
+    if (!ticle) {
+      throw new NotFoundException(ErrorMessage.TICLE_NOT_FOUND);
+    }
+
+    const summary = this.summaryRepository.create({
+      audioUrl: audioUrl,
+      ticle: ticle,
+    });
+    return await this.summaryRepository.save(summary);
+  }
+
+  async transcribeAudio(remoteFileName: string) {
+    const speechEndpoint = this.configService.get<string>('CLOVASPEECH_ENDPOINT');
+
+    const body = {
+      dataKey: remoteFileName,
+      params: { enable: true },
+      language: 'ko-KR',
+      completion: 'sync',
+      fulltext: true,
+    };
+
+    try {
+      const response = await fetch(speechEndpoint, {
+        method: 'POST',
+        headers: {
+          'Content-Type': 'application/json',
+          'X-CLOVASPEECH-API-KEY': this.configService.get<string>('CLOVASPEECH_API_KEY'),
+        },
+        body: JSON.stringify(body),
+      });
+
+      const data = await response.json();
+      return data.text;
+    } catch (error) {
+      console.error('Failed to transcribe audio:', error);
+      throw new InternalServerErrorException(ErrorMessage.FAILED_TO_TRANSCRIBE_AUDIO);
+    }
+  }
+
+  async summaryAudio(text: string) {
+    const studioEndpoint = this.configService.get<string>('CLOVASTUDIO_ENDPOINT');
+
+    const body = {
+      texts: [text],
+      segMinSize: 300,
+      includeAiFilters: true,
+      autoSentenceSplitter: true,
+      segCount: -1,
+      segMaxSize: 1000,
+    };
+
+    try {
+      const response = await fetch(studioEndpoint, {
+        method: 'POST',
+        headers: {
+          'Content-Type': 'application/json',
+          'X-NCP-CLOVASTUDIO-API-KEY': this.configService.get<string>('CLOVASTUDIO_API_KEY'),
+          'X-NCP-APIGW-API-KEY': this.configService.get<string>('CLOVASTUDIO_APIGW_API_KEY'),
+        },
+        body: JSON.stringify(body),
+      });
+
+      if (!response.ok) {
+        const errorText = await response.text();
+        console.error('Error response:', errorText);
+        throw new Error(`API error: ${response.status} - ${response.statusText}`);
+      }
+
+      const data = await response.json();
+
+      return data.result.text;
+    } catch (error) {
+      console.error('Failed to summarize audio:', error);
+      throw new InternalServerErrorException(ErrorMessage.FAILED_TO_SUMMARY_AUDIO);
+    }
+  }
+
+  async getSummaryText(ticleId: number) {
+    const summary = await this.summaryRepository.findOne({
+      where: { ticle: { id: ticleId } },
+    });
+    return summary.summaryText;
+  }
+
+  async updateSummaryText(summary: Summary, summaryText: string[]) {
+    summary.summaryText = summaryText;
+    await this.summaryRepository.save(summary);
+  }
+}

packages/types/src/errorMessages.ts

@@ -14,6 +14,8 @@ export const ErrorMessage = {
   FAILED_TO_CREATE_TICLE: '티클 생성에 실패했습니다',
   CANNOT_REQUEST_OWN_TICLE: '자신이 발표자인 티클에는 신청할 수 없습니다',
   TICLE_ALREADY_REQUESTED: '이미 신청한 티클입니다',
+  FAILED_TO_TRANSCRIBE_AUDIO: 'CLOVA SPEECH로 텍스트 변환에 실패했습니다',
+  FAILED_TO_SUMMARY_AUDIO: 'CLOVA STRUDIO로 요약에 실패했습니다',
   CANNOT_DELETE_OTHERS_TICLE: '다른 사람의 티클은 삭제할 수 없습니다',
   CANNOT_START_TICLE: '티클을 시작할 수 없습니다',
   CANNOT_END_TICLE: '티클을 종료할 수 없습니다',