pth/api.py

import http.client
import json
import base64
import os
import random
import sys

# --- 配置 ---
API_HOST = "api2.aigcbest.top"
# 警告：请不要在公共场合暴露您的 API Key。此 Key 仅为示例。
API_KEY = "Bearer sk-adCUrxA2Itnm24VHdNODmeFATufbQEDqz76993UwT6qi4WIc"
DEBUG_RESPONSE_FILENAME = "api_last_response.json"

# 可用的音色列表
VOICES = ["alloy", "ash", "ballad", "coral", "echo", "fable", "nova", "onyx", "sage", "shimmer"]

# 输出目录
OUTPUT_DIR = "/data/liyangzhuo/a"

def ensure_output_directory():
    """确保输出目录存在"""
    if not os.path.exists(OUTPUT_DIR):
        os.makedirs(OUTPUT_DIR, exist_ok=True)
        print(f"创建输出目录: {OUTPUT_DIR}")

def select_two_different_voices():
    """从音色列表中随机选择两个不同的音色"""
    selected_voices = random.sample(VOICES, 2)
    return selected_voices[0], selected_voices[1]

def call_audio_api(text_content, voice, person_type):
    """调用音频API生成语音"""
    # 构建prompt，要求模型按照对话场景复述内容
    prompt = f"Please retell the following text in a natural dialogue scene to a high standard of naturalness, using appropriate rhyme and mood based on the following. Please repeat the content accurately and do not add anything:\n\n{text_content}"
    
    payload_dict = {
        "model": "gemini-2.5-flash-all",
        "modalities": ["text", "audio"],
        "audio": {"voice": voice, "format": "wav"},
        "messages": [{"role": "user", "content": prompt}]
    }
    payload = json.dumps(payload_dict)

    headers = {
        'Accept': 'application/json',
        'Authorization': API_KEY,
        'Content-Type': 'application/json'
    }

    conn = None
    try:
        print(f"正在为{person_type}生成音频 (音色: {voice})...")
        conn = http.client.HTTPSConnection(API_HOST)
        
        conn.request("POST", "/v1/chat/completions", payload, headers)
        res = conn.getresponse()
        response_body = res.read()
        
        print(f"响应状态码: {res.status} {res.reason}")

        response_text = response_body.decode('utf-8')
        response_json = json.loads(response_text)

        # 保存调试响应
        debug_filename = f"api_response_{person_type}.json"
        with open(debug_filename, "w", encoding="utf-8") as f:
            json.dump(response_json, f, indent=2, ensure_ascii=False)

        if res.status == 200 and 'error' not in response_json:
            try:
                message = response_json.get('choices', [{}])[0].get('message', {})
                audio_info = message.get('audio', {})
                
                audio_base64_data = audio_info.get('data')
                transcript = audio_info.get('transcript')

                if audio_base64_data:
                    audio_binary_data = base64.b64decode(audio_base64_data)
                    return audio_binary_data, transcript
                else:
                    print(f"警告: 未找到{person_type}的音频数据")
                    return None, None

            except (IndexError, AttributeError) as e:
                print(f"解析{person_type}响应时出错: {e}")
                return None, None
        else:
            print(f"{person_type}请求失败:")
            if 'error' in response_json:
                print(json.dumps(response_json['error'], indent=2, ensure_ascii=False))
            return None, None

    except Exception as e:
        print(f"{person_type}API调用过程中发生错误: {e}")
        return None, None
    finally:
        if conn:
            conn.close()

def process_json_file(json_file_path):
    """处理JSON文件中的所有对话条目"""
    try:
        with open(json_file_path, 'r', encoding='utf-8') as f:
            data = json.load(f)
        
        # 如果是单个对象，转为列表
        if isinstance(data, dict):
            data = [data]
        
        print(f"开始处理 {len(data)} 个对话条目...")
        
        for i, item in enumerate(data):
            try:
                item_id = item.get('id', f'item_{i}')
                person_a_text = item.get('person_a', '')
                person_b_text = item.get('person_b', '')
                
                if not person_a_text or not person_b_text:
                    print(f"跳过条目 {item_id}: 缺少person_a或person_b内容")
                    continue
                
                print(f"\n处理条目 {item_id} ({i+1}/{len(data)})...")
                
                # 为这个对话选择两个不同的音色
                voice_a, voice_b = select_two_different_voices()
                print(f"选择音色: person_a={voice_a}, person_b={voice_b}")
                
                # 调用API生成person_a的音频
                audio_data_a, transcript_a = call_audio_api(person_a_text, voice_a, "person_a")
                if audio_data_a:
                    filename_a = f"{item_id}_person_a.wav"
                    filepath_a = os.path.join(OUTPUT_DIR, filename_a)
                    with open(filepath_a, "wb") as f:
                        f.write(audio_data_a)
                    print(f"✅ person_a 音频已保存: {filepath_a}")
                else:
                    print(f"❌ person_a 音频生成失败")
                
                # 调用API生成person_b的音频
                audio_data_b, transcript_b = call_audio_api(person_b_text, voice_b, "person_b")
                if audio_data_b:
                    filename_b = f"{item_id}_person_b.wav"
                    filepath_b = os.path.join(OUTPUT_DIR, filename_b)
                    with open(filepath_b, "wb") as f:
                        f.write(audio_data_b)
                    print(f"✅ person_b 音频已保存: {filepath_b}")
                else:
                    print(f"❌ person_b 音频生成失败")
                    
            except Exception as e:
                print(f"处理条目 {item.get('id', i)} 时发生错误: {e}")
                continue
        
        print(f"\n✅ 完成处理JSON文件: {json_file_path}")
        
    except Exception as e:
        print(f"读取JSON文件时发生错误: {e}")

def main():
    """主函数"""
    if len(sys.argv) != 2:
        print("用法: python api.py <json_file_path>")
        print("示例: python api.py /path/to/conversations.json")
        sys.exit(1)
    
    json_file_path = sys.argv[1]
    
    if not os.path.exists(json_file_path):
        print(f"错误: 文件不存在 - {json_file_path}")
        sys.exit(1)
    
    # 确保输出目录存在
    ensure_output_directory()
    
    # 处理JSON文件
    process_json_file(json_file_path)

if __name__ == "__main__":
    main()