Twilio REST API响应处理问题：抑郁症辅助程序开发遇阻

需求说明 #

• 开发抑郁症辅助治疗程序，需通过Twilio收集用户对9条指定问题的回复，并按1-9项顺序处理
• 已查阅Twilio官方文档并联系支持，仍存在两个核心问题：
  1. 编写的Python代码仅能发送问题，无法完成回复收集与按顺序处理的步骤
  2. 配置的Twilio Flow无法正常工作，且当前UI与官方演示视频不符

待收集回复的问题列表 #

注：第一条包含Twilio试用账号标识，实际部署可移除

1. Sent from your Twilio trial account - What are the positive results or outcomes you have achieved lately?
2. What are the strengths and resources you have available to you to get even more results and were likely the reason you got the results in the first question.
3. What are your current priorities? What do you and your team need to be focused on right now?
4. What are the benefits to all involved-you, your team and all other stakeholders who will be impacted by achieving your priority focus.
5. How can we (you and/or your team) move close? What action steps are needed?
6. What am I going to do today?
7. What am I doing tomorrow ?
8. What did I do yesterday?

已尝试的Python代码 #

# Download the helper library from https://www.twilio.com/docs/python/install
import os
from twilio.rest import Client
import logging
import csv
import psycopg2
from flask import Flask, request, redirect
from twilio.twiml.messaging_response import MessagingResponse

app = Flask(__name__)

logging.basicConfig(level=logging.DEBUG, format='%(asctime)s - %(levelname)s - %(message)s')

# Set environtment variables
DATABASE = os.environ["DATABASE"]
PASSWORD = os.environ["PASSWORD"]
PORT     = os.environ["PORT"]
USER     = os.environ["USER"]
HOST     = os.environ["HOST"]


# initialization TODO: move into env vars
MY_PHONE_NUMBER        = os.environ["MY_PHONE_NUMBER"]
TWILIO_PHONE_NUMBER    = os.environ["TWILIO_PHONE_NUMBER"]
TWILIO_ACCOUNT_SID     = os.environ["TWILIO_ACCOUNT_SID"]
TWILIO_AUTH_TOKEN      = os.environ["TWILIO_AUTH_TOKEN"]

# Configure Twillio
# Set environment variables for your credentials
# Read more at http://twil.io/secure
client = Client(TWILIO_ACCOUNT_SID, TWILIO_AUTH_TOKEN)
logging.debug(f"Connected to Twilio using MY_PHONE_NUMBER:{MY_PHONE_NUMBER},TWILIO_PHONE_NUMBER{TWILIO_PHONE_NUMBER}")

# Establish db connection
# use psycopg to connect to the db and create a table
conn = psycopg2.connect(
database=DATABASE, user=USER, password=PASSWORD, host=HOST, port=PORT)
conn.autocommit = True
cursor = conn.cursor()


# Step 1: Set up frequency, i.e. times to send messages

# Step 2: Load questions
questionsFile   = open('questions.csv')
questions = csv.reader(questionsFile)
logging.debug(f"message:{questions}")
message = "\n".join([question for row in questions for question in row])
logging.debug(f"message: {message}")

# Step 3: Send questions
# message = client.messages.create(
#   body=message,
#   from_=TWILIO_PHONE_NUMBER,
#   to=MY_PHONE_NUMBER
# )

# Step 4: Collect response
@app.route("/sms", methods=['GET', 'POST'])
def incoming_sms():
    """Send a dynamic reply to an incoming text message"""
    # Get the message the user sent our Twilio number
    body = request.values.get('Body', None)

    # Start our TwiML response
    resp = MessagingResponse()

    # Determine the right reply for this message
    if body == 'hello':
        resp.message("Hi!")
    elif body == 'bye':
        resp.message("Goodbye")

    return str(resp)


if __name__ == "__main__":
    app.run(debug=True)

# Step 5: Create a database table as the sheet name and Save responses in db
logging.debug(f'Step 2 creating table response')
# TODO: create 10 columns for saving responses (each response contains 10 answers)
sql = f'CREATE TABLE IF NOT EXISTS public.responses'
logging.debug(f'CREATE TABLE IF NOT EXISTS public.responses')
# cursor.execute(sql)
# conn.commit()

# Next steps:
# 1. Process positive and negative sentiment from responses
# 2. Calculuate total positive sentiment
# 3. Calculate  total negative sentiment
# 4. Plot positive sentiment vs. negative sentiment

核心问题拆解 #

代码层面问题 #

现有/sms路由仅能处理固定关键词（hello/bye），未实现：

• 按顺序向用户推送单个问题（而非一次性发送所有）
• 跟踪用户当前回答到第几个问题的状态
• 将用户回复关联到对应问题并存储到数据库

Twilio Flow问题 #

• 当前Flow界面与演示视频不一致，导致配置逻辑无法匹配
• 未实现多轮对话的状态跟踪，无法引导用户完成9个问题的回复

解决方案实现 #

1. 代码改造：实现多轮对话与回复收集 #

核心思路

• 使用用户手机号作为唯一标识，存储当前对话状态（已回答的问题序号）
• 每次收到回复后，先存储该回复到数据库，再推送下一个问题
• 所有问题回复完成后，发送结束提示

改造后代码片段

# 新增：存储用户对话状态（生产环境建议用数据库替代内存字典）
user_session = {}

# 加载问题列表（按顺序存储）
def load_questions():
    questions = []
    with open('questions.csv', 'r') as f:
        reader = csv.reader(f)
        for row in reader:
            questions.extend(row)
    # 确保是9个问题
    return questions[:9]

QUESTIONS = load_questions()

@app.route("/sms", methods=['GET', 'POST'])
def incoming_sms():
    resp = MessagingResponse()
    body = request.values.get('Body', '').strip()
    from_number = request.values.get('From', '')

    # 初始化用户会话
    if from_number not in user_session:
        user_session[from_number] = {'current_question': 0, 'responses': {}}
        # 发送第一个问题
        resp.message(QUESTIONS[0])
        return str(resp)
    
    current_session = user_session[from_number]
    current_idx = current_session['current_question']

    # 存储当前回复
    current_session['responses'][current_idx + 1] = body

    # 判断是否还有下一个问题
    if current_idx + 1 < len(QUESTIONS):
        current_session['current_question'] += 1
        next_question = QUESTIONS[current_session['current_question']]
        resp.message(next_question)
    else:
        # 所有问题完成，保存到数据库
        save_responses_to_db(from_number, current_session['responses'])
        resp.message("感谢你的回复，我们会尽快处理你的信息！")
        # 清除会话状态
        del user_session[from_number]
    
    return str(resp)

# 新增：保存回复到数据库函数
def save_responses_to_db(phone_number, responses):
    # 创建完整的表结构
    create_table_sql = """
    CREATE TABLE IF NOT EXISTS public.responses (
        id SERIAL PRIMARY KEY,
        phone_number VARCHAR(20) NOT NULL,
        question1 TEXT,
        question2 TEXT,
        question3 TEXT,
        question4 TEXT,
        question5 TEXT,
        question6 TEXT,
        question7 TEXT,
        question8 TEXT,
        question9 TEXT,
        created_at TIMESTAMP DEFAULT CURRENT_TIMESTAMP
    )
    """
    cursor.execute(create_table_sql)
    
    # 插入数据
    insert_sql = """
    INSERT INTO public.responses (phone_number, question1, question2, question3, question4, question5, question6, question7, question8, question9)
    VALUES (%s, %s, %s, %s, %s, %s, %s, %s, %s, %s)
    """
    # 按顺序提取1-9的回复，无内容则用空字符串填充
    response_values = [
        responses.get(1, ''), responses.get(2, ''), responses.get(3, ''),
        responses.get(4, ''), responses.get(5, ''), responses.get(6, ''),
        responses.get(7, ''), responses.get(8, ''), responses.get(9, '')
    ]
    cursor.execute(insert_sql, (phone_number, *response_values))

2. Twilio Flow配置修复 #

核心步骤

1. 进入Twilio控制台的Studio页面，创建新的Flow
2. 使用Send & Wait for Reply组件实现多轮对话：
   • 第一个组件发送第1个问题，等待用户回复
   • 添加Split Based On...组件，直接跳转到下一个Send & Wait for Reply组件（无需判断关键词，需收集任意文本回复）
   • 重复上述步骤，直到9个问题全部发送完成
3. 最后添加Send Message组件发送结束提示
4. 将Flow的触发方式设置为Incoming Message，关联到你的Twilio号码

注意事项

• 若当前UI与演示视频不同，优先参考Twilio Studio官方最新文档
• 使用Set Variables组件存储用户当前的问题序号，实现状态跟踪
• 所有回复可通过Flow的Widgets数据传递到你的Flask接口进行存储

后续优化建议 #

• 将user_session从内存字典改为数据库存储，避免服务重启丢失状态
• 添加错误处理逻辑（如用户长时间未回复时的提醒）
• 实现定时发送问题的功能（使用Twilio Scheduled Messages或第三方定时任务工具）

内容的提问来源于stack exchange，提问作者Evan Gertis