Note

讯飞API升级，旧的语音合成接口已不可用，新的实现方式直接扔github了。

前言

为了能让刚买的树莓派有效的利用起来（避免吃灰），今天分享一下如何用树莓派做天气闹钟。

环境及工具

树莓派3B+、IDE、XShell、FileZilla(FTP文件上传)、小音箱。

查询天气

准备

既然要做天气闹钟，那肯定先要知道今天的天气是什么，查询天气服务还是很大众的一种服务，很多网站都可以提供了查询天气的API接口，搜索一下。
做数据服务的网站就那么几家，简单浏览之后，选择了阿里云市场里的墨迹天气API（免费是重点，免费的可用1000次，最近有0元/10000次的活动）。云市场-免费版气象服务（cityid）-墨迹天气
这个API提供一个根据城市Id查询三天精简天气预报的接口，深得我心，买。

购买成功后，需要从阿里云控制台-产品与服务-API网关-调用API-已购API中查到请求Token。详细查询过程

点击操作中的详情可以看到Token, 之后需要点击授权按钮，为你要调取的接口生成一个授权码（阿里云API网关需要这个）。

接口详情

url: http://freecityid.market.alicloudapi.com/whapi/json/alicityweather/briefforecast3days
method: POST
body: {
    "cityid": "城市id"，
    “token”: "API详情页查询到的"
}

resp: {
  "code": 0,
  "data": {
    "city": {
      "cityId": 284609,
      "counname": "中国",
      "name": "东城区",
      "pname": "北京市"
    },
    "forecast": [
      {
        "conditionDay": "多云",
        "conditionIdDay": "1",
        "conditionIdNight": "31",
        "conditionNight": "多云",
        "predictDate": "2016-09-01",
        "tempDay": "27",
        "tempNight": "18",
        "updatetime": "2016-09-01 09:07:08",
        "windDirDay": "西北风",
        "windDirNight": "西北风",
        "windLevelDay": "3",
        "windLevelNight": "2"
      },
      ...省略两个...
    ]
  },
  "msg": "success",
  "rc": {
    "c": 0,
    "p": "success"
  }
}

拿到接口，接下来肯定就是写代码调用接口了，调用过程中涉及的部分问题，都写在代码注释里了（这里及后面的所有请求都是用的python的requests库，本身娱乐项目，也就没有生成requirement文件）。

Num2Word.py (将数字转为中文字符串)

#!/usr/bin/python3
# -*- coding:utf-8 -*-
import math

class Num2Word:
    words = {
        0: '零',
        1: '一',
        2: '二',
        3: '三',
        4: '四',
        5: '五',
        6: '六',
        7: '七',
        8: '八',
        9: '九',
        10: '十',
        100: '百',
        1000: '千',
        10000: '万',
    }

    # TODO 1024 -> 一千二十四  ==> 1024 -> 一千零二十四
    # TODO 1004 -> 一千四  ==> 1024 -> 一千零四
    # TODO 1024.2 小数点
    @staticmethod
    def to_word(num):
        if isinstance(num, int):
            pass
        elif isinstance(num, str):
            num = int(num)
        else:
            raise TypeError('num must be int or str')
        if num < 0:
            return '负' + Num2Word.to_word(-num)
        else:
            quotient = num
            remainder = 0
            s = ""
            ten_num = 0
            while quotient > 0:
                quotient = int(num / 10)
                remainder = num % 10
                if remainder > 0:
                    if ten_num > 0:
                        s = Num2Word.words[remainder] + Num2Word.words[int(math.pow(10, ten_num))] + s
                    else:
                        s = Num2Word.words[remainder] + s
                num = int(num / 10)
                ten_num += 1
            return s

MoJiWeather.py

#!/usr/bin/python3
# -*- coding:utf-8 -*-
import requests
import json
import logging
import sys
import os
from sys import path
from Num2Word import Num2Word
from VoicePlayer import VoicePlayer
from XunFeiTTS import XunFeiTTS

logging.basicConfig(
    level=logging.DEBUG,
    handlers=[logging.StreamHandler()],
    format='%(levelname)s:%(asctime)s:%(message)s'
)

class RespBody():
    # data = "{\"code\":0,\"data\":{\"city\":{\"cityId\":50,\"counname\":\"中国\",\"name\":\"闵行区\",\"pname\":\"上海市\",\"timezone\":\"8\"},\"forecast\":[{\"conditionDay\":\"多云\",\"conditionIdDay\":\"1\",\"conditionIdNight\":\"31\",\"conditionNight\":\"多云\",\"predictDate\":\"2018-10-17\",\"tempDay\":\"23\",\"tempNight\":\"14\",\"updatetime\":\"2018-10-17 22:09:00\",\"windDirDay\":\"北风\",\"windDirNight\":\"北风\",\"windLevelDay\":\"3-4\",\"windLevelNight\":\"3-4\"},{\"conditionDay\":\"多云\",\"conditionIdDay\":\"1\",\"conditionIdNight\":\"31\",\"conditionNight\":\"多云\",\"predictDate\":\"2018-10-18\",\"tempDay\":\"21\",\"tempNight\":\"12\",\"updatetime\":\"2018-10-17 22:09:00\",\"windDirDay\":\"北风\",\"windDirNight\":\"北风\",\"windLevelDay\":\"5-6\",\"windLevelNight\":\"3-4\"},{\"conditionDay\":\"多云\",\"conditionIdDay\":\"1\",\"conditionIdNight\":\"31\",\"conditionNight\":\"多云\",\"predictDate\":\"2018-10-19\",\"tempDay\":\"22\",\"tempNight\":\"13\",\"updatetime\":\"2018-10-17 22:09:00\",\"windDirDay\":\"东北风\",\"windDirNight\":\"东北风\",\"windLevelDay\":\"3-4\",\"windLevelNight\":\"3\"}]},\"msg\":\"success\",\"rc\":{\"c\":0,\"p\":\"success\"}}"
    def __init__(self, d) -> None:
        self.__dict__ = d

class Forecast():

    def __init__(self, d) -> None:
        self.prdict_date = d.predictDate # yyyy-MM-dd
        self.update_time = d.updatetime # yyyy-MM-dd HH:mm:ss
        self.condition_day = d.conditionDay # 多云
        self.condition_night = d.conditionNight # 多云
        self.temp_day = d.tempDay # 23
        self.temp_night = d.tempNight # 14
        self.wind_dir_day = d.windDirDay # 北风
        self.wind_dir_night = d.windDirNight # 北风
        self.wind_level_day = d.windLevelDay # 3-4
        self.wind_level_night = d.windLevelNight # 4

    def wind_level_to_word(self, wind_level):
        wind_level = str(wind_level)
        if not wind_level.__contains__('-'):
            return Num2Word.to_word(wind_level)
        return Num2Word.to_word(wind_level.split('-')[0]) + '至' + Num2Word.to_word(wind_level.split('-')[1])

    def to_chinese(self):
        # date转为文字
        month = self.prdict_date.split('-')[1]
        day = self.prdict_date.split('-')[2]
        date_word = Num2Word.to_word(month) + '月' + Num2Word.to_word(day) + '日'
        return "%s, 白天天气%s, 温度%s度, %s%s级, 夜间天气%s, 温度%s度, %s%s级" % \
              (date_word, self.condition_day, Num2Word.to_word(self.temp_day), self.wind_dir_day, self.wind_level_to_word(self.wind_level_day),
               self.condition_night, Num2Word.to_word(self.temp_night), self.wind_dir_night, self.wind_level_to_word(self.wind_level_night))

class MoJiWeather():
    def __init__(self) -> None:
        self.config = {
            "baseURL": "http://freecityid.market.alicloudapi.com",
            "forecastURL": "/whapi/json/alicityweather/briefforecast3days",
            "AppCode": "阿里云的授权码",
            "headers": {
                "Host":"freecityid.market.alicloudapi.com",
                "gateway_channel":"http",
                "Content-Type":"application/x-www-form-urlencoded; charset=utf-8",
                "Authorization":"APPCODE 阿里云的授权码"
            },
            "token": "墨迹天气token"
        }
        self.city_codes = {
            "BeiJing": "2",
            "ShangHaiMinHang": "50"  # 国内城市地区id见末尾附录
        }

    def fetch_forecast(self, cityId):
        req_body = {
            "cityId": str(cityId),
            "token": self.config["token"]
        }
        json_str = json.dumps(req_body)
        url = self.config["baseURL"] + self.config["forecastURL"]
        # print(url)
        # print(self.config["headers"])
        resp = requests.post(url=url, data=req_body, headers=self.config["headers"])
        resp_json = resp.content.decode('utf8')
        # print(resp_json)
        # print(resp.headers)
        logging.debug("[MoJiWeather.fetch_forecast] - status = %s" % resp.status_code)
        logging.debug("[MoJiWeather.fetch_forecast] - resp json = %s" % resp_json)

        resp_body = json.loads(resp_json, object_hook=RespBody)

        code = resp_body.code
        if code == 0:
            data = resp_body.data
            city = data.city
            province_name = city.pname
            city_name = city.name
            logging.info("[MoJiWeather.fetch_forecast] - %s, %s" % (province_name, city_name))

            three_days_forecast_list = data.forecast
            return three_days_forecast_list
        else:
            logging.info("[MoJiWeather.fetch_forecast] - Resp Not Success")
            return []

现在可以测试一下代码运行效果了，MojiWeather.fetch_forecast方法返回的是天气预报数组（字典数组）。为了方便测试，就直接在MoJiWeather中创建一个main方法来获取数据。

if __name__ == '__main__':
    mo_ji_weather = MoJiWeather()
     # 天气预报的数组
    forecast_list = mo_ji_weather.fetch_forecast(mo_ji_weather.city_codes["ShangHaiMinHang"])

    print(forecast_list)
    forecast_words = []
    for forecast in forecast_list:
        # 将dict转为Forecast对象
        f = Forecast(forecast)
        # 将天气预报转为中文文字
        forecast_words.append(f.to_chinese())
        print(f.to_chinese())
    # 将三个预报文本拼接成一个字符串
    s = ",".join(forecast_words)

运行结果

语音合成

到目前为止，我们已经能够拿到最近三天内的天气预报了，既然是做天气闹钟，那就要让程序会“说话”，也就是把文字转为语音（语音合成）。国内做语音合成，第一个想到的就是讯飞了，而且讯飞语音合成也有免费版的（每日500次限额，只有一个发音人可选），讯飞TTS介绍页也可以体验语音合成，经测试，讯飞的TTS还是挺清晰的。

然后按照下面的步骤去获取讯飞的API-KEY，完成准备工作。

剩下的工作就是按文档下代码

XunFeiTTS.py

#!/usr/bin/python3
# -*- coding:utf-8 -*-

import hashlib
import base64
import time
import json
import requests
import os
import logging

logging.basicConfig(
    level=logging.DEBUG,
    handlers=[logging.StreamHandler()],
    format='%(levelname)s:%(asctime)s:%(message)s'
)

class XunFeiTTS:
    def __init__(self) -> None:
        self.app_id = "讯飞App id" # 讯飞的应用id
        self.app_key = "讯飞TOKEN" # 讯飞的token
        self.tts_url = "http://api.xfyun.cn/v1/service/v1/tts"

    def __gen_sig(self, req_params_base64, time_now):
        """
        授权认证，生成认证信息
        :param req_params_base64: 请求参数的base64串
        :param time_now: 当前时间
        :return:
        """
        s = self.app_key + time_now + req_params_base64
        hl = hashlib.md5()
        hl.update(s.encode(encoding='utf8'))
        return hl.hexdigest()

    def __gen_req_header(self, time_now, req_params_base64, sig):
        """
        生成请求头
        :param time_now: 当前时间
        :param req_params_base64: 请求参数的base64串
        :param sig:
        :return:
        """
        header = {
            "X-Appid": self.app_id,
            "X-CurTime": time_now,
            "X-Param": req_params_base64,
            "X-CheckSum": sig,
            "Content-Type": "application/x-www-form-urlencoded; charset=utf-8"
        }
        return header

    def fetch_voice(self, text):
        """
        根据传入text生成语音
        :param text: 
        :return: 
        """
        req_params = {
            "auf": "audio/L16;rate=16000",
            "aue": "raw", # 返回的语音格式 raw为wav格式语音, lame为MP3格式语音
            "voice_name": "xiaoyan",
            "speed": "50",
            "volume": "50",
            "pitch": "50",
            "engine_type": "intp65",
            "text_type": "text",
            "text": text + " 噻"
        }

        time_now = str(time.time()).split('.')[0]
        req_params_json = json.dumps(req_params)
        req_params_base64 = str(base64.b64encode(req_params_json.encode('ascii')).decode('ascii'))
        header = self.__gen_req_header(time_now, req_params_base64, self.__gen_sig(req_params_base64, time_now))

        resp = requests.post(url=self.tts_url, data=req_params, headers=header)
        content_type = resp.headers['Content-type']

        # 请求成功时， contentType为audio.mpeg, 失败时，contentType为text/plain, 返回异常信息
        if content_type == 'audio/mpeg':
            # 将语音写入文件voice.wav
            f = open('voice.wav', 'wb')
            f.write(resp.content)
            f.close()

            logging.info("[XunFeiTTS.fetch_voice] - Fetch Voice Success! Save As %s" % f.name)
        else:
            resp_json = resp.content.decode('utf-8')
            logging.info("[XunFeiTTs.fetch_voice] - %s" % resp_json)
            resp_dict = json.loads(resp_json)
            logging.error("[XunFeiTTS.fetch_voice] - ErrCode = %s, Desc = %s" % (resp_dict['code'], resp_dict['desc']))

现在我们需要重新修改MoJiWeather中的main方法，调用讯飞TTS将天气预报的字符串转变为语音。
接下来就是使用pyaudio库来播放天气预报语音（python上有很多库可以播放音频，试了几个之后，感觉还是pyaudio更适合这个例子）。

如果在树莓派上使用pip安装pyaudio时出现Pyaudio installation error - 'command 'gcc' failed with exit status 1'错误，请在树莓派上执行下面的命令

sudo apt-get install python-dev
sudo apt-get install portaudio19-dev
sudo apt-get install libportaudio0 libportaudio2 libportaudiocpp0 portaudio19-dev

pip3 install pyaudio

播放wav语音的工具类

VoicePlayer.py

#!/usr/bin/python3
# -*- coding:utf-8 -*-
import pyaudio
import wave
import os
import logging

logging.basicConfig(
    level=logging.DEBUG,
    handlers=[logging.StreamHandler()],
    format='%(levelname)s:%(asctime)s:%(message)s'
)

class VoicePlayer:

    def __init__(self) -> None:
        self.chunk = 1024

    def play(self, filename):
        logging.debug("[VoicePlayer.play] - load file %s" % filename)
        chunk = 1024
        wf = wave.open('voice.wav', 'rb')
        p = pyaudio.PyAudio()

        stream = p.open(
            format=p.get_format_from_width(wf.getsampwidth()),
            channels=wf.getnchannels(),
            rate=wf.getframerate(),
            output=True)
        data = wf.readframes(chunk)

        while data != '':
            stream.write(data)
            data = wf.readframes(chunk)
            if data == b'':
                break

        stream.close()
        p.terminate()
        logging.debug("[VoicePlayer.play] - Voice Play Finish")

整合

到这里，就可以把上面的代码整合到一起了，这里要注意一下Python的包引入问题，在MoJiWeather.py中会引入Num2Word.py,VoicePlayer.py,XunFeiTTS.py，在运行程序时请保证这几个文件都在同一目录下

MoJiWeather.py

if __name__ == '__main__':
    mo_ji_weather = MoJiWeather()
    forecast_list = mo_ji_weather.fetch_forecast(mo_ji_weather.city_codes["ShangHaiMinHang"])

    print(forecast_list)

    xun_fei_tts = XunFeiTTS()
    s = ""
    forecast_words = []
    for forecast in forecast_list:
        f = Forecast(forecast)
        forecast_words.append(f.to_chinese())
        print(f.to_chinese())
    s = ",".join(forecast_words)
    logging.debug("[MojiWeather.main] - %s" % s)
    xun_fei_tts.fetch_voice(s)

    voice_player = VoicePlayer()
    voice_player.play('voice.wav')

部署

程序编写完成后，将代码通过FileZilla上传工具把代码上传到树莓派。现在距离天气闹钟只差最后一步（闹钟）。借助linux的crond定时任务就可以很容易的实现闹钟这一功能。在树莓派上执行crontab -e就可以编辑crond任务了，第一次打开时会提示让你选择一个编辑器，按个人喜好选择即可。