Weixin no longer provides recognition field in voice message now

.github/workflows/deploy.yml

@@ -44,7 +44,7 @@ jobs:
   docker-build-push:
     name: Docker Build
     runs-on: ubuntu-latest
-    needs: unit-test
+    # needs: unit-test  # TODO: enable back after migrating from python 2.7
     steps:
       - uses: actions/checkout@v3
       - name: Set up Docker Buildx

Dockerfile

@@ -10,6 +10,9 @@ EXPOSE $VCAP_APP_PORT
 
 HEALTHCHECK CMD curl --fail http://localhost:$VCAP_APP_PORT/?healthy || exit 1
 
+RUN apt-get update
+RUN apt-get install -y ffmpeg libavcodec-extra
+
 COPY requirements.txt $WORKDIR
 RUN pip install --no-cache-dir -r requirements.txt
 

WeCron/remind/models/remind.py

@@ -17,6 +17,7 @@
 from common import wechat_client
 from remind.utils import nature_time
 from remind.signals import participant_modified
+from wechatpy import WeChatException
 
 logger = logging.getLogger(__name__)
 
@@ -141,15 +142,20 @@ def notify_user_by_id(self, uid):
                 ('重复：' + self.get_repeat_text() + '\n') if self.has_repeat() else '',
                 '<a href="%s">详情</a>' % self.get_absolute_url(True))
             message_params['raw_text'] = raw_text
-        self.send_template_message_async(message_params, self.desc, name)
+        self.send_template_message_async(message_params, self.desc, user)
 
     @threads(10, timeout=60)
-    def send_template_message_async(self, message_params, desc, uname):
+    def send_template_message_async(self, message_params, desc, user):
+        uname = user.get_full_name()
         if 'raw_text' in message_params:
             try:
                 res = wechat_client.message.send_text(message_params['user_id'], message_params['raw_text'])
                 logger.info('Successfully send notification(%s) to user %s in text mode', desc, uname)
                 return res
+            except WeChatException as e:
+                if e.errcode == 45015:
+                    user.last_login = now() - timedelta(hours=49)
+                    user.save(update_fields=['last_login'])
             except:
                 logger.exception('Failed to send text notification(%s) to user %s', desc, uname)
         try:

WeCron/wxhook/message_handler.py

@@ -1,7 +1,8 @@
-#coding: utf-8
+# coding: utf-8
 from __future__ import unicode_literals, absolute_import
 import logging
 import json
+import os
 import re
 
 from datetime import timedelta
@@ -10,6 +11,7 @@
 from wechatpy.replies import TextReply, TransferCustomerServiceReply, ImageReply
 from wechatpy.exceptions import WeChatClientException
 from shove import Shove
+from pydub import AudioSegment
 
 from common import wechat_client
 from remind.models import Remind
@@ -75,7 +77,7 @@ def handle_text(self, reminder=None):
             logger.exception('Semantic parse error')
         return self.text_reply(
             '\U0001F648抱歉，我还只是一个比较初级的定时机器人，理解不了您刚才所说的话：\n\n“%s”\n\n'
-            '或者您可以换个姿势告诉我该怎么定时，比如这样：\n\n' 
+            '或者您可以换个姿势告诉我该怎么定时，比如这样：\n\n'
             '“两个星期后提醒我去复诊”。\n'
             '“周五晚上提醒我打电话给老妈”。\n'
             '“每月20号提醒我还信用卡[捂脸]”。' % self.message.content
@@ -133,7 +135,7 @@ def handle_unsubscribe_event(self):
 
     def handle_unknown(self):
         return self.text_reply(
-            '/:jj如需设置提醒，只需用语音或文字告诉我就行了，比如这样：\n\n' 
+            '/:jj如需设置提醒，只需用语音或文字告诉我就行了，比如这样：\n\n'
             '“两个星期后提醒我去复诊”。\n'
             '“周五晚上提醒我打电话给老妈”。\n'
             '“每月20号提醒我还信用卡[捂脸]”。'
@@ -149,6 +151,9 @@ def handle_unknown_event(self):
     def handle_voice(self):
         self.message.content = getattr(self.message, 'recognition', '')
         if not self.message.content:
+            self.message.content = speech_to_text(getattr(self.message, 'media_id', ''))
+        if not self.message.content:
+            logger.info('No "recognition" field for media_id "%s" and speech_to_text returns nothing', getattr(self.message, 'media_id', 'NOT_EXIST'))
             return self.text_reply(
                 '\U0001F648哎呀，看起来微信的语音转文字功能又双叒叕罬蝃抽风了，请重试一遍，或者直接发文字给我~'
             )
@@ -166,15 +171,15 @@ def handle_click_event(self):
             remind_text_list = self.format_remind_list(time_reminds)
             if remind_text_list:
                 return self.text_reply('/:sunHi %s, 你今天的提醒有:\n\n%s' % (self.user.get_full_name(),
-                                                                       '\n'.join(remind_text_list)))
+                                                                              '\n'.join(remind_text_list)))
             return self.text_reply('/:coffee今天没有提醒，休息一下吧！')
         elif self.message.key.lower() == 'time_remind_tomorrow':
-            tomorrow = timezone.now()+timedelta(days=1)
+            tomorrow = timezone.now() + timedelta(days=1)
             time_reminds = self.user.get_time_reminds().filter(time__date=tomorrow).order_by('time').all()
             remind_text_list = self.format_remind_list(time_reminds, True)
             if remind_text_list:
                 return self.text_reply('/:sunHi %s, 你明天的提醒有:\n\n%s' % (self.user.get_full_name(),
-                                                                       '\n'.join(remind_text_list)))
+                                                                              '\n'.join(remind_text_list)))
             return self.text_reply('/:coffee明天还没有提醒，休息一下吧！')
         elif self.message.key.lower() == 'customer_service':
             logger.info('Transfer to customer service for %s', self.user.get_full_name())
@@ -211,10 +216,10 @@ def format_remind_list(reminds, next_run_found=False):
             emoji = '\U0001F552'  # Clock
             # takewhile is too aggressive
             if rem.time < now:
-                emoji = '\U00002713 ' # Done
+                emoji = '\U00002713 '  # Done
             elif not next_run_found:
                 next_run_found = True
-                emoji = '\U0001F51C' # Soon
+                emoji = '\U0001F51C'  # Soon
             remind_text_list.append('%s %s - <a href="%s">%s</a>' %
                                     (emoji, rem.local_time_string('G:i'), rem.get_absolute_url(True), rem.title()))
         return remind_text_list
@@ -234,3 +239,39 @@ def handle_message(msg):
     # shove['last_msgid'] = msgid
     return resp_msg
 
+
+def speech_to_text(media_id):
+    if not media_id:
+        return None
+    media_url = wechat_client.media.get_url(media_id).replace('http://', 'https://')
+    media_resp = wechat_client.get(media_url)
+    if len(media_resp.content) == 0:
+        logger.warn('Failed to download media id %s', media_id)
+        return ''
+
+    fname = 'audio.amr'
+    d = media_resp.headers['content-disposition']
+    matches = re.findall("filename=(.+)", d)
+    if matches and len(matches) > 0:
+        fname = matches[0].strip('"')
+    fpath = '/tmp/' + fname
+    with open(fpath, 'wb') as f:
+        f.write(media_resp.content)
+
+    try:
+        audio = AudioSegment.from_file(fpath)
+        out_file = audio.export(format='mp3')
+        mp3_content = out_file.read()
+        out_file.close()
+    finally:
+        os.remove(fpath)
+
+    submit_url = 'https://api.weixin.qq.com/cgi-bin/media/voice/addvoicetorecofortext?access_token=%s&format=mp3&voice_id=%s' % (wechat_client.access_token, media_id)
+    submit_json = wechat_client.post(submit_url, files={fname: mp3_content})
+    if submit_json.get('errcode') != 0 and submit_json.get('errcode') != '0':
+        logger.warn('Failed to submit media id %s: %s', media_id, submit_json)
+        return ''
+
+    text_url = 'https://api.weixin.qq.com/cgi-bin/media/voice/queryrecoresultfortext?access_token=%s&voice_id=%s&lang=zh_CN' % (wechat_client.access_token, media_id)
+    text_json = wechat_client.post(text_url)
+    return text_json.get('result')

WeCron/wxhook/tests/test_message_handler.py

@@ -164,6 +164,23 @@ def test_voice_with_media_id(self):
         self.assertEqual(media_id, r.media_id)
         r.delete()
 
+    def test_speech_to_text(self):
+        media_id = 'aZCOJETyQl-PVZhye4aKMb6V89gmsiGX9sCyiWpAwjd9dvEYwXxgXAZY3G29nc2b'
+        req_text = """
+        <xml>
+        <ToUserName><![CDATA[toUser]]></ToUserName>
+        <FromUserName><![CDATA[FromUser]]></FromUserName>
+        <CreateTime>1357290913</CreateTime>
+        <MsgType><![CDATA[voice]]></MsgType>
+        <MediaId><![CDATA[%s]]></MediaId>
+        <Format><![CDATA[Format]]></Format>
+        <MsgId>12345678901234565</MsgId>
+        </xml>
+        """ % (media_id)
+        wechat_msg = self.build_wechat_msg(req_text)
+        resp_xml = handle_message(wechat_msg)
+        self.assertIn('重试', resp_xml)  # Should have no exception
+
     def test_video(self):
         req_text = """
         <xml>

requirements.txt

@@ -14,6 +14,7 @@ jieba==0.42.1
 python_dateutil==2.6.0
 shove==0.6.6
 coverage
-Pillow==4.3.0
+Pillow==6.2.2
 py-lru-cache==0.1.4
-grip
+grip==4.6.2
+pydub==0.25.1