grpc - 修改后的 Google pushtotalk.py 在缺少属性“DeviceRequestHandler”时崩溃
问题描述
我在带有最新 Raspbian 的 Raspberry Pi 3b+ 上运行。我创建了一个 Python 2.7 虚拟环境来测试一些东西。我使用以下说明安装了 Google Assistant Api： https ://developers.google.com/assistant/sdk/guides/library/python/embed/install-sample - 除了音频，我使用的是 Respeaker 4-mic hat因为我已经可以正常工作了。
当我运行 pushtotalk.py 的示例代码时，它工作正常，除了 CTRL-C 不再起作用（必须关闭终端窗口才能杀死它）。
我做了一些小的（或者我认为是小的）更改，当我运行代码时，我得到了一个奇怪的错误。
我的代码版本：
# Original pushtotalk.py file Copyright (C) 2017 Google Inc.
# modified by @captstephan for T3 project
#
# my imports for the mechanical functions, motor drivers, etc.
from Raspi_PWM_Servo_Driver import PWM
from voice_engine.source import Source
from voice_engine.channel_picker import ChannelPicker
from voice_engine.kws import KWS
from voice_engine.doa_respeaker_4mic_array import DOA
from pixels import pixels
# imports from the original Google pushtotalk.py file
import concurrent.futures
import json
import logging
import os
import os.path
import pathlib2 as pathlib
import sys
import time
import uuid
import click
import grpc
import google.auth.transport.grpc
import google.auth.transport.requests
import google.oauth2.credentials
from google.assistant.embedded.v1alpha2 import (
embedded_assistant_pb2,
embedded_assistant_pb2_grpc
)
from tenacity import retry, stop_after_attempt, retry_if_exception
import assistant_helpers
import audio_helpers
import browser_helpers
import device_helpers
# set up Google Assistant variables (from pushtotalk.py)
ASSISTANT_API_ENDPOINT = 'embeddedassistant.googleapis.com'
END_OF_UTTERANCE = embedded_assistant_pb2.AssistResponse.END_OF_UTTERANCE
DIALOG_FOLLOW_ON = embedded_assistant_pb2.DialogStateOut.DIALOG_FOLLOW_ON
CLOSE_MICROPHONE = embedded_assistant_pb2.DialogStateOut.CLOSE_MICROPHONE
PLAYING = embedded_assistant_pb2.ScreenOutConfig.PLAYING
DEFAULT_GRPC_DEADLINE = 60 * 3 + 5
# set up items for motor hats
# Initialise the PWM device using the default address
pwm = PWM(0x6F)
# set max and min, servo0=Horiz, servo1=vert
servoMin0 = 155 # Min pulse length out of 4096
servoMid0 = 370
servoMax0 = 585 # Max pulse length out of 4096
servoMin1 = 410 # Min pulse length out of 4096
servoMid1 = 530
servoMax1 = 650 # Max pulse length out of 4096
pwm.setPWMFreq(60) # Set frequency to 60 Hz
# class assignment from pushtotalk.py file:
class SampleAssistant(object):
"""Sample Assistant that supports conversations and device actions.
Args:
device_model_id: identifier of the device model.
device_id: identifier of the registered device instance.
conversation_stream(ConversationStream): audio stream
for recording query and playing back assistant answer.
channel: authorized gRPC channel for connection to the
Google Assistant API.
deadline_sec: gRPC deadline in seconds for Google Assistant API call.
device_handler: callback for device actions.
"""
def __init__(self, language_code, device_model_id, device_id,
conversation_stream, display,
channel, deadline_sec, device_handler):
self.language_code = language_code
self.device_model_id = device_model_id
self.device_id = device_id
self.conversation_stream = conversation_stream
self.display = display
# Opaque blob provided in AssistResponse that,
# when provided in a follow-up AssistRequest,
# gives the Assistant a context marker within the current state
# of the multi-Assist()-RPC "conversation".
# This value, along with MicrophoneMode, supports a more natural
# "conversation" with the Assistant.
self.conversation_state = None
# Force reset of first conversation.
self.is_new_conversation = True
# Create Google Assistant API gRPC client.
self.assistant = embedded_assistant_pb2_grpc.EmbeddedAssistantStub(
channel
)
self.deadline = deadline_sec
self.device_handler = device_handler
def __enter__(self):
return self
def __exit__(self, etype, e, traceback):
if e:
return False
self.conversation_stream.close()
def is_grpc_error_unavailable(e):
is_grpc_error = isinstance(e, grpc.RpcError)
if is_grpc_error and (e.code() == grpc.StatusCode.UNAVAILABLE):
logging.error('grpc unavailable error: %s', e)
return True
return False
@retry(reraise=True, stop=stop_after_attempt(3),
retry=retry_if_exception(is_grpc_error_unavailable))
def assist(self):
"""Send a voice request to the Assistant and playback the response.
Returns: True if conversation should continue.
"""
continue_conversation = False
device_actions_futures = []
self.conversation_stream.start_recording()
logging.info('Recording audio request.')
def iter_log_assist_requests():
for c in self.gen_assist_requests():
assistant_helpers.log_assist_request_without_audio(c)
yield c
logging.debug('Reached end of AssistRequest iteration.')
# This generator yields AssistResponse proto messages
# received from the gRPC Google Assistant API.
for resp in self.assistant.Assist(iter_log_assist_requests(),
self.deadline):
assistant_helpers.log_assist_response_without_audio(resp)
if resp.event_type == END_OF_UTTERANCE:
logging.info('End of audio request detected.')
logging.info('Stopping recording.')
self.conversation_stream.stop_recording()
if resp.speech_results:
logging.info('Transcript of user request: "%s".',
' '.join(r.transcript
for r in resp.speech_results))
if len(resp.audio_out.audio_data) > 0:
if not self.conversation_stream.playing:
self.conversation_stream.stop_recording()
self.conversation_stream.start_playback()
logging.info('Playing assistant response.')
self.conversation_stream.write(resp.audio_out.audio_data)
if resp.dialog_state_out.conversation_state:
conversation_state = resp.dialog_state_out.conversation_state
logging.debug('Updating conversation state.')
self.conversation_state = conversation_state
if resp.dialog_state_out.volume_percentage != 0:
volume_percentage = resp.dialog_state_out.volume_percentage
logging.info('Setting volume to %s%%', volume_percentage)
self.conversation_stream.volume_percentage = volume_percentage
if resp.dialog_state_out.microphone_mode == DIALOG_FOLLOW_ON:
continue_conversation = True
logging.info('Expecting follow-on query from user.')
elif resp.dialog_state_out.microphone_mode == CLOSE_MICROPHONE:
continue_conversation = False
if resp.device_action.device_request_json:
device_request = json.loads(
resp.device_action.device_request_json
)
fs = self.device_handler(device_request)
if fs:
device_actions_futures.extend(fs)
if self.display and resp.screen_out.data:
system_browser = browser_helpers.system_browser
system_browser.display(resp.screen_out.data)
if len(device_actions_futures):
logging.info('Waiting for device executions to complete.')
concurrent.futures.wait(device_actions_futures)
logging.info('Finished playing assistant response.')
self.conversation_stream.stop_playback()
return continue_conversation
def gen_assist_requests(self):
"""Yields: AssistRequest messages to send to the API."""
config = embedded_assistant_pb2.AssistConfig(
audio_in_config=embedded_assistant_pb2.AudioInConfig(
encoding='LINEAR16',
sample_rate_hertz=self.conversation_stream.sample_rate,
),
audio_out_config=embedded_assistant_pb2.AudioOutConfig(
encoding='LINEAR16',
sample_rate_hertz=self.conversation_stream.sample_rate,
volume_percentage=self.conversation_stream.volume_percentage,
),
dialog_state_in=embedded_assistant_pb2.DialogStateIn(
language_code=self.language_code,
conversation_state=self.conversation_state,
is_new_conversation=self.is_new_conversation,
),
device_config=embedded_assistant_pb2.DeviceConfig(
device_id=self.device_id,
device_model_id=self.device_model_id,
)
)
if self.display:
config.screen_out_config.screen_mode = PLAYING
# Continue current conversation with later requests.
self.is_new_conversation = False
# The first AssistRequest must contain the AssistConfig
# and no audio data.
yield embedded_assistant_pb2.AssistRequest(config=config)
for data in self.conversation_stream:
# Subsequent requests need audio data, but not config.
yield embedded_assistant_pb2.AssistRequest(audio_in=data)
@click.command()
@click.option('--api-endpoint', default=ASSISTANT_API_ENDPOINT,
metavar='<api endpoint>', show_default=True,
help='Address of Google Assistant API service.')
@click.option('--credentials',
metavar='<credentials>', show_default=True,
default=os.path.join(click.get_app_dir('google-oauthlib-tool'),
'credentials.json'),
help='Path to read OAuth2 credentials.')
@click.option('--project-id',
metavar='<project id>',
help=('Google Developer Project ID used for registration '
'if --device-id is not specified'))
@click.option('--device-model-id',
metavar='<device model id>',
help=(('Unique device model identifier, '
'if not specifed, it is read from --device-config')))
@click.option('--device-id',
metavar='<device id>',
help=(('Unique registered device instance identifier, '
'if not specified, it is read from --device-config, '
'if no device_config found: a new device is registered '
'using a unique id and a new device config is saved')))
@click.option('--device-config', show_default=True,
metavar='<device config>',
default=os.path.join(
click.get_app_dir('googlesamples-assistant'),
'device_config.json'),
help='Path to save and restore the device configuration')
@click.option('--lang', show_default=True,
metavar='<language code>',
default='en-US',
help='Language code of the Assistant')
@click.option('--display', is_flag=True, default=False,
help='Enable visual display of Assistant responses in HTML.')
@click.option('--verbose', '-v', is_flag=True, default=False,
help='Verbose logging.')
@click.option('--input-audio-file', '-i',
metavar='<input file>',
help='Path to input audio file. '
'If missing, uses audio capture')
@click.option('--output-audio-file', '-o',
metavar='<output file>',
help='Path to output audio file. '
'If missing, uses audio playback')
@click.option('--audio-sample-rate',
default=audio_helpers.DEFAULT_AUDIO_SAMPLE_RATE,
metavar='<audio sample rate>', show_default=True,
help='Audio sample rate in hertz.')
@click.option('--audio-sample-width',
default=audio_helpers.DEFAULT_AUDIO_SAMPLE_WIDTH,
metavar='<audio sample width>', show_default=True,
help='Audio sample width in bytes.')
@click.option('--audio-iter-size',
default=audio_helpers.DEFAULT_AUDIO_ITER_SIZE,
metavar='<audio iter size>', show_default=True,
help='Size of each read during audio stream iteration in bytes.')
@click.option('--audio-block-size',
default=audio_helpers.DEFAULT_AUDIO_DEVICE_BLOCK_SIZE,
metavar='<audio block size>', show_default=True,
help=('Block size in bytes for each audio device '
'read and write operation.'))
@click.option('--audio-flush-size',
default=audio_helpers.DEFAULT_AUDIO_DEVICE_FLUSH_SIZE,
metavar='<audio flush size>', show_default=True,
help=('Size of silence data in bytes written '
'during flush operation'))
@click.option('--grpc-deadline', default=DEFAULT_GRPC_DEADLINE,
metavar='<grpc deadline>', show_default=True,
help='gRPC deadline in seconds')
@click.option('--once', default=False, is_flag=True,
help='Force termination after a single conversation.')
def main(api_endpoint, credentials, project_id,
device_model_id, device_id, device_config,
lang, display, verbose,
input_audio_file, output_audio_file,
audio_sample_rate, audio_sample_width,
audio_iter_size, audio_block_size, audio_flush_size,
grpc_deadline, once, *args, **kwargs):
# Inserted the following code to set up the snowboy keyword activation using "Hey T3"
src = Source(rate=16000, channels=4, frames_size=320)
ch1 = ChannelPicker(channels=4, pick=1)
kws = KWS()
doa = DOA(rate=16000)
src.link(ch1)
ch1.link(kws)
src.link(doa)
pixels.listen()
pwm.setPWM(0, 0, 370)
pwm.setPWM(1, 0, 640)
# When snowboy detects the custom keyword, set the camera position to near direction of voice
def on_detected(keyword):
position = doa.get_direction()
pixels.wakeup(position)
print('detected {} at direction {}'.format(keyword, position))
if position >= 30 and position <= 180:
pwm.setPWM(0, 0, 175)
pwm.setPWM(1, 0, 500)
elif position > 180 and position <= 330:
pwm.setPWM(0, 0, 560)
pwm.setPWM(1, 0, 500)
elif position > 330 or position < 30:
pwm.setPWM(0, 0, 370)
pwm.setPWM(1, 0, 6200)
else:
pwm.setPWM(0, 0, 370)
pwm.setPWM(1, 0, 640)
# end of stuff I inserted
# Setup logging.
logging.basicConfig(level=logging.DEBUG if verbose else logging.INFO)
# Load OAuth 2.0 credentials.
try:
with open(credentials, 'r') as f:
credentials = google.oauth2.credentials.Credentials(token=None,
**json.load(f))
http_request = google.auth.transport.requests.Request()
credentials.refresh(http_request)
except Exception as e:
logging.error('Error loading credentials: %s', e)
logging.error('Run google-oauthlib-tool to initialize '
'new OAuth 2.0 credentials.')
sys.exit(-1)
# Create an authorized gRPC channel.
grpc_channel = google.auth.transport.grpc.secure_authorized_channel(
credentials, http_request, api_endpoint)
logging.info('Connecting to %s', api_endpoint)
# Configure audio source and sink.
audio_device = None
if input_audio_file:
audio_source = audio_helpers.WaveSource(
open(input_audio_file, 'rb'),
sample_rate=audio_sample_rate,
sample_width=audio_sample_width
)
else:
audio_source = audio_device = (
audio_device or audio_helpers.SoundDeviceStream(
sample_rate=audio_sample_rate,
sample_width=audio_sample_width,
block_size=audio_block_size,
flush_size=audio_flush_size
)
)
if output_audio_file:
audio_sink = audio_helpers.WaveSink(
open(output_audio_file, 'wb'),
sample_rate=audio_sample_rate,
sample_width=audio_sample_width
)
else:
audio_sink = audio_device = (
audio_device or audio_helpers.SoundDeviceStream(
sample_rate=audio_sample_rate,
sample_width=audio_sample_width,
block_size=audio_block_size,
flush_size=audio_flush_size
)
)
# Create conversation stream with the given audio source and sink.
conversation_stream = audio_helpers.ConversationStream(
source=audio_source,
sink=audio_sink,
iter_size=audio_iter_size,
sample_width=audio_sample_width,
)
if not device_id or not device_model_id:
try:
with open(device_config) as f:
device = json.load(f)
device_id = device['id']
device_model_id = device['model_id']
logging.info("Using device model %s and device id %s",
device_model_id,
device_id)
except Exception as e:
logging.warning('Device config not found: %s' % e)
logging.info('Registering device')
if not device_model_id:
logging.error('Option --device-model-id required '
'when registering a device instance.')
sys.exit(-1)
if not project_id:
logging.error('Option --project-id required '
'when registering a device instance.')
sys.exit(-1)
device_base_url = (
'https://%s/v1alpha2/projects/%s/devices' % (api_endpoint,
project_id)
)
device_id = str(uuid.uuid1())
payload = {
'id': device_id,
'model_id': device_model_id,
'client_type': 'SDK_SERVICE'
}
session = google.auth.transport.requests.AuthorizedSession(
credentials
)
r = session.post(device_base_url, data=json.dumps(payload))
if r.status_code != 200:
logging.error('Failed to register device: %s', r.text)
sys.exit(-1)
logging.info('Device registered: %s', device_id)
pathlib.Path(os.path.dirname(device_config)).mkdir(exist_ok=True)
with open(device_config, 'w') as f:
json.dump(payload, f)
device_handler = device_helpers.DeviceRequestHandler(device_id)
@device_handler.command('action.devices.commands.OnOff')
def onoff(on):
if on:
logging.info('Turning device on')
else:
logging.info('Turning device off')
@device_handler.command('com.example.commands.BlinkLight')
def blink(speed, number):
logging.info('Blinking device %s times.' % number)
delay = 1
if speed == "SLOWLY":
delay = 2
elif speed == "QUICKLY":
delay = 0.5
for i in range(int(number)):
logging.info('Device is blinking.')
time.sleep(delay)
with SampleAssistant(lang, device_model_id, device_id,
conversation_stream, display,
grpc_channel, grpc_deadline,
device_handler) as assistant:
# If file arguments are supplied:
# exit after the first turn of the conversation.
if input_audio_file or output_audio_file:
assistant.assist()
return
# changed the wait for keypress to a wait for keyword using the snowboy module
# If no file arguments supplied:
# keep recording voice requests using the microphone
# and playing back assistant response using the speaker.
# When the once flag is set, don't wait for a trigger. Otherwise, wait.
wait_for_user_trigger = not once
while True:
if wait_for_user_trigger:
#click.pause(info='Press Enter to send a new request...
kws.set_callback(on_detected)
continue_conversation = assistant.assist()
# wait for user trigger if there is no follow-up turn in
# the conversation.
wait_for_user_trigger = not continue_conversation
# If we only want one conversation, break.
if once and (not continue_conversation):
break
if __name__ == '__main__':
main()
我收到以下错误：
回溯（最后一次调用）：文件“/usr/lib/python2.7/runpy.py”，第 174 行，在 _run_module_as_main “ main ”、fname、loader、pkg_name）文件“/usr/lib/python2.7/ runpy.py”，第 72 行，在 run_globals 文件中的 _run_code 执行代码“/home/pi/T3google.py”，第 501 行，在 main() 文件中“/home/pi/env/local/lib/python2.7/ site-packages/click/core.py”，第 722 行，通话中 return self.main(*args, **kwargs) File "/home/pi/env/local/lib/python2.7/site-packages/click/core.py", line 697, in main rv = self.invoke (ctx) 文件“/home/pi/env/local/lib/python2.7/site-packages/click/core.py”，第 895 行，调用 return ctx.invoke(self.callback, **ctx.params ）文件“/home/pi/env/local/lib/python2.7/site-packages/click/core.py”，第 535 行，在调用返回回调（*args，**kwargs）文件“/home/pi /T3google.py"，第 425 行，在主 device_handler = device_helpers.DeviceRequestHandler(device_id) AttributeError: 'module' 对象没有属性 'DeviceRequestHandler'
我还是 python 的新手，但是来自 Google 的示例代码可以工作，而我插入的代码片段可以在另一个文件中独立工作。
我检查了 device_helpers.py 和 DeviceRequestHandler 是一个将 device_id 作为输入的类。如果我在 pushtotalk.py 中调用它，它可以正常工作，但在修改后的代码中则不行。
有什么想法吗？
在此先感谢，斯蒂芬
解决方案
我因其他原因关闭，第二天回来并重新启动，问题就消失了。可能需要重新启动才能完成安装，但我错过了。该部分现在工作正常，无法完全按预期工作。
推荐阅读
- html - 使用 Angular 时，HTML“值”字段未预填充表单
- django - 如何将值从 Django 模板传递到视图
- laravel - 如何在多对多关系laravel中使用条件？
- c# - 使用 SOAP，如何在一个请求中传递多个 ID？
- c++ - TableView 不会将 QSqlQueryModel 加载到其中
- php - 如何将 Sengrid API 与 CI 集成为一个库
- python - Python 中的 KMeans - 转换变量后保留样本权重
- javascript - Inversify 容器在运行时未定义
- dart - 如果填充宽度，如何使行进入新行
- python - 如何在数据透视表中添加计算比率？