def aksk_connect(ak, sk, url, params, http_method, config=None):
"""
根据url，返回json
:param ak: ak
:param sk: sk
:param url: 完整请求url
:param params: 请求参数， dict
:param http_method: 请求方法，'POST' or 'GET', 其他会报错
:param config: SisConfig(), 配置超时和代理
:return: http返回结果转化为json
"""
sis_config = config
if sis_config is None:
sis_config = SisConfig()
if not isinstance(sis_config, SisConfig):
error_msg = 'the param \'config\' in aksk_connect must be SisConfig class'
logger.error(error_msg)
raise ClientException(error_msg)
signed_headers = _get_signed_headers(ak, sk, url, params, http_method)
time_out = (sis_config.get_connect_timeout(),
sis_config.get_read_timeout())
resp = http_utils.http_connect(url, signed_headers, params, http_method,
time_out, sis_config.get_proxy())
json_result = http_utils.parse_resp(resp)
if resp is not None:
resp.close()
return json_result
def _on_message(ws, message):
result = json.loads(message)
_check_result(result)
result_type = result['resp_type']
trace_id = result['trace_id']
if result_type == 'START':
self._callback.on_start(trace_id)
elif result_type == 'EVENT':
event = result['event']
if event == 'EXCEEDED_AUDIO':
logger.warn(
'the duration of the audio is too long, the rest won\'t be recognized'
)
self._status = 'end'
elif event == 'EXCEEDED_SILENCE':
logger.error(
'silent time is too long, the audio won\'t be recognized'
)
self._callback.on_error(
'silent time is too long, the audio won\'t be recognized'
)
self._status = 'error'
elif event == 'VOICE_END':
logger.warn(
'detect voice end, the rest won\'t be recognized')
self._status = 'end'
elif result_type == 'RESULT':
self._callback.on_response(result)
elif result_type == 'END':
self._status = 'end'
self._callback.on_end(trace_id)
else:
logger.error('%s don\'t belong to any type' % result_type)
def _generate_request_proxy(proxy):
if proxy is None:
return proxy
if not isinstance(proxy, list) or (not len(proxy) == 2
and not len(proxy) == 4):
logger.error(
'Proxy must be list, the format is [host, port] or [host, port, username, password]'
)
raise ClientException(
'Proxy must be list, the format is [host, port] or [host, port, username, password]'
)
proxy_str = str(proxy[0]) + ':' + str(proxy[1])
if len(proxy) == 2:
proxy = {
'http': 'http://' + proxy_str,
'https': 'https://' + proxy_str
}
else:
proxy = {
'http':
'http://' + str(proxy[2]) + ':' + str(proxy[3]) + '@' + proxy_str,
'https':
'https://' + str(proxy[2]) + ':' + str(proxy[3]) + '@' + proxy_str
}
return proxy
def encode_file(file_path):
if not os.path.exists(file_path):
logger.error('The Path %s doesn\'t exist' % file_path)
raise ClientException('The Path %s doesn\'t exist' % file_path)
with open(file_path, 'rb') as f:
data = f.read()
base64_data = str(base64.b64encode(data), 'utf-8')
return base64_data
def get_token(user_name,
password,
domain_name,
region,
url=None,
config=SisConfig()):
"""
获取token
:param user_name: 用户名
:param password: 密码
:param domain_name: 账户名，一般等同用户名
:param region: 区域，如cn-north-4
:param url: 请求token的url，可使用默认值
:param config 配置信息
:return: 请求的token
"""
if url is None:
url = 'https://iam.' + region + '.myhuaweicloud.com/v3/auth/tokens'
if not isinstance(config, SisConfig):
error_msg = 'the param \'config\' in token_service must be SisConfig class'
logger.error(error_msg)
raise ClientException(error_msg)
time_out = (config.get_connect_timeout(), config.get_read_timeout())
proxy = config.get_proxy()
auth_data = {
"auth": {
"identity": {
"password": {
"user": {
"name": user_name,
"password": password,
"domain": {
"name": domain_name
}
}
},
"methods": ["password"]
},
"scope": {
"project": {
"name": region
}
}
}
}
headers = {'Content-Type': 'application/json'}
req = http_utils.http_connect(url, headers, auth_data, 'POST', time_out,
proxy)
if 'X-Subject-Token' not in req.headers:
logger.error('Error occurs in getting token, %s' % req.text)
raise ClientException('Error occurs in getting token, %s' % req.text)
token = req.headers['X-Subject-Token']
return token
def generate_scheme_host_uri(url):
if url.find('//') == -1 or url.find('com') == -1:
error_msg = '%s is invalid' % url
logger.error(error_msg)
raise ClientException(error_msg)
split1s = url.split('//')
split2s = split1s[1].split('com')
scheme = split1s[0] + '//'
host = split2s[0] + 'com'
uri = split2s[1]
return scheme, host, uri
def create(self, request):
"""
创建热词表
:param request: 热词表请求
:return: 热词表id
"""
if not isinstance(request, HotWordRequest):
logger.error('the parameter in \'create_hot_word(request)\' should be HotWordRequest class')
raise ClientException('the parameter in \'create_hot_word(request)\' should be HotWordRequest class')
url = self._service_endpoint + '/v1/' + self._project_id + '/asr/vocabularies'
params = request.construct_params()
result = aksk_service.aksk_connect(self._ak, self._sk, url, params, 'POST', self._sis_config)
return result
def construct_params(self):
paras = dict()
paras['encode_type'] = self._encode_type
paras['sample_rate'] = self._sample_rate
if self._data is None and self._url is None:
logger.error('data and url can\'n be both None')
raise ClientException('data and url can\'n be both None, you can choose set one parameter')
if self._data is not None and self._url is not None:
logger.warn('when data and url are not both None, only data can take effect.')
if self._data is not None:
paras['data'] = self._data
else:
paras['url'] = self._url
return paras
def assessment_video(self, request):
"""
多模态评测接口
:param request: 多模态评测请求
:return: 响应结果，返回为json格式
"""
if not isinstance(request, PaVideoRequest):
error_msg = 'the parameter in \'assessment_video(request)\' should be PaVideoRequest class'
logger.error(error_msg)
raise ClientException(error_msg)
url = self._service_endpoint + '/v1/' + self._project_id + '/assessment/video'
params = request.construct_parameter()
result = aksk_service.aksk_connect(self._ak, self._sk, url, params, 'POST', self._sis_config)
return result
def get_short_response(self, request):
"""
一句话识别接口
:param request: 一句话识别请求AsrCustomShortRequest
:return: 一句话识别响应结果，返回为json格式
"""
if not isinstance(request, AsrCustomShortRequest):
error_msg = 'the parameter in \'get_short_response(request)\' should be AsrCustomShortRequest class'
logger.error(error_msg)
raise ClientException(error_msg)
url = self._service_endpoint + '/v1/' + self._project_id + '/asr/short-audio'
params = request.construct_params()
result = aksk_service.aksk_connect(self._ak, self._sk, url, params,
'POST', self._sis_config)
return result
def _check_result(result):
result_str = json.dumps(result)
if 'error_code' in result and 'error_msg' in result:
logger.error(result_str)
# 睡眠2s保证关闭前，发送端有充足时间收到关闭请求。
self._status = 'close'
self._callback.on_error(result_str)
time.sleep(2)
self._ws.close()
raise ClientException(result_str)
if 'resp_type' not in result:
self._status == 'error'
error_msg = 'result doesn\'t contain key resp, result is %s' % result_str
logger.error(error_msg)
self._callback.on_error(error_msg)
raise ClientException(error_msg)
def get_asr_response(self, request):
"""
短语音识别接口
:param request: 短语音识别请求 AsrRequest
:return: 短语音识别响应，json格式
"""
if not isinstance(request, AsrRequest):
logger.error(
'the parameter in \'get_asr_response(request)\' should be AsrRequest class'
)
raise ClientException(
'the parameter in \'get_asr_response(request)\' should be AsrRequest class'
)
url = self._service_endpoint + '/v1.0/voice/asr/sentence'
params = request.construct_params()
result = aksk_service.aksk_connect(self._ak, self._sk, url, params,
'POST', self._sis_config)
return result
def submit_job(self, request):
"""
录音文件识别，提交任务接口
:param request: 录音文件识别请求
:return: job_id
"""
if not isinstance(request, AsrCustomLongRequest):
error_msg = 'the parameter in \'submit_job(request)\' should be AsrCustomLongRequest class'
logger.error(error_msg)
raise ClientException(error_msg)
url = self._service_endpoint + '/v1/' + self._project_id + '/asr/transcriber/jobs'
params = request.construct_parameter()
result = aksk_service.aksk_connect(self._ak, self._sk, url, params,
'POST', self._sis_config)
if 'job_id' not in result:
error_msg = 'The result of long audio transcription doesn\'t contain key job_id, result is ' % result
logger.error(error_msg)
raise ClientException(error_msg)
return result['job_id']
def parse_resp(resp):
"""
requests响应转化为json格式
:param resp: requests请求返回的响应
:return: json
"""
if resp is None or resp.text is None or resp.text == '':
return None
text = resp.text
try:
result = json.loads(text)
except Exception as e:
error_msg = 'Parsing json failed, the text is %s' % text
logger.error(error_msg)
raise ClientException(error_msg)
if 'error_code' in result and 'error_msg' in result:
error_msg = json.dumps(result)
logger.error(error_msg)
raise ServerException(result['error_code'], result['error_msg'])
return result
def get_ttsc_response(self, request):
"""
定制语音合成接口
:param request: 定制语音合成请求，TtsCustomRequest
:return: 请求结果，json格式
"""
if not isinstance(request, TtsCustomRequest):
logger.error('the parameter in \'get_ttsc_response(request)\' should be TtsCustomRequest class')
raise ClientException('the parameter in \'get_ttsc_response(request)\' should be TtsCustomRequest class')
url = self._service_endpoint + '/v1/' + self._project_id + '/tts'
params = request.construct_params()
result = aksk_service.aksk_connect(self._ak, self._sk, url, params, 'POST', self._sis_config)
if 'result' not in result:
error_msg = 'The result of tts customization is invalid. Result is %s ' % json.dumps(result)
logger.error(error_msg)
raise ClientException(error_msg)
if request.get_saved():
base_str = result['result']['data']
io_utils.save_audio_from_base64str(base_str, request.get_saved_path())
result['is_saved'] = True
result['saved_path'] = request.get_saved_path()
return result
def _on_error(ws, error):
logger.error(error)
self._status = 'error'
self._callback.on_error(error)
def http_connect(url,
header,
data,
http_method='POST',
time_out=5,
proxy=None):
"""
post请求，带有header信息（用于认证）
:param url: -
:param header: 头部
:param data: post数据
:param time_out: 超时
:param proxy: 代理
:param http_method: http方法，目前支持put、delete、post、get
:return: http请求的response
"""
if isinstance(data, dict):
data = json.dumps(data)
if proxy is not None:
proxy = _generate_request_proxy(proxy)
else:
proxy = {'http': None, 'https': None}
# 加入重试机制
count = 0
resp = None
while count < NUM_MAX_RETRY:
try:
if http_method == 'POST':
resp = requests.post(url,
headers=header,
data=data,
timeout=time_out,
verify=False,
proxies=proxy)
elif http_method == 'GET':
resp = requests.get(url,
headers=header,
params=data,
timeout=time_out,
verify=False,
proxies=proxy)
elif http_method == 'PUT':
resp = requests.put(url,
headers=header,
data=data,
timeout=time_out,
verify=False,
proxies=proxy)
elif http_method == 'DELETE':
resp = requests.delete(url,
headers=header,
params=data,
timeout=time_out,
verify=False,
proxies=proxy)
else:
logger.error('%s is invalid' % http_method)
raise ClientException('%s is invalid' % http_method)
break
except requests.exceptions.RequestException as e:
logger.error(
'Error occurs in %s, the client will retry 5 times. Error message is %s'
% (http_method, e))
count += 1
if resp is None:
logger.error('%s Response is empety, url is %s' % (http_method, url))
raise ClientException('%s Response is empety, url is %s' %
(http_method, url))
return resp
def _check_request(request):
if not isinstance(request, RasrRequest):
error_msg = 'The parameter of request in RasrClient should be RasrRequest class'
logger.error(error_msg)
raise ClientException(error_msg)
def __init__(self,
user_name,
password,
domain_name,
region,
project_id,
callback,
config=SisConfig(),
service_endpoint=None,
token_url=None,
retry_sleep_time=1):
"""
实时语音转写client初始化
:param user_name: 用户名
:param password: 密码
:param domain_name: 账户名，一般等同用户名
:param region: 区域，如cn-north-4
:param project_id: 项目ID，可参考https://support.huaweicloud.com/api-sis/sis_03_0008.html
:param callback: 回调类RasrCallBack，用于监听websocket连接、响应、断开、错误等
:param service_endpoint: 终端节点，一般使用默认即可
:param token_url: 请求token的url，一般使用默认即可
:param retry_sleep_time: 当websocket连接失败重试的间隔时间，默认为5s
"""
if service_endpoint is None:
self._service_endpoint = 'wss://sis-ext.' + region + '.myhuaweicloud.com'
else:
self._service_endpoint = service_endpoint
if token_url is None:
self._token_url = 'https://iam.' + region + '.myhuaweicloud.com/v3/auth/tokens'
else:
self._token_url = token_url
if not isinstance(callback, RasrCallBack):
logger.error('The parameter callback must be RasrCallBack class')
raise ClientException(
'The parameter callback must be RasrCallBack class')
if not isinstance(config, SisConfig):
logger.error('The parameter config must by SisConfig class')
raise ClientException(
'The parameter config must by SisConfig class')
self._project_id = project_id
self._config = config
# token 缓存必须在client进行，才可以在多线程中生效。
now_time = time.time()
self._token = None
if user_name in user_dict and user_name in time_dict:
token = user_dict[user_name]
save_time = time_dict[user_name]
if now_time - save_time < 5 * 3600:
logger.info('use token cache')
self._token = token
if self._token is None:
self._token = token_service.get_token(user_name,
password,
domain_name,
region,
url=self._token_url,
config=self._config)
user_dict[user_name] = self._token
time_dict[user_name] = now_time
self._callback = callback
self._status = 'pre_start'
self._request = None
self._retry_sleep_time = retry_sleep_time
def on_error(self, error):
logger.error(error)
def _connect(self, url):
def _check_result(result):
result_str = json.dumps(result)
if 'error_code' in result and 'error_msg' in result:
logger.error(result_str)
# 睡眠2s保证关闭前，发送端有充足时间收到关闭请求。
self._status = 'close'
self._callback.on_error(result_str)
time.sleep(2)
self._ws.close()
raise ClientException(result_str)
if 'resp_type' not in result:
self._status == 'error'
error_msg = 'result doesn\'t contain key resp, result is %s' % result_str
logger.error(error_msg)
self._callback.on_error(error_msg)
raise ClientException(error_msg)
def _on_open(ws):
logger.info('websocket open')
self._status = 'start'
self._callback.on_open()
def _on_message(ws, message):
result = json.loads(message)
_check_result(result)
result_type = result['resp_type']
trace_id = result['trace_id']
if result_type == 'START':
self._callback.on_start(trace_id)
elif result_type == 'EVENT':
event = result['event']
if event == 'EXCEEDED_AUDIO':
logger.warn(
'the duration of the audio is too long, the rest won\'t be recognized'
)
self._status = 'end'
elif event == 'EXCEEDED_SILENCE':
logger.error(
'silent time is too long, the audio won\'t be recognized'
)
self._callback.on_error(
'silent time is too long, the audio won\'t be recognized'
)
self._status = 'error'
elif event == 'VOICE_END':
logger.warn(
'detect voice end, the rest won\'t be recognized')
self._status = 'end'
elif result_type == 'RESULT':
self._callback.on_response(result)
elif result_type == 'END':
self._status = 'end'
self._callback.on_end(trace_id)
else:
logger.error('%s don\'t belong to any type' % result_type)
def _on_close(ws):
logger.info('websocket close')
self._status = 'close'
self._callback.on_close()
def _on_error(ws, error):
logger.error(error)
self._status = 'error'
self._callback.on_error(error)
# 重试机制
headers = {'X-Auth-Token': self._token}
sslopt = {"cert_reqs": ssl.CERT_NONE}
retry_count = 5
for i in range(retry_count):
self._status = 'pre_start'
self._ws = websocket.WebSocketApp(url,
headers,
on_open=_on_open,
on_close=_on_close,
on_message=_on_message,
on_error=_on_error)
self._thread = threading.Thread(
target=self._ws.run_forever,
args=(None, sslopt,
self._config.get_connect_lost_timeout() + 1,
self._config.get_connect_lost_timeout()))
self._thread.daemon = True
self._thread.start()
connect_count = int(self._config.get_connect_timeout() /
connect_sleep_time)
for j in range(connect_count):
if self._status != 'pre_start':
break
else:
time.sleep(connect_sleep_time)
if self._status == 'start':
break
else:
logger.error(
'connect meets error, it will retry %d times, now it is %d'
% (retry_count, i + 1))
time.sleep(self._retry_sleep_time)
if self._status == 'pre_start' or self._status == 'close' or self._status == 'error' or self._status == 'end':
logger.error('websocket connect failed， url is %s' % url)
raise ClientException('websocket connect failed， url is %s' % url)