Python client library to quickly get started with the various Watson APIs services.
Table of Contents
- You need an IBM Cloud account.
To install, use pip or easy_install:
pip install --upgrade watson-developer-cloudor
easy_install --upgrade watson-developer-cloudNote the following:
a) If you run into permission issues try:
sudo -H pip install --ignore-installed six watson-developer-cloudFor more details see #225
b) In case you run into problems installing the SDK in DSX, try
!pip install --upgrade pip
Restarting the kernel
For more details see #405
The examples folder has basic and advanced examples. The examples within each service assume that you already have service credentials.
Note: Conversation V1 is deprecated and will be removed in the next major release of the SDK. Use Assistant V1 or Assistant V2.
Watson Assistant V2 API is released in beta. For details, see the "Introducing Watson Assistant" blog post.
If you run your app in IBM Cloud, the SDK gets credentials from the VCAP_SERVICES environment variable.
Watson services are migrating to token-based Identity and Access Management (IAM) authentication.
- With some service instances, you authenticate to the API by using IAM.
- In other instances, you authenticate by providing the username and password for the service instance.
Note: Authenticating with the X-Watson-Authorization-Token header is deprecated. The token continues to work with Cloud Foundry services, but is not supported for services that use Identity and Access Management (IAM) authentication. See here for details.
To find out which authentication to use, view the service credentials. You find the service credentials for authentication the same way for all Watson services:
- Go to the IBM Cloud Dashboard page.
- Either click an existing Watson service instance or click Create resource > AI and create a service instance.
- Copy the
urland eitherapikeyorusernameandpassword. Click Show if the credentials are masked.
IBM Cloud is migrating to token-based Identity and Access Management (IAM) authentication. IAM authentication uses a service API key to get an access token that is passed with the call. Access tokens are valid for approximately one hour and must be regenerated.
You supply either an IAM service API key or an access token:
- Use the API key to have the SDK manage the lifecycle of the access token. The SDK requests an access token, ensures that the access token is valid, and refreshes it if necessary.
- Use the access token if you want to manage the lifecycle yourself. For details, see Authenticating with IAM tokens.
# In the constructor, letting the SDK manage the IAM token
discovery = DiscoveryV1(version='2018-08-01',
url='<url_as_per_region>',
iam_apikey='<iam_apikey>',
iam_url='<iam_url>') # optional - the default value is https://iam.bluemix.net/identity/token# after instantiation, letting the SDK manage the IAM token
discovery = DiscoveryV1(version='2018-08-01', url='<url_as_per_region>')
discovery.set_iam_apikey('<iam_apikey>')# in the constructor, assuming control of managing IAM token
discovery = DiscoveryV1(version='2018-08-01',
url='<url_as_per_region>',
iam_access_token='<iam_access_token>')# after instantiation, assuming control of managing IAM token
discovery = DiscoveryV1(version='2018-08-01', url='<url_as_per_region>')
discovery.set_iam_access_token('<access_token>')from watson_developer_cloud import DiscoveryV1
# In the constructor
discovery = DiscoveryV1(version='2018-08-01', url='<url_as_per_region>', username='<username>', password='<password>')# After instantiation
discovery = DiscoveryV1(version='2018-08-01', url='<url_as_per_region>')
discovery.set_username_and_password('<username>', '<password>')Tested on Python 2.7, 3.4, 3.5, and 3.6.
Version 1.0 focuses on the move to programmatically-generated code for many of the services. See the changelog for the details.
DetailedResponse which contains the result, headers and HTTP status code is now the default response for all methods.
from watson_developer_cloud import AssistantV1
assistant = AssistantV1(
username='xxx',
password='yyy',
url='<url_as_per_region>',
version='2018-07-10')
response = assistant.list_workspaces(headers={'Custom-Header': 'custom_value'})
print(response.get_result())
print(response.get_headers())
print(response.get_status_code())See the changelog for the details.
This version includes many breaking changes as a result of standardizing behavior across the new generated services. Full details on migration from previous versions can be found here.
To set client configs like timeout use the with_http_config() function and pass it a dictionary of configs.
from watson_developer_cloud import AssistantV1
assistant = AssistantV1(
username='xxx',
password='yyy',
url='<url_as_per_region>',
version='2018-07-10')
assistant.set_http_config({'timeout': 100})
response = assistant.message(workspace_id=workspace_id, input={
'text': 'What\'s the weather like?'}).get_result()
print(json.dumps(response, indent=2))For ICP(IBM Cloud Private), you can disable the SSL certificate verification by:
service.disable_SSL_verification()Custom headers can be passed in any request in the form of a dict as:
headers = {
'Custom-Header': 'custom_value'
}For example, to send a header called Custom-Header to a call in Watson Assistant, pass
the headers parameter as:
from watson_developer_cloud import AssistantV1
assistant = AssistantV1(
username='xxx',
password='yyy',
url='<url_as_per_region>',
version='2018-07-10')
response = assistant.list_workspaces(headers={'Custom-Header': 'custom_value'}).get_result()If you would like access to some HTTP response information along with the response model, you can set the set_detailed_response() to True. Since Python SDK v2.0, it is set to True
from watson_developer_cloud import AssistantV1
assistant = AssistantV1(
username='xxx',
password='yyy',
url='<url_as_per_region>',
version='2018-07-10')
assistant.set_detailed_response(True)
response = assistant.list_workspaces(headers={'Custom-Header': 'custom_value'}).get_result()
print(response)This would give an output of DetailedResponse having the structure:
{
'result': <response returned by service>,
'headers': { <http response headers> },
'status_code': <http status code>
}You can use the get_result(), get_headers() and get_status_code() to return the result, headers and status code respectively.
The Text to Speech service supports synthesizing text to spoken audio using web sockets with the synthesize_using_websocket. The Speech to Text service supports recognizing speech to text using web sockets with the recognize_using_websocket. These methods need a custom callback class to listen to events. Below is an example of synthesize_using_websocket. Note: The service accepts one request per connection.
from watson_developer_cloud.websocket import SynthesizeCallback
class MySynthesizeCallback(SynthesizeCallback):
def __init__(self):
SynthesizeCallback.__init__(self)
def on_audio_stream(self, audio_stream):
return audio_stream
def on_data(self, data):
return data
my_callback = MySynthesizeCallback()
service.synthesize_using_websocket('I like to pet dogs',
my_callback,
accept='audio/wav',
voice='en-US_AllisonVoice'
)- requests
python_dateutil>= 2.5.3- responses for testing
- Following for web sockets support in speech to text
websocket-client0.48.0
See CONTRIBUTING.md.
This library is licensed under the Apache 2.0 license.