Watson Developer Cloud Python SDK

Deprecated builds

Python client library to quickly get started with the various [Watson APIs][wdc] services.

Before you begin

You need an [IBM Cloud][ibm-cloud-onboarding] account. We now only support python 3.5 and above

Installation

To install, use pip or easy_install:

pip install --upgrade ibm-watson

easy_install --upgrade ibm-watson

Note the following: a) Versions prior to 3.0.0 can be installed using:

pip install --upgrade watson-developer-cloud

b) If you run into permission issues try:

sudo -H pip install --ignore-installed six ibm-watson

For more details see #225

c) In case you run into problems installing the SDK in DSX, try

!pip install --upgrade pip

Restarting the kernel

For more details see #405

Examples

The [examples][examples] folder has basic and advanced examples. The examples within each service assume that you already have service credentials.

Running in IBM Cloud

If you run your app in IBM Cloud, the SDK gets credentials from the [VCAP_SERVICES][vcap_services] environment variable.

Authentication

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.

Getting credentials

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 in your resource list or click Create resource > AI and create a service instance.
Click on the Manage item in the left nav bar of your service instance.

On this page, you should be able to see your credentials for accessing your service instance.

Supplying credentials

There are three ways to supply the credentials you found above to the SDK for authentication.

Credential file

With a credential file, you just need to put the file in the right place and the SDK will do the work of parsing and authenticating. You can get this file by clicking the Download button for the credentials in the Manage tab of your service instance.

The file downloaded will be called ibm-credentials.env. This is the name the SDK will search for and must be preserved unless you want to configure the file path (more on that later). The SDK will look for your ibm-credentials.env file in the following places (in order):

The top-level directory of the project you're using the SDK in
Your system's home directory

As long as you set that up correctly, you don't have to worry about setting any authentication options in your code. So, for example, if you created and downloaded the credential file for your Discovery instance, you just need to do the following:

discovery = DiscoveryV1(version='2019-04-30')

And that's it!

If you're using more than one service at a time in your code and get two different ibm-credentials.env files, just put the contents together in one ibm-credentials.env file and the SDK will handle assigning credentials to their appropriate services.

If you would like to configure the location/name of your credential file, you can set an environment variable called IBM_CREDENTIALS_FILE. This will take precedence over the locations specified above. Here's how you can do that:

export IBM_CREDENTIALS_FILE="<path>"

where <path> is something like /home/user/Downloads/<file_name>.env.

Environment Variables

Simply set the environment variables using _ syntax. For example, using your favourite terminal, you can set environment variables for Assistant service instance:

export ASSISTANT_APIKEY="<your apikey>"
export ASSISTANT_AUTH_TYPE="iam"

The credentials will be loaded from the environment automatically

assistant = AssistantV1(version='2018-08-01')

Manually

If you'd prefer to set authentication values manually in your code, the SDK supports that as well. The way you'll do this depends on what type of credentials your service instance gives you.

IAM

IBM Cloud has migrated 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 a bearer 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.
Use a server-side to generate access tokens using your IAM API key for untrusted environments like client-side scripts. The generated access tokens will be valid for one hour and can be refreshed.

Supplying the API key

from ibm_watson import DiscoveryV1
from ibm_cloud_sdk_core.authenticators import IAMAuthenticator

# In the constructor, letting the SDK manage the token
authenticator = IAMAuthenticator('apikey',
                                 url='<iam_url>') # optional - the default value is https://iam.cloud.ibm.com/identity/token
discovery = DiscoveryV1(version='2019-04-30',
                        authenticator=authenticator)
discovery.set_service_url('<url_as_per_region>')

Generating bearer tokens using API key

from ibm_watson import IAMTokenManager

# In your API endpoint use this to generate new bearer tokens
iam_token_manager = IAMTokenManager(apikey='<apikey>')
token = iam_token_manager.get_token()

Supplying the bearer token

from ibm_watson import DiscoveryV1
from ibm_cloud_sdk_core.authenticators import BearerTokenAuthenticator

# in the constructor, assuming control of managing the token
authenticator = BearerTokenAuthenticator('your bearer token')
discovery = DiscoveryV1(version='2019-04-30',
                        authenticator=authenticator)
discovery.set_service_url('<url_as_per_region>')

Username and password

from ibm_watson import DiscoveryV1
from ibm_cloud_sdk_core.authenticators import BasicAuthenticator

authenticator = BasicAuthenticator('username', 'password')
discovery = DiscoveryV1(version='2019-04-30', authenticator=authenticator)
discovery.set_service_url('<url_as_per_region>')

No Authentication

from ibm_watson import DiscoveryV1
from ibm_cloud_sdk_core.authenticators import NoAuthAuthenticator

authenticator = NoAuthAuthenticator()
discovery = DiscoveryV1(version='2019-04-30', authenticator=authenticator)
discovery.set_service_url('<url_as_per_region>')

MCSP

To use the SDK through a third party cloud provider (such as AWS), use the MCSPAuthenticator. This will require the base endpoint URL for the MCSP token service (e.g. https://iam.platform.saas.ibm.com) and an apikey.

from ibm_watson import AssistantV2
from ibm_cloud_sdk_core.authenticators import MCSPAuthenticator

# In the constructor, letting the SDK manage the token
authenticator = MCSPAuthenticator('apikey', 'token_service_endpoint')
assistant = AssistantV2(version='2023-06-15',
                        authenticator=authenticator)
assistant.set_service_url('<url_as_per_region>')

Python version

Tested on Python 3.9, 3.10, and 3.11.

Questions

If you have issues with the APIs or have a question about the Watson services, see Stack Overflow.

Configuring the http client (Supported from v1.1.0)

To set client configs like timeout use the set_http_config() function and pass it a dictionary of configs. See this documentation for more information about the options. All options shown except method, url, headers, params, data, and auth are configurable via set_http_config(). For example for a Assistant service instance

from ibm_watson import AssistantV1
from ibm_cloud_sdk_core.authenticators import IAMAuthenticator

authenticator = IAMAuthenticator('your apikey')
assistant = AssistantV1(
    version='2021-11-27',
    authenticator=authenticator)
assistant.set_service_url('https://api.us-south.assistant.watson.cloud.ibm.com')

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))

Use behind a corporate proxy

To use the SDK with any proxies you may have they can be set as shown below. For documentation on proxies see here

See this example configuration:

from ibm_watson import AssistantV1
from ibm_cloud_sdk_core.authenticators import IAMAuthenticator

authenticator = IAMAuthenticator('your apikey')
assistant = AssistantV1(
    version='2021-11-27',
    authenticator=authenticator)
assistant.set_service_url('https://api.us-south.assistant.watson.cloud.ibm.com')

assistant.set_http_config({'proxies': {
  'http': 'http://10.10.1.10:3128',
  'https': 'http://10.10.1.10:1080',
}})

Sending custom certificates

To send custom certificates as a security measure in your request, use the cert property of the HTTPS Agent.

from ibm_watson import AssistantV1
from ibm_cloud_sdk_core.authenticators import IAMAuthenticator

authenticator = IAMAuthenticator('your apikey')
assistant = AssistantV1(
    version='2021-11-27',
    authenticator=authenticator)
assistant.set_service_url('https://api.us-south.assistant.watson.cloud.ibm.com')

assistant.set_http_config({'cert': ('path_to_cert_file','path_to_key_file')})

Disable SSL certificate verification

For ICP(IBM Cloud Private), you can disable the SSL certificate verification by:

service.set_disable_ssl_verification(True)

Or can set it from extrernal sources. For example set in the environment variable.

export <service name>_DISABLE_SSL=True

Setting the service url

To set the base service to be used when contacting the service

service.set_service_url('my_new_service_url')

Or can set it from extrernal sources. For example set in the environment variable.

export <service name>_URL="<your url>"

Sending request headers

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 ibm_watson import AssistantV1
from ibm_cloud_sdk_core.authenticators import IAMAuthenticator

authenticator = IAMAuthenticator('your apikey')
assistant = AssistantV1(
    version='2018-07-10',
    authenticator=authenticator)
assistant.set_service_url('https://gateway.watsonplatform.net/assistant/api')

response = assistant.list_workspaces(headers={'Custom-Header': 'custom_value'}).get_result()

Parsing HTTP response information

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 ibm_watson import AssistantV1
from ibm_cloud_sdk_core.authenticators import IAMAuthenticator

authenticator = IAMAuthenticator('your apikey')
assistant = AssistantV1(
    version='2018-07-10',
    authenticator=authenticator)
assistant.set_service_url('https://gateway.watsonplatform.net/assistant/api')

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.

Getting the transaction ID

Every SDK call returns a response with a transaction ID in the X-Global-Transaction-Id header. Together the service instance region, this ID helps support teams troubleshoot issues from relevant logs.

Suceess

from ibm_watson import AssistantV1

service = AssistantV1(authenticator={my_authenticator})
response_headers = service.my_service_call().get_headers()
print(response_headers.get('X-Global-Transaction-Id'))

Failure

from ibm_watson import AssistantV1, ApiException

try:
    service = AssistantV1(authenticator={my_authenticator})
    service.my_service_call()
except ApiException as e:
    print(e.global_transaction_id)
    # OR
    print(e.http_response.headers.get('X-Global-Transaction-Id'))

However, the transaction ID isn't available when the API doesn't return a response for some reason. In that case, you can set your own transaction ID in the request. For example, replace <my-unique-transaction-id> in the following example with a unique transaction ID.

from ibm_watson import AssistantV1

service = AssistantV1(authenticator={my_authenticator})
service.my_service_call(headers={'X-Global-Transaction-Id': '<my-unique-transaction-id>'})

Using Websockets

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 ibm_watson.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'
                                  )

Cloud Pak for Data

If your service instance is of CP4D, below are two ways of initializing the assistant service.

1) Supplying the username, password and authentication url

The SDK will manage the token for the user

from ibm_watson import AssistantV1
from ibm_cloud_sdk_core.authenticators import CloudPakForDataAuthenticator

authenticator = CloudPakForDataAuthenticator(
    '<your username>',
    '<your password>',
    '<authentication url>', # should be of the form https://{icp_cluster_host}{instance-id}/api
    disable_ssl_verification=True) # Disable ssl verification for authenticator

assistant = AssistantV1(
    version='<version>',
    authenticator=authenticator)
assistant.set_service_url('<service url>') # should be of the form https://{icp_cluster_host}/{deployment}/assistant/{instance-id}/api
assistant.set_disable_ssl_verification(True) # MAKE SURE SSL VERIFICATION IS DISABLED

2) Supplying the access token

from ibm_watson import AssistantV1
from ibm_cloud_sdk_core.authenticators import BearerTokenAuthenticator

authenticator = BearerTokenAuthenticator('your managed access token')
assistant = AssistantV1(version='<version>',
                        authenticator=authenticator)
assistant.set_service_url('<service url>') # should be of the form https://{icp_cluster_host}/{deployment}/assistant/{instance-id}/api
assistant.set_disable_ssl_verification(True) # MAKE SURE SSL VERIFICATION IS DISABLED

Logging

Enable logging

import logging
logging.basicConfig(level=logging.DEBUG)

This would show output of the form:

DEBUG:urllib3.connectionpool:Starting new HTTPS connection (1): iam.cloud.ibm.com:443
DEBUG:urllib3.connectionpool:https://iam.cloud.ibm.com:443 "POST /identity/token HTTP/1.1" 200 1809
DEBUG:urllib3.connectionpool:Starting new HTTPS connection (1): gateway.watsonplatform.net:443
DEBUG:urllib3.connectionpool:https://gateway.watsonplatform.net:443 "POST /assistant/api/v1/workspaces?version=2018-07-10 HTTP/1.1" 201 None
DEBUG:urllib3.connectionpool:Starting new HTTPS connection (1): gateway.watsonplatform.net:443
DEBUG:urllib3.connectionpool:https://gateway.watsonplatform.net:443 "GET /assistant/api/v1/workspaces/883a2a44-eb5f-4b1a-96b0-32a90b475ea8?version=2018-07-10&export=true HTTP/1.1" 200 None
DEBUG:urllib3.connectionpool:Starting new HTTPS connection (1): gateway.watsonplatform.net:443
DEBUG:urllib3.connectionpool:https://gateway.watsonplatform.net:443 "DELETE /assistant/api/v1/workspaces/883a2a44-eb5f-4b1a-96b0-32a90b475ea8?version=2018-07-10 HTTP/1.1" 200 28

Low level request and response dump

To get low level information of the requests/ responses:

from http.client import HTTPConnection
HTTPConnection.debuglevel = 1

Dependencies

[requests]
python_dateutil >= 2.5.3
[responses] for testing
Following for web sockets support in speech to text
- websocket-client 1.1.0
ibm_cloud_sdk_core >= 3.16.2

Contributing

See [CONTRIBUTING.md][contributing].

License

This library is licensed under the [Apache 2.0 license][license].