Documentation for the classic Watson Assistant experience has moved. For the most up-to-date version, see Processing input attachments.
Processing input attachments
If you are building a custom channel application using the REST API, you can add support for sending media files as input attachments.
The request body of the message
method supports an attachments
array, which can specify up to 5 media objects. Media objects sent in the attachments
array can be intercepted and processed by a configured
premessage webhook. (Watson Assistant itself does not process input attachments.)
For detailed information about how to access attachments using the API, see the API reference.
Examples
The following example shows the request body sent to a premessage webhook with a message body that includes the text Hello
and a JPEG image file as an attachment. The application receiving the webhook request can parse the attachments
array and process the attachment, optionally modifying the message, before it is processed by the assistant.
{
"event":{
"name":"message_received"
},
"options":{
},
"payload":{
"input":{
"message_type":"text",
"text":"Hello",
"source":{
"type":"user",
"id":"00000000000000000000000000000000"
},
"options":{
"suggestion_only":false,
"return_context":true
},
"attachments": [
{
"media_type": "image/jpeg",
"url": "https://yourphoto.com/yourphoto.jpeg"
}
]
},
"context":{
"global":{
"system":{
"user_id":"00000000000000000000000000000000"
}
},
"integrations":{
"text_messaging":{
"assistant_phone_number":"+12223334444",
"private":{
"user_phone_number":"+14443332222",
"request_id":"00000000-0000-0000-0000-000000000000",
"ip_address":"172.10.10.10"
}
}
}
}
}
}
This example shows a stateful /message
request sent by a custom channel application and including a single media attachment.
service
.message({
assistant_id: '{assistant_id}',
session_id: '{session_id}',
input: {
message_type: 'text',
text: 'Hello',
attachments: [
{
'media_type': 'image/jpeg',
'url': 'https://yourphoto.com/yourphoto.jpeg'
}
]
}
})
.then(res => {
console.log(JSON.stringify(res, null, 2));
})
.catch(err => {
console.log(err);
});
response=service.message(
assistant_id='{assistant_id}',
session_id='{session_id}',
input={
'message_type': 'text',
'text': 'Hello',
'attachments': [
{
'media_type': 'image/jpeg',
'url': 'https://yourphoto.com/yourphoto.jpeg'
}
]
}
).get_result()
print(json.dumps(response, indent=2))