Dialogflow "export" via API: Missing required json file agent.json

Question

I'm trying to restore a Dialogflow agent using the Python SDK (google-cloud-dialogflow=2.7.1, google-api-core=2.0, Python 3.7):

import os
import base64
import google.cloud.dialogflow as dialogflow

# Authenticate and open session
os.environ["GOOGLE_APPLICATION_CREDENTIALS"] = 'test-a.json'
DIALOGFLOW_PROJECT_ID = 'xxxxxxxx'
SESSION_ID = '#'
session_client = dialogflow.SessionsClient()
session = session_client.session_path(DIALOGFLOW_PROJECT_ID, SESSION_ID)

# Get the agent
agent = dialogflow.AgentsClient()

# Read and encode zip file
encoded_file = base64.b64encode(open("test_A.zip", "rb").read())

# Restore agent
request = {
        "parent": f'projects/{DIALOGFLOW_PROJECT_ID}',
        "agent_content": encoded_file
    }
agent.restore_agent(request=request)

Here's what I get:

google.api_core.exceptions.InvalidArgument: 400 com.google.apps.framework.request.BadRequestException: Invalid agent zip. Missing required json file agent.json

But the file agent.json is in the zip file. In fact, the zip file is the exact file I downloaded from the Dialogflow's console export option:

Answer 1

You get this error because your passing base64 encoded data to field agent_content. As per RestoreAgentRequest(), agent_content accepts type: bytes.

class google.cloud.dialogflow_v2.types.RestoreAgentRequest(mapping=None, *, ignore_unknown_fields=False, **kwargs)

Bases: proto.message.Message

The request message for [Agents.RestoreAgent][google.cloud.dialogflow.v2.Agents.RestoreAgent].

parent

• Required. The project that the agent to restore is associated with. Format: projects/.
• Type: str

agent_uri

• The URI to a Google Cloud Storage file containing the agent to restore. Note: The URI must start with “gs://”.

Type: str

agent_content

• Zip compressed raw byte content for agent.

• Type: bytes

To fix this just remove the conversion to base64. See code below:

import os
import base64
import google.cloud.dialogflow as dialogflow

# Authenticate and open session
os.environ["GOOGLE_APPLICATION_CREDENTIALS"] = 'your-service-account.json'
DIALOGFLOW_PROJECT_ID = 'your-project'

# Get the agent
agent = dialogflow.AgentsClient()
zip_file="/full/path/of_file.zip"
# Read and encode zip file
with open(zip_file, 'rb') as file_data:
    bytes_content = file_data.read()

check_type=type(bytes_content)

print(f"bytes_content: {check_type}" )
# Restore agent
request = {
        "parent": f'projects/{DIALOGFLOW_PROJECT_ID}',
        "agent_content": bytes_content
    }
agent.restore_agent(request=request)

agent.restore_agent() will return an operation afterwards.