dfagent 🤖

The dfagent is a package for the handling of Dialogflow agents. You can for example retrieve training examples and save into a preferred format, or you can use it to update an intent by simply feeding it training examples you stored in a preferred format.

Overview

• dfagent/ contains all the core code to extend dfagent.

Install

To install the dfagent package you only need to run install.sh script.

Usage

Once dfagent is installed you can simply import it in your code.

Save training phrases [remote]

The following snippet illustrates a simple example to get and save training examples from an online Dialogflow agent.

To create a Dialogflow agent you only need that

import dfagent

agent = dfagent.DialogFlowAgent(
    local_path_or_url='my_gcp_project_id',
    service_account='path/to/sa.json',
    content_type='json',
    output_format='default'
)

Then you can get a list of dialogflow examples for saving as follows

examples = agent.get_training_examples()
agent.save_training_examples(examples, output_dir='path/to/dir')

Update intent with new training phrases [remote]

In the following is a snippet that illustrates an example to update a remote Dialogflow agent using training phrases you stored as a raw text file. Remember that dfagent can be extended to support any input or output file format.

Once you instante a df agent

import dfagent

agent = dfagent.DialogFlowAgent(
    local_path_or_url='my_gcp_project_id',
    service_account='path/to/sa.json',
    input_format='default',
)

You can update your remote Dialoflow agent in that way

response, raw_examples, df_examples = agent.add_training_examples(
    intent_name='help.cooking',
    input_dir_or_file='path/to/phrases.train',
    lang='en'
)

From local or zip

In case you already have exported your Dialogflow on your local computer, you can give as local_path_or_url the path to the zip or unzipped exported agent.

import dfagent

agent = dfagent.DialogFlowAgent(
    local_path_or_url='path/to/myagent.zip',
    ...
)