rasa-api

rasa-api is a node package that allows you to easily configure and train your NLP through Rasa HTTP API.

Note that rasa-api uses promises, so be aware to always catch at least once when implementing your code.

Please refer to the official HTTP API documentation to know about available endpoints.

Installation

npm install rasa-api

Initialization

const Rasa = require('rasa-api')

const rasa = new Rasa('<ENDPOINT-URL>', '<YOUR-PROJECT-NAME>','<YOUR-SERVER-ACCESS-TOKEN>', '<TIMEOUT-IN-SECONDS>')

Parse

rasa.parse('Wake me up when september end!').then((res) => {
  console.log(res)
}).catch((error) => {
  console.error(error)
})

Training

You must provide data in JSON format. You can refer to Rasa Documentation in order know about the data structure.

rasa.train({
  "common_examples": [{
    "text": "hey",
    "intent": "greet",
    "entities": []
  }, {
    "text": "i'm looking for a place in the north of town",
    "intent": "restaurant_search",
    "entities": [
      {
        "start": 31,
        "end": 36,
        "value": "north",
        "entity": "location"
      }
    ]
  }],
  "regex_features" : [],
  "lookup_tables"  : [],
  "entity_synonyms": []
}).then(res => {
  console.log(res)
}).catch((error) => {
  console.error(error)
})

Evaluate

You must provide test dataset in JSON format. You can refer to Rasa Documentation in order to learn about the data structure.

rasa.evaluate({
  "common_examples": [{
    "text": "hey",
    "intent": "greet",
    "entities": []
  }],
  "regex_features" : [],
  "lookup_tables"  : [],
  "entity_synonyms": []
}).then(res => {
  console.log(res)
})

Status/Version/Config

//
let info = 'version' // You can pass one of the following ['version', 'status', 'config']
rasa.get(info).then(res => {
  console.log(res)
}).catch((error) => {
  console.error(error)
})

Delete models

//
let model = 'model_20190221-114950'
rasa.delete(model).then(res => {
  console.log(res)
}).catch((error) => {
  console.error(error)
})