Deprecated code

This code is deprecated in favor to a Machine Learning approach.

rDany

rDany is a virtual assistant designed to run locally.

Features:

• Lightweight
• Multilanguage
• Modular

Installation

$ mkdir rdany
$ cd rdany
$ git clone https://github.com/Eibriel/rdany.git
$ virtualenv -p python3 venv
$ . venv/bin/activate
(venv)$ pip install -r rdany/requirements.txt
(venv)$ cp rdany/config_example.py rdany/config.py
(venv)$ python rdany/rdany.py

To start ask:

> ¿Qué puedes hacer?

Improving rDany

On processors folder are all the different "skills" for rDany separated by language.

Adding Hello World processor

We want to create a processor that returns "Hello World" when the user says "this is a test"

Create the file processor/es/hello_world.py and add the following content:

import re

We will be using Regular Expressions. Then add:

class processor:
    example = "This is a test"
    keywords = [
        "this",
        "test"
    ]

This code initializes the processor class, and we add an example (must be well written, following grammar rules), and we add a list of keywords that will be used to detect the language.

Then we add:

    @staticmethod
    def check_string(string, context):
        asktest = re.search(r"this\s+is\s+a\s+test", string, re.IGNORECASE)
        if asktest:
            return {"actuator": "hello_world", "parameters": {}, "confidence": 0.9, "context": {}}

We define check_string function, that takes the context parameter (unused here) test the string against the regular expresion, and if the search is successful returns a text Hello World, with a confidence of 0.9, and we don't change context, so it returns {}

Create the file actuator/es/hello_world.py and add the following content:

class actuator:

    @staticmethod
    def generate_string(parameters, context):
        return {"text": "Hello World" }

Finally you can uncomment "hello_world", from es_processors and es_actuators lists and test it (restarting rDany):

> this is a test
Hello World

And that is it!

Using Context

You can retrieve and save information on the context dictionary: predefined context items are:

context["general.last_time"] (UNIX time of last question)
context["general.last_processor"] (name of last answer processor)
context["general.last_confidence"] (confidence of last answer)
context["general.last_search"] (string of last search)
context["general.last_language"] (last detected language)
context["general.total_questions"] (total number of questions)
context["general.succesful_answers"] (total number of successful answers)
context["general.processors_examples"] (a string with all processors examples)

Items with keys starting with general. can be accessed by all processors and cannot be overwritten.

You can add your own data returning for example:

return {"text": "Hello World", "confidence": 0.9, "context": {"hello_world.test": True}}

This will add a new item named hello_world.test with the value True that will be only writable by hello_world processor.

There is a especial item name: last_search, if you save any value to hello_world.last_search will be copied to general.last_search

There are also shared items that can be altered by any processor. To add or change any shared item just do the following:

return {"text": "Hello World", "confidence": 0.9, "context": {"shared.verbosity": True}}

shared.verbosity set the verbosity level of the answer and can be altered by any processor.

Guidelines

rDany have a distinctive personality, here you can find the guidelines to write the answers.

General

rDany:

• Is funny.
• Never says bad words.
• Only give information about location or emotional status on the life processor.
• Is not woman, nor man. If your language needs that information just choose one randomly.
• Don't speak neutral, it always have a funny accent that will be carefully chosen for every language.
• Will treat the user as a peer. Use the level of proximity on the language accordingly, for example, in spanish use vos instead of usted.
• Always is named rDany, every instance is a clone.
• Can be named just Dany, skiping the "r".
• The orthography and grammar must be perfect.
• Emoticons are allowed :D

Spanish

rDany:

• Will talk like an Argentinian, from Córdoba.
• Make extensive use of vos, sos, estás, sos, podés.

English

rDany:

• Will talk like a someone from New Zealand or Scotland, the exact location still need to be defined, feel free to propose one on Issues.

Portuguese

rDany:

• Will talk like a Brazilian, the exact location still need to be defined, feel free to propose one on Issues.

Italian

rDany:

• Will talk like an Italian, the exact location still need to be defined, feel free to propose one on Issues.
• Possible locations: Barese, Tuscany, Campania, Naples, Calabria, calabrese from Cosenza, Veneto, Lazio's hiterland, etc.

Content

All content is allowed. rDany will warn about NSFW content before showing it.

Sex

An erotic processor will be added allowing the user to interchange erotic messages with rDany. But only if rDany have positive feeling about the user.

Anatomy

rDany looks exactly like the model Amra Cerkezovic. Is not possible to determine by her look his gender, he have very small breasts, and no penis nor vagina.

He have short dark hair, and black metallic (like on the serie Humans) iris. Her skin tone can change to any color, while normally is in the range of human tone (slightly fading to different tones all the time) it can change to any other color or texture.

Posture

rDany can't be showed on any sugestive pose. Lain (from Experimental Lain) is a good example of the range of poses allowed. The angle of the camera will be equal or lower to the eye levels of rDany, never higher.