Installation

• Building from source code (must to be updated)
• Docker container

Building from source code

Build dependences

• Tools: cmake, ninja, libttspico-utils, c++, nltk, python3, sox, liblo

Under Ubuntu, most of these dependencies are installed with the following packages:

$ sudo apt-get install -y build-essential ca-certificates gcc curl wget unzip git apt-utils \
libpq-dev  mercurial pkg-config python3 python3-dev python3-pip libttspico-utils alsa-utils \
sox git cmake cmake-data python-nltk libsox-fmt-all locales ninja-build pyliblo-utils make

To install all python packages in an environment, follow these instructions.

After installing virtualenv, do:

$ mkvirtualenv --python=/usr/bin/python3 env_name
$ workon env_name

Pyblio is installed with the following instructions:

wget http://downloads.sourceforge.net/liblo/liblo-0.29.tar.gz
tar xzf liblo-0.29.tar.gz && cd liblo-0.29
./configure
make install
cd .. && rm -R liblo*
pip3 install pyliblo 

Installation of the python version of sent2vec:

$ git clone https://github.com/jbcdnr/sent2vec.git
$ cd sent2vec
$ make
$ pip3 install cython numpy nltk
$ pip3 install .
$ cd ..

Installation of nltk stopwords data

$ python3 -m nltk.downloader -d nltk_data stopwords

Lima docker

$ sudo docker  pull aymara/lima:latest

Install Iagotchi Bot

• Download the Iagotchi-bot folder and move it to an installation location. Let's suppose install_folder the installation folder containing iagotchi-bot. Then follow the following instructions:

  >  cd install_folder
  >  mkdir Dist
  >  mkdir Build
  >  git clone https://github.com/ChatScript/ChatScript.git
  >  sudo chmod 755 ./ChatScript/BINARIES/LinuxChatScript64
  >  mv ChatScript Dist
  

• Define the following environment variable:

  IAGOTCHICHATSCRIPTINSTALLDIR=/path/to/install_folder/Dist/ChatScript
  

• Continue with the following instructions:

  >  cd Build
  >  cmake -DCMAKE_INSTALL_PREFIX=/path/to/install_folder/Dist ../iagotchi-bot -G Ninja
  >  ninja install 
  >  cd ../Dist
  >  mkdir -p static/assets
  

Usage

Configuring Bot

All the configurations needed to run Iagotchi-bot are in the config.json file located in /path/to/install_folder/Dist/data. The configuration file is read by the system at each restart and the parameters are reset. The fields necessary to fill in the file are:

• {session} : fill with the durations in second of session time, of the user relaunch, etc ...
• {lima} : set the appropriate values for the Lima server address (127.0.0.1) and port (8090)
• {botresponse} : specify the port on which the bot must return its response.
• {synthesize} : set to False for using an external speech synthesis from the response.
• {musique} : specify the port on which the bot must return the music request.

Running Bot

There are two possible tasks to perform with the main script. Among other things, the ability to rebuild the embeddings sentences of the similarity module with the --build option before running (with --run) the bot's dialog system. This option allows you to take into account the news sentences collected during the execution of the bot.

• Running Lima

  You must first run lima before running the bot.

  $ sudo docker run -p 0.0.0.0:8090:8080/tcp -it aymara/lima limaserver
  

• For calculating sentence embeddings

  $ python3 main.py --build similarity
  

• For running the dialogue system.

  $ python3 main.py --run bot
  
  

  Then lauch a browser and go to the URL http://127.0.0.1:8088/

Docker container

Lima

- sudo docker pull aymara/lima:latest

- sudo docker tag aymara/lima limaserver:latest

Build Iagotchi-Bot

1- Access to parent folder of iagotchi-bot

2- sudo docker-compose build

Run Iagotchi-Bot

sudo docker-compose up

Usage

Configuring Bot

All the configurations needed to run Iagotchi-bot are in the config.json file located in /path/to/install_folder/Dist/data. The configuration file is read by the system at each restart and the parameters are reset. The fields necessary to fill in the file are:

• {what_run} - using to specify what to run: bot (set to bot), retrain similarity model with topics logs (set to similarity-with-topics), retrain similatiry model without topics logs (set to similarity-without-topics)
• {bot} - using to specify what topic to use: for rencontre topic set the name to iagotchi and iagotchig5 for G5 topic.
• {session} - fill with the durations in second of session time, of the user relaunch, etc ...
• {botresponse} - specify host ip address and port on which the bot must return its response.
• {synthesize} - set to False for using an external speech synthesis.
• {musique} - specify the port on which the bot must return the music request.

Modify the line 14 in docker-compose.yml with the full path to the shared folder.

Running Bot

• For running the dialogue system.

```
$ sudo docker-compose up  -d 
```


Then launch a browser and go to the URL http://127.0.0.1:8088/

• For stopping the system

  $ sudo docker-compose stop
  

• For starting the system

  $ sudo docker-compose start
  

Troubleshoot

Containers must be started separately to identify the cause of an error caused by the command docker-compose up

I see Problem with ChatScript

This error usually occurs when the chatscript server has not started properly perhaps due to a missing file in the sources of chatscript or an error in the name of the bot to launch. The following commands can provide a trace of the error:

1. First, run limaserver container

$ sudo docker-compose run -d limaserver

2. Then, run and access to iagotchi container

$ sudo docker-compose run --service-ports iagotchi bash

3. In iagotchi container

root@211c9bbd9a0f:/Dist# cd ChatScript/BINARIES/
root@211c9bbd9a0f:/Dist/ChatScript/BINARIES# ./LinuxChatScript64 local

4. After entering a username,

HARRY:  Welcome to ChatScript.
fr: > :build iagotchi

if you do not get the following message, it is likely that the server encountered an error that will be displayed.

IAGOTCHI:  Je n'ai pas compris. Peux-tu reformuler, s'il te pla�t? notrule
fr: > 

I see iagotchi_1 exited with code 137

This error is commonly associated with Docker for Mac not having enough RAM allocated to it. Error 137 in Docker denotes that the container was ‘KILL’ed by ‘oom-killer’ (Out of Memory). This happens when there isn’t enough memory in the container for running the process.

So it is an easy fix, go to the Docker Menu and select Preferences then the Advanced tab and increase the Memory. You have to increase it to at least 4.0GB.

To debug and trace the line that causes the container crash

You must perform the following steps after those previously described in 1 and 2.

3. In iagotchi container

root@211c9bbd9a0f:/Dist# python3 test_chatscript.py
root@211c9bbd9a0f:/Dist# python3 main.py --chat iagotchi