Python-based skill server for the hermes MQTT protocol
Project description
Hermes Skill Server
Intent handling based on modular skills written in python. Intents are supplied from the voice assistant via Hermes MQTT protocol.
Voice Assistant <=== mqtt ===> hss-server <===> hss-skill
Compatible with the Rhasspy voice assistant.
Installation
The server is preferably installed within a virtualenv, and requires python >=3.6.
/home/s710 $> mkdir hss
/home/s710 $> cd hss
/home/s710/hss $> mkdir hss
/home/s710/hss $> python3.7 -m venv /home/s710/hss/venv
/home/s710/hss $> source venv/bin/activate
(venv) /home/s710/hss $> pip install hss_server
Afterwards the server can be run:
(venv) /home/s710/hss $> hss-server
INFO:hss: Hermes Skill Server v1.0.0
INFO:hss: Using config dir '/home/s710/.config/hss_server'
INFO:hss: Loading skills from '/home/s710/.config/skills'
INFO:hss_server.skillserver: Loading skills ...
INFO:hss_server.collection: Initializing skills ...
INFO:hss_server.collection: Skill 'hss-skill-s710-weather' loaded
INFO:hss_server.collection: Loaded 1 skill
INFO:hss_server.skillserver: Connecting to MQTT server ...
INFO:hss_server.mqtt: Connected to 10.0.50.5:1883
INFO:hss_server.mqtt: Publishing TTS to topic 'hermes/tts/say'
INFO:hss_server.mqtt: Subscribing to topic 'hermes/intent/#' ...
After the initial start, hss-server
creates its configuration file ([USER_CONFIG_DIR]/hss_server/config.ini
). The config file will contain the location where skills are installed, which by default is [USER_CONFIG_DIR]/hss_server/skills
.
On Linux, the USER_CONFIG_DIR
will be ~/.config
, on MacOS it will be ~/Library/Application Support
.
Updating
Just simply use pip
again to update:
/home/s710 $> cd hss
/home/s710/hss $> source venv/bin/activate
(venv) /home/s710/hss $> pip install hss_server --upgrade
Features
The server opens a connection to the given MQTT broker, and listens on the intent-topics (by default: hermes/intent/#
).
Also, all available skills from the skills directory will be loaded, each skill as own process with its own virtualenv.
For every incoming intent which is published via MQTT, the skill-server tries to find a matching skill, and, if found, hands the intent over to the skill so it can be handled. If the skill implements a response text, then this text will be returned via MQTT to the TTS topic (by default: hermes/tts/say
).
Each skill is running in its own python virtualenv, with its own python dependencies and configuration. Skills can be installed easily from a git repository using the hss-cli
tool.
Options
Usage:
$ ./hss-server [OPTIONS]
Options:
--host=[HOST] MQTT host to connect to (default: localhost)
--port=[PORT] MQTT port (default: 1883)
--topic=[TOPIC] MQTT topic to listen on (default: hermes/intent/#)
--tts-topic=[TOPIC] MQTT topic to publish TTS to (default: hermes/tts/say)
--tts-url=[URL] URL to post TTS to. If set, TTS is not sent via MQTT (default: None)
--start-port=[PORT] Starting port for RCP communication (default: 51000)
--config-dir=[DIR] Location where the server's config.ini is located (default: user config dir)
--log-file=[FILE] Log file to write log entries to (default: console)
--debug Enable debug log output
--help Show this help and exit
--version Show version and exit
The options host
/port
/topic
/tts-topic
are needed for MQTT communication with the voice assistant.
The option --tts-url
switches the text-to-speech output from MQTT to HTTP, in case the voice assistant does not support MQTT-based TTS messages. For rhasspy <= 2.4, this should be set to http://[RHASSPYHOST]:12101/api/text-to-speech
.
The --start-port
option denotes the beginning of the dynamic RPC port range, and should not be changed unless there are issues with other services.
CLI
The hss-cli
tool is used to:
- list all installed skills
- install a skill
- uninstall a skill
- update one or all installed skills
Usage
Usage:
$ ./hss-cli [OPTIONS]
Options:
--list List all installed skills
--install --url=[URL] Install a new skill using [URL]. [URL] must be a valid GIT link.
--update (--skill=[NAME]) Update an already installed skill named [NAME].
If skill is ommited, ALL skills will be updated.
--uninstall --skill=[NAME] Uninstall an already installed skill named [NAME]
--help Show this help and exit
--version Show version and exit
Installing
When installing skills, the GIT repository URL must be given. The repository name is considered to be the skill-name, and will be the subdirectory name within the skills-directory.
Installing a skill involves the following steps:
- cloning the remote repository
- creating a virtualenv
- installing dependencies given by the skill developer (
requirements.txt
) - asking the user for configuration parameters, if the skill provides the
config.ini.default
file
The server must be restarted after installing a new skill.
Updating
Updating one or more skills is as easy as pulling changes from the remote GIT repository of the skill.
In addition, hss-cli
will compare the existing config.ini
(if it exists) with a new config.ini.default
, to detect newly added configuration parameters, and then prompt the user for the parameters.
Uninstalling
Uninstalling simply leads to the deletion of the skill's subfolder within the skill-directory. No other actions involved.
Skill development
In order to develop your own skill, check out the hss_skill
package at HSS - Skill. It will give detailed instructions of how to develop skills for the hermes skill server.
Project details
Release history Release notifications | RSS feed
Download files
Download the file for your platform. If you're not sure which to choose, learn more about installing packages.
Source Distribution
Built Distribution
Hashes for hss_server-0.1.2-py3-none-any.whl
Algorithm | Hash digest | |
---|---|---|
SHA256 | 31c7ca28a763c01b305f0e7b8413e267de2f2f6954f70b1370d4a6653caac9de |
|
MD5 | ac2f51ab4621c12d687e1bc413866045 |
|
BLAKE2b-256 | 83c2e0f0de007552b478221a40556c44a4bb74e9e0dd0571b7550b8b632ee2f4 |