Kimitsune · starze · Jul 13, 2020 · Jul 13, 2020 · Jul 13, 2020 · Jul 13, 2020
diff --git a/.env b/.env
diff --git a/.env.sample b/.env.sample
@@ -0,0 +1,59 @@
+###################################################################################################
+# Generic Settings (devices, sso, market, account)
+###################################################################################################
+
+MYCROFT_DOMAIN=asdf.asdf
+SSO_BASE_URL=sso.asdf.asdf
+OAUTH_BASE_URL=sso.asdf.asdf
+
+SELENE_ENVIRONMENT=prod
+
+DB_HOST=db
+DB_NAME=mycroft
+DB_PORT=5432
+DB_USER=selene
+
+REDIS_HOST=redis
+REDIS_PORT=6379
+
+###################################################################################################
+# Market API Skills Setup (skills_setup)
+###################################################################################################
+
+GITHUB_USER=<todo>
+GITHUB_PASSWORD=<todo>
+
+###################################################################################################
+# Device API (devices)
+###################################################################################################
+
+EMAIL_SERVICE_HOST=
+EMAIL_SERVICE_PORT=
+EMAIL_SERVICE_USER=
+EMAIL_SERVICE_PASSWORD=
+
+# The speech to text engine requires a key to Google's STT API.
+GOOGLE_STT_KEY=
+
+# The weather skill requires a key to the Open Weather Map API
+OWM_KEY=
+OWM_URL=https://api.openweathermap.org/data/2.5
+
+# The Wolfram Alpha skill requires an API key to the Wolfram Alpha API
+WOLFRAM_ALPHA_KEY=
+WOLFRAM_ALPHA_URL=https://api.wolframalpha.com
+
+###################################################################################################
+# Single Sign On API (sso)
+###################################################################################################
+
+# The password reset functionality sends an email to the user with a link to reset their password. 
+# Selene uses SendGrid to send these emails so a SendGrid account and API key are required.
+SENDGRID_API_KEY=
+
+# Access to the Github API is required to support logging in with your Github account. 
+# Details can be found here <https://developer.github.com/v3/guides/basics-of-authentication/>
+GITHUB_CLIENT_ID=
+GITHUB_CLIENT_SECRET=
+
+
diff --git a/.gitignore b/.gitignore
@@ -1,2 +1,3 @@
+.env
 storage
-generated
+generated
diff --git a/README.md b/README.md
@@ -6,7 +6,7 @@ Heavily based on the original repo by Dimitris Karakasilis (jimmykarily) at http
 
 ## Goal
 
-Deploy an offline home assistant server as easely as possible.
+Deploy an offline home assistant server as easily as possible.
 
 ## Why
 
@@ -21,20 +21,80 @@ The project that is closer to the desired result is Mycroft (https://mycroft.ai/
 - Internet connection to download docker images and dependencies
 - A good CPU, STT AND TTS are very CPU intensive. (Using a (nvidia/CUDA ?) GPU would speed up the process a lot but I don't know if it works with this setup yet.)
 
-## Getting started
+## Prepare configuration
+
+You can use `.env.sample` as a starting point to create your specific `.env` file. 
+ ```sh
+cp .env.sample .env
+```
+
+To generate all needed passwords in `./generated` you can run:
+```sh
+./setup.sh (only needed once)  
+```
+
+If you do not have a valid domain you can use a fake one like `asdf.asdf` as `MICROFT_DOMAIN`.
+
+Then you should edit your `/etc/hosts` to something like that:
+```
+127.0.0.1 sso.asdf.asdf
+127.0.0.1 account.asdf.asdf
+127.0.0.1 api.asdf.asdf
+127.0.0.1 market.asdf.asdf
+127.0.0.1 home.asdf.asdf
+127.0.0.1 asdf.asdf
+```
+
+## Start Mycroft-Backend
+
+```sh
+docker-compose up -d  
+```
+And everything is ready to go. The setup phase can take a while.  
 
-Edit and complete :  
-
-- config.env
-- config_external_accounts.env
-- .env
+Now you should be able to access your very own Selene Backend (i.e. browse to `https://home.asdf.asdf`)
 
-The .env file is only used to build the images, but the variable it contains must have the same value as the one in config.env.  
 
-Run ./setup.sh (only needed once)  
+## Start Mycroft-Core (client)
+
+```sh
+cd ./client
+docker-compose up -d
+```
+
+The host-connections of `./client/storage/mycroft.conf` are linked to the backend docker containers. If you want to start the client on a different machine you need to adjust these ones and the network link in `./client/docker-compose.yml`.
+
+## Register Device and enjoy your personal assistant
+
+During first startup, Mycroft-Core should tell you a pairing code. Browse to `https://home.asdf.asdf` and register your client.
+
+After the registration you should be able to talk to Mycroft and even access Backend-APIs like WolframAlpha (if you registered a valid API-Key).
+
+
+## Security hint
+
+The reverse proxy of Microft-Backend currently uses a self-signed certificate so we needed to disable client side SSL-Verification. If you are using a reverse proxy delivering a valid certificate we strongly recommend you to re-enable SSL-Verification.
+
+To enable SSL-Verification you simply need to remove the entire `entrypoint` section in `./client/docker-compose.yml` and rebuild the container.
+
+## Development
+
+After changing the domain settings in `.env` you need to clean build everything to 'burn' the new domain settings into the images
+
+```sh
+docker-compose build --no-cache --pull
+```
+
+A completely fresh startup can be achieved by
+
+```sh
+docker-compose down -v
+rm -rf ./generated
+./setup.sh
+docker-compose build --no-cache --pull
+docker-compose up -d
+```
 
-Run docker-compose up -d  
-And everything is ready to go. The setup phase can take a while.  
 ## Troubleshooting
 
 #### Problem: No skills available in the marketplace.  
@@ -44,10 +104,10 @@ And everything is ready to go. The setup phase can take a while.
 - Use uWSGI for both mozilla-tts and deepspeech server instead of directly exposing the flask dev server through nginx.
 - Make the deepspeech server container use the mainstream repo now that it has been updated.
 - core-version should not be hardcoded in docker-compose.yml (get latest somehow?)
-- Add more variables to the .env files. (There are some left in docker-compose.yml)
 - Make TTS and STT use the GPU. Not very difficult to do, but they seem to use CUDA and for now I do not plan on installing anything proprietary on my servers.
 - Include a fix for the problems in the troubleshooting section.  
 - Rewrite the nginx config to expose api only on the required domains ? to make it easier to split this on multiple computers without docker swarm ?  
+- Use .env Variables in wakeword setup
 
 ## Note
 

diff --git a/client/docker-compose.yml b/client/docker-compose.yml
@@ -0,0 +1,27 @@
+version: '3'
+services:
+  mycroft:
+    image: mycroftai/docker-mycroft
+    restart: "on-failure"
+    volumes:
+      - ./storage:/root/.mycroft
+      - ${XDG_RUNTIME_DIR}/pulse/:${XDG_RUNTIME_DIR}/pulse/
+      - ${HOME}/.config/pulse/:/root/.config/pulse/
+    environment:
+      PULSE_SERVER: "unix:${XDG_RUNTIME_DIR}/pulse/native"
+    ports:
+      - "8181:8181"
+    devices:
+      - "/dev/snd:/dev/snd"
+    entrypoint:
+      - bash
+      - -c
+      # Deactivate ssl verification to communicate with servers having a self signed certificate
+      - "sed -i 's/data=data, json=json_body, timeout=(3.05, 15)/data=data, json=json_body, timeout=(3.05, 15), verify=False/' /opt/mycroft/mycroft/api/__init__.py &&
+         /opt/mycroft/startup.sh
+        "
+
+networks:
+  default:
+    external:
+      name: mycroft_offline_web
diff --git a/client/storage/mycroft.conf b/client/storage/mycroft.conf
@@ -0,0 +1,15 @@
+{
+  "max_allowed_core_version": 20.2,
+  "server": {
+    "url": "https://selene-ui-proxy",
+    "version": "v1",
+    "update": true,
+    "metrics": false
+  },
+  "stt": {
+    "module": "deepspeech_server",
+    "deepspeech_server": {
+      "uri": "http://deepspeech:8080/stt"
+    }
+  }
+}
diff --git a/config.env b/config.env
diff --git a/config_external_accounts.env b/config_external_accounts.env
diff --git a/deepspeech/Dockerfile b/deepspeech/Dockerfile
@@ -2,7 +2,11 @@ FROM ubuntu:18.04
 
 # https://pypi.org/project/deepspeech-server/
 
-RUN apt-get update && apt-get install -y python3.7 python python3-pip git wget python3.7-dev && pip3 install deepspeech && pip3 install deepspeech-server && wget https://github.com/mozilla/DeepSpeech/releases/download/v0.7.1/deepspeech-0.7.1-models.pbmm && wget https://github.com/mozilla/DeepSpeech/releases/download/v0.7.1/deepspeech-0.7.1-models.scorer
-RUN wget -O config.json https://raw.githubusercontent.com/kiratsune/deepspeech-server/master/config.sample.json
-
+RUN apt-get update && \
+    apt-get install -y python3.7 python python3-pip git wget python3.7-dev && \
+    pip3 install deepspeech && \
+    pip3 install deepspeech-server && \
+    wget https://github.com/mozilla/DeepSpeech/releases/download/v0.7.1/deepspeech-0.7.1-models.pbmm && \
+    wget https://github.com/mozilla/DeepSpeech/releases/download/v0.7.1/deepspeech-0.7.1-models.scorer
+RUN wget -O config.json https://raw.githubusercontent.com/MainRo/deepspeech-server/master/config.sample.json
 CMD ["deepspeech-server", "--config", "config.json"]