Caution
Any environment variable prefixed with REACT_APP_
set in the building environment will be accessible by anyone on Pioneer front-end.
To deploy Pioneer on Vercel click on the button bellow:
-
To do so define
REACT_APP_MAINNET_NODE_SOCKET
,REACT_APP_MAINNET_QUERY_NODE
,REACT_APP_MAINNET_QUERY_NODE_SOCKET
, andREACT_APP_MAINNET_MEMBERSHIP_FAUCET_URL
.For example, for the Joystream mainnet:
REACT_APP_MAINNET_NODE_SOCKET=wss://rpc.joystream.org:9944 REACT_APP_MAINNET_QUERY_NODE=https://query.joystream.org/graphql REACT_APP_MAINNET_QUERY_NODE_SOCKET=wss://query.joystream.org/graphql
Additionally a membership faucet, a notification back-end, or an Avatar upload service can be configured with e.g:
REACT_APP_MAINNET_MEMBERSHIP_FAUCET_URL=https://faucet.joystream.org/member-faucet/register`
REACT_APP_MAINNET_BACKEND=https://api-7zai.onrender.com`
REACT_APP_AVATAR_UPLOAD_URL=https://atlas-services.joystream.org/avatars
Important
faucet.joystream.org
requires a specific hCaptcha key to be set in: REACT_APP_CAPTCHA_SITE_KEY
.
-
yarn run build
Will build Pioneer inside the
packages/ui/build
folder.
QUERY_NODE_ENDPOINT=https://<value>
: Query node to fetch from (in most cases this should be:https://query.joystream.org/graphql
).PIONEER_URL=https://<value>
: The URL of the your Pioneer web application.STARTING_BLOCK=<value>
: The block to start fetching the events from (in most cases this should be the current block).EMAIL_SENDER=<value>
: The address to send e-mail with.
Replace the <value>
by the actual values and save the configuration in a file.
To both register users and notify them an email provider needs to be set up. At the moment only SendGrid and Mailgun are supported but custom SMTP configuration is coming soon.
To set up SendGrid:
- Sign up for a SendGrid account.
- Verify your Sender e-mail address Identity.
- Create your SendGrid API key with full access "Mail Send" permissions.
- Add this API key to the file mentioned earlier this value is called
SENDGRID_API_KEY=<value>
To set up Mailgun:
- Sign up for a Mailgun account.
- Add and verify your domain.
- Get your Mailgun private API key from your dashboard.
- Add this API key to the file mentioned earlier this value is called
MAILGUN_API_KEY=<value>
- Also add
MAILGUN_DOMAIN=<value>
to the file and if this is an EU domainMAILGUN_API_URL=https://api.eu.mailgun.net
should be added too.
To deploy on Render click on the button bellow:
Warning
Only Sendgrid is currently supported with the Render deployment.
-
The following deployment instructions are relying on the
joystream/pioneer-backend
docker image. Therefore Docker is a requirement. -
Store the database URL in the file mentioned above. It should look like this:
DATABASE_URL=postgresql://{username}:{password}@{host}:5432/{database name}
-
At this point it should have a all the following configuration.
DATABASE_URL=postgresql://{username}:{password}@{host}:5432/{database name} QUERY_NODE_ENDPOINT=https://<value> PIONEER_URL=https://<value> STARTING_BLOCK=<value> EMAIL_SENDER=<value>
In addition it should also have your e-mail provider configuration with either
SENDGRID_API_KEY
, orMAILGUN_API_KEY
,MAILGUN_DOMAIN
and maybeMAILGUN_API_URL=https://api.eu.mailgun.net
Finally
APP_SECRET_KEY=<secret value>
should be added to the configuration. It's best to pick this value at random and not to store it anywhere else. -
E.g on a private server with the environment variable - mentioned above - set in a file at the path:
/path/to/.env
, the API can be run with:docker run -d -p 80:80 --restart=always --name=api --env-file=/path/to/.env joystream/pioneer-backend
-
E.g still on a private server the notify script can be run every 10 minutes by setting the following line in the crontab:
*/10 * * * * docker run --name=notify --env-file=/path/to/.env joystream/pioneer-backend notify
A "maintenance screen" can temporarily replace the app, in order to occasionally prevent users from using Pioneer (like during sensitive runtime uprade for example).
To activate it: simply set the environment variable REACT_APP_IS_UNDER_MAINTENANCE=true
, then rebuild Pioneer.
To deactivate it: set the environment variable REACT_APP_IS_UNDER_MAINTENANCE=false
(or completely remove this variable), finally rebuild Pioneer.
Undesirable images can be moderated in the following ways:
The most straight forward way is to simply define REACT_APP_BLACKLISTED_IMAGES
with whitespace (e.g " ", "\n", "\t", ...) separated image urls.
REACT_APP_BLACKLISTED_IMAGES="https://example.com/x.png https://example.com/y.png"
Pioneer instances build with this configuration will not display: https://example.com/x.png
nor https://example.com/y.png
.
However even with a small amount of images this method becomes tedious.
A longer term solution is to fetch a blacklist from a url. This url can be defined in IMAGE_SAFETY_BLACKLIST_URL
. The supported response types are:
-
https://example.com/x.png https://example.com/y.png
-
In this case
{ "results": [ { "id": "0", "value": "https://example.com/x.png" }, { "id": "1", "value": "https://example.com/y.png" }, ] }
IMAGE_SAFETY_BLACKLIST_JSON_PATH
will have to be define in order to target the property to extract. In this example:For JSON responses with particularly complex structures, a JSONPath can be set inIMAGE_SAFETY_BLACKLIST_JSON_PATH=value
IMAGE_SAFETY_BLACKLIST_JSON_PATH
.
Note
If both method 1 and 2 are used, both blacklists are merged together.
If needed IMAGE_SAFETY_BLACKLIST_HEADERS
can define the headers (\n
separated) passed to the request made to IMAGE_SAFETY_BLACKLIST_URL
. e.g:
IMAGE_SAFETY_BLACKLIST_HEADERS="Authorization: Bearer API_KEY"
Given an api key keyXYZ
, a base appXYZ
, and a table tableXYZ
with the following structure:
This configuration:
IMAGE_SAFETY_BLACKLIST_JSON_PATH=image
IMAGE_SAFETY_BLACKLIST_URL="https://api.airtable.com/v0/appXYZ/tableXYZ?api_key=keyXYZ&filterByFormula={blacklisted}"
will result in blacklisting https://example.com/x.png
and https://example.com/z.png
on future builds.
Pioneer supports an image reporting user interface, which can be enable in two ways:
This is the easiest and safer method. Just define a url in REACT_APP_IMAGE_REPORT_FORM_URL
. Users will follow it to report images on an external service (form, instant messaging, email, etc...).
Importantly the reported image url and the page it was reported on can be set in the url. e.g:
-
REACT_APP_IMAGE_REPORT_FORM_URL="mailto:you@example.com?subject=Report image&body=Reported: {image}\nFound at: {context}"
-
Given the Airtable form id:
formXYZ
:REACT_APP_IMAGE_REPORT_FORM_URL="https://airtable.com/formXYZ?prefill_image={image}&prefill_page={context}&hide_image=true&hide_page=true"
In both example above {image}
and {context}
will be substituted with the actual url of the reported image and the page it was reported on.
However to avoid taking users out of Pioneer. A url where to POST a image urls to report, can be set in REACT_APP_IMAGE_REPORT_API_URL
. In most cases the structure of the body of the request should also be defined in REACT_APP_IMAGE_SAFETY_REPORT_BODY_TEMPLATE
. Otherwise the image url will be post.
Similarly to the blacklist api, a REACT_APP_IMAGE_SAFETY_REPORT_HEADERS
can define the headers (\n
separated) passed to the request made to REACT_APP_IMAGE_REPORT_API_URL
. e.g:
REACT_APP_IMAGE_SAFETY_REPORT_HEADERS="Authorization Bearer API_KEY\nContent-Type:application/json"
Warning
Unlike fetching the blacklist which happens on the build step, this happens on the front-end. Therefore (just as with any environment variable prefixed with REACT_APP_
) any sensitive information set in REACT_APP_IMAGE_REPORT_API_URL
or REACT_APP_IMAGE_SAFETY_REPORT_HEADERS
will be accessible by anyone.
Caution
No CAPTCHA nor any other way to securely prevent some users from spamming the third party API is currently supported.
Given an api key keyXYZ
(with write access to the table), a base appXYZ
, and a table tableXYZ
with the following structure:
src | page |
---|---|
https://example.com/x.png | https://pioneerapp.xyz/#/forum/thread/1 |
This configuration:
REACT_APP_IMAGE_REPORT_API_URL="https://api.airtable.com/v0/appXYZ/tableXYZ?api_key=keyXYZ"
REACT_APP_IMAGE_SAFETY_REPORT_BODY_TEMPLATE={"records":[{"fields":{"src":"{image}","page":"{context}"}}]}
Will allow users to report image urls and the page they were reported from, into the tableXYZ
.
Caution
Because this is making tableXYZ
public by making keyXYZ
accessible to anyone. The actually blacklisted url, should probably be stored somewhere else (where tableXYZ
cannot write). Otherwise this would allow anyone to blacklist any image.