webhookd

README

webhookd

A very simple webhook server to launch shell scripts.

At a glance

Installation

Run the following command:

$ go install github.com/ncarlier/webhookd@latest

Or download the binary regarding your architecture:

$ sudo curl -s https://raw.githubusercontent.com/ncarlier/webhookd/master/install.sh | bash
or
$ curl -sf https://gobinaries.com/ncarlier/webhookd | sh

Or use Docker:

$ docker run -d --name=webhookd \
  -v ${PWD}/scripts:/scripts \
  -p 8080:8080 \
  ncarlier/webhookd

Note: The official Docker image is lightweight and allows to run simple scripts but for more advanced needs you can use the ncarlier/webhookd:edge-distrib image. For example, with this distrib image, you can interact with your Docker daemon using Docker CLI or Docker Compose.

Or use APT:

Finally, it is possible to install Webhookd using the Debian packaging system through this custom repository.

Note: Custom configuration variables can be set into /etc/webhookd.env file. Sytemd service is already set and enable, you just have to start it with systemctl start webhookd.

Configuration

Webhookd can be configured by using command line parameters or by setting environment variables.

Type webhookd -h to display all parameters and related environment variables.

All configuration variables are described in etc/default/webhookd.env file.

Usage

Directory structure

Webhooks are simple scripts within a directory structure.

By default inside the ./scripts directory. You can override the default using the WHD_SCRIPTS environment variable or -script parameter.

Example:

/scripts
|--> /github
  |--> /build.sh
  |--> /deploy.sh
|--> /push.js
|--> /echo.sh
|--> ...

Note: Webhookd is able to run any type of file in this directory as long as the file is executable. For example, you can execute a Node.js file if you give execution rights to the file and add the appropriate #! header (in this case: #!/usr/bin/env node).

You can find sample scripts in the example folder. In particular, examples of integration with Gitlab and Github.

Webhook URL

The directory structure define the webhook URL.

You can omit the script extension. If you do, webhookd will search for a .sh file. If the script exists, the output the will be streamed to the HTTP response.

The streaming technology depends on the HTTP request:

• Server-sent events is used when:
  • Using GET verb
  • Using text/event-stream in Accept request header
• Chunked Transfer Coding is used otherwise.

Example:

The script: ./scripts/foo/bar.sh

#!/bin/bash

echo "foo foo foo"
echo "bar bar bar"

Output using POST or GET (Chunked Transfer Coding):

$ curl -v -XPOST http://localhost:8080/foo/bar
< HTTP/1.1 200 OK
< Content-Type: text/plain; charset=utf-8
< Transfer-Encoding: chunked
< X-Hook-Id: 7
foo foo foo
bar bar bar

Output using GET and Accept header (Server-sent events):

$ curl -v --header "Accept: text/event-stream" -XGET http://localhost:8080/foo/bar
< HTTP/1.1 200 OK
< Content-Type: text/event-stream
< Transfer-Encoding: chunked
< X-Hook-Id: 8
data: foo foo foo

data: bar bar bar

Webhook parameters

You have several ways to provide parameters to your webhook script:

• URL request parameters are converted to script variables
• HTTP headers are converted to script variables
• Request body (depending the Media Type):
  • application/x-www-form-urlencoded: keys and values are converted to script variables
  • text/* or application/json: payload is transmit to the script as first parameter.

Note: Variable name follows "snakecase" naming convention. Therefore the name can be altered. ex: CONTENT-TYPE will become content_type.

Webhookd adds some additional parameters to the script:

• hook_id: hook ID (auto-increment)
• hook_name: hook name
• hook_method: HTTP request method
• x_forwarded_for: client IP
• x_webauth_user: username if authentication is enabled

Example:

The script:

#!/bin/bash

echo "Hook information: name=$hook_name, id=$hook_id, method=$hook_method"
echo "Query parameter: foo=$foo"
echo "Header parameter: user-agent=$user_agent"
echo "Script parameters: $1"

The result:

$ curl --data @test.json -H 'Content-Type: application/json' http://localhost:8080/echo?foo=bar
Hook information: name=echo, id=1, method=POST
Query parameter: foo=bar
Header parameter: user-agent=curl/7.52.1
Script parameter: {"message": "this is a test"}

Webhook timeout configuration

By default a webhook has a timeout of 10 seconds. This timeout is globally configurable by setting the environment variable: WHD_HOOK_TIMEOUT (in seconds).

You can override this global behavior per request by setting the HTTP header: X-Hook-Timeout (in seconds).

Example:

$ curl -H "X-Hook-Timeout: 5" http://localhost:8080/echo?foo=bar

Webhook logs

As mentioned above, web hook logs are stream in real time during the call. However, you can retrieve the logs of a previous call by using the hook ID: http://localhost:8080/<NAME>/<ID>

The hook ID is returned as an HTTP header with the Webhook response: X-Hook-ID

Example:

$ # Call webhook
$ curl -v http://localhost:8080/echo?foo=bar
...
< HTTP/1.1 200 OK
< Content-Type: text/event-stream
< X-Hook-Id: 2
...
$ # Retrieve logs afterwards
$ curl http://localhost:8080/echo/2

If needed, you can also redirect hook logs to the server output (configured by the WHD_HOOK_LOG_OUTPUT environment variable).

Post hook notifications

The output of the script is collected and stored into a log file (configured by the WHD_HOOK_LOG_DIR environment variable).

Once the script is executed, you can send the result and this log file to a notification channel. Currently, only two channels are supported: Email and HTTP.

Notifications configuration can be done as follow:

$ export WHD_NOTIFICATION_URI=http://requestb.in/v9b229v9
$ # or
$ webhookd --notification-uri=http://requestb.in/v9b229v9

Note: Only the output of the script prefixed by "notify:" is sent to the notification channel. If the output does not contain a prefixed line, no notification will be sent.

Example:

#!/bin/bash

echo "notify: Hello World" # Will be notified
echo "Goodbye"             # Will not be notified

You can override the notification prefix by adding prefix as a query parameter to the configuration URL.

Example: http://requestb.in/v9b229v9?prefix="foo:"

HTTP notification

Configuration URI: http://example.org

Options (using query parameters):

• prefix: Prefix to filter output log

The following JSON payload is POST to the target URL:

{
  "id": "42",
  "name": "echo",
  "text": "foo\nbar...\n",
  "error": "Error cause... if present",
}

Note: that because the payload have a text attribute, you can use a Mattermost, Slack or Discord webhook endpoint.

Email notification

Configuration URI: mailto:foo@bar.com

Options (using query parameters):

• prefix: Prefix to filter output log
• smtp: SMTP host to use (by default: localhost:25)
• username: SMTP username (not set by default)
• password: SMTP password (not set by default)
• conn: SMTP connection type (tls, tls-insecure or by default: plain)
• from: Sender email (by default: noreply@nunux.org)
• subject: Email subject (by default: [whd-notification] {name}#{id} {status})

Authentication

You can restrict access to webhooks using HTTP basic authentication.

To activate basic authentication, you have to create a htpasswd file:

$ # create passwd file the user 'api'
$ htpasswd -B -c .htpasswd api

This command will ask for a password and store it in the htpawsswd file.

By default, the daemon will try to load the .htpasswd file. But you can override this behavior by specifying the location of the file:

$ export WHD_PASSWD_FILE=/etc/webhookd/users.htpasswd
$ # or
$ webhookd --passwd-file /etc/webhookd/users.htpasswd

Once configured, you must call webhooks using basic authentication:

$ curl -u api:test -XPOST "http://localhost:8080/echo?msg=hello"

Signature

You can ensure message integrity (and authenticity) by signing HTTP requests.

Webhookd supports 2 signature methods:

• HTTP Signatures
• Ed25519 Signature (used by Discord)

To activate request signature verification, you have to configure the trust store:

$ export WHD_TRUST_STORE_FILE=/etc/webhookd/pubkey.pem
$ # or
$ webhookd --trust-store-file /etc/webhookd/pubkey.pem

Public key is stored in PEM format.

Once configured, you must call webhooks using a valid signature:

# Using HTTP Signature:
$ curl -X POST \
  -H 'Date: <req-date>' \
  -H 'Signature: keyId=<key-id>,algorithm="rsa-sha256",headers="(request-target) date",signature=<signature-string>' \
  -H 'Accept: application/json' \
  "http://localhost:8080/echo?msg=hello"
# or using Ed25519 Signature:
$ curl -X POST \
  -H 'X-Signature-Timestamp: <timestamp>' \
  -H 'X-Signature-Ed25519: <signature-string>' \
  -H 'Accept: application/json' \
  "http://localhost:8080/echo?msg=hello"

You can find a small HTTP client in the "tooling" directory that is capable of forging HTTP signatures.

TLS

You can activate TLS to secure communications:

$ export WHD_TLS=true
$ # or
$ webhookd --tls

By default webhookd is expecting a certificate and key file (./server.pem and ./server.key). You can provide your own certificate and key with -tls-cert-file and -tls-key-file.

Webhookd also support ACME protocol. You can activate ACME by setting a fully qualified domain name:

$ export WHD_TLS=true
$ export WHD_TLS_DOMAIN=hook.example.com
$ # or
$ webhookd --tls --tls-domain=hook.example.com

Note: On *nix, if you want to listen on ports 80 and 443, don't forget to use setcap to privilege the binary:

sudo setcap CAP_NET_BIND_SERVICE+ep webhookd

License

The MIT License (MIT)

See LICENSE to see the full text.

---