Tobias Salzmann

My way of using Signal-CLI-Rest-API to send Editable Messages in Home Assistant

Tue, 27 May 2025 00:00:00 +0000

There are multiple ways to send a Signal message in home assistant.
This post describes my way of sending Editable Signal messages using the signal-cli-rest-api.
I choose to use the REST integration to build a script that is able to send messages while can also edit sent messages.
The default way using notify does not allow you to edit messages, so this is a workaround to send messages that are updatable.

Prerequisites

A running signal-cli-rest-api instance. (preferably secured with a reverse proxy and authentication).

Changes in `configuration.yaml`

rest_command:
    signal_cli:
        url: "http://:signal-rest-api/v2/send"
        method: "POST"
        payload: "{{payload}}"
        username: "hass"
        password: ""
        content_type: "application/json"

Notice: The username and password are optional, but I choose to protect the signal-cli-rest-api with a reverse proxy and basic authentication.

Script

Create a new script in your home assistant.

alias: Signal Message
description: Send a message using the Signal CLI Rest API
fields:
  number:
    name: Number
    description: Phone Number to send message from
    required: true
    example: "+01123456789"
    selector:
      text: {}
  recipients:
    name: Recipients
    description: |
      List of recipients to add

      Need to be in the form of

      +01123456789
      group.123456789
    required: true
    selector:
      text:
        multiple: true
  message:
    name: Message (Optional)
    description: Message to send
    required: false
    example: Hello World
    selector:
      text:
        multiline: true
  attachments:
    name: Attachments (Optional)
    description: |
      Attachments to add

      Need to be in the form of

      <BASE64 ENCODED DATA>
      data:<MIME-TYPE>;base64<comma><BASE64 ENCODED DATA>
      data:<MIME-TYPE>;filename=<FILENAME>;base64<comma><BASE64 ENCODED DATA>
    required: false
    selector:
      text:
        multiple: true
  load_edit_timestamp_helper:
    name: Load Edit Timestamp from Helper (Optional)
    description: >-
      Load the Edit Timestamp value from the specified helper (This can be used
      to edit a previously sent message)
    required: false
    selector:
      entity:
        filter:
          domain: input_text
  save_edit_timestamp_helper:
    name: Save Timestamp to Helper (Optional)
    description: Set this helper to the Timestamp of the sent message
    required: false
    selector:
      entity:
        filter:
          domain: input_text
sequence:
  - variables:
      base_payload:
        recipients: "{{recipients}}"
      message_payload: |-
        {% if message and message != "" %}
          {"message": "{{message}}"}
        {% else %}
          {}
        {% endif %}
      attachments_payload: |-
        {% if attachments|length > 0 %}
          {"base64_attachments": {{attachments}}}
        {% else %}
          {}
        {% endif %}
      edit_timestamp_payload: |-
        {% if load_edit_timestamp_helper %}
          {"edit_timestamp": {{states(load_edit_timestamp_helper) | int}}}
        {% else %}
          {}
        {% endif %}
      payload: >
        {{ base_payload | combine({"number": number}) | combine(message_payload)
        | combine(attachments_payload) | combine(edit_timestamp_payload) }}
  - action: rest_command.signal_cli
    response_variable: response
    data:
      payload: "{{payload | to_json}}"
  - if:
      - condition: template
        value_template: "{{response['status'] != 201}}"
    then:
      - stop: Status was not 201
        error: true
  - if:
      - condition: template
        value_template: "{{ save_edit_timestamp_helper and save_edit_timestamp_helper != \"\"}}"
    then:
      - action: input_text.set_value
        metadata: {}
        target:
          entity_id: "{{ save_edit_timestamp_helper }}"
        data:
          value: "{{response['content']['timestamp']}}"
mode: single

Usage

From an automation or script you can call the script like this:

action: script.signal_message
metadata: {}
data:
  recipients:
    - group....
  number: "+0123456789"
  message: Hello World

Editing Messages

In order to edit a message you need to use the save_edit_timestamp_helper to save the timestamp of the sent message. When you call the script again you can use the load_edit_timestamp_helper to load the timestamp of the message you want to edit.

Example Automation:

alias: Send Signal Attachment
triggers: []
conditions: []
actions:
  - action: script.signal_message
    metadata: {}
    data:
      recipients:
        - group....
      number: "+123456789"
      message: Hello World
      save_edit_timestamp_helper: input_text.signal_edit_timestamp
    enabled: true
  - delay: "00:00:05"
  - action: script.signal_message
    metadata: {}
    data:
      recipients:
        - group....
      number: "+123456789"
      message: Hello Earth
      load_edit_timestamp_helper: input_text.signal_edit_timestamp
    enabled: true
mode: single

Attachments

A big caveat is that attachments cannot be updated!

You can add attachments to the message by using the attachments field. I wanted to send videos, so I a base64 command to the configuration.yaml:

shell_command:
  base64: "base64 "

Then I used following automation:

alias: Send Signal Attachment
triggers: []
conditions: []
actions:
  - action: shell_command.base64
    metadata: {}
    data:
      args: "-i /config/www/previews/cam2.20250526-102212.mp4"
    response_variable: response
  - if:
      - condition: template
        value_template: "{{response['returncode'] != 0 }}"
    then:
      - stop: Return code was not 0
        error: true
  - action: script.signal_message
    metadata: {}
    data:
      recipients:
        - group....
      number: "+123456789"
      message: CAM2 Preview
      attachments:
        - "{{response['stdout']}}"
    enabled: true
mode: single

Plex in docker with binds

Fri, 23 Mar 2018 00:00:00 +0000

The problem is that Plex runs as the plex user and therefore is not possible to access the bindings.

Find /etc/services.d/plex/run and change the line

exec s6-setuidgid plex /bin/sh -c 'LD_LIBRARY_PATH=/usr/lib/plexmediaserver /usr/lib/plexmediaserver/Plex\ Media\ Server'

exec s6-setuidgid root /bin/sh -c 'LD_LIBRARY_PATH=/usr/lib/plexmediaserver /usr/lib/plexmediaserver/Plex\ Media\ Server'

Responsive golang webapp without JavaScript

Fri, 13 Oct 2017 00:00:00 +0000

tl;dr

It is possible to flush the http.ResponseWriter with the help of http.Flusher and create a responsive feedback for users that dont have JavaScript enabled.

Introduction

Recently I developed a mini website that has a database backend with a search form.
The website should be usable for users that do not have JavaScript enabled (Crazy right? ;-)), so no rest API, and no fancy JavaScript frameworks.
I noticed that the search function could take some time, and most of the users (nowadays) expect pages to load instantly.
Thus, it can happen that some users start clicking the search button over and over again, resulting in aborting the current request and asking for a new search. This can lead to massive io overload, since one user triggers (accidently) multiple searches.
Of course we can limit the resources per user, to avoid multiple search instances, however the ux is still horrible.
I wanted something that notifies the user that the search is actually happening.

Disabling the button

Well I could disable the button as soon as the user clicks it, or display a spinner or something similar. However this will likely reqire JavaScript…

Responding directly with a page

The best solution for me is to send a ‘i am working’ page directly after the user triggers the search and keeping the connection open, this allowes me to send an update as soon as the search is done. This update message can be used to hide the working dialog and display the search results.

http.Flusher

By using http.Flusher it is possible to directly flush the contents:

func workHandler(w http.ResponseWriter, req *http.Request) {
    var flusher http.Flusher
    if f, ok := w.(http.Flusher); ok {
        flusher = f
    }
    w.Write([]byte("I am working\n"))
    if flusher != nil {
    	flusher.Flush()
    }
    time.Sleep(1000 * time.Millisecond)
    w.Write([]byte("Still working...\n"))
        if flusher != nil {
    	flusher.Flush()
    }
    w.Write([]byte("I am done"))
}

This allows us to send the header first, maybe some information about the progress a bit later, and the results on the end.
With some CSS it is also possible to show a progressbar, done so in the sample project you can find here.

Generating an rest api with go-gen-api

Wed, 23 Aug 2017 00:00:00 +0000

tl;dr

I generated an api with go-gen-api to speed up my development process.

Introduction

I often work with databases and rest services. Mostly I need to connect both services, doing some tasks on the database via the rest api, e.g. creating, updating or deleting records.
I had like 7 Tables and for each Table I needed to create a create, update, delete CRUD operation (Users, Groups, AccessRoles, Tokens, Resources, …). I don’t know how you create these CRUD operations but I was sick of writing the same code always again:

func Create(w http.ResponseWriter, r *http.Request)  {
    if r.Method != "POST" || r.Method != "PUT" {
    	throw some error
	}
    if err := json.unmarshal(r.Body, &newUserStruct); err != nil {
		throw some error
    }
    if err := validateUserStruct(newUserStruct); err != nil {
        throw some error
    }
    if err := createNewUserOnTable(&newUserStruct); err != nil {
        throw some error
    }
    send everything worked response
}

Writing a generator

So the objective is clear: Write an generator that auto generates all those crud operations

Thoughts

It must be easy, no primary keys, no forign keys, no m2n
Generate go files before the main projects needs them.
Most of the code is the same so use some kind of template, there is text/template, use that
Split the API into two parts RESTAPI and API, so we can use the generator for non rest api projects aswell

Coding

The main challenge was the structuring and templating, during coding I noticed that tables have null values, thats why the generated structs always use pointers.
Also how should the update work? The generator does not know what the primary key is (and I dont want him to know, remember? It should be simple). The solution is to split the request in a find and update field, meaning that the application will use the find field to find the records it needs to update and using the update field to set the values.
Another feature that is required for mostly all requirements is the Hooks feature. It enables you to modify the requested field before it gets passed to the database and at the same time it allows you to set a custom response for the query.

Finalizing

After working a few hours on the generator and I decided it is good enough for my needs, I stitched an example. (You can find it here). I realized that the code is still, well, a lot. At least I saved the whole Marshaling/Unmarshaling and Database operations, so it makes life a bit easier.

So what can we do now?

We have an API that can be used for Database manipulation
We have an REST-API that has an create, delete, update and get function for each specified table.

Sample of generation

type User struct {
	ID       int
	Name     string
	Password string
}

type Group struct {
	ID       int
	Name     string
}

type GroupMember struct {
	ID       int
	UserID   int
	GroupID  int
}

err := gogenapi.Generate(&gogenapi.Config{
	Structs:      []interface{}{&User{}, &Group{}, &GroupMember{}},
	OutputPath:   "gogenapi",
})
if err != nil {
	panic(err)
}

Sample of usage

package main

import (
	"database/sql"
	"log"
	"gogenapi/user" // Path to generated user
	"github.com/gorilla/mux"
	_ "github.com/mattn/go-sqlite3"
)

func main() {
	db, err := sql.Open("sqlite3", "user.sqlite")
	router := mux.NewRouter()
	restAPI := user.NewRestAPI(router.PathPrefix("/user").Subrouter(), user.New(db))
	err = http.ListenAndServe(":8000", router)
	if err != nil {
		log.Fatal(err)
	}
}

Result

You can find the project here and the example here.

Getting the Skype for business status without any api

Wed, 12 Jul 2017 00:00:00 +0000

If you use an Skype for Business onprem server and you want to get the status of all users easily use this database script on your \RTCLOCAL table.

SELECT LOWER(UserAtHost) AS UserAtHost, Status=
	CASE
		WHEN Availability BETWEEN 0 AND 2999 THEN Availability
		WHEN Availability BETWEEN 3000 AND 4499 THEN 'Available'
		WHEN Availability BETWEEN 4500 and 5999 THEN 'Available - Idle'
		WHEN Availability BETWEEN 6000 and 7499 THEN 'Busy'
		WHEN Availability BETWEEN 7500 and 8999 THEN 'Busy - Idle'
		WHEN Availability BETWEEN 9000 and 11999 THEN 'Do not Disturb'
		WHEN Availability BETWEEN 12000 and 14999 THEN 'Be right back'
		WHEN Availability BETWEEN 15000 and 17999 THEN 'Away'
		WHEN Availability >= 18000 THEN 'Offline'
	END,
	LastPubTime
FROM rtc.dbo.Resource Resource
	RIGHT JOIN (
		SELECT
			Instance.PublisherId,
			SUBSTRING(Instance.Data, CHARINDEX('<availability>', Instance.Data) + 14, CHARINDEX('</availability>', Instance.Data) - CHARINDEX('<availability>', Instance.Data) - 14) AS Availability,
			Instance.LastPubTime
		FROM (
				SELECT PublisherId, cast(substring(Data, 0, 256) AS varchar(256)) AS Data, LastPubTime FROM rtcdyn.dbo.PublishedInstance WHERE ContainerNum = 2 AND CategoryId = 4
				UNION ALL
				SELECT PublisherId, cast(substring(Data, 0, 256) AS varchar(256)) AS Data, LastPubTime FROM rtc.dbo.PublishedStaticInstance WHERE ContainerNum = 2 AND CategoryId = 4
			) AS Instance
		WHERE
			CHARINDEX('aggregateState', Data) > 0
) AS UserAndAvailability ON Resource.ResourceId = PublisherId
/*WHERE UserAtHost = 'somebody@example.com' COLLATE SQL_Latin1_General_CP1_CI_AS*/
ORDER BY UserAtHost, LastPubTime DESC

Recap: GoDaddy Bug - Certificates issued without proper domain validation

Sun, 12 Mar 2017 00:00:00 +0000

In late December 2016 I found a bug regarding GoDaddy’s domain validation, I was using Microsoft’s Azure portal as usually and created some certificates for domains. I noticed that some were issued with valid certificates for domains that did not even point to a valid server.
After doing some testing I found that the validation process was broken:

Specify a local address (127.0.0.1) as an A record in the domain dns settings, and you will get a valid certificate

However, after reporting the issue it turns out there is a more: Invalid validation also occurs on an misconfigured web server, that echos GoDaddy’s validation token.

GoDaddy’s validation process looks like this (simplified):

Generate a token and ask the owner to put this token in a textfile (or similar) on their webroot
GoDaddy looks if the token is in that file
If the token was found the valdiation is successful

However if you set up a server that simply echos all data it gets. Such as the Apache Error page…

… the token will be printed and the validation succeeds.

Thats why thousands of certificates have been issued without a propper validation.

Report

I thought (at this time) that Microsoft validates the owner/server and just pulls the certificates directly from GoDaddy. I was not aware that Microsoft is just a reseller. Thats why I reported the bug to Microsoft initially.

Some time passed and Microsofts response was something like “it is not a bug”.
OK, well then I explain some scenarios. MITM attacks, identity theft,…
Time passes and passes…
They responded and and told me they forwarded the issue to GoDaddy, GoDaddy will contact me. So far so good.

Long story short: GoDaddy never contacted me.

By browsing HN, I found out that the issue was allready fixed.

Some more time passed (3 months later) and I signed up on Cobalt and requested an invite to GoDaddy’s program, explaining that I reported the issue a while back.

I explained the Issue in detail again. And recived a bounty.

Timeline


19 Dec 2016	Reported Issue to Microsofts security team
27 Dec 2016	Recived first response
3 Jan 2017	MSRC Case opened
11 Jan 2017	GoDaddy fixes bug, Wayne Tyer (CEO of GoDaddy) explains issue on moz.dev.security.policy
17 Jan 2017	Asked GoDaddy HQ for the status -> No response
10 Mar 2017	GoDaddy adds me to Cobalt program
12 Mar 2017	Explained issue to GoDaddy over Cobalt
17 Apr 2017	Recived $3000 bounty

Press mentions

Jekyll - Low Cost Static pages on Azure

Sun, 21 Aug 2016 16:52:00 +0000

This article gives you instructions on howto host a low cost, static website on Microsofts cloud platform Azure.

Create a new Web App

Create a new Web App using portal.azure.com.
Choose a subdomain and make sure you use/create a App Service plan with the Free (F1) tier.

Preparing the environment

Install Jekyll

As soon as the deployment is done you need to install Jekyll. Goto your Web-App settings, look for Extensions in the Developmenttools section.
Click on Add, search for Jekyll and install the extension.

By the time of writing the installation of the Jekyll extension fails, However it is installed!

Source, Destination and Build folders

Goto:

https://<your-name>.scm.azurewebsites.net/dev/

This brings up the *new* App Serivce Editor where you can edit and upload files.

Bookmark this site, we will use this editor later to create and edit posts.

Move your mouse over WWWROOT on the left sidebar and create three folders:

src - for our source files
dst - for the files generated by Jekyll
build - for our build script (that triggers Jekyll)

To save resources in the free App-Service plan we only build the website on demand, therefore we need a build script that runs as soon as we changed content.

Create a new file in the build folder and name it index.php.
Paste following contents:

<?php
    $jekyll = 'D:\home\SiteExtensions\JekyllExtension\Commands\Ruby-2.2.3\ruby-2.2.2-i386-mingw32\bin\jekyll.bat';
    $srcDir = 'D:\home\site\wwwroot\src';
    $dstDir = 'D:\home\site\wwwroot\dst';
    // ==================================================

    header('Content-Type: text/plain');
    header('Cache-Control: no-cache');
    ini_set('output_buffering', 'off');
    ini_set('zlib.output_compression', false);
    ini_set('implicit_flush', true);
    set_time_limit(0);
    chdir(dirname($jekyll));
    $handle = popen(basename($jekyll).' build -s "'.$srcDir.'" -d "'.$dstDir.'" 2>&1', 'r');
    while(!feof($handle))
    {
        echo fread($handle, 1024);
        flush();
    } 
    pclose($handle);
?>

Now we need to change the root of our Web-App to the newly created dst folder:
Go to your Web-App Application settings and scroll to the Virtual applications and directories section.
Change the physical path from site\wwwroot to site\wwwroot\dst.

Additionally we create a new entry with /some-random-string and site\wwwroot\build. This will point to our build script, so use a truly random (long) sequence of letters and numbers (e.g. lQFtbZ6c1LuOkjhP0JWR), keep this secret!

Back in the App Service Editor we link to this build folder by adding a run.suffix key to the .settings/settings.json file.
If you do not see the file you might create it first. Add the following content:

{
    "run.suffix": "/some-random-string"
}

Uploading sources

Time to upload the content, I assume you have a Jekyll site allready, if not you can download a sample site somewhere.
There are multiple methods to upload the site, I recomend using FTP. You can set an ftp username and password in the settings panel of your Web-App (Settings->Deployment credentials).

Upload all sources to wwwroot/src.

Building your site

Goto the App Service Editor and hit the Play Button, a new window will open and execute the build script.
Your site should now be ready.

Creating a CDN for more speed

Since we have a static website, why not cache everything with a fast cdn?

Create a new CDN-Profile with a provider that suites your needs
Create an Endpoint and point the Origin Type to Web-App, Orgin Hostname to your Web-App instance.
Remember that it could take some time after the Endpoint is created, and available.
Add your domain to the Endpoint

Keep in mind that the CDN providers charge you by bandwith, meaning more traffic means more costs. In this scenario the only cost factor is the CDN, however you could skip the CDN and drop the costs to 0. But keep in mind, that the Free Service has bandwith quotas and your site might go offline.

Enabling SSL for the site

By the time of writing it is not possible to assign a certificate to the custom CDN Domain. However Microsoft plans to add support for this on Q4 2016.

Install Jekyll Extensions

So your site relies on some jekyll extensions such as jemoji or github-pages?
Well there is a solution, goto Kudu (Web-App Settings) and open a shell.
Change the directory to D:\home\SiteExtensions\JekyllExtension\Commands\Ruby-2.2.3\ruby-2.2.2-i386-mingw32\bin and install the extension via gem:

gem install jemoji

Windows Subsystem for Linux: Changing the Shell

Wed, 03 Aug 2016 14:22:22 +0000

So, you installed the Windows 10 anniversary update and rushed directly to install the Windows Subsystem for Linux?

If you wonder how to change the bash shell to another shell there is a simple way of doing it.

chsh is not enough

Running chsh is not enough, we need a little script that starts our desired shell for us.

However, to keep the good practice we are setting the shell using chsh and execute the shell later:

chsh -s /bin/sh $LOGNAME

Executing the new shell from .bashrc

We add this litte script to our ~/.bashrc:

if [ $PPID == 1 ]; then
    SHELL=$(getent passwd $LOGNAME | cut -d: -f7)
    if [ -e $SHELL ]; then
            $SHELL
            exit
    else
            echo "$SHELL" not found!
    fi
fi

It runs our shell that was set by chsh.

Note for the root user it is enough to use chsh, since sudo runs the default shell by itself.

Open Command Window for Files

Mon, 30 May 2016 14:17:25 +0000

You might know a feature in Windows that allows you to open a command window in the current folder:

I wanted a similar command for files, so I created one.
It is using nothing than the integrated tools of Windows.
Using a bit of vbscript and escaping:
https://gist.github.com/Eun/227d4bd5a8ccde1a9a5d1a42ccab01ce

Result:

HTML: t1 a copy and paste safe table

Mon, 30 May 2016 14:17:25 +0000

I recently found a problem in user experience on Firefox (and some Chrome versions).
I was using a normal table syntax like this:

<table>
	<tr>
		<td>Your Information</td>
		<td>Some Information details...</td>
	</tr>
	<tr>
		<td>Some More Important Information</td>
		<td>Moar information...</td>
	</tr>
</table>

As I tried to copy the contents of the second cell in my table, I got following data in my clipboard:

	Some Information details

Notice the TAB char infront of the text?

This appeared in Firefox only (for me).

What is causing this problem?

If you don’t know, there are three ways to select a text.

Double / Triple click.
Double / Triple click on the text and magically all text is selected.
If you copy the text now you won’t see the issue.
Start selecting from the left side of the text.
If you start selecting the text from the left side and end at the last character you will likely see the issue.
Start selecting from the right side of the text.
Same as 2., but start on the right side.

On 2. & 3. the problem is appearing.

You can try it out on this fiddle.

Apparently if you start selecting in another cell or outside the table you will get this problem.

What to do now?

Normally it shouldn’t be a web developers job to fix these kind of issues, but in terms of user experience it is important to fix these kind of issues to keep your project ux friendly.

I needed a simple table with 2 columns so it was relatively easy to provide a fix.
Additionally I needed a solution for coloring the background of each tr diffrently.
Some older browsers does not support nth-child, so I added a fix for that as well.

Here is an example of the css:
https://jsfiddle.net/Lhxpj690/

Tobias Salzmann

My way of using Signal-CLI-Rest-API to send Editable Messages in Home Assistant

Prerequisites

Changes in configuration.yaml

Script

Usage

Editing Messages

Attachments

Plex in docker with binds

Responsive golang webapp without JavaScript

tl;dr

Introduction

Disabling the button

Responding directly with a page

http.Flusher

Generating an rest api with go-gen-api

tl;dr

Introduction

Writing a generator

Thoughts

Coding

Finalizing

So what can we do now?

Sample of generation

Sample of usage

Result

Getting the Skype for business status without any api

Recap: GoDaddy Bug - Certificates issued without proper domain validation

Report

Timeline

Press mentions

Jekyll - Low Cost Static pages on Azure

Create a new Web App

Preparing the environment

Install Jekyll

Source, Destination and Build folders

Uploading sources

Building your site

Creating a CDN for more speed

Enabling SSL for the site

Install Jekyll Extensions

Windows Subsystem for Linux: Changing the Shell

chsh is not enough

Executing the new shell from .bashrc

Open Command Window for Files

HTML: t1 a copy and paste safe table

What is causing this problem?

What to do now?

Changes in `configuration.yaml`