Category: Linux

Using VI to insert spaces at the begining of every line

In the Vi text editor, you can insert two spaces at the beginning of each line using a command called “substitution.” Here’s how you can do it:

when in vi with your text displayed :

Press the Esc key to ensure you’re in command mode.

Enter the following command to perform the substitution:

:%s/^/  /

Press the Enter key to execute the substitution command.

Explanation of the command:

%s indicates that the substitution should be done throughout the entire file.
^ represents the beginning of each line.
/ is the space you want to insert at the beginning of each line. Make sure to type two spaces.


Now, each line in the file should start with two spaces. You can verify this by scrolling through the file or using other search and navigation commands in Vi.

JSON text filtering with jq

I frequently have to intereact with some json files , since bash is the main tool i use day to day i must master the usage of jq, but i noticed that the vast majority of totorials on jq usage just sucks.
I’m going to write some kind of memo to remember the main command that i use most of the time
lets take this dummy contact.json file

{
  "contacts": [
    {
      "name": "John Smith",
      "phone": "(123) 456-7890",
      "email": "john.smith@example.com",
      "sex": "male",
      "position": "Manager",
      "birthday": "1985-06-23",
      "devices": [
        {
          "type": "laptop",
          "model": "MacBook Pro",
          "serial_number": "ABCD1234",
          "issued_date": "2022-01-15"
        },
        {
          "type": "phone",
          "model": "iPhone 12",
          "serial_number": "EFGH5678",
          "issued_date": "2022-01-15"
        }
      ]
    },
    {
      "name": "Jane Doe",
      "phone": "(234) 567-8901",
      "email": "jane.doe@example.com",
      "sex": "female",
      "position": "Sales Representative",
      "birthday": "1990-02-14",
      "devices": [
        {
          "type": "laptop",
          "model": "Dell XPS 13",
          "serial_number": "IJKL9012",
          "issued_date": "2022-03-01"
        }
      ]
    },
    {
      "name": "Bob Johnson",
      "phone": "(345) 678-9012",
      "email": "bob.johnson@example.com",
      "sex": "male",
      "position": "IT Specialist",
      "birthday": "1982-11-30",
      "devices": [
        {
          "type": "laptop",
          "model": "ThinkPad X1 Carbon",
          "serial_number": "MNOP3456",
          "issued_date": "2022-02-10"
        },
        {
          "type": "phone",
          "model": "Samsung Galaxy S21",
          "serial_number": "QRST7890",
          "issued_date": "2022-02-10"
        }
      ]
    }
  ]
}

to list the name of all the male employes in the json file
cat contact.json | jq '.contacts[] | select(.sex=="male") | .name'

this one is truly awesome , display the name of all employes that are more than 35 years old using their birthday.
cat contact.json | jq '.contacts[] | select((.birthday | strptime("%Y-%m-%d") | mktime) <= (now - 35*365*24*60*60)) | .name'

this one is going to list all the phone and their issued date
cat contact.json | jq '.contacts[].devices[] | select(.type=="phone") | {model: .model, issued_date: .issued_date}'

To display the devices issued after 2022-02-12 and their owner’s name, you can use the following jq command:
cat contact.json | jq '.contacts[] | {owner: .name, devices: [.devices[] | select(.issued_date > "2022-02-12") ] } | select(.devices != [])'

ffmpeg : Only keep one language audio track

When downloading movies from the internet, you may notice that some titles include the word “MULTI.” This indicates that the file contains multiple language options.
However, despite the relatively low cost of disk space nowadays, I prefer not to store unnecessary audio tracks that I will never listen to.

That is when the invaluable program named ffmpeg come to the rescue to eliminate these useless tracks.

You can use the following FFmpeg command to keep only the English audio track and copy the rest of the tracks into an MKV file:

ffmpeg -i “in.mkv” -max_muxing_queue_size 1024 -map 0:v -c:v copy -map 0:a:m:language:en -c:a copy -map 0:s? -c:s copy "out.mkv"

Here’s a breakdown of what each part of the command does:

ffmpeg: This is the command that starts the FFmpeg program.
-i “in.mkv”: This specifies the input file, “in.mkv”, that FFmpeg will be working with.
-max_muxing_queue_size 1024: usefull because sometime ffmpeg crash when this is not present.
-map 0:v: This selects the video stream from the input file (stream 0) to include in the output file.
-c:v copy: This copies the video stream from the input file to the output file without any encoding or re-encoding.
-map 0:a:m:language:en: This selects the audio stream from the input file (stream 0), but only if it has the “en” language code. This is useful if there are multiple audio tracks in the input file and you only want to include the English track in the output file.
-c:a copy: This copies the selected audio stream from the input file to the output file without any encoding or re-encoding.
-map 0:s?: This selects any subtitle streams from the input file (stream 0), if they exist. The ? makes the subtitle stream optional, so if there are no subtitles in the input file, this won’t cause an error.
-c:s copy: This copies any selected subtitle streams from the input file to the output file without any encoding or re-encoding.
“out.mkv”: This specifies the output file that FFmpeg will create, “out.mkv”.

There are many language codes available in FFmpeg for selecting audio and subtitle streams based on language. Here are some examples:

English: en
Spanish: es
French: fr
German: de
Italian: it
Portuguese: pt
Russian: ru
Arabic: ar
Chinese: zh
Japanese: ja
Korean: ko
Hindi: hi
Urdu: ur

Using the REST api to read sensors on Home assistant

During one of my little project of making timelapses videos with some IP cameras, I had the following problem : How do you stop capturing new images at night when the camera see nothing,
After all , the sun rise and sun set time are changing everyday ! and i didn’t see a simple way to calculate this.
then i remembered the sun.sun integration in Home assistant.

If i had a simple way to query this integration in my script , i could very simply stop my script when the sun was no longer present in the sky !

The Home assistant REST API

In any home assistant installation there is access to a REST API that lets you do a lot a things , but in my case, I don’t want much , I just want to know if the sun is above or below the horizon.

  1. Get a Authorization: Bearer token
    you first have to generate a token to authenticate your request . You have to go to your user section , this is the circle a the bottom of the toolbar, then at the bottom of the page , you can create a long term token ,
    Please take note of this token because Home assitant can only display it one time , if you loose it , you must recreate an other one.
  2. Then in my script i can use this
  3. it will return either above_horizon or below_horizon, I can then use this in my script to stop the capture when it’s below_horizon

I used jq filter the json result , but , if you can’t or don’t want to install it , you can replace it with this simple awk

Orange Pi Zero 2 : A new Raspberry Pi contender ?

The new Orange Pi Zero 2 Single Board computer

If you were searching for a very small and capable SBC to run some linux app on arm cpu , you were already able to buy the original Orange PI zero ,
but the board has some obvious flaws like .

  • 100mbps ethernet
  • very unreliable wifi
  • no display connector
  • Unreliable power connector (MicroB)

Now after a few years, the company responsible for the original board has created a successor with updated specs.
The size of the board is larger than the first , but this is still one of the smallest SBC available on the market at 60 x 53 mm

Bigger than the original OrangePi Zero

Here are the complete specs :

  • SoC – Allwinner H616 64-bit quad-core Arm Cortex-A53 processor @ up to 1.5 GHz with Arm Mali G31 MP2 GPU
  • Memory – 512MB or 1GB DDR3
  • Storage – MicroSD card slot
  • Wired Networking – Gigabit Ethernet
  • Wireless Networking – Dual-band 802.11 b/g/n/ac WiFi 5 + Bluetooth 5.0
  • USB – 1x USB 2.0 host port, 2x USB 2.0 interfaces via 13-pin header
  • 13-pin header with 2x USB, analog stereo audio output, CVBS video output, IR input, 3x GPIOs
  • 26-pin IO header with I2C, SPI, UART, multiple GPIOs
  • Debugging – 3-pin UART header for serial console
  • Power Supply – USB Type-C port 5V
  • Dimensions – (60 x 53 mm)

All this for a board that cost less than 20euros !
looking at the cpu, we have a quad core that might be able to match the performance of a raspberry pi4 but when you look at the cost , it’s more in competition with the Raspberry PI zero W witch has a lot less features than this board and is way less powerful !

I can completely see that board screwed behind a screen and used has an emulation machine .
Or a IP security camera server using MotionEyeOs
Or connected to a ISP router to run pihole (but i would still prefer the original pi zero for this , this one is too powerful for pihole).
Maybe a domotic server like Domoticz or Jeedom , the possibilities are endless !

The main flaw of this new board is the USB port . It’s only a USB2.0 port , and it will limit pretty severely the kind of projects that you can realistically do on this board,
Even if USB2 still has a 300mbps bandwidth. I would choose a board with USB3 to run a server like OpenMediaVault.

you can buy this board here

Using bash and gatttool to get readings from Xiaomi Mijia LYWSD03MMC Temperature Humidity sensor

There is a new inexpensive Temperature and humidity sensor by xiaomi.
This time is no longer round ,

Xiaomi Mijia LYWSD03MMC Bluetooth 4.2 Temperature Humidity sensor

if you like me would like to get the temperature and humidity data from time to time to import in a graphing tool like grafana, there is a simple solution using classic bash tools and gatttool.
First you have to indentify the mac-address of your little sensor, for this , just make sure the sensor is in range of your linux device ans launch a
hcitool lescann

this command will spit out all the bluetooth devices in range
just find the line with the name of the device and copy the mac address A4:C1:38:8C:77:CA LYWSD03MMC

then you must start a little bash script like this :

#!/bin/bash
bt=$(timeout 15 gatttool -b A4:C1:38:8C:77:CA --char-write-req --handle='0x0038' --value="0100" --listen)
if [ -z "$bt" ]
then
echo "The reading failed"
else
echo "Got data"
echo $bt temphexa=$(echo $bt | awk -F ' ' '{print $12$11}'| tr [:lower:] [:upper:] )

humhexa=$(echo $bt | awk -F ' ' '{print $13}'| tr [:lower:] [:upper:])
temperature100=$(echo "ibase=16; $temphexa" | bc)
humidity=$(echo "ibase=16; $humhexa" | bc)
echo "scale=2;$temperature100/100"|bc
echo $humidity
fi

this is a skeleton that you can improve , but for now it pretty much work like that ,

first it use gatttol to connect to the sensor and listen for 15 sec
During these fifteen seconds , you can be pretty much sure to receive at least some data like this :

Characteristic value was written successfully Notification handle = 0x0036 value: 58 03 47 a0 0b Notification handle = 0x0036 value: 55 03 47 a0 0b

this tell me that during the 15 sec of connection i received the information that i need two times.
what i’m after are the value 58 03 for the temperature , and 47 for the humidity.

the temperature is little endian format its mean that the two group must be inverted before decoding the data. i invert the values with awk and decode them using bc.
bc doesn’t like when the hex values are not capitalised so tr is used to do that.

bc then give you the temperature multiplied by 100. relaunch bc to divice per 100,

For the humidity its simpler , you get the value in one step without inverting anyting

Then you can do what you need with theses two vars , insert then in some database etc ..

there is room from improvement: for example this script is not capable of decoding negative temperature.
i will post an improved version when i figure out how to do it

Using a old intel cpu cooler on the raspberry pi 4.

Intel cpu cooler on raspberry pi

The raspberry pi 4 has one of the fastest CPU on a SBC , but it doesn’t come with any cooler ,
The raspberry is indeed capable of running without a cooler , but if you do anything other than idling, or something very lightweight like pihole, you are going to thermal throttle.

Since i have a old intel cpu cooler laying around i figured i could adapt it to cool my raspberry pi.

A old intel cpu cooler without the fan.

the cooler is way to big to be used the way it was designed. so i need to cut the cooler with a metal saw to make a new flat face , and put part of the cooler on the raspberry pi CPU.

sawing the intel cooler in two parts

after cutting the cooler you can see that i was far from having a flat surface to put in contact with the cpu.

To flatten the surface , I used a sanding disc fixed on my drill

After sanding the bumps , I obtained a kind of flat surface ,
i’m aware that it’s not truly flat , so to remedy that fact , i putted some thermal paste on the cpu surface.
The thermal compound will fill any irregularities enabling a good contact between the cpu and cooler.

Intel cpu cooler on raspberry pi
raspberry with cooler installed

Then to maintain the cooler flat on the cpu surface , i added a rubber band around the raspberry .
It will help put pressure on the radiator keeping it well pressed on the raspberry cpu

With this cooler the idle temp is aroud 50c and jump to 65c under load (ffmepg encoding)
The thermal throttling temperature of 85c is nevec reached

Installing Jellyfin on a Rasberry PI4 (Rasbian Buster)

The openSSL problem is now corrected , you should not be obliged to do what is described in this page

As soon as the raspberrypi foundation announcement of the new raspberry pi was made. I was sure it was going to be a great headless media server for my network.
with the Rasberry pi 3b and the severe lack of bandwidth on the ethernet and the USB2 ports , using the device as a media server wasn’t really practical.
Now that i received my Raspberry pi 4 B 2GB edition i looked at what was available in term of good media server and I heard about Jellyfin . a pretty features full media server .

I first installed the release of raspbian buster lite available on the raspberry pi website , then activated the ssh by creating an empty ssh file in the boot partition, ran sudo apt update && sudo apt upgrade and was soon ready for the next step

After i was ready to install Jellyfin , i went to the official jellyfin website and started reading the installation instructions.
Raspbian is a tuned version of debian so I chose the instructions regarding debian and launched the following commands :

After downloading more or less 400MB, the installation was completed ,
but the server wasn’t responding on the 8096 port !
after launching the command netstat -tlp on my raspberry I figured out the port wasn’t even opened.

Weirdly systemctl status jellyfin.service was telling me the service was active and running ,
shortly after that i located the log of the application ( /var/log/jellyfin/jellyfin )

/var/log/jellyfin/jellyfin

[2019-06-30 12:40:10.020 +01:00] [INF] Jellyfin version: "10.3.5.0"
[2019-06-30 12:40:10.113 +01:00] [INF] Arguments: ["/usr/lib/jellyfin/bin/jellyfin.dll", "--restartpath=/usr/lib/jellyfin/restart.sh", "--ffmpeg=/usr/lib/jellyfin-ffmpeg/ffmpeg", ""]
[2019-06-30 12:40:10.147 +01:00] [INF] Operating system: "Linux" "4.19.50.7"
[2019-06-30 12:40:10.148 +01:00] [INF] Architecture: Arm
[2019-06-30 12:40:10.156 +01:00] [INF] 64-Bit Process: False
[2019-06-30 12:40:10.156 +01:00] [INF] User Interactive: True
[2019-06-30 12:40:10.157 +01:00] [INF] Processor count: 4
[2019-06-30 12:40:10.158 +01:00] [INF] Program data path: "/var/lib/jellyfin"
[2019-06-30 12:40:10.159 +01:00] [INF] Web resources path: "/usr/lib/jellyfin/bin/jellyfin-web/src"
[2019-06-30 12:40:10.159 +01:00] [INF] Application directory: "/usr/lib/jellyfin/bin/"

this log wasn’t helping at all it was just the same 10 line repeating over and over again, i just knew something prevented the launch of the application ,it was a starting loop, but the log wasn’t verbose enough for useful information to be logged.

After unsuccessfully trying to remove and reinstall jellyfin i tried to run the application manually and not from her service.
i first stopped the service by launching


Then launched the app by

and BINGO , i got the message No usable version of the libssl was found

after a little googlefu I figured-out the libssl installed by default in raspbian buster was too new and was causing an issue with jellyfin

my version was ssl 1.1.1c and i needed a 1.0.x version

So , to force my system to use a old version of this library without having to recompile half the OS , i created the file : /etc/apt/sources.list.d/jessie.list
and written inside the following lines :

i then launched the commands

after theses last commands , your raspberrypi is now ready to launch Jellyfin.
you can do it with


you can now access the webserver of your media server on the tcp port 8096