How to create a CSV dynamically with a script?

About

This howto will show you how you can create any resource dynamically with a script.

In this example, we will create a CSV but you can create any type of resource on the fly.

Steps

Pre-requisites

You should have followed the bash script howto to understand how to execute and pass arguments via a command

The bash script that creates the csv

The bash script that will be executed is:

a hello-csv sample script
that
- accepts the number of lines generated as first argument (default to 1)
- is located in the command directory of the howto connection
- outputs a CSV

We can print the content with the data cat command

tabul data cat command/hello-csv@howto

#!/usr/bin/env bash

# The first argument is the number of lines generated, default to 1
LINES=${1:-1}

# Print CSV headers
echo "Id;Greeting"

# Generate the lines
for ((i=1; i<=LINES; i++)); do
  if [ "$i" -gt 1 ]; then
    echo "$i;Hello World $i times"
    continue
  fi
  echo "$i;Hello World $i time"
done

Simple execution

We can execute it to test that it works.

tabul data print --strict-selection '(command/hello-csv@howto)@tmp'
# The quotes are only mandatory in bash because parenthesis are a bash token (ie subshell)
# We execute it with a strict-selection flag so that we get an error if the executable file is not found.

(command/hello-csv@howto)@tmp
lines
--------------------
Id;Greeting
1;Hello World 1 time

Generating 3 lines

To pass the argument, we have created the following command manifest

tabul data cat command/hello-csv--command.yml@howto

kind: command
spec:
  # the runtime data uri
  data-uri: (command/hello-csv@howto)@tmp
  # The data definition (attributes)
  data-def:
    # A list of arguments passed to the command
    arguments:
      - 3 # The first argument, the number of lines
    # Create a file with the CSV extension
    stdout-data-uri: 'execute/${execution_start_time}-${executable_logical_name}.csv@tmp'

Executing it yield 3 lines

tabul data print --strict-selection command/hello-csv--command.yml@howto
# We execute it with a strict-selection flag so that we get an error if the file is not found.

(command/hello-csv@howto)@tmp
Id;Greeting
---------------------
1;Hello World 1 time
2;Hello World 2 times
3;Hello World 3 times

Defining the CSV structure

Unfortunately, we can see that the CSV does not split correctly the columns because the separator is a semicolon ; and not a the default colon , character.

To define the csv structure, we created the following csv manifest

tabul data cat command/hello-csv--csv.yml@howto

kind: csv
spec:
  # the command manifest
  data-uri: hello-csv--command.yml@md
  data-def:
    # the character separator
    delimiter-character: ;
    # the columns to define the data type
    columns:
      - name: Id
        type: integer
      - name: Greeting
        type: varchar

Executing it yield now a CSV with 2 columns

tabul data print --strict-selection command/hello-csv--csv.yml@howto
# We execute it with a strict-selection flag so that we get an error if the file is not found.

execute/20251110-204747-850-hello-csv.csv@tmp
Id   Greeting
--   -------------------
 1   Hello World 1 time
 2   Hello World 2 times
 3   Hello World 3 times

Conclusion

This example shows how you can create dynamic resources.

It's the way to hack Tabulify and because it's based on programing language, the sky is your only limit.