使用 KCL 编写简单配置

1. Introduction

KCL is a simple and easy-to-use configuration language, where users can simply write the reusable configuration code.

In this first codelab, we will learn how to write a simple config with KCL.

Learning this codelab only requires basic programming knowledge, and experience with python will make it even easier.

What We Will Learn

1. Write simple key-value configuration in a programmable way
2. Write simple logic in KCL code
3. Write collections in KCL code
4. Test and debug with KCL code
5. Use built-in support in KCL code
6. Share and reuse KCL code
7. Write config with dynamic input arguments

2. Write Key-Value Pairs

Generate a simple config by creating a my_config.k, we can fill in the following code without strict format which describes the configuration of deploy.

cpu = 256
memory = 512
image = "nginx:1.14.2"
service = "my-service"

In the code above, cpu and memory are declared as int value, while image and service are string literal.

Run with KCL, we will see the generated data in yaml format as below:

KCL command:

kcl my_config.k

Stdout:

cpu: 256
memory: 512
image: nginx:1.14.2
service: my-service

The exported variable is immutable by default so that once it is declared, we can't modify it some where else.

3. Write Simple Logic

Sometimes we want to write a logic in configuration, then we can use:

• Mutable and non-exported variable starting with '_'
• If-else statement

A non-exported variable means it will not appear in the output YAML, and it can be assigned multiple times.

Here is a sample to show how to adjust the resource with conditions.

KCL command:

kcl my_config.k

_priority = 1 # a non-exported and mutable variable
_cpu = 256 # a non-exported and mutable variable

if _priority == 1:
    _cpu = 256
elif _priority == 2:
    _cpu = 512
elif _priority == 3:
    _cpu = 1024
else:
    _cpu = 2048

cpu = _cpu
memory = _cpu * 2
image = "nginx:1.14.2"
service = "my-service"

Run with KCL, we will see the generated data in yaml format as below:

kcl my_config.k

Stdout:

cpu: 256
memory: 512
image: nginx:1.14.2
service: my-service

.. note:: KCL has rich support of operators and string member functions, please read manual and specification for more details.

4. Write Collections

We can use collections to represent complex data types. The collections which are already supported are:

• list
• dict

_priority = 1  # a non-exported and mutable variable
_cpu = 256  # a non-exported and mutable variable

if _priority == 1:
    _cpu = 256
elif _priority == 2:
    _cpu = 512
elif _priority == 3:
    _cpu = 1024
else:
    _cpu = 2048

cpu = _cpu
memory = _cpu * 2
command = ["nginx"] # a list
labels = {run = "my-nginx"} # a dict
image = "nginx:1.14.2"
service = "my-service"

Run with kcl, we will see the generated data as yaml format as below:

KCL command:

kcl my_config.k

Stdout:

cpu: 512
memory: 1024
command:
    - nginx
labels:
    run: my-nginx
image: nginx:1.14.2
service: my-service

Check manual and specification out for more about collection date type and member functions.

5. Append Items Into Collections

We can combine logical expressions, comprehensions, slices, unions and other characteristics to dynamically add elements to the collection

_priority = 1 # a non-exported and mutable variable
_cpu = 256 # a non-exported and mutable variable
_env = "pre-prod"

if _priority == 1:
    _cpu = 256
elif _priority == 2:
    _cpu = 512
elif _priority == 3:
    _cpu = 1024
else:
    _cpu = 2048

cpu = _cpu
memory = _cpu * 2
_command = ["nginx"] # a list
_command = _command + ["-f", "file"]  # Append items into command using + operator to contact two lists
command = [c.lower() for c in _command]  # Take each element in the list to lowercase
_labels = {
    run = "my-nginx"
    if _env:
        env = _env  # Append a dict key-value pair when the _env is not None/Undefined or empty using if expressions
} # a dict
labels = _labels
image = "nginx:1.14.2"
service = "my-service"

Run with kcl, we will see the generated data as yaml format as below:

kcl my_config.k

Stdout:

cpu: 256
memory: 512
command:
- nginx
- -f
- file
labels:
  run: my-nginx
image: nginx:1.14.2
service: my-service

6. Write Assert

To make code testable and robust, we can verify config data with assertions.

_priority = 1 # a non-exported and mutable variable
_cpu = 256 # a non-exported and mutable variable

if _priority == 1:
    _cpu = 256
elif _priority == 2:
    _cpu = 512
elif _priority == 3:
    _cpu = 1024
else:
    _cpu = 2048

cpu = _cpu
memory = _cpu * 2
command = ["nginx"] # a list
labels = {run = "my-nginx"} # a dict
image = "nginx:1.14.2"
service = "my-service"
assert "env" in labels, "env label is a must"
assert cpu >= 256, "cpu cannot be less than 256"

Run with KCL, we will see eval failure with an error message as output as below:

kcl my_config.k

Stderr:

Assertion failure: env label is a must.

After adding env:pre-prod pair into labels, we will get the output as:

cpu: 512
memory: 1024
command:
    - nginx
labels:
    run: my-nginx
    env: pre-prod
image: nginx:1.14.2
service: my-service

7. Use Handy Built-in Support

What's more, we can use built-in functions to help we debug or simplify coding.

_priority = 1  # a non-exported and mutable variable
_cpu = 256  # a non-exported and mutable variable

if _priority == 1:
    _cpu = 256
elif _priority == 2:
    _cpu = 512
elif _priority == 3:
    _cpu = 1024
else:
    _cpu = 2048

_name = "nginx"
# exported variables
cpu = _cpu
memory = _cpu * 2
command = [_name] # a list
labels = {
    run = "my-{}".format(_name)
    env = "pre-prod"
} # a dict
image = "{}:1.14.2".format(_name) # string format
service = "my-service"

# debugging
print(labels) # debugging by print

# test
assert len(labels) > 0, "labels can't be empty" # uselen() to get list length
assert "env" in labels, "env label is a must"
assert cpu >= 256, "cpu cannot be less than 256"

This sample shows how we use format(), len(), print() function to help customize the config.

Run with KCL, we will see the generated data in yaml format as below:

KCL command:

kcl my_config.k

Stdout:

cpu: 512
memory: 1024
command:
    - nginx
labels:
    run: my-nginx
    env: pre-prod
image: nginx:1.14.2
service: my-service
run: my-nginx
env: pre-prod

Note: more built-in functions and modules can be seen in spec/module

8. Reuse Variables in Another Module

To make our code well-organized, we can simply separate our code to my_config.k and my_config_test.k.

Config data defined in my_config.k,

_priority = 1  # a non-exported and mutable variable
_cpu = 256  # a non-exported and mutable variable

if _priority == 1:
    _cpu = 256
elif _priority == 2:
    _cpu = 512
elif _priority == 3:
    _cpu = 1024
else:
    _cpu = 2048
_name = "nginx"

# exported variables
cpu = _cpu
memory = _cpu * 2
command = [_name] # a list
labels = {
    run = "my-{}".format(_name)
    env = "pre-prod"
} # a dict
image = "{}:1.14.2".format(_name) # string format
service = "my-service"

And test code defined in my_config_test.k, in which we can import my_config.k:

import my_config

# debugging
print(my_config.labels) # debugging by print

# test
assert len(my_config.labels) > 0, "labels can't be empty" # use len() to get list length
assert "env" in my_config.labels, "env label is a must"
assert my_config.cpu >= 256, "cpu cannot be less than256"

9. Config with Input Arguments

Sometimes we need to get external input via parameters dynamically from the end user or platform.

In this case, we can pass in priority and env on demand:

• Pass in arguments: -D priority=1 -D env=pre-prod
• Get value by option keyword in KCL code

_priority = option("priority") # a non-exported and mutable variable
_env = option("env") # a non-exported and mutable variable
_cpu = 256 # a non-exported and mutable variable

if _priority == 1:
    _cpu = 256
elif _priority == 2:
    _cpu = 512
elif _priority == 3:
    _cpu = 1024
else:
    _cpu = 2048

_name = "nginx"
# exported variables
cpu = _cpu
memory = _cpu * 2
command = [_name] # a list
labels = {
    run = "my-{}".format(_name)
    env = _env
} # a dict
image = "{}:1.14.2".format(_name) # string format
service = "my-service"

Run with KCL, we will see the generated data in yaml format as below:

kcl my_config.k -D priority=2 -D env=pre-prod

Stdout:

cpu: 512
memory: 1024
command:
    - nginx
labels:
    run: my-nginx
    env: pre-prod
image: nginx:1.14.2
service: my-service

10. Simplify Logic Expression using Dict

When we need to write complex logic, we can use dict to simplify the writing of logic.

_priority = option("priority") # a non-exported and mutable variable
_env = option("env") # a non-exported and mutable variable
_priorityCpuMap = {
    "1" = 256
    "2" = 512
    "3" = 1024
}
# Using a dict to simplify logic and the default value is 2048
_cpu = _priorityCpuMap[_priority] or 2048
_name = "nginx"
# exported variables
cpu = _cpu
memory = _cpu * 2
command = [_name] # a list
labels = {
    run = "my-{}".format(_name)
    env = _env
} # a dict
image = "{}:1.14.2".format(_name) # string format
service = "my-service"

Run with KCL, we will see the generated data in yaml format as below:

KCL command:

kcl my_config.k -D priority=2 -D env=pre-prod

Stdout:

cpu: 512
memory: 1024
command:
    - nginx
labels:
    run: my-nginx
    env: pre-prod
image: nginx:1.14.2
service: my-service

11. The Final Step

Congratulations!

We have completed the first lesson about KCL, we have used KCL to replace our key-value text file to get better programming support.

Please check schema codelab out now to learn how to write an advanced config collaboratively with KCL schema mechanism.