Config reference

Config schema

Top-level directives:#

Version#

key: version

type: semver string

Specify minimum required lets version to run this config.

Example:

version: '0.0.20'

Shell#

key: shell

type: string

required: true

Specify shell to use when running commands

Example:

shell: bash

Global env#

key: env

type: map string => string or map with execution mode

Specify global env for all commands.

Env can be declared as static value or with execution mode:

Example:

shell: bash
env:
MY_GLOBAL_ENV: "123"
MY_GLOBAL_ENV_2:
sh: echo "`id`"
MY_GLOBAL_ENV_3:
checksum: [Readme.md, package.json]

Global eval_env#

Deprecated

key: eval_env

type: mapping string => string

Since env now has sh execution mode, eval_env is deprecated.

Specify global eval_env for all commands.

Example:

shell: bash
eval_env:
CURRENT_UID: echo "`id -u`:`id -g`"

Global before#

key: before

type: string

Specify global before script for all commands.

Example:

Run redis with docker-compose using log level ERROR

shell: bash
before:
function @docker-compose() {
docker-compose --log-level ERROR $@
}
export XXX=123
commands:
redis: |
echo $XXX
@docker-compose up redis

Mixins#

key: mixins

type: list of string

Allows to split lets.yaml into mixins (mixin config files).

To make lets.yaml small and readable it is convenient to split main config into many smaller ones and include them

Example:

# in lets.yaml
...
shell: bash
mixins:
- test.yaml
commands:
echo:
cmd: echo Hi
# in test.yaml
...
commands:
test:
cmd: echo Testing...

And lets test works fine.

Ignored mixins#

It is possible to specify mixin file which could not exist. It is convenient when you have git-ignored file where you write your own commands.

To make lets read this mixin just add - prefix to filename

For example:

shell: bash
mixins:
- -my.yaml

Now if my.yaml exists - it will be loaded as a mixin. If it is not exist - lets will skip it.

Commands#

key: commands

type: mapping

required: true

Mapping of all available commands

Example:

commands:
test:
description: Test something

Command directives:#

description#

key: description

type: string

Short description of command - shown in help message

Example:

commands:
test:
description: Test something

cmd#

key: cmd

type:
- string
- array of strings
- map of string => string (experimental)

Actual command to run in shell.

Can be either:

  • a string (also a multiline string)
  • an array of strings - it will allow to append all arguments passed to command as is (see bellow)
  • a map of string => string - this will allow run commands in parallel (experimental)

Example single string:

commands:
test:
description: Test something
cmd: go test ./... -v

Example multiline string:

commands:
test:
description: Test something
cmd: |
echo "Running go tests..."
go test ./... -v

Example array of strings:

commands:
test:
description: Test something
cmd:
- go
- test
- ./...

When run with cmd as array of strings:

lets test -v

the -v will be appended, so the resulting command to run will be go test ./... -v

cmd can be a map (it is experimental feature).

Example of map of string => string

commands:
run:
description: Test something
cmd:
app: npm run app
nginx: docker-compose up nginx
redis: docker-compsoe up redis

There are two flags --only and --exclude you can use with cmd map.

There must be used before command name:

lets --only app run

work_dir#

key: work_dir

type: string

Specify work directory to run in. Path must be relative to project root. Be default command's workdir is project root (where lets.yaml located).

Example:

commands:
run-docs:
description: Run docusaurus documentation live
work_dir: docs
cmd: npm start

shell#

key: shell

type: string

Specify shell to run command in.

Any shell can be used, not only sh-compatible, for example python.

Example:

shell: bash
commands:
run-sh:
shell: /bin/sh
cmd: echo Hi
run-py:
shell: python
cmd: print('hi')

after#

key: after

type: string

Specify script to run after the actual command. May be useful, when we want to cleanup some resources or stop some services

after script is guaranteed to execute if specified, event if cmd exit code is not 0

Example:

commands:
redis:
description: Run redis
cmd: docker-compose up redis
after: docker-compose stop redis
run:
description: Run app and services
cmd:
app: node server.js
redis: docker-compose up redis
after: |
echo Stopping app and redis
docker-compose stop redis

depends#

key: depends

type: array of string or array or object

Specify what commands to run before the actual command. May be useful, when you have one shared command. For example, lets say you have command build, which builds docker image.

Example:

commands:
build:
description: Build docker image
cmd: docker build -t myimg . -f Dockerfile
test:
description: Test something
depends: [build]
cmd: go test ./... -v
fmt:
description: Format the code
depends: [build]
cmd: go fmt

build command will be executed each time you run lets test or lets fmt

Override arguments in depends command#

It is possible to override arguments or env for any commands declared in depends.

For example we want:

  • build command to be executed with --verbose flag in test depends.
  • alarm command to be executed with Something is happening arg and LEVEL: INFO env in test depends.
commands:
greet:
cmd: echo Hi developer
alarm:
options: |
Usage: lets alarm <msg>
env:
LEVEL: DEBUG
cmd: echo Alarm ${LETSOPT_MSG}
build:
description: Build docker image
options: |
lets build [--verbose]
cmd: |
if [[ -n ${LETSOPT_VERBOSE} ]]; then
echo Building docker image
fi
docker build -t myimg . -f Dockerfile
test:
description: Test something
depends:
- greet
- name: alarm
args: Something is happening
env:
LEVEL: INFO
- name: build:
args: [--verbose]
cmd: go test ./... -v

Running lets test will output:

# lets test
# Hi developer
# Something is happening
# Building docker image
# ... continue building docker image

options#

key: options

type: string (multiline string)

One of the most cool things about lets than it has built in docopt parsing. All you need is to write a valid docopt for a command and lets will parse and inject all values for you.

More info http://docopt.org

When parsed, lets will provide two kind of env variables:

  • LETSOPT_<VAR>
  • LETSCLI_<VAR>

How does it work?

Lets see an example:

commands:
echo-env:
description: Echo env vars
options:
Usage: lets [--log-level=<level>] [--debug] <args>...
Options:
<args>... List of required positional args
--log-level,-l Log level
--debug,-d Run with debug
cmd: |
echo ${LETSOPT_ARGS}
app ${LETSCLI_DEBUG}

So here we have:

args - is a list of required positional args

--log-level - is a key-value flag, must be provided with some value

--debug - is a bool flag, if specified, means true, if no specified means false

In the env of cmd command there will be available two types of env variables:

lets echo-env --log-level=info --debug one two three

Parsed and formatted key values

echo LETSOPT_ARGS=${LETSOPT_ARGS} # LETSOPT_ARGS=one two three
echo LETSOPT_LOG_LEVEL=${LETSOPT_LOG_LEVEL} # LETSOPT_LOG_LEVEL=info
echo LETSOPT_DEBUG=${LETSOPT_DEBUG} # LETSOPT_DEBUG=true

Raw flags (useful if for example you want to pass --debug as is)

echo LETSCLI_ARGS=${LETSCLI_ARGS} # LETSCLI_ARGS=one two three
echo LETSCLI_LOG_LEVEL=${LETSCLI_LOG_LEVEL} # LETSCLI_LOG_LEVEL=--log-level info
echo LETSCLI_DEBUG=${LETSCLI_DEBUG} # LETSCLI_DEBUG=--debug

env#

key: env

type: mapping string => string or map with execution mode

Env is as simple as it sounds. Define additional env for a command:

Env can be declared as static value or with execution mode:

Example:

commands:
test:
description: Test something
env:
GO111MODULE: "on"
GOPROXY: https://goproxy.io
MY_ENV_1:
sh: echo "`id`"
MY_ENV_2:
checksum: [Readme.md, package.json]
cmd: go build -o lets *.go

eval_env#

Deprecated

key: eval_env

type: mapping string => string

Since env now has sh execution mode, eval_env is deprecated.

Same as env but allows you to dynamically compute env:

Example:

commands:
test:
description: Test something
eval_env:
CURRENT_UID: echo "`id -u`:`id -g`"
CURRENT_USER_NAME: echo "`id -un`"
cmd: go build -o lets *.go

Value will be executed in shell and result will be saved in env.

checksum#

key: checksum

type: array of string | mapping string => array of string

Checksum used for computing file hashes. It is useful when you depend on some files content changes.

In checksum you can specify:

  • a list of file names
  • a list of file regexp patterns (parsed via go path/filepath.Glob)

or

  • a mapping where key is name of env variable and value is:
    • a list of file names
    • a list of file regexp patterns (parsed via go path/filepath.Glob)

Each time a command runs, lets will calculate the checksum of all files specified in checksum.

Result then can be accessed via LETS_CHECKSUM env variable.

If checksum is a mapping, e.g:

commands:
build:
checksum:
deps:
- package.json
doc:
- Readme.md

Resulting env will be:

  • LETS_CHECKSUM_DEPS - checksum of deps files
  • LETS_CHECKSUM_DOC - checksum of doc files
  • LETS_CHECKSUM - checksum of all checksums (deps and doc)

Checksum is calculated with sha1.

If you specify patterns, lets will try to find all matches and will calculate checksum of that files.

Example:

shell: bash
commands:
app-build:
checksum:
- requirements-*.txt
cmd: |
docker pull myrepo/app:${LETS_CHECKSUM}
docker run --rm myrepo/app${LETS_CHECKSUM} python -m app

persist_checksum#

key: persist_checksum

type: bool

This feature is useful when you want to know that something has changed between two executions of a command.

persist_checksum can be used only if checksum declared for command.

If set to true, each run all calculated checksums will be stored to disk.

After each subsequent run lets will check if new checksum and stored checksum are different.

Result of that check will be exposed via LETS_CHECKSUM_CHANGED and LETS_CHECKSUM_[checksum-name]_CHANGED env variables.

IMPORTANT: New checksum will override old checksum only if cmd has exit code 0

LETS_CHECKSUM_CHANGED will be true after the very first execution, because when you first run command, there is no checksum yet, so we are calculating new checksum - that means that checksum has changed.

Example:

commands:
build:
persist_checksum: true
checksum:
deps:
- package.json
doc:
- Readme.md

Resulting env will be:

  • LETS_CHECKSUM_DEPS - checksum of deps files

  • LETS_CHECKSUM_DOC - checksum of doc files

  • LETS_CHECKSUM - checksum of all checksums (deps and doc)

  • LETS_CHECKSUM_DEPS_CHANGED - is checksum of deps files changed

  • LETS_CHECKSUM_DOC_CHANGED - is checksum of doc files changed

  • LETS_CHECKSUM_CHANGED - is checksum of all checksums (deps and doc) changed

ref#

key: ref

type: string

Experimental feature

NOTE: ref is not compatible (yet) with any directives except args. Actually ref is a special syntax that turns command into reference to command. It may be changed in the future.

Allows to run command with predefined arguments. Before this was implemented, if you had some commmand and wanted same command but with some predefined args, you had to use so called lets-in-lets hack.

Before:

commands:
ls:
cmd: [ls]
ls-table:
cmd: lets ls -l

Now:

commands:
ls:
cmd: [ls]
ls-table:
ref: ls
args: -l

args#

key: args

type: string

Experimental feature

args is used only with ref and allows to set additional positional args to referenced command. See ref example.

Edit this page