Awesome
Jean Paul Start
Hell is the .bats
-
Jean-Paul Start is verbose about plugin loading, batch loading and parsing, task execution, ...
-
Be sure to set your logging level to
INFO
if needed
Read this in other languages: Français (by sayanel@github).
Installation
pip install git+https://github.com/Arubinu/jeanpaulstart.git
Usage as a CLI
python -m jeanpaulstart -f path/to/a/batch.yml
Usage from the API
Basic batch run
import jeanpaulstart
jeanpaulstart.run_from_filepath('path/to/a/batch.yml')
Batch run inside a GUI
Tasks can be run step by step, so progress can be forwarded to the user
Here, ui
stands for any class that represents a window or dialog
import jeanpaulstart
executor = jeanpaulstart.executor_from_filepath('path/to/a/batch.yml')
while not executor.has_stopped:
ui.set_status_message(executor.next_task.name)
status, messages = executor.step()
if status != jeanpaulstart.OK:
ui.set_status_message(' '.join([messages[-1], status]))
ui.set_progress(executor.progress)
if executor.success:
ui.set_status_message('Execution OK !')
ui.set_progress(0.0)
Batch Redaction
Batch are redacted in YAML or JSON, in a way similar to Ansible playbooks
A batch describes
- a name
- an icon path
- tags
- an environment (through variables, can be omitted)
- tasks to perform
---
name: Jean Bauchefort
icon_path: $ENVIRONMENT/_config/jeanpaulstart/icons/jean-bauchefort.png
tags:
- production
environment:
PYTHONPATH:
- C:\cube\python\Lib\site-packages
- R:\deploy\cube
tasks:
- name: Running application
raw:
command: python -m jeanbauchefortui
...
For a complete example please read example-batch.yml
User Tags
A YAML file represents how usernames (based on os.getpass.getuser()
) are associated with batch tags
Example
---
production:
- jp.sartre
- p.deproges
graphist_base:
- y.montand
- j.hallyday
rigging:
- s.weaver
- j.rochefort
...
Batches with 'production' tag will show for jp.sartre
- Special character
$
allows to reference a group withing another group
everyone:
- $production
- $graphist_base
- $rigging
- m.polnareff
...
Tasks
Each task is identified by a command name
Tasks are written as plugins, in the package jeanpaulstart.tasks
, they must conform to what the plugin loader expects
Copy
Copies a file
If destination exists and force: no
, nothing will happen
- name: Name of task
copy:
src: /path/to/source.ext
dest: /path/to/destination.ext
force: [yes|no]
File
Creates a file
Use state: directory
if you want to create a directory
Use state: absent
if you want to remove a file, or a directory and its subdirectories
- name: Name of task
file:
path: /path/to/file
state: [directory|absent]
Include Tasks
Runs another batch file
Current environment is passed to executed batch, modifications made by that batch to the environment are not kept
- name: Name
include_tasks:
file: path/to/batch/file.yml
Ini File
Modifies a config file (.ini)
- name: Task Name
ini_file:
src: /path/to/ini/file.ini
state: [present|absent]
section: sectionName
option: optionName
value: valueValue
Pip / Pip3
Runs pip install
/ pip3 install
Please make sure that pip
/ pip3
is in your $PATH
Parameter state
is not mandatory, defaults to present
- name: Task Name
pip:
name: PySide
state: [present|forcereinstall]
- name: Task Name
pip3:
name: PySide
state: [present|forcereinstall]
- name: Task Name
pip:
name: git+http://some/url.git
Raw
Run a command in the terminal
Parameter async
spawns a new process (Popen()
)
Parameter open_terminal
opens a new terminal window
async
defaults to True
open_terminal
defaults to False
If the exit code of the command is 0, jeanpaulstart.OK
is return, else the exit code is returned
If async: yes
, jeanpaulstart.OK
is immediately returned
- name: Launch djv_view
raw:
command: "\"C:\\Program Files\\djv-1.1.0-Windows-64\\bin\\djv_view.exe\""
async: [yes|no]
open_terminal: [yes|no]
Template
Copy a file and applies Jinja2 templating, based on environment variables
Parameter force: yes
will overwrite destination if it already exists
- name: Task Name
template:
src: /path/to/source.ext.j2
dest: /path/to/dest.ext
force: [yes|no]
Url
Opens the given url in the default browser
- name: Task Name
url: http://some.url/
Tasks flags
All tasks have 4 flags that can be set besides the command and its arguements.
register_status
defaults toFalse
orno
catch_exception
defaults toFalse
orno
exit_if_not_ok
defaults toTrue
oryes
when
defaults to"True"
'Register status'
register_status
memorises the status returned by the task, and stores this
status in the environment variable $JPS_REGISTERED_STATUS
, immediately available for the next task
The executor returns the last registered status
'Catch exception' and 'exit if not ok'
All tasks must return OK
, maybe something else if something went wrong.
Most tasks should not raise any exception (exception should be dealt with when writing the task plugin), but it can happen (e.g when a socket is busy, a file missing, ...)
Those two flags are used to control batch execution, when task execution doesn't return OK
,
or when an Exception is raised.
If catch_exception
is set to True
or yes
, exit_if_not_ok
is forced to False
or no
Here is a table of what happens on different cases
Task return | catch_exception=True <br>exit_if_not_ok=False | catch_exception=False <br>exit_if_not_ok=True |
---|---|---|
OK | execution continues | execution continues |
not OK but no Exception is raised | execution continues | batch is exited<br>status is returned |
Exception raised | exception is catched<br>TASK_ERROR_IGNORED is returned<br>execution continues | exception is raised |
When
The when
clause allows to specify a python expression that will be evaluated
If that expression doesn't evaluates to True
, task is skipped
Environment variables can be used. For example
tasks:
- name: Render Image Sequence
raw:
command: $BLENDER_BIN --enable-autoexec ...
async: False
open_terminal: True
register_status: True
- name: Make mov
raw:
command: python -m preview ....
when: $JPS_REGISTERED_STATUS == 'OK'
register_status: True
The "Make mov" task will be run only if the task "Render Image Sequence" returns OK
If the when
clause evaluates to False
and register_status
is on, no status will be
registered