hopla 1.0.4

|Travis|_ |Coveralls|_ |Python27|_ |Python34|_ |PyPi|_

Contents

• Overview

• Usage

• Perspectives

Easy to use pure-Python scheduler. Visit also the API documentation.

Overview

With the increasing amount of data to be treated, efficient scaling strategies are necessary. This observation made me start ‘hopla’ which provides:

• a scheduler that produces human readable outputs.

• a converter that enables to execute kilometer command lines

• workers that enable local or cluster executions.

Usage

The proposed module has been currently developped to execute Python scripts. Consider the following demonstration script that lists an input folder (the latter is available in the ‘hopla.demo.my_ls_script’ module):

#! /usr/bin/env python
##########################################################################
# Hopla - Copyright (C) AGrigis, 2015
# Distributed under the terms of the CeCILL-B license, as published by
# the CEA-CNRS-INRIA. Refer to the LICENSE file or to
# http://www.cecill.info/licences/Licence_CeCILL-B_V1-en.html
# for details.
##########################################################################

# Bredala import
try:
    import bredala
    bredala.USE_PROFILER = False
    bredala.register("os.path", names=["listdir"])
except:
    pass

# System import
from datetime import datetime
import argparse
import os

# Parameters to keep trace
__hopla__ = ["runtime", "inputs", "outputs"]


def is_directory(dirarg):
    """ Type for argparse - checks that directory exists.
    """
    if not os.path.isdir(dirarg):
        raise argparse.ArgumentError(
            "The directory '{0}' does not exist!".format(dirarg))
    return dirarg


parser = argparse.ArgumentParser(description="List a directory.")
required = parser.add_argument_group("required arguments")
required.add_argument(
    "-d", "--dir", dest="dir", required=True, metavar="PATH",
    help="a valid directory to be listed.", type=is_directory)
parser.add_argument(
    "-b", "--fbreak", dest="fbreak", action="store_true",
    help="a activated raise a ValueError.")
required.add_argument(
    "-l", "--mylist", dest="mylist", nargs="+", type=int, required=True,
    help="a list that will be printed.")
parser.add_argument(
    "-v", "--verbose", dest="verbose", type=int, choices=[0, 1, 2], default=0,
    help="increase the verbosity level: 0 silent, [1, 2] verbose.")
args = parser.parse_args()

directory = args.dir
break_flag = args.fbreak
mylist = args.mylist
runtime = {
    "timestamp": datetime.now().isoformat()
}
inputs = {
    "directory": directory,
    "break_flag": break_flag,
    "mylist": mylist
}
outputs = None
if break_flag:
    raise ValueError("BREAK ACTIVATED.")
files = os.listdir(directory)
print("[res] --------", mylist)
if args.verbose > 0:
    print("[res] --------", files)
outputs = {
    "files": files
}

Note the ‘__hopla__’ list that specifies which parameters will be dispalyed in the scheduler execution log. This mechanism is usefull to keep trace of important script elements. The scaled execution of this script on two CPUs is realized using a simple call:

# System import
from pprint import pprint

# Hopla import
from hopla.converter import hopla
import hopla as root

# Define script parameters
apath = os.path.join(os.path.abspath(os.path.dirname(root.__file__)), "demo")
script = os.path.join(apath, "my_ls_script.py")

# Local execution
status, exitcodes = hopla(
    script, hopla_iterative_kwargs=["d", "b"], verbose=0, l=[1, 2],
    b=[False, True, False], d=[apath, apath, apath], hopla_verbose=1,
    hopla_cpus=2)
pprint(status)
pprint(exitcodes)

After the execution call (through the hopla function), exit codes are inspected. The ‘hopla_verbose’ has been set to one, some logging information has been displayed:

2016-08-02 15:59:26,562 - INFO - Using 'hopla' version '1.0.2'.
2016-08-02 15:59:26,562 - INFO - For exitcode values:
    = 0 - no error was produced.
    > 0 - the process had an error, and exited with that code.
    < 0 - the process was killed with a signal of -1 * exitcode.
2016-08-02 15:59:26,927 - INFO - job_0.inputs = {'directory': '/home/ag239446/git/hopla/hopla/demo', 'break_flag': False, 'mylist': [1, 2]}
2016-08-02 15:59:26,928 - INFO - job_0.exitcode = 0
2016-08-02 15:59:26,928 - INFO - job_0.cmd = ['/home/ag239446/git/hopla/hopla/demo/my_ls_script.py', '--dir', '/home/ag239446/git/hopla/hopla/demo', '--mylist', '1', '2', '--verbose', '1']
2016-08-02 15:59:26,928 - INFO - job_0.outputs = {'files': ['my_ls_script.py', 'demo.py']}
2016-08-02 15:59:26,928 - INFO - job_0.runtime = {'timestamp': '2016-08-02T15:59:26.926153'}
2016-08-02 15:59:26,928 - INFO - job_1.inputs = {'directory': '/home/ag239446/git/hopla/hopla/demo', 'break_flag': True, 'mylist': [1, 2]}
2016-08-02 15:59:26,929 - INFO - job_1.exitcode = 1 - 'Traceback (most recent call last):
  File "/home/ag239446/git/hopla/hopla/workers.py", line 70, in worker
    exec(ofile.read(), job_status)
  File "<string>", line 65, in <module>
ValueError: BREAK ACTIVATED.
'
2016-08-02 15:59:26,929 - INFO - job_1.cmd = ['/home/ag239446/git/hopla/hopla/demo/my_ls_script.py', '-b', '--dir', '/home/ag239446/git/hopla/hopla/demo', '--mylist', '1', '2', '--verbose', '1']
2016-08-02 15:59:26,929 - INFO - job_1.outputs = None
2016-08-02 15:59:26,929 - INFO - job_1.runtime = {'timestamp': '2016-08-02T15:59:26.926772'}
2016-08-02 15:59:26,979 - INFO - job_2.inputs = {'directory': '/home/ag239446/git/hopla/hopla/demo', 'break_flag': False, 'mylist': [1, 2]}
2016-08-02 15:59:26,979 - INFO - job_2.exitcode = 0
2016-08-02 15:59:26,979 - INFO - job_2.cmd = ['/home/ag239446/git/hopla/hopla/demo/my_ls_script.py', '--dir', '/home/ag239446/git/hopla/hopla/demo', '--mylist', '1', '2', '--verbose', '1']
2016-08-02 15:59:26,979 - INFO - job_2.outputs = {'files': ['my_ls_script.py', 'demo.py']}
2016-08-02 15:59:26,979 - INFO - job_2.runtime = {'timestamp': '2016-08-02T15:59:26.969334'}
{'job_0': 0, 'job_1': 1, 'job_2': 0}

Perspectives

It will be nice to generalize some concepts (ie., accept different kind of scripts).