ctypes-unicode-proclaunch 0.1.0a1

A minimal, robust, and Unicode-aware process launching library using ctypes.

ctypes-unicode-proclaunch

A minimal, robust, and Unicode-aware process launching library using ctypes.

• No:
  • Shell involvement: Arguments are not interpolated or globbed by any shell, ensuring "what you pass is what you get".
  • Python subprocess abstractions: No Popen/communicate/universal_newlines/encoding drama.
• Yes:
  • Native system call usage for maximum control and correctness.
    • CreateProcessW (NT)
    • fork and execve (POSIX)
  • Unicode on both Python 2 and 3.
  • Fully typed.

Features

• Precise Process Spawning: Direct system call wrappers, not wrappers around wrappers.
• Escape-proof Argument Passing: Arguments are not interpreted or glob-expanded.
• Full Unicode Support: Give and receive Unicode paths/arguments everywhere.
• Explicit Environment: Can supply a Unicode environment dictionary. Does not mutate global environment variables.
• Handles Redirection: Pass open file descriptors for stdin, stdout, stderr just like with the C API.

Installation

pip install ctypes-unicode-proclaunch

Usage

Process Launched

A Python script print_argv.py:

# coding=utf-8
from __future__ import print_function
import sys

if __name__ == '__main__':
    for i, arg in enumerate(sys.argv):
        print('sys.argv[%d]=%s' % (i, arg))

NT

# coding=utf-8
from ctypes_unicode_proclaunch import launch, wait

# No `cmd.exe`'s handling of special characters:
process_handle_1 = launch([u'python', u'print_argv.py', u'%USERNAME%'])
wait(process_handle_1)
# sys.argv[0]=print_argv.py
# sys.argv[1]=%USERNAME%

process_handle_2 = launch([u'python', u'print_argv.py', u'Hello', u'&', u'python', u'print_argv.py', u'injected'])
wait(process_handle_2)
# sys.argv[0]=print_argv.py
# sys.argv[1]=Hello
# sys.argv[2]=&
# sys.argv[3]=python
# sys.argv[4]=print_argv.py
# sys.argv[5]=injected

process_handle_3 = launch([u'python', u'print_argv.py', u'*.txt'])
wait(process_handle_3)
# sys.argv[0]=print_argv.py
# sys.argv[1]=*.txt

POSIX

# coding=utf-8
from ctypes_unicode_proclaunch import launch, wait

pid_1 = launch([u'python', u'print_argv.py', u'$HOME'])
wait(pid_1)
# sys.argv[0]=print_argv.py
# sys.argv[1]=$HOME

pid_2 = launch([u'python', u'print_argv.py', u'"test"', u'>', u'$file'])
wait(pid_2)
# sys.argv[0]=print_argv.py
# sys.argv[1]="test"
# sys.argv[2]=>
# sys.argv[3]=$file

pid_3 = launch([u'python', u'print_argv.py', u'*.txt'])
wait(pid_3)
# sys.argv[0]=print_argv.py
# sys.argv[1]=*.txt

Custom Environment and File Redirection

# coding=utf-8
from ctypes_unicode_proclaunch import launch, wait
from read_unicode_environment_variables_dictionary import read_unicode_environment_variables_dictionary

# Doesn't modify `os.environ`
unicode_environment_variables_dictionary = read_unicode_environment_variables_dictionary()
unicode_environment_variables_dictionary[u'COVERAGE_FILE'] = u'test_coverage.sqlite3'

# We pass the file descriptors of the opened files for redirection.
# Just as in the Unix C API.
# The launched processes DO NOT respect the encodings you specified if you opened the files in text mode.
# This means that you should always open files in binary mode (`'rb'`, `'wb'`, etc.) to make that explicit.
with open('test_stdin.txt', 'rb') as f_0, open('test_stdout.txt', 'wb') as f_1, open('test_stderr.txt', 'wb') as f_2:
    pid_or_process_handle = launch(
        [u'python', u'-m', u'coverage', u'run', u'test.py'],
        environment=unicode_environment_variables_dictionary,
        stdin_file_descriptor=f_0.fileno(),
        stdout_file_descriptor=f_1.fileno(),
        stderr_file_descriptor=f_2.fileno()
    )
    wait(pid_or_process_handle)

Contributing

Contributions are welcome! Please submit pull requests or open issues on the GitHub repository.

License

This project is licensed under the MIT License.