temps
is a python module containing context managers for creating and
cleaning up temporary files and directories.
This package is alpha and its API is not stable.
Why do I never use the tempfile
module?
- I am responsible for removing files and dirs tempfile creates, which creates lots of boilerplate whenever I use it.
- There is no context manager for temp dirs.
- When using
mkstemp()
, I get an open file descriptor, not a file object, that I have to close. - The context manager for temp files contains statements like this from the tempfile docs: "Whether the name can be used to open the file a second time, while the named temporary file is still open, varies across platforms"
- I do not get to choose the file perms of files and dirs I create.
What do I like about this module?
- It has a context manager for creating a temp dir and another for temp files.
- The context manager cleans up the dir or file upon context exit, not upon file closure.
- No ambiguity about whether you can or cannot open a file twice.
- You can set the permissions of the temp file or dir to what you want.
- It is very clear what the implementation is:
- directories are created and the path is returned.
- files are not created, since you'll want to do that in a
with open(filename) ...
statement, and the path is returned. - directories and files are cleaned up by the context managers.
- file and dir names are generated using the uuid module, which presumably will avoid race conditions.
Feel free to make a pull request on github.
- Probably Python 2.7 (since that is the only version it has been tested with.)
Download and install using pip:
pip install temps
Using github, one can clone and install a specific version of the package:
cd ~
git clone git@github.com:todddeluca/temps.git
cd temps
python setup.py install
Creating a working dir for subprocesses:
with temps.tmpdir() as workdir:
with open(os.path.join(workdir, 'datafile'), 'wb') as fh:
fh.write(data)
subprocess.call('compute.sh {}'.format(workdir), shell=True)
with open(os.path.join(workdir, 'outfile')) as fh:
print fh.read()
Creating a temp file for a transform and upload:
with temps.tmpfile() as transformed_path:
transform(input_path, transformed_path)
upload(transformed_path, destination)
The default values when parameters are not specified, are stored in variables that are set using environment variables if available or a default value otherwise. Here is a table listing the variable, the environment variable checked, and the default value:
Variable, ENV_VAR, Default
TEMPS_DIR, TEMPS_DIR, os.cwd()
TEMPS_PREFIX, TEMPS_PREFIX, ''
TEMPS_SUFFIX, TEMPS_SUFFIX, ''
TEMPS_MODE, TEMPS_MODE, '0777'