Fix Crash when Calculation of size of deleted dirs is asked for

[Atticus1806]

When calling the current code it crashes with:

Calculate size of affected directories? (Y/n): 
---------------------------------------------------------------------------
TypeError                                 Traceback (most recent call last)
Cell In[2], line 1
----> 1 tk.cleaner.cleanup_unused(mode="dryrun")

File ~/src/sisyphus/sisyphus/cleaner.py:316, in cleanup_unused(load_from, job_dirs, mode)
    314     job_dirs = list_all_graph_directories()
    315 to_remove = search_for_unused(job_dirs, verbose=True)
--> 316 remove_directories(to_remove, 'Not used in graph', mode=mode, force=False)

File ~/src/sisyphus/sisyphus/cleaner.py:235, in remove_directories(dirs, message, move_postfix, mode, force)
    233 with tempfile.NamedTemporaryFile() as tmp_file:
    234     for directory in dirs:
--> 235         tmp_file.write(directory + "\x00")
    236     tmp_file.flush()
    237     command = 'du -sch --files0-from=%s' % tmp_file.name

File /usr/lib/python3.10/tempfile.py:622, in _TemporaryFileWrapper.__getattr__.<locals>.func_wrapper(*args, **kwargs)
    620 @_functools.wraps(func)
    621 def func_wrapper(*args, **kwargs):
--> 622     return func(*args, **kwargs)

TypeError: a bytes-like object is required, not 'str'

This fixes this error. Tbh. I am not sure how this has worked before?

[Atticus1806]

Currently using python 3.10, at first I thought the default was maybe changed but according to this the default is the same in 3.6:
https://docs.python.org/3.6/library/tempfile.html

[michelwi]

I am wondering if this is the correct solution. Here we write a list of directories to a file. The directories should be delimited with zero-bytes "\x00" to work with the option --files0-from (cf. https://man7.org/linux/man-pages/man1/du.1.html)

If I see it correctly, the default mode is binary write and you change it to text write. With your proposal I think the lines are delimited with the characters "\x00". So then the du command below should not find the corresponding folders?

I think rather should we change what is written to binary:

Suggested change

	with tempfile.NamedTemporaryFile(mode="w") as tmp_file:
	for directory in dirs:
	tmp_file.write(directory + "\x00")
	with tempfile.NamedTemporaryFile() as tmp_file:
	for directory in dirs:
	tmp_file.write(directory.encode() + b"\x00")

(But please test both variants)

[Atticus1806]

so with my variant I got some result (which is hard to confirm whether correct but looked reasonable)
will test yours now

[michelwi]

it seems that \x will make the next two characters be interpreted as a binary value. So your solution works fine.

[Atticus1806]

i just tested your variant aswell, both produce the same results. Feel free to merge whatever you prefer.

[michelwi]

it seems that \x will make the next two characters be interpreted as a binary value. So your solution works fine.