如何在python3中获得glob表达式的反斜杠转义版本

Question

I am building commands in Python, including a glob expression that also contain whitespace characters.

I read that the suggested way is using shlex.quote , but it only works for actual resoved pathnames, because enclosing the expression within ' or " also blocks the shell glob expansion .

Simplest possible example:

touch 'a file.txt'
ls 'a file.*' # this gives an error because the glob is not resolved
# shlex.quote achieves the same result
python3 -c "import os; from shlex import quote; os.system('ls '+quote('a file.*'))"    
ls a\ file.* # this is what I need

I know that I can use any replacement technique to replace spaces ' ' with '\ ' in that string (eg regexps to name one), but I thought that should there already exist a suitable function, I should first look for it rather than writing my own code...

The real case

The commands I am composing, should affect multiple files whose pathname is like (think of it separated by underscores '_' ):

dir/WrongID_Another ID With Spaces_other-stuff-with-hyphens.bmp

I want to obtain commands like:

rename "s/^WrongID_/RightId_/" dir/WrongID_Another\ ID\ With\ Spaces_*.*

... where dir , WrongID , Another ID With Spaces , RightID , are read from a table

Many such commands are written to a single file, a script which I will then feed to bash.

Answer 1

Are you perhaps looking for

while read -p dir wrong rest right; do
    rename "s/^${wrong}_/${right}_/" "$dir/${wrong}_${rest}_"*.*
done <<\____
    one    few     other   many
    two    theyre  stuff   there
    three  grammer correct grammar
____

The braces are necessary to disambiguate ${wrong}_ (the variable wrong followed by an underscore) from $wrong_ (the variable wrong_ with an underscore).

I don't think Python is necessary or useful here, though as you have found, shlex.quote() does provide the functionality you are asking for. If you want to drive this from Python, it's much better to avoid the shell entirely.

import subprocess
import glob

for dir, wrong, rest, right in (
        ('one',   'few',     'other',   'many'),
        ('two',   'theyre',  'stuff',   'there'),
        ('three', 'grammer', 'correct', 'grammar')):
    subprocess.run(['rename', 's/^{0}_/{1}_/'.format(wrong, right)] +
        glob.glob('{0}/{1}_{2}_*.*'.format(dir, wrong, rest))])