How to convert this string to an interable list of Python files in Bash? - bash

Question

I have a string as follows:
string = '[ "file1.py", "file2.py", "file3.py", "file4.py" ]'
How to convert it to an iterable list of individual filenames so that I can run a for loop on it as follows:
for filen in fileArr
do
echo $filen
done
Expected output:
file1.py
file2.py
file3.py
file4.py
So far I have just removed the first and last square brackets using string=${string:1:${#string}-2} but I still have quotations and commas to remove. Is there a clean and simple way to achieve this?

Answer 1

You can use the tr command (more information about the tr command here) to eliminate the start bracket, the end bracket, and the double quotations. Also, you can substitute the coma for a tab so you can later iterate the over the result.
Here the code:
$ s='[ "file1.py", "file2.py", "file3.py", "file4.py" ]'
$ s2=$(echo $s | tr "[" " " | tr "]" " " | tr "\"" " " | tr "," "\\t")
$ for x in $s2; do echo $x ; done
file1.py
file2.py
file3.py
file4.py
More in-depth, the s2 statement uses:
s2=$( --> Assigns the output of the full command to s2
echo $s --> Prints the content of s to the pipes
| tr "[" " " --> Substitute the start bracket by spaces
| tr "]" " " --> Substitute the end bracket by spaces
| tr "\"" " " --> Substitute the double quotations by spaces
| tr "," "\\t" --> Substitute the comma by a tab
)
Notice that this code is valid for the kind of input you provided but it will not work if the filenames contain spaces.
EDIT:
Another solution is possible using substring replacement instead of using the tr command.
Here the code:
$ s2=${s//\[/} # Erase the start brackets
$ s3=${s2//\]/} # Erase the end brackets
$ s4=${s3//\"/} # Erase the double quotations
$ s5=${s4//,/ } # Substitute the comma by a tab
$ for x in $s5; do echo $x ; done
file1.py
file2.py
file3.py
file4.py
As the previous solution, notice that the code will not work if the filenames contain spaces (since they will be considered separated entries).
EDIT 2:
As #Z4-tier pointed out, the first option can be re-written in a more compact way using the -d option available in tr. This option erases the given characters. Also, the obtained string after the parsing might not be iterable if the Internal Field Separator (IFS) is set incorrectly. Although I think that the previous solutions cover most of the cases, you might consider setting and restoring the IFS value if it was set to something else than the default value.
Hence, you could write:
$ s='[ "file1.py", "file2.py", "file3.py", "file4.py" ]'
$ s2=$(echo $s | tr -d "[]\"" | tr "," "\\t")
$ IFS=$' '
$ for x in $s2; do echo $x ; done
file1.py
file2.py
file3.py
file4.py
$ IFS= #restore your IFS value

Answer 2

tr <string1> <string2> will replace any occurance of a character in string1 with the character appearing in the same index position from string2, so this can be done with 1 pipe.
tr can also be used as tr -d <string1> where any occurance of a character in string1 is deleted.
s='[ "file1.py", "file2.py", "file3.py", "file4.py" ]'
s2=$(echo $s | tr -d '[]",')
This is not iterable like you might think. It is still one string:
bash-$ echo "${s2[0]}"
file1.py file2.py file3.py file4.py
Try this:
IFS=$' '
bash-$ my_array=($(echo $s | tr -d '[]",'))
bash-$ echo "${my_array[0]}"
file1.py
bash-$ for k in "${my_array[#]}"; do echo $k; done
file1.py
file2.py
file3.py
file4.py
This sets IFS (the internal field separator) to a space character to take advantage of the spaces that are left in the string after it gets run through tr. It uses a subshell to translate s into what we called s2 above (now using my_array to indicate that it's not a string...), surrounds it with (...) to create an indexed array (this is where IFS is important).