Regex pattern for quoted numbers and commas - regex

Question

I'm trying to find the correct regex to search a file for double quoted numbers separated by a comma. For example I'm trying to find "27,422,734" and then replace it in a text editor to correct the comma to be every 4 numbers so the end result would be "2742,2734"
I've tried a few examples I found on SO but none are helping me with this scenario like
"[^"]+"
'\d+'
while the above do find matches, I don't know how to deal with the commas and how what to replace that with.
Thanks for any help!

Answer 1

I found an even shorter solution (works with gnu-sed):
colonmv () {
echo $# | sed 's/,//g' | sed -r ':a;s/\B[0-9]{4}\>/,&/;ta'
}
But attention, the first sed command eats every comma, not just between digits, so improve it or filter your input before.
The second command uses the :a trick.
Read 4 digits, followed by a non digit (>) replace with the same plus comma, when a replacement took place, jump back from ta to :a and repeat.
Now, let's see colonmv in the wild:
colonmv '"A 3-grouped, pretty long number: 5,127,422,734 and an ungrouped one 5678905567789065778"'
"A 3-grouped pretty long number: 51,2742,2734 and an ungrouped one 567,8905,5677,8906,5778"

Answer 2

There might be better way of doing but I propose the following approach:
INPUT:
$ cat to_transform.txt
abc "27,422,734" def"27,422,734" def
ltu "123,734" abc "345,678,123,734" vtu
xtz "345,678,123,734" vtu "345,678,123,734"
u "1" a
"123"
iu"abc"a "123,734"
CMD:
$ paste -d' ' <(grep -oP '(?<=")(:?\d+,\d+)+(?=")' to_transform.txt) <(grep -oP '(?<=")(:?\d+,\d+)+(?=")' to_transform.txt | sed -e 's/,//g;:loop s/\([0-9]\{4\}\)\($\|,\)/\2,\1/g; s/,,/,/g; /\([0-9]\{5\}\)/b loop') | awk '{cmd="sed -i 0,/"$1"/s/" $1 "/" $2 "/ to_transform.txt"; system(cmd)}'
OUTPUT:
$ cat to_transform.txt
abc "2742,2734" def"2742,2734" def
ltu "12,3734" abc "3456,7812,3734" vtu
xtz "3456,7812,3734" vtu "3456,7812,3734"
u "1" a
"123"
iu"abc"a "12,3734"
CODE DETAILS AND EXPLANATIONS:
<(grep -oP '(?<=")(:?\d+,\d+)+(?=")' to_transform.txt) will extract each number to be processed from the input file, the regex used here use lookbehind/lookahead to enforce the surrounded by quotes condition, (:?\d+,\d+)+ is used to extract the numbers like 27,422,734.
the sed command will getting the output from the grep command will then do the following operations:
SED DETAILS:
s/,//g #remove all , in the number
:loop #create a label to loop
s/\([0-9]\{4\}\)\($\|,\)/\2,\1/g #add a coma after every chain of 4 characters starting by the end of the string/or from the latest coma added
s/,,/,/g #remove duplicate comas added by the previous step if any
/\([0-9]\{5\}\)/b loop #if there are at least 5 digits present successively in the string loop and continue the processing.
Temporary output after the paste operation:
27,422,734 2742,2734
27,422,734 2742,2734
123,734 12,3734
345,678,123,734 3456,7812,3734
345,678,123,734 3456,7812,3734
345,678,123,734 3456,7812,3734
123,734 12,3734
Last but not least the awk command will read this file and run some sed command to replace every element of the first column by the corresponding value in the second command: awk '{cmd="sed -i 0,/"$1"/s/" $1 "/" $2 "/ to_transform.txt"; system(cmd)}'.

Answer 3

Precondition: Your input conforms to "[0-9,]*" and is a "#,###"-format correct number.
#!/bin/bash
colonmv () {
echo $1 | sed -r 's/,([0-9]{3})+/\1/g;' | \
rev | sed -r 's/[^0-9]?([0-9]{4})/\1,/g;s/,"$/"/;s/.*/"&/' | rev
}
colonmv '"734"'
colonmv '"2,734"'
colonmv '"22,734"'
colonmv '"422,734"'
colonmv '"7,422,734"'
colonmv '"27,422,734"'
colonmv '"127,422,734"'
colonmv '"5,127,422,734"'
Test:
colonmv.sh
"734""
"2734"
"2,2734"
"42,2734"
"742,2734"
"2742,2734"
"1,2742,2734"
"51,2742,2734"