请求修正sed命令：修改每行指定匹配模式间的字符

Hey there! Let's sort out that sed command for you. The main issue with your current attempt is that you're using sed's line range syntax (/1,/,/,2/) when you actually need to target a character range within each line — sed treats /a/,/b/ as selecting lines from the first line containing a to the first line containing b, not the text between a and b on a single line. That's why your command wasn't working as expected.

Here's the corrected sed command that does exactly what you need: #

sed -E ':a s/(1, \{[^}]*),([^}]*\}, 2)/\1 \2/; ta; s/(1, \{) /\1/' myfile.txt

Let's break down how it works: #

1. -E: Enables extended regular expressions, making the pattern cleaner to write.
2. :a: Defines a loop label named a (we'll use this to repeat the replacement until all commas are gone).
3. s/(1, \{[^}]*),([^}]*\}, 2)/\1 \2/:
   • Captures the part from 1, { up to a comma, then the rest of the content inside {} until }, 2.
   • Replaces the captured comma with a space. This targets every comma inside the curly braces between 1, and ,2.
4. ta: If the previous replacement was successful (meaning there are still commas left), jump back to label a and repeat the process.
5. s/(1, \{) /\1/: Cleans up the extra space right after the opening { (since your original lines have {, which becomes { after the first replacement, this fixes it to {).

Testing it with your sample input: #

Running this command on your myfile.txt will produce exactly the output you're expecting:

abc, def, 1, {jsdfsd kfgdsf lgfgd}, 2, pqr, stu
abc, def, 1, {jsdfsqwe k fdfsfl}, 2, pqr, stu
abc, def, 1, {asdasdj kgfdgdf ldsfsdf}, 2, pqr, stu
abc, def, 1, {jsds kfdsf fdsl}, 2, pqr, stu

If you prefer a more concise alternative (using Perl, which handles nested/inline ranges more intuitively), you could use:

perl -pe 's/(1, \{)(.*?)(\}, 2)/$1 . $2 =~ s|,| |gr . $3/ge' myfile.txt

This uses Perl's regex substitution with embedded code to replace all commas in the captured {} content with spaces in one go.

内容的提问来源于stack exchange，提问作者Arun Srinivasan