Stack Overflow首帖：如何用Grep/Sed提取跨多行的{}或[]间内容？

Hey there! Welcome to Stack Overflow—glad you're here asking questions as a Bash beginner, that's the best way to learn. Let's tackle your problem of pulling out text between curly braces {} and square brackets [], even when that content spans multiple lines.

Using Grep (Perl-Compatible Mode) #

Grep usually works line-by-line, but we can tweak it to handle multi-line content with the -z flag (treats the entire input as a single "stream" using null bytes as separators) and -P (enables Perl-compatible regular expressions).

Extract content between curly braces {} #

grep -zPo '(?<=\{).*?(?=\})' your_input_file.txt

• -z: Processes the whole file as one continuous block instead of line-by-line
• -P: Lets us use lookaround assertions to target the content between delimiters
• (?<=\{): Positive lookbehind—locks onto the position right after a {
• .*?: Non-greedy match (stops at the first closing } instead of the last one, avoiding over-matching)
• (?=\}): Positive lookahead—locks onto the position right before a }

Extract content between square brackets [] #

Just swap the braces for brackets in the regex:

grep -zPo '(?<=\[).*?(?=\])' your_input_file.txt

Bonus: Keep the surrounding braces/brackets #

If you want to include the delimiters in the output, skip the lookarounds and match the entire block directly:

# For braces
grep -zPo '\{.*?\}' your_input_file.txt

# For brackets
grep -zPo '\[.*?\]' your_input_file.txt

Using Sed (Multi-Line Mode) #

Sed can handle multi-line content by reading lines into its "pattern space" until it finds the closing delimiter. Here's how to set it up:

Extract content between curly braces {} #

sed -n '/{/{:loop;N;/}/!b loop;s/.*{\(.*\)}.*/\1/p}' your_input_file.txt

• -n: Suppresses default output so we only print the content we care about
• /\{/: Starts processing when a line containing { is found
• :loop: Creates a label to loop back to if we haven't found the closing delimiter yet
• N: Reads the next line into the pattern space (combines lines)
• /}/!b loop: If the pattern space doesn't have a }, jump back to loop and read another line
• s/.*{\(.*\)}.*/\1/p: Replaces everything before { and after } with just the middle content, then prints it

Extract content between square brackets [] #

We need to escape the brackets since they're special characters in sed:

sed -n '/\[/\{:loop;N;/\]/!b loop;s/.*\[\(.*\)\].*/\1/p}' your_input_file.txt

A Quick Heads-Up #

These commands work great for non-nested braces/brackets. If you have nested structures (like { outer { inner } }), regular expressions (and thus grep/sed) will struggle because they can't handle recursive patterns. For nested cases, you'd want to use a tool like awk with custom logic or a language with proper parsing capabilities—but that's a more advanced topic for later!

Hope these solutions help you out—happy scripting!

内容的提问来源于stack exchange，提问作者sebXIII1