Why isn't this sed command working for the regex?

Question

I want to remove the multi-line comments in Java using sed command. The regex is working if we use character ranges like a-zA-Z0-9 etc. but not in case of word character like \\w\\S .

Test-File content:

hello world
/* abcd
efgh*/
world hello

Command used :

sed -i -e  "s/\/\*[\s\S]\*\///" <file>

Expected results:

hello world
world hello

Actual results:

hello world
 abcd
efgh*/
world hello

Answer 1

You can use grep with a regex ( -E ) and inverted matches ( -v ), ie:

grep -Ev '(\/\*|\*\/)' < text.txt

Output:

hello world
world hello

Answer 2

It is possible with sed, but not so easy.

sed ':x ; /^\/\*/ { N ; s/.*\*\/// ; /^$/d ; bx }' file

• :x is a label
• /^\\/\\*/ is /*
• N append line from the input to the pattern space
• s/.*\\*\\/// replace any {...}*/
• /^$/d remove empty line
• bx jumps unconditional to :x

Find more here in the documentation

• sed: Branching and flow control
• sed: Multiline technique