如何用sed文件中的某些模式替换单词？

Question

I know how to replace a word with sed. Using

sed -i 's/[old_text]\b/[new_text]/g'

could replace old_text with new_text.

What I want to do is to replace nchar(10) or nchar (10) or nchar ( 10)... with char(10) in a DB schema.

CREATE TABLE Human
(
    ID int,
    Name nchar(10)  --> should be replace with char(10)
);

The format is quite hard to be determined.

I don't know how to set the sed regular expression to archive this.

Answer 1

You can make spaces optional everywhere:

sed -i.bak 's/\<nchar *( *\([0-9]*\) *)/char(\1)/g' file.sql

• \\< is used for word boundary.

Answer 2

Or rather you can open the file in vi and use following in vi cmd mode

%s/nchar/char/g

This replaces in entire file at once

All occurrences of nchar will be replaced in one go.

And to check before replacing all

%s/nchar/char/gc

y for yes

n for not replacing

a for replacing rest all

Answer 3

Using sed and extended regex -r (for clarity). Also -i means "inplace" so no need to create a different output file

sed -i sqlfile -r -e "s/nchar\s*\(\s*([0-9]+)\s*\)/char(\1)/g"