sed find with a regex and replace does not work [duplicate] - regex

Question

I'm trying to refine my code by getting rid of unnecessary white spaces, empty lines, and having parentheses balanced with a space in between them, so:
int a = 4;
if ((a==4) || (b==5))
a++ ;
should change to:
int a = 4;
if ( (a==4) || (b==5) )
a++ ;
It does work for the brackets and empty lines. However, it forgets to reduce the multiple spaces to one space:
int a = 4;
if ( (a==4) || (b==5) )
a++ ;
Here is my script:
#!/bin/bash
# Script to refine code
#
filename=read.txt
sed 's/((/( (/g' $filename > new.txt
mv new.txt $filename
sed 's/))/) )/g' $filename > new.txt
mv new.txt $filename
sed 's/ +/ /g' $filename > new.txt
mv new.txt $filename
sed '/^$/d' $filename > new.txt
mv new.txt $filename
Also, is there a way to make this script more concise, e.g. removing or reducing the number of commands?

Answer 1

If you are using GNU sed then you need to use sed -r which forces sed to use extended regular expressions, including the wanted behavior of +. See man sed:
-r, --regexp-extended
use extended regular expressions in the script.
The same holds if you are using OS X sed, but then you need to use sed -E:
-E Interpret regular expressions as extended (modern) regular expressions
rather than basic regular regular expressions (BRE's).

Answer 2

You have to preceed + with a \, otherwise sed tries to match the character + itself.
To make the script "smarter", you can accumulate all the expressions in one sed:
sed -e 's/((/( (/g' -e 's/))/) )/g' -e 's/ \+/ /g' -e '/^$/d' $filename > new.txt
Some implementations of sed even support the -i option that enables changing the file in place.

Answer 3

Sometimes, -r and -e won't work.
I'm using sed version 4.2.1 and they aren't working for me at all.
A quick hack is to use the * operator instead.
So let's say we want to replace all redundant space characters with a single space:
We'd like to do:
sed 's/ +/ /'
But we can use this instead:
sed 's/ */ /'
(note the double-space)

Answer 4

May not be the cleanest solution. But if you want to avoid -E and -r to remain compatible with both versions of sed, you can do a repeat character cc* - that's 1 c then 0 or more c's == 1 or more c's.
Or just use the BRE syntax, as suggested by #cdarke, to match a specific number or patternsc\{1,\}. The second number after the comma is excluded to mean 1 or more.

Answer 5

This might work for you:
sed -e '/^$/d' -e ':a' -e 's/\([()]\)\1/\1 \1/g' -e 'ta' -e 's/ */ /g' $filename >new.txt

Answer 6

on the bash front;
First I made a script test.sh
cat test.sh
#!/bin/bash
while IFS='' read -r line || [[ -n "$line" ]]; do
echo "Text read from file: $line"
SRC=`echo $line | awk '{print $1}'`
DEST=`echo $line | awk '{print $2}'`
echo "moving $SRC to $DEST"
mv $SRC $DEST || echo "move $SRC to $DEST failed" && exit 1
done < "$1"
then we make a data file and a test file aaa.txt
cat aaa.txt
<tag1>19</tag1>
<tag2>2</tag2>
<tag3>-12</tag3>
<tag4>37</tag4>
<tag5>-41</tag5>
then test and show results.
bash test.sh list.txt
Text read from file: aaa.txt bbb.txt
moving aaa.txt to bbb.txt