UNIX for Dummies Questions & Answers

Hi All

I have a simple text file and I want to be able to replace any alpha character and comma combination with any other symbol of my choice

here is the text file I want to replace:

Code:

pear
apple
ban,ana
grape
",g1234,"
te,st1

here is how it should look afterwards:

Code:

pear
apple
ban*ana
grape
",g1234,"
te*st1

Notice the commas after the " double-quotes have not been replaced (I want to preserve these)

I want to use sed

so far I have managed to locate the correct strings with the regular expression

Example

Code:

cat list1.txt | sed 's/[a-zA-Z],/ I DONT KNOW WHAT TO PUT HERE /g'

Code:

sed 's/[a-zA-Z],/

(this bit works - but what do you put in the replace part???

Please help ! -

Greg

Hi, here is one possibility with the sample you provided:

Code:

awk -F\" '{gsub(/,/,"*",$1)}1' OFS=\" infile

*Edit oops I noticed you are looking for a sedway...

How about this for starters:

Code:

sed 's/^\([^"]*\),\([^"]*\)$/\1*\2/' infile

or perhaps this:

Code:

sed '/"/!s/,/*/g' infile

You can use \(...\) to save text which matches part of the search expression and then \<N> (Nth bracketed expression) in the replacement text, e.g.

Code:

sed 's/\([a-zA-Z]\),/\1/

Thanks Scrutinizer and CarloM

Both methods work -

Scrutinizer's does the 'replace' which is what I'm after - so thanks very much for that Scrutinizer ...

Thanks CarloM - this replace function will definitely come in useful ...


Scrutizer - I will attempt also to not allow replaces in the case of

01/01/01,,,,,,,01/01/01

I.e. new requirement: 1 or more commas should not be replaced between 2 dates in the file. I will run some tests to make sure I can trap these occurrences.

Cheers !
Greg

What are you actually trying to do? If it's parsing an RFC4180-style CSV file it can be rather complicated to handle all the different cases.