sed substitution matches too many inputs

Question

This is my command:

echo "Test" | sed -f <(sed -e 's/.*/s,&,gI/' mydic)

The file mydic contains 2 columns delimited by commas (,)

a,AlphabetA  
.  
.   
.    
e,AlphabetE   
.   
.   
s,AlphabetS  
.   
t,AlphabetT   
test,testedd   
.   
.   
zebra,zebraaaa

The expect result is testedd, but I get AlphabetTAlphabetEAlphabetSAlphabetT.

Peter.O · Accepted Answer · 2012-04-10 16:30:35Z

up vote 2 down vote accepted

echo Test |sed -f <(sed 's/\(.*\),\(.*\)/s,\\<\1\\>,\2,gI/' mydic)

\< and \> indicate the start and end of a word, respectively.

answered Apr 10 '12 at 16:30

Peter.O
8,01531649

That works. Thank you – Runicer Apr 10 '12 at 16:36

ams · Answer 2 · 2012-04-10 15:55:16Z

up vote 1 down vote

You've asked it to replace "e" with "AlphabetE", which it's done correctly. I suspect you meant it to only consider whole words?

Try this:

echo "test" | sed -f <(sed -e 's/.*/s,\\<&\\>,gI/' mydic)

Edit: I noticed you had in fact dealt with the case issue.

answered Apr 10 '12 at 15:55

ams
1,36428

	Yes I meant the whole word and your command result is sed only the first alphabet. Result: AlphabetT>est – Runicer Apr 10 '12 at 16:01
	Ah, I see what I did wrong. Yes, @Peter has made the right adjustment. – ams Apr 10 '12 at 19:00

2 Answers