Regex - Matching Comma Separated List.

[birdeboy]

Hi All,

I am trying to write a shell script that uses a regular expression to match one or more '[a-z]{3}=[0-9]' in a comma separated list which does not end in a comma:

MATCHES:
abc=1
ajk=2,wer=3
bnd=7,djd=9,jkl=4
...etc

NON-MATCHES
abc=1,
dfs=2,ath=3,
ABC=10,
Abc=3
ddc 3
agh=2 , dfd=2
abc
adg3
ab
...etc

Any help will be much appreciated.

[Rubberman]

For a task like this, I will usually write a quick C or Java program to do it. This is basically a parsing job, and bash isn't a great language for parsers.

[ghostdog74]

@op, why is agh=2 , dfd=2 not a match

For a task like this, I will usually write a quick C or Java program to do it. This is basically a parsing job, and bash isn't a great language for parsers.

why would you want to spend time coding C/Java and compiling it when there are tools like awk , Perl , Python or even bash's regular expression can do the job?

awk -F"," '!/,$/{
 for(i=1;i<=NF;i++){
  if ( $0~/[a-z][a-z][a-z]=[0-9]/) {
     print $0
     break
  }
 }
}' file

[Rubberman]

Matter of personal preference I expect. Some jobs call for awk, sed, grep/egrep and all those other great Unix tools. This one was simple enough that I could write a parser in C or C++ just about as quickly. Others would prefer your solution. So, whatever works.

[ghostdog74]

Matter of personal preference I expect. Some jobs call for awk, sed, grep/egrep and all those other great Unix tools. This one was simple enough that I could write a parser in C or C++ just about as quickly. Others would prefer your solution. So, whatever works.

well, awk is written in C with the purpose of parsing files/strings with ease. why then would you want to recreate another "parser" for this particular case?. I am not saying you can't do that, but just not the obvious way to do things.

[Rubberman]

well, awk is written in C with the purpose of parsing files/strings with ease. why then would you want to recreate another "parser" for this particular case?. I am not saying you can't do that, but just not the obvious way to do things.

Awk is a general-purpose tool - it handles lots of stuff at the cost of complexity and (possibly) performance. I'm talking about a narrow-purpose tool that does only one thing, but very efficiently. Again, my point is that this is a decision for the user and what tools they are most comfortable with. For some of these regex pattern matching problems, getting something like awk to do just what you want can take a lot more time than just doing it oneself. Not necessarily - each situation has to be evaluated on its own (de)merits.

And speaking of regex evaluators, I have designed and implemented several, and have ported (and debugged) Henry Spencer's regex library (used in nedit) to a number of platforms. I have also designed and implemented numerous parsers over the years as well as rule-processing languages that can handle extremely complex pattern matching over very large domains (used in manufacturing optimization problems). So, yes I am comfortable with low-level languages such as C, and I recognize that the original poster of this thread might not be. I don't disagree with you, but I am just pointing out that there are a number of valid approaches to solving this problem.

[Rubberman]

And FWIW, I have a great deal of respect for all three of A(ho), W(einberg), and K(ernighan). They are all brilliant lights in the Computer Science field and have each inspired me in my career.