Shell Programming and Scripting

I wanted to use GAWK's 'word boundary' feature but can't
get it to work. Doesn't GAWK support \<word\>?

Sample record:

Code:

Title                   Bats in the fifth act of Chushingura (top);
                        the world of the bell - the story of Anchin and Kiyohime (bottom)                               
Series Title            Sketches by Yoshitoshi     
Title-Alternative       Yoshitoshi ryakuga: Komori no godanme (top); Kane no sekai (bottom)

Shouldn't /^\<Title\>/ work to remove "Title-Alternative"? It doesn't. I have to use this:
$1 ~ /^Title$/

Bubnoff

because you are escaping \> and \< whereas in your data, there is no < >

Thanks for answering ghostdog74, however, I'm still a bit unclear on
what you mean. I am aware that I do not have the gt lt characters in the data, I was trying to use GAWK's word boundary operators.

According to the documentation ( GAWK: Effective ...etc. )the regex operators:

\< and \> can be used to indicate word boundaries. They do, but they
use a space as the delimiter ( if I would've RTFMed a bit closer I
would've saved myself this confusion ).

eg. "Title-Alternative" will be true but "TitleAlternative" will be false.

This still makes no sense. How is this working?

I originally thought I could remove "Title-Alternative" by using the word
boundary operators like:

\<Title\>

But since Title-Alternative has a hyphen it's still legal ( why exactly I can't say ). This regex will
remove "titleAlternative" which is closer to the example in the docs, but won't remove "Title-Alternative".

So I think my problem was not fully understanding the way GAWK's W.B.
operators worked ( still don't ).

I am new to AWK and am wondering how others would pull "Title" from
a record that looks similar to what is in my above post.

Code:

 gawk '$1 ~ /^Title$/{print}'

The above works. Any insight into these crazy Word Boundary operators in
GAWK would be much appreciated.

Thanks -

Bub

From GNU Regexp Operators - The GNU Awk User's Guide
So hyphens are out.

Thanks Ygor!

I'm embarrassed to say I read this section at least twice, today alone, and didn't catch that. Its times like these when a person should just step
away from the screen, grab a cup o' joe and go for a walk.

Bub

my bad. didn't see your requirement properly. if you want to get whole words, there is no need for regular expression. Just go through each word and test for it

Code:

awk '{
 for(i=1;i<=NF;i++){
   if( $i == "Title"){ # or ~ /^Title$/
         ........
   }
 }
}

'

Forgot to mention the case possibilities -

Each test has to take into account possible capitalization ( or lack thereof ). So actually, I've been using:

Code:

/^[Tt]itle$/

The background to this is that I have around one hundred Dublin Core
records to analyze and the elements I'm testing for are always in
field $1 with the values in fields $2 or $3. "Title" is one of around 15 DC elements I'm testing for plus or minus the screwy ones people insist on adding. Some catalogers capitalize and other do not.

I could use another Gnu Awk feature though:

Code:

 gawk -v IGNORECASE=1 '$1 == "title"{print $1}' test.notes

- as you suggest, instead of with regex -

 gawk '$1 ~ /^[Tt]itle$/{print}' test.notes

Regex spelling is quicker though.

To distinguish title from:
"Title Alternative" or "title alternative", I am using:

Code:

 gawk '$1 ~ /^[Tt]itle$/&&$2 !~ /[Aa]lternative/{print $1}' test.notes

Thanks for the replies -

Bub