Delete words that aren't in all caps 1

[dodge20]

I would like to delete all the words in a file that aren't in all caps. I would like to keep any numbers that are in the file also. How would I go about doing this?

Dodge20

 

[dodge20]

I suppose I should mention what I have tried.

tr -d '[:lower:]' < input_file

But this only removes the lowercase letters I want to remove words that aren't entirely in caps. So Proper case words should be removed also.

I also tried

 sed -e "s/[a-z]\{2,\}//g"

But this did the same as above.

Dodge20

 

[Annihilannic]

An interesting requirement... why do you need to do this out of curiosity?

Something like this perhaps:

[tt]$ echo 'But this only removes the lowercase letters I want to remove words that arent entirely in caps. So Proper case words should be removed also.' | perl -pe 's/\b *[[:lower:]]+\b//g'
But I. So Proper.
$[/tt]

Annihilannic.

 

[dodge20]

The reason I need this is I have a huge file with a bunch of addresses in it along with a bunch of other junk. The 'other junk' is all in lower or proper case and i need to get rid of it. All of the addresses are in all Caps.

So, you script isn't quite right. From your input "I" would be the only word I would want since it is the only one in all caps.

example:

This is My Address
123 FAKE STREET
MYTOWN, MYSTATE 55555
More junk here
and here

The output should be
MYNAME 123 FAKE STREET
MYTOWN, MYSTATE 55555

There can be any number of lines of junk between the addresses.

Dodge20

 

[Annihilannic]

Oh, sorry, misread the question. Can't you just grep -v '[[:lower:]]' to do that? i.e. remove any lines containing a lowercase character.

Annihilannic.

 

[dodge20]

Thanks - both answers were great.

Dodge20

 

[chipperMDW]

Or, with GNU sed:

$ sed -e 's/\b\w*[a-z]\w*\b//g' <<EOF
> This is My Address
> 123 FAKE STREET
> MYTOWN, MYSTATE 55555
> More junk here
> and here
> Junk Junk Junk CAPITAL Junk Junk
> EOF
   
123 FAKE STREET
MYTOWN, MYSTATE 55555
  
 
   CAPITAL
$

That'll preserve the all-capital words that appear on lines with lowercase characters.

 

[macd68]

More granular and special purpose (imho).

#include <stdio.h>
#include <string.h>
#include <ctype.h>




int main(void) {
int p;

                while ( (p = getc(stdin)) != EOF) {
                      if (islower(p)) {continue;}
                      if (isspace(p) || ispunct(p) ||  p == '\n') { printf("%c",p);}
                      if (isdigit(p) || isupper(p)) {printf("%c",p);}
               }
return 0;
}

 

[quirkasaurus]

Try this:

sed -e 's/[^A-Z0-9 ]//g'