Hello Everyone,

I'm hoping i can get some assistance using Sed to do the following.;
"Remove all lines that do NOT contain a period followed by 3 characters (either numbers or letters or spaces...)Basicly, trying to cleanup a windows directory listing so that only files ending with .*** are
left.

I have been able to successfully delete all lines that do NOT contain
a period with the following syntax,

sed '/\./'!d

but I'm unable to insert the necessary wildcards to keep lines that
contain periods immediately followed by three characters..


Thanks for any help you can provide,

Regards,

You can use the -n option to suppress printing unless explicitly done with the 'p' command, and then make a rule which matches the desired line and upon matching calls 'p'.

Thanks for the quick response Matthew.. Yes that is another option but what I'm confused about is what the rule should look like. The only known character is the period. The 3 trailing characters can be anything...

For example;

\directory\subdir\file1.txt - Want to keep this line
\directory\subdir\file1.xls - Want to keep this line

\directory\subdir\subdir.1 - Want to delete this line.
\directory\subdir\subdir - Want to delete this line.

Following the suggestion by matthewg42 you are near to the solution. To build the regular expression you are looking for, take in mind the special meaning of . (dot): it matches any single character. Three consecutive dots match any sequence of three characters.

What about "\directory\subdir\file1.his.txtt" - gotta watch out for "corner" cases.
Ensure the data you are checking is appropriate - say, only at the end of the input.

sed -n '/\....$/p' filename > newfilename

literal "." + any 3 characters--all at the end of the line

Hi Colucix.. I didn't know that about the dot. Thanks, but in my case
I had to escape it with a backslash to get the literal dot... how do you "unescape" and add the remaining dots.


Regards,

Bad, bad pix- we were trying to get the OP to answer his own q.
As pix showed, the escape is for only the one character.

Well, I'm sorry. My homework detector was silent, and it seemed that OP was a real person--one who might go on to learn greater things.

I'll be better next time......unless I'm not..

Just leave them unescaped! You want to match a sequence of 4 characters: a literal dot followed by three other characters and by an end of line (as syg00 pointed out). You have to build a regular expression with 4 + 1 items: a literal (escaped) dot immediately followed by three (unescaped) dots plus the special character which means "end-of-line". Pixellany already posted the solution (just a final twist of the knife... ).

Also note that another method to match literal characters is to embed them in a "character list" using square brackets. So the regular expression could be written
A character list can be used to match any of the listed characters. For example [agk] match any single occurrence of a, g or k. You can also use "intervals" like [0-5] to match any single digit from 0 to 5, or even [a-zA-Z] to match one single alphabetic character, be it upper- or lower-case. Hope this helps.