Weird! The ^Ms that are getting left behind on the end of the first names suggests that the contents of the data file is weird, though given the source that's no real surprise.
The reason for having the no warnings 'uninitialized' is to avoid the need for a special case to deal with location lines that don't contain the room number. The regex is saying:
m[ ([^/]+) / # capture the phone number before the slash (?: #optionally capture the room number if it is exists (\S+) # capture all the non-spaces between the / \s+ #and one or more spaces )? # but only if there are two sets of non-spaces # separated by a one or more spaces before EOL (\S+) #capture the building code. $ ]x;
If the room number isn't present, the second capture ($2) will be undefined, hence the need to suppress the warning. However, that you are getting three warnings means that all 3 captures are undefined (ie. the regex failed to match), which suggests that the data in the file is formatted somewhat differently to the sample data you posted.
Without being able to see the actual contents of the file it is a little difficult to diagnose the problem.
Perhaps you could run this one liner on the data file to dump the first few lines in hex and post the output here?
perl -nle" exit if $. == 15; print unpack 'H*', $_" testdata.txt
"Think for yourself!" - Abigail
|Replies are listed 'Best First'.|
Re: Re: Re: Re: Need a better way to count input lines
by Theo (Priest) on May 07, 2004 at 22:19 UTC
by BrowserUk (Patriarch) on May 07, 2004 at 22:22 UTC
by Theo (Priest) on May 07, 2004 at 22:36 UTC