in reply to looking for speed!! large file search and extract

one-liner: 
c:\> perl -n -e "print if /^abcde.+PARTNAME$/" c:\somefile.txt>k:\1\so
+mefile.txt

[download]
or 
c:\> perl -n -e "print if /^abcde/ && /PARTNAME$/" c:\somefile.txt>k:\
+1\somefile.txt

[download]
whatever is faster.

Update: The second one is approx 50% faster.
I tried with a file of 73MB and 900.000 lines, where every second line matches.
one-liner 1 takes 11 seconds, one-liner 2 takes 6 seconds.

Update:
one-liner using 
substr()

c:\> perl -n -e "print if substr($_,0,5) eq q(abcde) && substr($_,-9) 
+eq qq(PARTNAME\n)" c:\somefile.txt>k:\1\somefile.txt

[download]

Replies are listed 'Best First'.
Re^2: looking for speed!! large file search and extract
by Roy Johnson (Monsignor) on Jan 12, 2005 at 16:03 UTC
    I recommend looking at substr and (if on Unix) the egrep utility, too.
    Caution: Contents may have been coded under pressure.
[reply]
      Thanx for answer but now have to make a small change
      I only want to extract the lines on condition that the
      line directly above it starts and ends with the
      following 5 chararacters "xyzdf"
      basically looking for 2 line match
      thanx
[reply]

        A couple minor points (and maybe I'm a bit too new to PM to make the comments):

        1. You probably should have this in a new question, not a reply on the previous question, since it's now a new question.
        2. You probably should try something yourself, and then come back if it doesn't work. Or even if it does - share your answer and get feedback on it.
        One WTDI is to use a simplified state machine: 
        my $match;
while (<FH>)
{
  print C $_ if $match and /^abcde.*PARTNAME$/;
  $match = /xyzdf$/;
}

        [download]
        This will set $match to true if the current line matches xyzdf at the end, false otherwise. The next time through the loop, we only check your second-line regexp if $match is already true (that is, the previous line matched the other regexp).

[reply]
[d/l]

        then change holli's command line statement from perl -n -e "print if /^abcde/ && /PARTNAME$/" c:\somefile.txt>k:\1\somefile.txt to perl -n -e "print if /^xyzdf/ && /xyzdf$/" c:\somefile.txt>k:\1\somefile.txt

        If you don't understand this I really recommend you read perlre

        Update: Somehow missed reading "line directly above", so ignore the rest...except for reading perlre that's always a good idea if you haven't

        "Cogito cogito ergo cogito sum - I think that I think, therefore I think that I am." Ambrose Bierce

[reply]
[d/l]
[select]
        if i get your comment right, this could be: 
        c:\> perl -n -e "print $last, $_ if /xyzdf$/ && $last; $last=  /^xyzdf
+/ ? $_ : ''" file1>file2

        [download]
        Assuming file1 looks like 
        abc
xyzdf def
hij xyzdf
klm
xyzdf nop
qrs xyzdf

        [download]
        file2 will end up as 
        xyzdf def
hij xyzdf
xyzdf nop
qrs xyzdf

        [download]
        Is that what you want?

        Update:
        if not, post some sample data and the desired output.
[reply]
[d/l]
[select]

In Section Seekers of Perl Wisdom