Kevin Zembower wrote:
>
> I'm trying to analyze web logs records which look like this:
>
> 2004-03-28 00:38:31 d7.facsmf.utexas.edu - W3SVC1 DB db.jhuccp.org GET
> /dbtw-wpd/exec/dbtwpcgi.exe
> XC=%2Fdbtw-wpd%2Fexec%2Fdbtwpcgi.exe&BU=http%3A%2F%2Fdb.jhuccp.org%2Fpopinform%2Fbasic.html&QB0=AND&QF0=Abstract+%7C+KeywordsMajor+%7C+KeywordsMinor+%7C+Notes+%7C+EngTitle+%7C+TT+%7C+FREAb+%7C+SPAAb&QI0=China%0D%0A&QB1=AND&QF1=Author+%7C+CN&QI1=&MR=10&TN=popline&RF=ShortRecordDisplay&DF=LongRecordDisplay&DL=1&RL=1&NP=0&AC=QBE_QUERY&x=37&y=4
> 200 0 21248 814 19391 80 HTTP/1.1
> Mozilla/4.0+(compatible;+MSIE+6.0;+Windows+NT+5.1;+.NET+CLR+1.0.3705) -
> http://db.jhuccp.org/popinform/basic.html
>
> In this record, in the tenth space-delimited field, which starts "XC=%2Fdbtw"
> there are variables which start with "QF" followed by a number, for instance
> "QF0=Abstract+%7C+KeywordsMajor+%7C+KeywordsMinor+%7C+Notes+%7C+EngTitle+%7C+TT+%7C+FREAb+%7C+SPAAb&"
> This indicates that the fields to be searched in the database are "Abstract
> KeywordsMajor KeywordsMinor..." The same numbered "QI" variable, in this case
> "QI0=China%0D%0A" indicates searching for "China" in these fields.
>
> For every "QF" record, there should be a corresponding "QI" record with the
> same number, although the value might be blank, as in "QF1=Author+%7C+CN&QI1=&".
> This section of the above example indicates that a search should be performed
> in the "Author" and "CN" fields, but the value for "QI1" is blank, so it matches
> everything.
>
> My program, which I've pasted in below my signature, tries to find a "QF"
> value, matches it to a list of fieldnames ("If the list of fields to be
> searched contains the 'Abstract' field, it should be considered a 'subject'
> search") then grabs the corresponding "QI" value, to print it out. However,
> I can never match anything beyond the digit. In my program below, the line:
>
> print "Match successful!\n" if ($query =~ /QI$1/);
You are using $1 from a previous match so it will work here however the
current match, if it is successful, will change the value of $1.
> works, but the next three lines:
>
> $query =~ /QI$1=(.*?)&/;
This will always fail.
> $subject = $1;
You should only use the dollar digit variables when the match succeeded.
> print "Subject: $1\n" if ($debug);
>
> never matches anything.
>
> I've been working on this, on and off, the last two days. Any suggestions or
> pointing out my boneheaded errors is gratefully appreciated. Any other
> overall suggestions on my coding are welcomed. This script seems to run very
> slowly, due probably to all the complex regex.
>
> centernet:/opt/analog/logdata/db # cat listqueries3.pl
> #!/usr/local/bin/perl
use warnings;
use strict;
> $debug = 1;
my $debug = 1;
Or perhaps even:
use constant DEBUG => 1;
> while (<>) {
> next unless (/TN=popline/i); #Just analyze the records for the POPLINE database
Are you sure you need a case insensitive match? You should also include
delimiters to ensure a correct match and the parentheses are not
required because the precedence is unambiguous.
next unless /&TN=popline&/i;
Or:
next unless /\bTN=popline\b/i;
> $subject = $author = $docno = $title = $journalsource = $keywords = $languages =
> $popreporttopic = $refereed = $year = "";
>
> if (/^.* .* .* .* .* .* .* GET [^ ]*dbtwpcgi\.exe .*QI0=[^&]*&.*QI1=[^&]*&.*/){
That regular expression is VERY inefficient. Each '.* ' will try to
match as much as possible and will then have to backtrack. For example,
compare the output of these two one-liners:
perl -Mre=debug -le'$_="a b c d e f g h i";/^.* .* .* d/&&print'
perl -Mre=debug -le'$_="a b c d e f g h i";/^\S* \S* \S* d/&&print'
> if (/QI2/) { $type = "A"; } else { $type = "B"; }
> ($date, $time, $source, $junk, $junk, $host, $FQDN, $method, $file, $query,
> $junk) = split;
Instead of using the variable $junk for fields you don't want to keep
you should use the undef function.
my ($date, $time, $source, undef, undef, $host, $FQDN, $method, $file,
$query) = split;
next unless $method eq 'GET' and substr($file, -12) eq 'dbtwpcgi.exe';
my $type = $query =~ /QI2/ ? 'A' : 'B';
> while ($query =~ m/QF(\d+)=(\S*?)&/ig) {
> print "fieldnumber = :$1:\tfieldname = $2\n" if ($debug);
> if ($2 =~ /abstract/i) {
> print "Abstract found!\n" if ($debug);
> print "Query: $query\n" if ($debug);
> print "Match successful!\n" if ($query =~ /QI$1/);
> $query =~ /QI$1=(.*?)&/;
> $subject = $1;
> print "Subject: $1\n" if ($debug);
> } elsif ($2 =~ /author/i) {
> $query =~ /QI$1=(\S*?)&/;
> $author = $1;
> } elsif ($2 =~ /endtitle/i) {
> $query =~ /QI$1=(\S*?)&/;
> $title = $1;
> }
> } #while there are more matches for QFn fields
This may work better:
my %hash = ( abstract => '', author => '', endtitle => '' );
$hash{ lc $2 } = $3
while $query =~
/QF(\d+)=[^&]*\b(abstract|author|endtitle)\b(?=.*?&QI\1=([^&]*))/ig;
> $outstring =
> "$type\t$date\t$time\t$subject\t$author\t$title\t$journalsource\t$keywords\t$languages\t$popreporttopic\t$refereed\t$year\n";
my $outstring = join( "\t",
$type, $date, $time, @hash{ qw( abstract author endtitle )
},
$journalsource, $keywords, $languages, $popreporttopic,
$refereed, $year
) . "\n";
> print translate($outstring);
Calling a subroutine _may_ slow down the loop (or it may not.)
> }# if it's a request for a database query
> }# while there are more lines in the input file
>
> sub translate() {
> $_ = $_[0];
> s/%22/\"/g;
> s/%2C/,/g;
> s/%20/ /g;
> s#%2F#/#g;
> s/%3D/=/g;
> s/%3B/;/g;
> s/%26/&/g;
> s/%0D//g;
> s/%0A//g;
> s/\+/ /g;
> s/%29/)/g;
> s/%28/(/g;
> s/%27/\' /g;
You don't have to escape " or ' in a double quoted string when the
delimiters are /.
> s/%2b/+/g;
> s/%7C/|/g;
> s/%3A/:/g;
> #Debbie request all boolean logical words and sumbols be replaced with '|'
> s/\b(and)\b/|/ig;
> s/\b(or)\b/|/ig;
The parentheses are not required because you are not grouping or
capturing anything.
s/\band\b/|/ig;
s/\bor\b/|/ig;
> s/&/|/g;
> s[/][|]g;
For single character replacement the tr/// operator is faster.
tr!&/!|!;
> $_;
> }
John
--
use Perl;
program
fulfillment
--
To unsubscribe, e-mail: [EMAIL PROTECTED]
For additional commands, e-mail: [EMAIL PROTECTED]
<http://learn.perl.org/> <http://learn.perl.org/first-response>