#!/usr/bin/perl # ############################################################################# # # Dieses CGI-Skript durchsucht den HTML-Dateibereich nach einem oder # mehreren Suchbegriffen. Es werden nur Dateien mit den Endungen '.htm' # bzw. '.html' beruecksichtigt. # # Es hat folgende Parameter # # 'searchvalue' = Suchbegriff(e), getrennt durch Leerzeichen # 'type' = all - UND-Verkn. / any - OR-Verkn. # 'dirs' = zu durchsuchende Verzeichnisse # 'exdirs' = auszusparende Verzeichnisse # Jedes Verzeichnis bei 'dirs' und 'exdirs' muss mit "/" # beginnen; mehrere Verzeichnisse sind durch ";" zu trennen # # Generelles Problem: Jede HTML-Datei wird komplett in den Speicher gelesen ############################################################################# use strict; $|=1; # Init Variable my $BASE = $ENV{'DOCUMENT_ROOT'}; my $LOGFILE = $BASE . "/search.log"; my $listdirs = 0; my $exdirs = 0; my $numberreturned = 0; my @directories = (); my @excldirs = (); my @search = (); my $dir = ''; my %ein = (); # fuer Formulareingaben my $input = ''; my $name = ''; my $value = ''; # Get the input read(STDIN,$input,$ENV{CONTENT_LENGTH}); # Split the name-value pairs foreach (split("&",$input)) { /(.*)=(.*)/; $name = $1; $value = $2; $value =~ s/\+/ /g; $value =~ s/%([a-fA-F0-9][a-fA-F0-9])/pack("C", hex($1))/eg; $ein{$name}=$value; } # Get Parameters @search = split(/ /, $ein{'searchvalue'}); $numberreturned = 0; $listdirs = 0; if (defined($ein{'dirs'})) { @directories = split(/;/,$ein{'dirs'}); $listdirs = 1; } $exdirs = 0; if (defined($ein{'exdirs'})) { @excldirs = split(/;/,$ein{'exdirs'}); $exdirs = 1; } &WriteLog; # Print HTML-Header print "Content-type: text/html", "\r\n\r\n"; print "", "\n"; print "Webserver-Suche", "\n"; print "", "\n"; print "\n"; print "

Suchergebnis

\n"; &Formular; print "

\n"; if ($listdirs) { foreach $dir (@directories) { &CheckFiles($BASE."$dir"); } } else { &CheckFiles($BASE); } print "\n"; if ($numberreturned == 0) { print "

Leider nichts gefunden!\n"; } else { print "

$numberreturned Fundstelle"; if ($numberreturned != 1) { print "n"; }; print "\n"; } print "\n"; exit(0); sub Formular { print "

\n"; print "

\n"; } sub WriteLog { #write details of use to log file my $site = ''; $site = $ENV{'REMOTE_ADDR'}; if (defined($ENV{'REMOTE_HOST'})) { $site = $site . "(" . $ENV{'REMOTE_HOST'} . ")"; } if (-f $LOGFILE) { open(OUTFILE,">$LOGFILE") or return; } else { open(OUTFILE,">>$LOGFILE") or return; } flock(OUTFILE, 2); # exclusive lock print OUTFILE $ein{'searchvalue'}, ":", $ein{'type'}, ":"; print OUTFILE localtime(time), ",", $site, "\n"; flock(OUTFILE, 8); # unlock close(OUTFILE); } sub CheckFiles { # read files in directory, passed as first paramater my $path = ''; my $fullFilename = ''; my @files = (); my $line = ''; my $title = ''; my $header = ''; my $search = ''; my $tmp = ''; my $tmpath = ''; my $recurse = 0; my $found = 0; $path = shift; return unless (-d $path); opendir(ROOT,$path) or return; @files = readdir(ROOT); closedir(ROOT); foreach (@files) { next if /^\.|\.\.$/; $fullFilename = "$path/$_"; if (-d $fullFilename) { $recurse = 1; if ($exdirs) { $tmp = $fullFilename; $tmp =~ s/$BASE//; foreach $tmpath (@excldirs) { if ($tmp eq $tmpath) { $recurse = 0; last; } } } if ($recurse) { CheckFiles($fullFilename); } next; } if ($fullFilename =~ m/\.htm/i) { open(FILE, $fullFilename) or return; $line = join(' ',); # Ganze Datei wird zu einer Zeile close(FILE); $line =~ s/\n//g; # Newlines weg $line =~ s/ä/ä/g; # Umlaute konvertieren $line =~ s/ö/ö/g; $line =~ s/ü/ü/g; $line =~ s/Ä/Ä/g; $line =~ s/Ö/Ö/g; $line =~ s/Ü/Ü/g; $line =~ s/ß/ß/g; $title = "No Title"; if($line =~ m!(.+)!i) { $title = $1; }; if ($ein{'type'} eq 'all' ) { $found = 1; foreach $search (@search) { if ($line !~ m/$search/i) { $found = 0; last; } } } else # type = any { $found = 0; foreach $search (@search) { if ($line =~ m/$search/i) { $found = 1; last; } } } if ($found) { $tmp = $fullFilename; $tmp =~ s/$BASE//; print "

$title\n"; $numberreturned++; } } } }