#!/usr/bin/perl # 4/13/96 Michael Smith # for WebGlimpse # see http://glimpse.cs.arizona.edu/webglimpse for more information # returns full search box (with preference to the referring page) ######################################################################### # The following variables are changed by wginstall $WEBGLIMPSE_HOME = "/home/bugs/bugstopper-www"; $CGIBIN = "cgi-bin"; $GLIMPSE_LOC = "/home/bugs/bugstopper-www/glimpse-4.1-bin-Linux-2.0.30-i486/bin/glimpse"; $CONVERT_LOC = "/home/bugs/bugstopper-www/glimpse-4.1-bin-Linux-2.0.30-i486/bin/wgconvert"; # End list of changed variables # lib directory $WEBGLIMPSE_LIB = "$WEBGLIMPSE_HOME/lib"; # we don't need to know the settings; all done by webglimpse $nh_pre=".nh."; $REMOTEDIR = ".remote"; $MAPFILE = ".wgmapfile"; # name of config file $CONFIGFILE = "archive.cfg"; # start the header print "Content-type: text/html\n\n"; # ENV variables # path info contains the archive directory $archivepwd = $ENV{'PATH_INFO'}; if ($archivepwd =~ m/^\s*\&/) { # Don't let anybody open specific descriptors. &err_file ($archivepwd); die ("UNREACHABLE REACHED"); } # Get virtual host if any $vhost = ""; open (F, "<$archivepwd/$CONFIGFILE") || &err_file ("$archivepwd/$CONFIGFILE"); while (($_ = ) && ($vhost == "")) { if (/^vhost/) { ($nullvar, $vhost) = split(/\s/); } } close(F); if ($vhost eq "default") { $vhost = ""; } ######################################################################### ######################################################################### ### SITE CONF STUFF ######################################################################### ######################################################################### $wgConfPath = "$WEBGLIMPSE_HOME/$vhost.wgsiteconf"; $prefix = "^DirectoryIndex|^UserDir|^Alias|^ScriptAlias|^DocumentRoot"; $DirectoryIndex=""; $UserDir=""; $DocumentRoot=""; # @AliasList=(); # @ScriptAliasList=(); # @ServerCache=(); $Port=""; $Server=""; $ServerAddress=""; ######################################################################### ### END SITE CONF STUFF ######################################################################### #--------------------------------- # make my libraries more important unshift(@INC, "$WEBGLIMPSE_LIB"); require "config.pl"; &siteconf_ReadConf(); print "\n"; # Check that a query has been made $query = $ENV{'QUERY_STRING'}; # Strip the variables out from the query string, # and assign them into variables, prefixed by 'QS_' foreach $pspec (split (/\&/, $query)) { $pname = ''; $pvalue = ''; ($pname, $pvalue) = (split (/=/, $pspec)); # Decode form results (hex characters, spaces etc) $pvalue = &www_form_urldecode($pvalue); $pname = &www_form_urldecode($pname); if ($pname =~ /^[a-zA-Z0-9_]*$/ ) { $varname = "QS_$pname"; $$varname = $pvalue; } } # query is the file $file = $QS_file; $file =~ s/\'//g; # print "file: $file
\n"; $show_neighborhood=0; if($QS_shownh){ $show_neighborhood=$QS_shownh; } # go read the config file in the archive if(&TestConfig($archivepwd)==0){ &err_conf; } ($title, $archiveurl, $traverse_type, $explicit_only, $numhops, $nhhops, $local_limit, $remote_limit, $addboxes, @urllist) = ReadConfig($archivepwd); # look for the title for this page $thistitle = &lookup_title($file); ### TO DO -- error if we can't find the file's title; all indexed files should ### have at least 'No title' if($file ne ""){ print "Full search from $thistitle\n"; }else{ print "Full search\n"; } print "\n\n"; print "\n"; print "\n"; print "\n"; print "\n"; print "\n\n"; print "\n\n"; print "\n"; print "\n"; print "

\n"; print "\n"; print "	\n"; print "\n"; print "WebGlimpse Search
\n"; print " \n"; if($file ne ""){ $link = &siteconf_LocalFile2Url($file); print "\n"; print "Search:\n"; print "\n"; print "The neighborhood of $thistitle\n"; print "The full archive: $title\n"; }else{ print "Search on the entire archive.\n"; } print "
\n"; print "String to search for: \n"; print "\n"; print " \n"; print "\n"; print "Case sensitive\n"; print " \n"; # print " \n"; print "Partial match\n"; print " \n"; print "Jump to line\n"; print " \n"; print "\n"; print "misspellings allowed\n"; print " \n"; print "\n"; print "Return only files modified within the last \n"; print "days.\n"; print " \n"; print "Maximum number of files returned:\n"; print "\n"; print " Maximum number of matches per file returned:\n"; print "\n"; print " \n"; print "\n"; print "
\n"; print "\n"; print "\n"; print "Glimpse and \n"; print "WebGlimpse, Copyright © 1996, \n"; print "University of Arizona\n"; print "\n"; print "

\n"; print "\n\n

"; $path_info = $ENV{'PATH_INFO'}; if($file ne ""){ if($show_neighborhood!=0){ # now add the files in that neighborhood print "

Pages in the neighborhood of $file:\n"; $neighborhood="$file"; # add the prefix $neighborhood=~ s/([^\/]+)$/$nh_pre$1/; #CALL CONVERT HERE BEFORE OPENING NEIGHBOURHOOD FILE --> bgopal oct/6/96 #$cmd = "$CONVERT_LOC -U -in -H $path_info < $neighborhood 2>&1 |"; #DON'T USE -U SINCE I WANT ALL INFORMATION... $cmd = "$CONVERT_LOC -in -H $path_info < $neighborhood 2>&1 |"; if (!open(NH, $cmd )) { print "No neighborhood file found.\n"; } else { print "

){ #$output = $_; $tempoutput = $_; chomp $tempoutput; @outputarray = split(" ", $tempoutput); $output = @outputarray[0]; shift outputarray; $link = @outputarray[0]; shift outputarray; $title = join(" ", @outputarray); next if ($output =~ /^$archivepwd\/$file$/); # skip if it's the file itself ### TO DO -- change output to filename # # get the title #$title = &lookup_title($output); # # get the URL #$link = &siteconf_LocalFile2Url($output); # # if there is no title, use the 'real' path if($title eq "No Title"){ $title = $link; } else { if($title eq ""){ $title = $link; } } print "

$title\n"; } close(NH); } }else{ print "

\n"; print "List the neighborhood of \"$thistitle\"
\n"; print "

\n"; } } print "\n"; print "\n"; ######################################################################### sub err_conf{ print "Cannot find archive\n"; print "\n"; print "

Specified directory $archivepwd is not an archive directory.

\n"; print "Configuration file was not found.\n"; print "\n"; exit -1; } sub err_file{ local ($file) = @_; print "Can't read file\n"; print "\n"; print "

Specified file $file is not readable.

\n"; print "\n"; exit -1; } sub www_form_urldecode { # Added 10/18/97 as per Peter Bigot --GB local($_) = @_; # Reverse the encoding: plus goes to space, then unhex encoded chars s/\+/ /g; s/%([A-Fa-f0-9]{2})/pack("c",hex($1))/ge; return $_; } sub lookup_title{ local($file) = @_; local($intitle, $title); if (($file =~ m/^\s*-\s*$/) || ($file =~ m/^\s*\&/)) { # Don't let anybody open stdin, or specific descriptors. &err_file ($file); die ("UNREACHABLE REACHED"); } if (open(IN, "<$file")) { $intitle = 0; line: while () { chomp; if((/\(.*)$/i)) { $intitle = 1; $title = $1; } elsif ($intitle) { $title .= " $_"; } if ($intitle && $title =~ s#.*##i) { last line; } } } # if there's no title, just return "", let webglimpse write 'No title'. # if($title eq ""){ # $title="No title"; # } # Trim blank space as per M. Ernst. 1/25/98 --GB $title =~ s/^\s+//; $title =~ s/\s+$//; return $title; } ######################################################################## #### SITE CONFIGURATION FUNCTIONS #### ######################################################################## ######################################################################## sub siteconf_ReadConf { local(@thearray); open (WMCONF, "<$wgConfPath") || &err_file ($wgConfPath); # hmm, I am not sure if it's a bug. If you have 2 of # DirectoryIndex, UserDir or DocumentRoot, we use the last one. # load up the HomeDirArray # NOT NEEDED HERE # while(@thearray = getpwent()){ # $HomeDir{@thearray[0]} = @thearray[7]; # } while () { if (/^DirectoryIndex[\s]*([\S]*)/i) { $DirectoryIndex = $1; } elsif (/^UserDir[\s]*([\S]*)$/i) { $UserDir = $1; } elsif (/^DocumentRoot[\s]*([\S]*)$/i) { $DocumentRoot = $1; } elsif (/^Alias[\s]*([\S]*)[\s]*([\S]*)$/i) { # do nothing. Not needed. # push(@AliasList, $2); } elsif (/^ScriptAlias[\s]*([\S]*)[\s]*([\S]*)$/i) { # do nothing. Not needed. # push(@ScriptAliasList, $2); } elsif (/^Port[\s]*([\S]*)$/i) { $Port = $1; } elsif (/^Server[\s]*([\S]*)$/i) { $Server = $1; } } if ($DirectoryIndex eq "") { $DirectoryIndex = "index.html"; } local($name,$aliases,$dm3,$dm4,$addrs) = gethostbyname($Server); local($alias); ### SERVER CACHE STUFF NOT NEEDED # $ServerCache{$Server} = $addrs; # $ServerCache{$name} = $addrs; # foreach $alias (@aliases) { # $ServerCache{$alias} = $addrs; # } $ServerAddress = $addrs; } sub siteconf_LocalFile2Url { local($file) = @_; local($alias, $homedir, $url); if ($Port eq "80") { $portPart = ""; } else { $portPart = ":$Port"; } if ($file =~ /^$DocumentRoot(.*)/) { $url = "http://$Server$portPart/$1"; return $url; } # We are NOT going for longest match. foreach $alias (keys %Alias) { $homedir = $Alias{$alias}; if ($file =~ /^$homedir(.*)$/) { $url = "http://$Server$portPart$alias/$1"; return $url; } } return ""; }

\n"; print "List the neighborhood of \"$thistitle\"\n"; print "

Specified directory $archivepwd is not an archive directory.

Specified file $file is not readable.

\n"; print "List the neighborhood of \"$thistitle\"
\n"; print "