#!/usr/bin/perl
#
# webxref.cgi - web interface to run webxref
#
# Written by Earl Fogel, March 1996
# using the cgi-lib.pl library
#
# March 1998 - updated to use webxref version 4
# Jan-Feb 1999 - added depth, group, user options
#              - added basepath and baseurl
require "common.pl";
require "cgi-lib.pl";

$| = 1;   # un-buffered output (required if we do an exec)
$ENV{'PATH'} = '/bin:/usr/bin';
$virtual = 0;	# show fields to check virtual hosts

 MAIN: 
{
# Read in all the variables set by the form
  if (&ReadParse(*in) || defined($ARGV[0])) {
    &ProcessForm;
  } else {
    &PrintForm;
  }
}

sub ProcessForm {

  $in{query} = $ARGV[0] if defined($ARGV[0]);
  $in{depth} = 0 unless $in{depth} > 0;

#  CgiDie("Sorry, you must type in a valid URL.")
#      unless $in{query} =~ /\S/;

  $query = purify($in{query});

  if ($in{baseurl}) {
    $baseurl = purify($in{baseurl});
  } else {
    $baseurl = "http://$ENV{SERVER_NAME}/";
  }

  if ($in{basepath}) {
    $basepath = purify($in{basepath});
  } else {
    $basepath = $ENV{DOCUMENT_ROOT};
  }
  CgiDie("Can't find base directory ($basepath) for $baseurl")
    unless -d $basepath;

  print &PrintHeader;

  # if it's not a url, assume it's a local file
  if ($query !~ m#://#i) {
    $query =~ s#^/www##;
    $query = "/$query" if ($query !~ m#^/#i);
    $query = "$baseurl$query";
  }

  # We can only handle local files
  if ($query !~ s#^$baseurl/##i) {
    CgiDie("Sorry, I can only check files on $ENV{SERVER_NAME}\n");
  }

  $args = "";
  $args .= "-long "       	      if $in{long};
  $args .= "-verbose -verbose "    	      if $in{progress};
  $args .= "-dots "               unless $in{progress} || $in{dots};
  $args .= "-nohttp "             unless $in{http};
  $args .= "-fluff "                  if $in{orphans};
  $args .= "-group "              unless $in{group};
  $args .= "-user "               unless $in{user};
  $args .= "-depth $in{depth} "       if defined $in{depth};
  $args .= "-islocal $baseurl ";
  $args .= "-timeout $in{timeout} "   if $in{timeout};
  $args .= "-debug "                  if $in{debug};
  $args .= "-html ";
  $args .= "'$query'";

  print "cd $basepath; /var/www/bin/webxref4 $args<p>\n"
    if $in{debug};

  system "cd $basepath; /var/www/bin/webxref4 $args 2>&1"
    || CgiDie("webxref4 $args failed: $?");
}


sub PrintForm {
  print &PrintHeader;
  print_header("Check Links on $ENV{SERVER_NAME}");
  print <<"EOF";
This form uses webxref to check links on web pages on $ENV{SERVER_NAME}.
<p>
Enter the URL to your file in the following field
(relative to http://$ENV{SERVER_NAME}/).
<p>

<form method="get">

File to check: <input name="query" size=40 value="index.html">
<input type="submit" value="Submit">

<p>
<dl>
<dt><strong>What to Display:</strong>
<dd>
<input name="progress" type="checkbox" checked>
Show progress indicator
<dd>
<input name="long" type="checkbox">
Show additional reports
<dd>
<input name="orphans" type="checkbox">
Show unreferenced files
</dl>
<dl>
<dt><strong>What to Check:</strong>
<dd>
<input name="depth" type=radio value="" checked> Check just this file<br>
<input name="depth" type=radio value=1> Check all files in this directory<br>
<input name="depth" type=radio value=2> Check sub-directories<br>
<input name="depth" type=radio value=3> Check sub-sub-directories<br>
<input name="depth" type=radio value=4> Check sub-sub-sub-directories<br>

<p>
<dd>
These options may slow things down:<br>
<input name="http" type="checkbox">Check links to remote documents<br>
<input name="user" type="checkbox" checked>Check files owned by other people<br>
<input name="group" type="checkbox">Check files owned by other groups<br>
<p>

<input type="hidden" name="dots" value="">
<input type="hidden" name="debug" value="0">
<input type="hidden" name="timeout" value="5">

EOF

  print <<"EOF" if $virtual;
Note: You can use this script to check links on any web site hosted
on www.usask.ca, including virtual hosts.  To check a virtual host,
put it's name in the url for this page (ie. use
http://ccs.usask.ca/cgi-bin/webxref.cgi).

<dd>
Use these options to check virtual domains.<br>
Base URL: <input name="baseurl" size=35 value="http://$ENV{SERVER_NAME}/"><br>
Base Path: <input name="basepath" size=35 value="$ENV{DOCUMENT_ROOT}/"><br>
EOF

  print <<"EOF" ;
</dl>
</form>

EOF
    print_footer();

}