#!/usr/bin/perl

# musings.cgi - a CGI script to search a swish index of my musings

# Eric Lease Morgan <eric_morgan@infomotions.com>

# 2004/11/21 - added screen as well as print css; I don't really understand it!
# 2004/11/09 - marked up subjects; added suggestions
# 2004/11/07 - first investigations; based on gallery.pl


# define the index file to search
use constant DICTIONARY => '/usr/local/apache/htdocs/musings/etc/musings.dict';
use constant INDEX      => 'etc/musings.idx';
use constant MENU       => 'etc/contents.inc';
use constant NAVBAR     => 'etc/navigation.inc';
use constant PREFIX     => '?cmd=search&amp;query=subject%3D';


###############################################################################################
# no configuration should be necessary below this line, unless you want to configure the output

# define the necessary modules
use CGI;
use CGI::Carp qw(fatalsToBrowser);
use Musings::Article;
use Musings::Subject;
use strict;
use SWISH::API;
use Text::Aspell;

# initalize CGI I/O
my $cgi = CGI->new;

# initialize global variables;
my ($hit_list, $number_of_hits);

# initialilze the HTML output
my $html  = $cgi->header;
my @style = ({-src => 'etc/musings.css', -media => 'screen'}, {-src => 'etc/musings-print.css', -media => 'print'});
$html .= $cgi->start_html(-title=>"Infomotions' Musings on Information and Librarianship", -style => \@style);
$html .= &read_navigation;
$html .= &read_menu;

# get the command
my $cmd = $cgi->param('cmd');

# branch according to the command
if (! $cmd) {

  # simply get the home page
  $html .= '<div class="content">';
  $html .= $cgi->h1("Infomotions' Musings on Information and Librarianship");
  $html .= $cgi->p('This is an (incomplete) collection of the things I\'ve written -- my musings.');
  $html .= $cgi->p('My day job resides in an academic environment, and I believe it is my responsibility to act academically. A lot of the time, this means thinking, studying, writing, sharing, and repeating the process. The exciting thing about writing, and while it may sound corny, is that is transcends both space and time. It is entirely possible for someone to write something down and have those written words to be communicated to other people on the other side of the world days, weeks, or even years later.');
  $html .= $cgi->p('I believe it is important to share one\'s ideas freely. That is why I have never completely signed away my copyrights and retained the rights to post my articles on my own website. This collection is a manifestation of that idea. I have been practicing "green" open access publishing for more than a dozen years.');
  $html .= $cgi->p('To these ends I am sharing the texts in this collection with you. It contains pre-edited, formally published articles, travel logs, descriptions of software applications, and the hand-outs of workshops and presentations. Feel free to use the items in this collection as you see fit, but please don\'t call the works your own. Place the blame and/or credit where the blame and/or credit is due.');
  $html .= '</div>';
}


elsif ($cmd eq 'title') {

  my @articles = Musings::Article->get_articles(sort => 'name');
  my $list;
  foreach (@articles) { $list .= $cgi->li($cgi->a({href => $_->article_path}, $_->article_title))}

  $html .= '<div class="content">';
  $html .= $cgi->h1("Infomotions' Musings on Information and Librarianship");
  $html .= $cgi->h2("Title index");
  $html .= $cgi->ul({class => 'detaillist'}, $list);
  $html .= '</div>';

}


elsif ($cmd eq 'date') {

  my @articles = Musings::Article->get_articles(sort => 'date');
  my $list;
  foreach (@articles) { $list .= $cgi->li($cgi->a({href => $_->article_path}, $_->article_title). ' (' . $_->article_date . ')')}

  $html .= '<div class="content">';
  $html .= $cgi->h1("Infomotions' Musings on Information and Librarianship");
  $html .= $cgi->h2("Date index");
  $html .= $cgi->ul({class => 'detaillist'}, $list);
  $html .= '</div>';

}


elsif ($cmd eq 'subject') {

  my @subjest = Musings::Subject->get_subjects(sort => 'name');
  my $list;
  foreach (@subjest) {
  
    $list .= $cgi->li($_->subject_term);
    my @article_ids = $_->article_ids;
    $list .= $cgi->start_ul;
    foreach my $a (@article_ids) {
    
      my $article = Musings::Article->new(id => $a);
      $list .= $cgi->li($cgi->a({href => $article->article_path}, $article->article_title));
    
    }
    $list .= $cgi->end_ul;
    
  }
  
  $html .= '<div class="content">';
  $html .= $cgi->h1("Infomotions' Musings on Information and Librarianship");
  $html .= $cgi->h2("Subject index");
  $html .= $cgi->ul({class => 'detaillist'}, $list);
  $html .= '</div>';

}


elsif ( $cmd eq 'help' ) {

  # simply disply the search help
  $html .= &help();
  
}


elsif ( $cmd eq 'search' ) {
  
  # check for query
  if (! $cgi->param('query')) {
  
    # error
    $html .= '<div class="content"><p><strong>Opps!</strong> No query was supplied.</p></div>';
    
    # quit
    &gracefulExit();
  
  }
  
  # search the index
  &search();
  
}

else {

  # error; unknown command
  $html .= "<b>Oops!</b> Unknown value for cmd ($cmd). Call Eric.";

}

# quit
&gracefulExit();


#############
# subroutines

# display the output and exit the application
sub gracefulExit {
  
  # add the footer to the output
  $html .= &footer;
  $html .= $cgi->end_html;
  
  # proces each macro
  $html =~ s/##NUMBEROFHITS##/$number_of_hits/g;
  $html =~ s/##HITLIST##/$hit_list/g;
  $html =~ s/##QUERY##/$cgi->param('query')/eg;
  
  # display the html, finally
  print $html;
  
  # quit, whew!
  exit;
  
}

# the heart of the matter
sub search {

  # get the result page
  $html .= &results();

  # open the index
  my $swish = SWISH::API->new(INDEX);
  
  # search the index
  my $results = $swish->Query($cgi->param('query'));
  
  # check for search results; need error checking
  $number_of_hits = $results->Hits;

  # initalize the hit list and a pointer
  $hit_list;
  
  if (! $number_of_hits) {
    
    # initalize dictionary
    my $dictionary = Text::Aspell->new;
    $dictionary->set_option('master', DICTIONARY);
    
    # parse the query
    my @query = split / /, $cgi->param('query');
    
    # initialize the new query
    my $new_query;
    
    # process each query word
    foreach my $q (@query) {
    
      # get suggestion
      my @suggestions = $dictionary->suggest($q);
      
      # build new query
      $new_query .= @suggestions[0] . ' ';

    }
    $new_query = substr($new_query, 0, -1);
    
    # add a suggestion to the output
    $hit_list = $cgi->p("Did you mean <a href='?cmd=search&query=$new_query'>$new_query</a>?");
  
    return;
    
  }
  
  # process each hit
  while (my $result = $results->NextResult) {
        
    # get the results
    my $title    = $result->Property ('swishtitle');
       $title    =~ s/ \/.*$//;
    my $note     = $result->Property('description');
    my $date     = $result->Property('date');
    my $source   = $result->Property('source');
    my $subjects = &swishify($result->Property('subject'));
    my $url      = $result->Property('swishdocpath');

    my $details;
    
    if ($note)     { $details .= $cgi->li("<b>Description</b>: $note") }
    if ($date)     { $details .= $cgi->li("<b>Date</b>: $date") }
    if ($source)   { $details .= $cgi->li("<b>Source</b>: $source") }
    if ($subjects) { $details .= $cgi->li("<b>Subject(s)</b>: $subjects") }
    if ($url)      { $details .= $cgi->li("<b>URL</b>: <a href='$url'>$url</a>") }
    
    $details = $cgi->ul({-class=>'detaillist'}, $details);
    
    $hit_list .= $cgi->li("<a href='$url'>$title</a>$details");
    
  }
  
  # finish the hit list
  $hit_list = $cgi->ol($hit_list);
  
}

# edit these subroutines to change the output's display

sub form {

  return <<EOF
<form action='./' method='get'>
<input type='hidden' name='cmd' value='search'>
Query: <input name='query' size='60' value='##QUERY##'><br>
<input type='submit' value='Go'> <small><a href='./search.pl?cmd=help'>Help</a></small>
</form>    
EOF

}


sub results {

  return <<EOF
<div class='content'>
<h1>Search results</h1>
<p>
Your search (##QUERY##) returned ##NUMBEROFHITS## hit(s):
</p>
##HITLIST##
</div>
EOF

}

sub help {

  return <<EOF
<div class='content'>
<h1>Help text</h1>
<p>
The search engine supports any combination of Boolean logic, nested queries, phrase, and field searches. Queries are case-insensitive. Examples include:
</p>
<center>
<table border='0'>
<tr>
  <td align='right' bgcolor='silver'><b>Types</b></td>
  <td bgcolor='silver'><b>Examples</b></td>
</tr>
<tr>
  <td align='right'>simple:</td>
  <td>red</td>
</tr>
<tr>
  <td align='right'>Boolean and:</td>
  <td>red and blue</td>
</tr>
<tr>
  <td align='right'>Boolean and:</td>
  <td>red blue</td>
</tr>
<tr>
  <td align='right'>Boolean or:</td>
  <td>red or blue</td>
</tr>
<tr>
  <td align='right'>Boolean not:</td>
  <td>red not blue</td>
</tr>
<tr>
  <td align='right'>nested:</td>
  <td>(red or blue) and green</td>
</tr>
<tr>
  <td align='right'>phrase:</td>
  <td>"light red"</td>
</tr>
<tr>
  <td align='right'>field:</td>
  <td>title = red</td>
</tr>
</table>
</center>
</div>
EOF

}

sub read_navigation {

  my $rv;
  
  open (NAV, '< ' . NAVBAR);
  while (<NAV>) { $rv .= $_ }
  close NAV;
  
  return $rv;

}

sub read_menu {

  my $rv;
  
  open (M, '< ' . MENU);
  while (<M>) { $rv .= $_ }
  close NAV;
  
  return $rv;

}


sub footer {

  return <<EOF;
<div class='footer'>
<p>
Author: Eric Lease Morgan &lt;<a href='mailto:eric_morgan\@infomotions.com'>eric_morgan\@infomotions.com</a>&gt;<br />
Date updated: 2004-11-18<br />
URL: <a href="http://infomotions.com/musings/">http://infomotions.com/musings/</a>
</p>
</div>
EOF

}


sub swishify {

  my @subjects = split ';', shift;
  my $rv;
  foreach my $s (@subjects) {
    
    my $term = $s;
    $s =~ s/ /%20/g;
    $rv .= "<a href='" . PREFIX . "$s'>$term</a>; ";
  
  }
  
  return $rv;
  
}