#!/usr/local/bin/perl 

sub man
{
  print << ;
  # -------Description--------------------------------------
  #
  # generate literature references list in HTML-format
  # from BibTeX-format. allows you to keep an organized
  # virtual bookshelf with only few additions to your
  # BibTeX-references.
  #
  # + uses title, author and year field to display
  #   information from reference
  # + uses nice icons :-) 
  # + uses comma separated entries in field 'topics'
  #   to categorize references
  # + uses URL in field 'sourceURL' to hyperlink the
  #   title to the source
  # + uses URL in field 'url' to hyperlink the book icon
  #   to a local copy (same field is used by bibsearch.pl
  #   and contains a full URL; bookshelf changes this to
  #   a relative URL, if it starts with users html-home)
  # + the resulting page provides a link to the BibTeX-file
  # + each entry may be linked to its BibTeX-reference
  # + the script also
  #   + converts german special characters 
  #   + converts occurrences of \$\\Pi\$ in titles
  #   + removes {}, \\em, \\bf and enclosing ""
  # + topics may be grouped into hierarchies based on 
  #   comments in the BibTex-source:
  #   % TOPICS
  #   % group1 : topic1, topic2, topic3
  #   % group2 : group1, topic4
  #   % ENDTOPICS
  #
  # - no real parser for BibTeX
  #   + multi-line fields allowed for author and title
  #   - important fields should be terminated by ','
  #     (year or topics as last field in entry will do ;-)
  #   - entries have to be delimited by empty lines
  #   +- text following empty line and not starting with
  #      @...{ is ignored (if e.g. an abstract field
  #      contains empty lines, insert spaces there)
  # 

}

require 'getopts.pl';

sub usage
{
  print << ;
  #
  # bookshelf [-hmscitlobT]
  #
  # -------Options------------------------------------------
  #
  # -h     : help (print this usage info)
  # -m     : man (print longer usage info)
  # -e     : edit BibTeX source file
  # -s     : split HTML (generate directory 
  #          with one file per topic)
  # -c     : classes (rudimentary support for supercategories)
  # -i     : info (link each entry to its BibTeX-reference)
  # -t     : text (no icons, implies -l)
  # -l     : show only one hyperlink per reference
  #          (url overrules sourceURL)
  # -o name: output to name.html (name/ with -s)
  # -b name: input from name
  # -T text: title (use text instead of Virtual Bookshelf)
  #
  # -------Defaults-----------------------------------------
  #
  # BibTeX    : \$HOME/public_html/bib/bookshelf.bib
  # HTML      : \$HOME/public_html/bib/bookshelf.html
  # topics directory (for -s): \$HOME/public_html/bib/bookshelf/
  # reference-file   (for -i): \$HOME/public_html/bib/bibtex.html
  #

}

# -d reserved for debugging purposes
do Getopts('u:o:b:T:cistlhmkde');

if ($opt_h) { &usage(); exit; }

if ($opt_m) { &usage(); &man(); exit; }

$user = $opt_u || $ENV{'USER'};

@pwent = getpwnam($user);
die "unknown user $user" unless @pwent;

$username = $pwent[6];
$username =~ s/,.*$//; # get rid of these (where do they come from??)
$userdir = $pwent[7];

##--- variable settings ---------------------
# you will certainly need to customize here..

chop($DATE = `date`);
chop($pwd = `pwd`);
$localhome = "$userdir/webpages";

$mailbase  = "ukc.ac.uk";
$wwwbase   = "http://www.cs.ukc.ac.uk/people/staff/";
$webhome   = "$wwwbase/$user";

# please don't rely on my icons directory ;-)
# use your own, *local* collection 
$iconsdir1 = "$webhome/icons";

if ($opt_b)
{
  $bibdir = $pwd; 
  $bibtex = $opt_b;
}
else
{
  $bibdir = "$localhome/bib"; 
  $bibtex = "bookshelf.bib";
}

if ($opt_e)
{
  $EDITOR = $ENV{'EDITOR'};
  if ($EDITOR =~ /%d.*%s/)
       { $EDITOR = sprintf($EDITOR,1,"$bibdir/$bibtex") } 
  else { $EDITOR = "$EDITOR $bibdir/$bibtex" }
  exec($EDITOR);
}

if ($opt_o)
{
  $shelfdir = $pwd; 
  $bookshelf = "$opt_o.html";
  $bookdir = "$opt_o" if $opt_s; 
}
else
{
  $shelfdir = "$localhome/bib"; 
  $bookdir = "bookshelf" if $opt_s; 
  $bookshelf = "bookshelf.html"; 
}

$INFO_FILE = "bibtex.html";

$shelftitle = $opt_T || "$username 's Virtual Bookshelf";

$PI = "<img alt=\"PI\" src=\"$iconsdir1/Pi.xbm\">";
$PLUS = "<img alt=\"+\" src=\"$iconsdir1/Plus.xbm\">";
$LAMBDA = "<img alt=\"lambda\" src=\"$iconsdir1/Lambda.xbm\">";
$FAST = "<img alt=\"FAST\" src=\"$iconsdir1/tfast.gif\">";
$shelfpic = "<img alt=\"\" src=\"$iconsdir1/tbook_shelf.gif\">";
$littleshelf = "<img alt=\"\" src=\"$iconsdir1/tbookshelf.gif\">";
$book_read = "<img alt=\"[+]\" border=0 src=\"$iconsdir1/tbook_read.gif\">";
$book_empty = "<img alt=\"[-]\" border=0 src=\"$iconsdir1/tbook_empty.gif\">";
$info = "<img alt=\"[BibTeX]\" border=0 src=\"$iconsdir1/tbook.gif\">";
#$info = "<img alt=\"[BibTeX]\" border=0 src=\"$iconsdir1/xman.gif\">";

##--- no need to customize below?? ----------

if ($opt_i)
{
  chdir($shelfdir);

  open(INFO,"> $INFO_FILE") || die "cannot write $INFO_FILE\n$!";

  print INFO <<HEADER;
  <html>
  <head>
  <title>BibTeX</title>
  </head>

  <body>

  <pre>
HEADER

  chdir($bibdir);

  open(BIB,"$bibtex") || die $!;

  $/='';
  $*=1;

  while(<BIB>)
  {
    if (/^@[^{]*{\s*([^,]*)\s*,/) { print INFO "% <a name=\"$1\">$1</a> <p>\n"; }
    print INFO;
  }

  close(BIB);

  print INFO <<FOOTER;
  </pre>

  <address>
  <a href="mailto:$user\@$mailbase">$user\@$mailbase</a>
  </address>
  </body>
  </html>
FOOTER

  close(INFO);
}

if ($opt_t)
{
  $opt_l = 1;

  $shelfpic = "";
  $littleshelf = "";
  $book_read = "";
  $book_empty = "";
}

# need to work with relative paths here
#
chdir($bibdir) || die "cannot chdir to $bibdir\n$!"; 

open(BIB,$bibtex) || die "cannot read $bibdir/$bibtex\n$!";

$/='';
$*=1;

while(<BIB>)
{
  $i++;
  if (/^@([^{]*){/)
  {
    $TYPE[$i] = $1;
    if (($TYPE[$i] =~ /string/i) || ($TYPE[$i] eq '')) { $i--; next; }
  }
  if ($opt_c && /%\s*TOPICS\s*%\s*((.|\n)*)\s*%\s*ENDTOPICS/)
  {
    &process_topics($1);
  }
  if ($opt_k && /%\s*KEYS\s*%\s*((.|\n)*)\s*%\s*ENDKEYS/)
  {
    &process_keys($1);
  }
  if (/%\s*DESCRIPTION\s*%\s*((.|\n)*)\s*%\s*ENDDESCRIPTION/)
  {
    &process_description($1);
  }
  if (/^%[^\n]*\n/) { $i--; next; }
  if (/^@[^{]*{\s*([^,]*)\s*,/) { $KEY[$i] = $1; print STDERR "$i: $1\n" if $opt_d }
  s/{\\"a}/&auml;/g;
  s/{\\"o}/&ouml;/g;
  s/{\\"u}/&uuml;/g;
  s/{\\"A}/&Auml;/g;
  s/{\\"O}/&Ouml;/g;
  s/{\\"U}/&Uuml;/g;
  s/{\\ss}/&szlig;/g;
  s/{\\rm//g;
  s/{\\sc([^}]*)}/$1/g;
  s/{\\em([^}]*)}/<em>$1<\/em>/g;
  s/{\\bf([^}]*)}/<b>$1<\/b>/g;
  s/\\-//g;
  if (/topics\s*=\s*{([^}]+)}/i) { $TOPICS[$i] = $1; }
  s/{|}//g;
  if (/editor\s*=\s*"\s*([^=]*)",\s*\S+\s*=/i) { $EDITOR[$i] = "[$1 (ed)]"; }
  elsif (/editor\s*=\s*([^=]*),\s*\S+\s*=/i) { $EDITOR[$i] = "[$1 (ed)]"; }
  if (/author\s*=\s*"\s*([^=]*)",\s*\S+\s*=/i) { $AUTHOR[$i] = $1; }
  elsif (/author\s*=\s*([^=]*),\s*\S+\s*=/i) { $AUTHOR[$i] = $1; }
  if (/\btitle\s*=\s*"\s*([^=]*)",\s*\S+\s*=/i) { $TITLE[$i] = $1; }
  elsif (/\btitle\s*=\s*([^=]*),\s*\S+\s*=/i) { $TITLE[$i] = $1; }
  if (/url\s*=\s*([^=]*),\s*\S+\s*=/) { $URL[$i] = $1; }
  if (/sourceURL\s*=\s*,\s*\S+\s*=/) { }
  elsif (/sourceURL\s*=\s*"\s*([^=]*)",\s*\S+\s*=/) { $SOURCE[$i] = $1; }
  elsif (/sourceURL\s*=\s*([^=]*),\s*\S+\s*=/) { $SOURCE[$i] = $1; }
  if (/abstractURL\s*=\s*"\s*([^=]*)",\s*\S+\s*=/) { $ABSTRACT[$i] = $1; }
  elsif (/abstractURL\s*=\s*([^=]*),\s*\S+\s*=/) { $ABSTRACT[$i] = $1; }
  if (/year\s*=\D*(\d+)\D/i) { $YEAR[$i] = $1; }
  $urls++ if ($SOURCE[$i] || $URL[$i]);
}

close(BIB);

foreach $entry (1..$i)
{
  $TITLE[$entry] =~ s/\$\\pi\$/$PI/g;
  $TITLE[$entry] =~ s/\\pired/$PI-Red/g;
  $TITLE[$entry] =~ s/\$\^\+\$/$PLUS/g;
  $TITLE[$entry] =~ s/\$\\lambda\$/$LAMBDA/g;
  $TITLE[$entry] =~ s/^fast #/$FAST/g;

  if ($opt_s) { $RELPATH = "../.."; }
       else   { $RELPATH = ".."; }

  if (defined($URL[$entry])) 
  { 
    unless ($opt_o)
    { $URL[$entry] =~ s|$wwwbase$user|$RELPATH|; }

#    $REF = "<a href=\"$URL[$entry]\">$book_read</a>"; 
  }
  else { $REF = "$book_empty"; }

  if (defined($TOPICS[$entry]))
  {
    @topics = split(',',$TOPICS[$entry]);
  }
  else { @topics = ("Miscellaneous") }

  if (defined($SOURCE[$entry]))
  {
    $TITLE = "<a href=\"$SOURCE[$entry]\">$TITLE[$entry]</a>";
  }
  elsif (defined($ABSTRACT[$entry]))
  {
    $TITLE = "<a href=\"$ABSTRACT[$entry]\">$TITLE[$entry]</a>";
  }
  else { $TITLE = "$TITLE[$entry]"; }

#  if ($opt_l && defined($URL[$entry]))
  if (defined($URL[$entry]))
  {
    $TITLE = "<a href=\"$URL[$entry]\">$TITLE[$entry]</a>";
#    $REF = $book_empty;
    $REF = $book_read;
  }

  if ($opt_i) 
  { 
#    $INFO = "<a href=\"$RELPATH/bib/$INFO_FILE#$KEY[$entry]\">$info</a>"; 
    $REF = "<a href=\"$RELPATH/bib/$INFO_FILE#$KEY[$entry]\">$REF</a>"; 
  }
#  else    
#  { $INFO = ""; }

  foreach $topic (@topics)
  {
    if (defined($sub{$topic}))
    { 
      print <<WARNING;
WARNING: entry in class not allowed
  $TITLE[$entry]
  [$topic]
WARNING
    }
    $TOPIC{$topic} .= <<ENTRY;
    <dt>$REF<b>$TITLE</b>
    <dd><em>$AUTHOR[$entry]$EDITOR[$entry]</em> ($TYPE[$entry], $YEAR[$entry])\n<p>

ENTRY
  }
}

# need to work with relative paths here
#
chdir($shelfdir) || die "cannot chdir to $shelfdir\n$!"; 

if ($opt_s) 
{ (mkdir($bookdir,0755) || die "cannot create $shelfdir/$bookdir\n$!") unless -e $bookdir; }

open(SHELF,"> $bookshelf") || die "cannot create $shelfdir/$bookshelf\n$!";

print SHELF <<HEADER;
<html>
<head>
<title>$shelftitle</title>
<base href="$wwwbase$user/bib/$bookshelf">
</head>

<body>
$shelfpic
<h2><a href=\"$bibtex\">$shelftitle</a></h2>
$i entries, $urls available online ($DATE)<p>
$DESCRIPTION
HEADER


print SHELF "<dl>\n";
if ($opt_s)
{
  foreach $topic (keys%TOPIC) 
  { 
    $tmp = $topic;
    # remove dangerous characters from filename
    $tmp =~ s|[^\w-]|_|g;
    $FILE{$topic} = "$bookdir/$tmp.html";

    $HREF{$topic} = $FILE{$topic};
  }
} else
{
  foreach $topic (keys%TOPIC) { $HREF{$topic} = "#$topic"; }
}

  foreach $topic (sort (keys%TOPIC,@classes)) 
  { 
    if (defined($super{$topic})) { next; }
    elsif (defined($sub{$topic})) 
    { &print_subtopics($topic); }
    else
    { print SHELF "<dt><a href=\"$HREF{$topic}\">$topic</a>\n"; }
  }

print SHELF "</dl>\n";

if ($opt_s)
{
  foreach $topic (sort keys%TOPIC)
  {
    open(TMP,"> $FILE{$topic}") || die "cannot create $FILE{$topic}\n$!";

    print TMP <<HERE;
    <html>
    <head>
    <title>Virtual Bookshelf: $topic</title>
    </head>

    <body>
    <h2><a href="../$bookshelf">$topic</a></h2>
    <em>$topic_keys{$topic}</em>

    <hr>

    <dl>
    $TOPIC{$topic}
    </dl>

    <hr>
    <address>
    <a href="mailto:$user\@$mailbase">
      $user\@$mailbase</a>
    </address>
    </body>
    </html>
HERE

    close(TMP);
  }
}
else
{
  foreach $topic (sort keys%TOPIC)
  {
    print SHELF <<HERE;
    <hr>
    <h2>$littleshelf <a name=\"$topic\">$topic</a></h2>\n
    <dl>
    $TOPIC{$topic}\n
    </dl>
HERE
}
}

print SHELF <<FOOTER;
<p>
<hr>
generated by 
<a href="http://www.cs.ukc.ac.uk/people/staff/cr3/toolbox/bookshelf">bookshelf</a>,
$DATE
<hr>
<address>
<a href="mailto:$user\@$mailbase">
  $user\@$mailbase</a>
</address>
</body>
</html>
FOOTER

close(SHELF);

sub process_description
{
  $DESCRIPTION = "<em>\n$_[0]\n</em>\n<hr>\n";
  $DESCRIPTION =~ s/\$MAIL/$user\@$mailbase/g;
  $DESCRIPTION =~ s/%//g;
}

sub process_keys
{
  local ( @TOPIC_LIST ) = split('%',$_[0]);

  foreach (@TOPIC_LIST)
  {
    if (/\s*(\S.*\S)\s*:\s*(.*)$/)
    {
      $topic = $1;
      if (defined($keys{$topic})) { $topic_keys{$topic} .= $2; }
      else
      {
        push(@key_topics,$topic);
        $topic_keys{$topic} = "<b>Keywords:</b> $2";
      }
      foreach $key (split(/,\s*/,$2))
      { $topic{$key} = $topic; }
    }
  }
}

sub process_topics
{
  local ( @TOPIC_LIST ) = split('%',$_[0]);

  foreach (@TOPIC_LIST)
  {
    if (/\s*(\S.*\S)\s*:\s*(.*)$/)
    {
      $class = $1;
      if (defined($sub{$class})) { $sub{$class} .= $2; }
      else
      {
        push(@classes,$class);
        $sub{$class} = $2;
      }
      foreach $topic (split(/,\s*/,$2))
      { $super{$topic} = $class; }
    }
  }
}

sub print_subtopics
{
  local ($topic) = shift;

  print SHELF "<dt>$topic<dl>\n";
  foreach $subtopic (sort split(/,\s*/,$sub{$topic}))
  {
    if (defined($sub{$subtopic}))
    { &print_subtopics($subtopic); }
    elsif (defined($HREF{$subtopic}))
    { print SHELF "<dt><a href=\"$HREF{$subtopic}\">$subtopic</a>\n"; }
    else
    { print SHELF "<dt>$subtopic\n"; }
  }
  print SHELF "</dl>\n";
}