LDP/LDP/builder/ldp_mk

#!/usr/bin/perl
#
# ldp_mk - create all various output forms needed for the LDP from SGML file
#
# gferg@sgi.com / Ferg
#
# LOG:
#     02Aug2000 - added cmd-line options and index creation (gferg)
#

if($ARGV[0] eq '') {
   &usage();
   exit(0);
}

my($_toolroot)    = $ENV{'SGML_TOOLROOT'} ||
                             '/export/sunsite/users/gferg/toolroot';
my($_jade)        = 'jade-1.2.1';
my($linuxdoc)     = 1;
my($create_index) = 1;
my($cmd, $fname, $fname_wo_ext, $txt_filter, $style, $s, $db_v) = '';

while(1) {

     if ($ARGV[0] eq "-style") {
         shift(@ARGV);
         $style = ($ARGV[0] + 0);
         shift(@ARGV);
     } elsif ($ARGV[0] eq "-txt") {
         shift(@ARGV);
         $txt_filter = $ARGV[0];
         shift(@ARGV);
     } elsif ($ARGV[0] eq "-no_index") {
         shift(@ARGV);
         $create_index = 0;
     } else {
         last;
     }
}
$fname = $ARGV[(@ARGV + 0) - 1];
$fname_wo_ext = $fname;
if( $fname =~ /\.sgml$/ ) {
    $fname_wo_ext =~ s/\.[\w]+$//;
} else {
    $fname .= "\.sgml";
}
if( !(-e "$fname") ) {
      print "\nldp_mk: ERROR - cannot find/read $fname\n";
      &usage();
      exit(0);
}

if( $txt_filter eq '' ) {
    $txt_filter = 'lynx';
}


# determine DTD 
#
open(FP_IN, "head $fname | grep -i '\!doctype' |") || 
               die "\nldp_mk: ERROR - cannot determine DTD for $fname\n";
$s = <FP_IN>;
close(FP_IN);

if( $s eq '' ) {
    die "\nldp_mk: cannot determine DTD for $fname\n";
}
if( $s =~ /linuxdoc/i ) {

    $linuxdoc = 1;
    print "\nldp_mk: $fname is LinuxDoc SGML\n";

} else {

    # determine docbook version
    #
    ($db_v) = ($s =~ /\s+V(\d)\./i);

    # do not override any various SGML catalog file settings
    #
    if( $ENV{'SGML_CATALOG_FILES'} eq '' ) {
        $s = ($db_v =~ /3/ ? "$_toolroot/dtd/docbook_31/catalog:" :
                             "$_toolroot/dtd/docbook_41/catalog:")   .
             "$_toolroot/dsssl/docbook/catalog:"  .
             "$_toolroot/jade-1.2.1/dsssl/catalog";
        $ENV{'SGML_CATALOG_FILES'} = $s;
    }

    if( !(-d "$fname_wo_ext") ) {
          mkdir("$fname_wo_ext", 0755);
    }

    $linuxdoc = 0;
    print "\nldp_mk: $fname is DocBook (vers. $db_v) SGML\n";
}


# create the SGML index...
#
if( $linuxdoc == 0 && $create_index == 1 ) {

    print "\nldp_mk: creating SGML index from $fname...\n";

    $cmd = "$_toolroot/mkindex/collateindex.pl -N -o index.sgml; " .
           "$_toolroot/$_jade/jade/jade -t sgml -V html-index " .
              "-d $_toolroot/dsssl/docbook/html/ldp.dsl#html " .
              ($style ne '' ? "-d $style " : '') .
              "$fname ; " .
           "$_toolroot/mkindex/collateindex.pl -g -t Index " .
              "-i doc-index -o index.sgml HTML.index;" .
           "rm -f HTML.index";
    system($cmd);
}


# create HTML version
#
print "\nldp_mk: creating HTML from $fname...\n";

if( $linuxdoc == 1 ) {

    $cmd = "$_toolroot/sgml-tools/bin/sgml2html -c latin $fname";

} else {

    $cmd = "$_toolroot/$_jade/jade/jade -t sgml -i html " .
              "-d $_toolroot/dsssl/docbook/html/ldp.dsl#html " .
              ($style ne '' ? "-d $style " : '') .
              "$fname; " .
           "mv -f \*.html $fname_wo_ext/;" .
           "$_toolroot/$_jade/jade/jade -t sgml -i html -V nochunks " .
              "-d $_toolroot/dsssl/docbook/html/ldp.dsl#html " .
              ($style ne '' ? "-d $style " : '') .
              " $fname > 00_" . "$fname_wo_ext.html";
}

system($cmd);
if( ($linuxdoc == 1 && !(-e "$fname_wo_ext.html"))
    ||
    ($linuxdoc == 0 && !(-e "$fname_wo_ext/index.html")) ) {
   print "\nldp_mk: WARNING - could not create HTML: $fname_wo_ext\n";
}


# create PLAIN TEXT version
#
print "\nldp_mk: creating plain text from $fname...\n";

if( $linuxdoc == 1 ) {

    $cmd = "$_toolroot/sgml-tools/bin/sgml2txt -c latin -f $fname";
    system($cmd);

} else {

    if( $txt_filter =~ /lynx/i ) {
        $cmd = "lynx -dump ";

    } elsif( $txt_filter =~ /w3m/i ) {
        $cmd = "$_toolroot/w3m/bin/w3m -S -cols 78 -dump ";

    } elsif( $txt_filter =~ /html2text/ ) {
        $cmd = "$_toolroot/html2text/bin/html2text -style pretty -nobs ";

    } else {
        print "\nldp_mk: txt_filter($txt_filter) unrecognized, using lynx\n";
        $cmd = "lynx -dump ";
    }

    $cmd .= " 00_" . "$fname_wo_ext" . ".html > $fname_wo_ext.txt";
    system($cmd);

    # patch the text file; especially needed w/lynx
    #
    if( -e "$fname_wo_ext.txt" ) {

       my($pat) = 'file://localhost/export/sunsite/users/gferg/howto/';

       open(HFILE, "+< $fname_wo_ext.txt") || 
                          die "\nldp_mk: cannot open: $fname_wo_ext.txt";
       my(@flines) = <HFILE>;

       for($i=0; $i < (@flines + 0); $i++) {
           $flines[$i] =~ s/$pat//g;
           $flines[$i] =~ s/00_$fname_wo_ext/$fname_wo_ext/g;
       }

       seek(HFILE,0,0);
       print HFILE @flines;
       truncate(HFILE, tell(HFILE));
       close(HFILE);
    }
}

if( !(-e "$fname_wo_ext.txt") ) {

      print "\nldp_mk: WARNING - could not create TXT: $fname_wo_ext.txt\n";

} else {

      # clean-up PLAIN TEXT version
      #
      $cmd = "cat $fname_wo_ext.txt|$_toolroot/sgml-tools/bin/cutblank >" .
             "ldp_mk_tmp; mv -f ldp_mk_tmp $fname_wo_ext.txt";
      system($cmd);
}


# create PDF/PS versions
#
# Note that we use the single-page HTML variant
#
print "\nldp_mk: creating PDF from $fname...\n";

if( $linuxdoc == 1 ) {
    system("$_toolroot/sgml_ld_1html $fname");
}

if( -e "00_$fname_wo_ext.html" ) {

    $cmd = "$_toolroot/htmldoc/bin/htmldoc --size universal -t pdf -f " .
           "$fname_wo_ext.pdf 00_" . "$fname_wo_ext.html; " .
           "$_toolroot/htmldoc/bin/htmldoc --size universal -t ps -f " .
           "$fname_wo_ext.ps  00_" . "$fname_wo_ext.html";
    system($cmd);

    if( !(-e "$fname_wo_ext.pdf") ) {
          print "\nldp_mk: WARNING - could not create $fname_wo_ext.pdf\n";
    }
    if( !(-e "$fname_wo_ext.ps") ) {
          print "\nldp_mk: WARNING - could not create $fname_wo_ext.ps\n";
    } else {
          $cmd = "gzip -f $fname_wo_ext.ps";
          system($cmd);
    }

} else {
    print "\nldp_mk: WARNING - HTML file 00_$fname_wo_ext.html not found; ",
          "skipping PDF/PS creation\n";
}


#
# perform the packaging steps
#

print "\nldp_mk: creating HTML package...\n";

if( $linuxdoc == 1 ) {
    $cmd = "tar -cvf $fname_wo_ext-html.tar $fname_wo_ext\*.html; " .
           "gzip -f $fname_wo_ext-html.tar";
} else {
    $cmd = "tar -cvf $fname_wo_ext-html.tar $fname_wo_ext/\*; " .
           "gzip -f $fname_wo_ext-html.tar";
}
system($cmd);

if( !(-e "$fname_wo_ext-html.tar.gz") ) {
   print "\nldp_mk: WARNING - could not create $fname_wo_ext-html.tar.gz\n";
}


print "\nldp_mk: creating SGML package...\n";

$cmd = "cp $fname ldp_mk_tmp; gzip -f ldp_mk_tmp; " .
       "mv -f ldp_mk_tmp.gz $fname_wo_ext.sgml.gz";
system($cmd);

if( !(-e "$fname_wo_ext.sgml.gz") ) {
   print "\nldp_mk: WARNING - could not create $fname_wo_ext.sgml.gz\n";
}


# if LinuxDoc, create DocBook SGML...
#
if( $linuxdoc == 1 ) {

    print "\nldp_mk: creating DocBook from LinuxDoc...\n";

    $ENV{'SGML_CATALOG_FILES'} =
                 "$_toolroot/dtd/docbook.cat:$_toolroot/$_jade/dsssl/catalog";

    $cmd = "$_toolroot/$_jade/sgmlnorm/sgmlnorm " .
                 "$_toolroot/ld2db/docbook.dcl $fname > EX_$fname ;" .
           "$_toolroot/$_jade/jade/jade -t sgml " .
                 "-c $_toolroot/ld2db/catalog "   .
                 "-d $_toolroot/ld2db/ld2db.dsl#db EX_$fname > DB_$fname ;" .
           "rm -f EX_$fname";
    system($cmd);

    if( !(-e "DB_$fname") ) {
        print "\nldp_mk: WARNING - could not create DocBook: DB_$fname\n";
    } else {
        $cmd = "gzip -f DB_$fname";
        system($cmd);
    }
}


print "\nldp_mk: completed...\n";

exit(0);



sub usage {

  print "\n\n",
        "usage: ldp_mk ",
        "-style <stylesheet> -txt <txt_filter> -no_index <file>.sgml\n\n";
        " where:\n",
        "       <stylesheet> - fullpath to a DSSSL stylesheet\n",
        "       <txt_filter> - one of the following: lynx, w3m, html2text\n";
}