#! /usr/bin/perl -w

# Script to turn PCRE man pages into HTML


# Subroutine to handle font changes and other escapes

sub do_line {
my($s) = $_[0];

$s =~ s/</&#60;/g;                   # Deal with < and >
$s =~ s/>/&#62;/g;
$s =~ s"\\fI(.*?)\\f[RP]"<i>$1</i>"g;
$s =~ s"\\fB(.*?)\\f[RP]"<b>$1</b>"g;
$s =~ s"\\e"\\"g;
$s =~ s/(?<=Copyright )\(c\)/&copy;/g;
$s;
}

# Subroutine to ensure not in a paragraph

sub end_para {
if ($inpara)
  {
  print TEMP "</PRE>\n" if ($inpre);
  print TEMP "</P>\n";
  }
$inpara = $inpre = 0;
$wrotetext = 0;
}

# Subroutine to start a new paragraph

sub new_para {
&end_para();
print TEMP "<P>\n";
$inpara = 1;
}


# Main program

$innf = 0;
$inpara = 0;
$inpre = 0;
$wrotetext = 0;
$toc = 0;
$ref = 1;

while ($#ARGV >= 0 && $ARGV[0] =~ /^-/)
  {
  $toc = 1 if $ARGV[0] eq "-toc";
  shift;
  }

# Initial output to STDOUT

print <<End ;
<html>
<head>
<title>$ARGV[0] specification</title>
</head>
<body bgcolor="#FFFFFF" text="#00005A" link="#0066FF" alink="#3399FF" vlink="#2222BB">
<h1>$ARGV[0] man page</h1>
<p>
Return to the <a href="index.html">PCRE index page</a>.
</p>
<p>
This page is part of the PCRE HTML documentation. It was generated automatically
from the original man page. If there is any nonsense in it, please consult the
man page, in case the conversion went wrong.
<br>
End

print "<ul>\n" if ($toc);

open(TEMP, ">/tmp/$$") || die "Can't open /tmp/$$ for output\n";

while (<STDIN>)
  {
  # Handle lines beginning with a dot

  if (/^\./)
    {
    # Some of the PCRE man pages used to contain instances of .br. However,
    # they should have all been removed because they cause trouble in some
    # (other) automated systems that translate man pages to HTML. Complain if
    # we find .br or .in (another macro that is deprecated).

    if (/^\.br/ || /^\.in/)
      {
      print STDERR "\n*** Deprecated macro encountered - rewrite needed\n";
      print STDERR "*** $_\n";
      die "*** Processing abandoned\n";
      }

    # Instead of .br, relevent "literal" sections are enclosed in .nf/.fi.

    elsif (/^\.nf/)
      {
      $innf = 1;
      }

    elsif (/^\.fi/)
      {
      $innf = 0;
      }

    # Handling .sp is subtle. If it is inside a literal section, do nothing if
    # the next line is a non literal text line; similarly, if not inside a
    # literal section, do nothing if a literal follows. The point being that
    # the <pre> and </pre> that delimit literal sections will do the spacing.
    # Always skip if no previous output.

    elsif (/^\.sp/)
      {
      if ($wrotetext)
        {
        $_ = <STDIN>;
        if ($inpre)
          {
          print TEMP "\n" if (/^[\s.]/);
          }
        else
          {
          print TEMP "<br>\n<br>\n" if (!/^[\s.]/);
          }
        redo;    # Now process the lookahead line we just read
        }
      }
    elsif (/^\.TP/ || /^\.PP/ || /^\.P/)
      {
      &new_para();
      }
    elsif (/^\.SH\s*("?)(.*)\1/)
      {
      # Ignore the NAME section
      if ($2 =~ /^NAME\b/)
        {
        <STDIN>;
        next;
        }

      &end_para();
      my($title) = &do_line($2);
      if ($toc)
        {
        printf("<li><a name=\"TOC%d\" href=\"#SEC%d\">$title</a>\n",
          $ref, $ref);
        printf TEMP ("<br><a name=\"SEC%d\" href=\"#TOC1\">$title</a><br>\n",
          $ref, $ref);
        $ref++;
        }
      else
        {
        print TEMP "<br><b>\n$title\n</b><br>\n";
        }
      }
    elsif (/^\.SS\s*("?)(.*)\1/)
      {
      &end_para();
      my($title) = &do_line($2);
      print TEMP "<br><b>\n$title\n</b><br>\n";
      }
    elsif (/^\.B\s*(.*)/)
      {
      &new_para() if (!$inpara);
      $_ = &do_line($1);
      s/"(.*?)"/$1/g;
      print TEMP "<b>$_</b>\n";
      $wrotetext = 1;
      }
    elsif (/^\.I\s*(.*)/)
      {
      &new_para() if (!$inpara);
      $_ = &do_line($1);
      s/"(.*?)"/$1/g;
      print TEMP "<i>$_</i>\n";
      $wrotetext = 1;
      }

    # A comment that starts "HREF" takes the next line as a name that
    # is turned into a hyperlink, using the text given, which might be
    # in a special font. If it ends in () or (digits) or punctuation, they
    # aren't part of the link.

    elsif (/^\.\\"\s*HREF/)
      {
      $_=<STDIN>;
      chomp;
      $_ = &do_line($_);
      $_ =~ s/\s+$//;
      $_ =~ /^(?:<.>)?([^<(]+)(?:\(\))?(?:<\/.>)?(?:\(\d+\))?[.,;:]?$/;
      print TEMP "<a href=\"$1.html\">$_</a>\n";
      }

    # A comment that starts "HTML" inserts literal HTML

    elsif (/^\.\\"\s*HTML\s*(.*)/)
      {
      print TEMP $1;
      }

    # A comment that starts < inserts that HTML at the end of the
    # *next* input line - so as not to get a newline between them.

    elsif (/^\.\\"\s*(<.*>)/)
      {
      my($markup) = $1;
      $_=<STDIN>;
      chomp;
      $_ = &do_line($_);
      $_ =~ s/\s+$//;
      print TEMP "$_$markup\n";
      }

    # A comment that starts JOIN joins the next two lines together, with one
    # space between them. Then that line is processed. This is used in some
    # displays where two lines are needed for the "man" version. JOINSH works
    # the same, except that it assumes this is a shell command, so removes
    # continuation backslashes.

    elsif (/^\.\\"\s*JOIN(SH)?/)
      {
      my($one,$two);
      $one = <STDIN>;
      $two = <STDIN>;
      $one =~ s/\s*\\e\s*$// if (defined($1));
      chomp($one);
      $two =~ s/^\s+//;
      $_ = "$one $two";
      redo;            # Process the joined lines
      }

    # .EX/.EE are used in the pcredemo page to bracket the entire program,
    # which is unmodified except for turning backslash into "\e".

    elsif (/^\.EX\s*$/)
      {
      print TEMP "<PRE>\n";
      while (<STDIN>)
        {
        last if /^\.EE\s*$/;
        s/\\e/\\/g;
        s/&/&amp;/g;
        s/</&lt;/g;
        s/>/&gt;/g;
        print TEMP;
        }
      }

    # Ignore anything not recognized

    next;
    }

  # Line does not begin with a dot. Replace blank lines with new paragraphs

  if (/^\s*$/)
    {
    &end_para() if ($wrotetext);
    next;
    }

  # Convert fonts changes and output an ordinary line. Ensure that indented
  # lines are marked as literal.

  $_ = &do_line($_);
  &new_para() if (!$inpara);

  if (/^\s/)
    {
    if (!$inpre)
      {
      print TEMP "<pre>\n";
      $inpre = 1;
      }
    }
  elsif ($inpre)
    {
    print TEMP "</pre>\n";
    $inpre = 0;
    }

  # Add <br> to the end of a non-literal line if we are within .nf/.fi

  $_ .= "<br>\n" if (!$inpre && $innf);

  print TEMP;
  $wrotetext = 1;
  }

# The TOC, if present, will have been written - terminate it

print "</ul>\n" if ($toc);

# Copy the remainder to the standard output

close(TEMP);
open(TEMP, "/tmp/$$") || die "Can't open /tmp/$$ for input\n";

print while (<TEMP>);

print <<End ;
<p>
Return to the <a href="index.html">PCRE index page</a>.
</p>
End

close(TEMP);
unlink("/tmp/$$");

# End