#! /usr/bin/perl -w
use strict;

# fix-roff-punct: Fix up punctuation usage in automatically-generated
# troff files (man pages).

# Authors:
#   Peter Moulder <pmoulder@mail.csse.monash.edu.au>
#
# Copyright (C) 2004 Monash University
#
# Gnu GPL v2+:
#
#   This program is free software; you can redistribute it and/or
#   modify it under the terms of the GNU General Public License as
#   published by the Free Software Foundation; either version 2 of the
#   License, or (at your option) any later version.
#
#   This program is distributed in the hope that it will be useful,
#   but WITHOUT ANY WARRANTY; without even the implied warranty of
#   MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the GNU
#   General Public License for more details.
#
#   You should have received a copy of the GNU General Public License
#   along with this program; if not, write to the Free Software
#   Foundation, Inc., 59 Temple Place, Suite 330, Boston, MA  02111-1307  USA


# Background: Humans use a number of dash-like characters:
#
#   - ASCII hyphen/minus needed for command-line options and other computer
#     input;
#   - hyphen (`one-to-one');
#   - en dash (`2000-2003');
#   - em dash -- like this.  [Not currently handled.]
#
# Troff input spells them as \-, -, \[en], \[em] respectively.  (See the
# groff_char.7 man page for a full list of such punctuation characters.)  If
# you run `man' with your LC_CTYPE indicating a rich character set like unicode
# (UTF-8 encoding), then it uses different output characters for each of the
# above.
#
# In particular, if your man page source has plain `-' when giving an example
# of a flag or command or other program input, then users won't be able to use
# mouse copy&paste from the formatted man page.

# This script is something of a hack: it is only big enough to handle a few man
# pages of interest (produced by pod2man).  You should manually check the
# changes it makes.

# Approach: we handle each line a word at a time, and typically make the same
# hyphen-vs-ASCII decision throughout the word.  We're a bit haphazard about
# word-splitting, but it's hard to find an example of where we'd be hurt by
# that, and by luck we would do the right thing for many gcc options like
# `-fconstant-string-class=\fICLASS-NAME\fR' (where CLASS-NAME should use a
# hyphen and the others should be ASCII hyphen-minus).
#
# Perl's /e (execute) flag for substitutions does just what we want
# for preserving non-word bits while transforming "words".
#
# We don't currently handle special things like `apt-get' that look like
# hyphenated english words but are actually program names.  In general the
# problem is AI complete, e.g. `apt-gettable' could be either hyphen (gettable
# by apt) or ASCII hyphen-minus (able to be processed by the `apt-get'
# program).
#
# We don't currently take hints from font choice.  (E.g. text in CR font should
# probably use ASCII hyphen-minus.)
#
# We currently only handle a couple troff requests and escapes (see groff.7).

sub frob ($);

my $yearRE = qr/(?:19[6-9]|20[013])[0-9]/;

sub frob ($) {
    my ($x) = @_;

    # Consider splitting into two words.
    if ($x =~ m{\A(.*?)(\\(?:[&/,~:d]|f[BRI]|s-?[0-9]+))(.*)\z}) {
	my ($before, $s, $after) = ($1, $2, $3);
	return frob($before) . $s . frob($after);
    }

    if ($x =~ m{\A(.*?)(\.+)\z}) {
	my $d = $2;
	return frob($1) . $d;
    }

    # `32-bit', `5-page'.
    if ($x =~ m{\A[0-9]+-[a-z]+\z}) {
	return $x;
    }

    # Year range: `(C) 1998-2003'.
    if ($x =~ m{\A$yearRE\\?-$yearRE\z}) {
	$x =~ s{\\?-}{\\[en]};
	return $x;
    }

    # ISO date.
    if ($x =~ m{\A$yearRE-[01][0-9]-[0-3][0-9]\z}) {
	return $x;
    }

    # Things likely to be computer input.
    if ($x =~ m{[0-9]|\.[a-zA-Z]|\A(?:[-/.]|\\-|\[.*\]\z)}) {
	$x =~ s/\\?-/\\-/g;
	return $x;
    }

    $x =~ s/\\?-/-/g;
    return $x;
}

while(<>) {
    if ($_ eq '.tr \(*W-|\(bv\*(Tr' . "\n") {
	# Get rid of pod2man's "helpful" munging of pipe symbol.
	next;
    }

    # Leave ASCII apostrophe unchanged (i.e. \[aq]) for examples.
    if (/\A\\\&    /) {
	s/'/\\[aq]/g;   # `\[aq]' = "ascii quote"
    }

    if (/\A\.IP /) {
	s/\\?-/\\-/g;
	s/\\s\\-1/\\s-1/g;
    }
    elsif (/\A\.IX /) {
	s/\\?-/-/g;
    }
    elsif (!/\A\. *(?:\\"|ds|if|ie)/) {
	# As an optimization, we process only words containing `-'.
	s{([.@/\\[:alnum:]]*-[-.@/\\[:alnum:]]*)}{frob($1)}ge;
    }
    print;
}