#!/usr/bin/perl

# normalize Markdown

use Getopt::Long;
use File::Slurp;
use Lingua::Sentence;

use strict;
use warnings;
use charnames ':full';

# setup sentence splitter
my $splitter = Lingua::Sentence->new("en");

my $force;
GetOptions ("force|f");

my $stem = shift;
$_ = read_file( $stem . '.md' );

# join non-headline multi-lines
s/([^\n=])\n/$1 /g;
#s/(?<=[\S^=])\h*\n(?=[\S^=])/ /g;

# split into sentences
$_ = $splitter->split($_);

# split after comma, and before and after ellipsis
#s/,\s(?=[^\v=]*\v)/,\n/mg;
s/\h+(\(\N{HORIZONTAL ELLIPSIS}\))/\n$1/mg;
s/(\(\N{HORIZONTAL ELLIPSIS}\))\h+/$1\n/mg;

write_file( $stem . '.mdwn', $_ );

print "DONE: $0 stem $stem\n";