#!/usr/bin/perl
# Program to convert simple typed-in data into Notebook input format
#
# Supports these POS's 
#       n/nu/nc/vi/vt/a(dj)/adv/int(erj)/phr/prep/pron/nom/[blank]
#
# 28/8/99; updated 28/4/00

######################################
## Set up biliographic details here ##
######################################

$start = "\%Start\:\n";
$end = "\%End\:";
$chap = "\%chap\:";
$cit = "\%cit\:\n\%def\:\n";
$sect = "\%section\:\n";

print "Year of text is?\n";
$textyear = <>;
chop($textyear);
$year = "\%year\:$textyear\n";

print "Name of author is?\n";
$authname = <>;
chop($authname);
$auth = "\%author\:$authname\n";

print "Story name is?\n";
$storname = <>;
chop($storname);
$story = "\%story\:$storname\n";

print "Book/Mag title is?\n";
$bmtitle = <>;
chop($bmtitle);
$book = "\%mag\/book\:$bmtitle\n";

print "Printing date is?\n";
$prindate = <>;
chop($prindate);
if ($prindate =~ /^19[0-9][0-9]$/)
{
    $prindate = "\($prindate\)";
}
$date = "\%date\:$prindate\n";

print "Provenance is?\n";
$prov = <>;
chop($prov);
$place = "\%place\:$prov\n";

print "Name of editor is?\n";
$edname = <>;
chop($edname);
$ed = "\%ed\:$edname\n";

print "Anthology title is?\n";
$anthname = <>;
chop($anthname);
$anth = "\%anthology\:$anthname\n";


print "Notes, if any!\n";
$notetext = <>;
chop($notetext);
$notes = "\%notes\:$notetext\n";

##################
## Main Program ##
##################

print "What file to read from?\n";
$infile = <>;
chop($infile);
$filecheck = &check_for_file;
while ($filecheck == 1)
{
    print "Can't find that file. What file to read from?\n";
    $infile = <>;
    chop($infile);
    $filecheck = &check_for_file;
}
open(FILE1, "$infile") || die "\"Cannot find file $infile\"\n";

$outfile = "$infile\.in";
&check_for_outfile;

print " $year $auth $story $book $date $anth\n";
print "Is this all correct? (y/n)\n";
$answer = <>;
chop($answer);

if ($answer eq 'y')
{
    print "Writing output to \"$outfile\"\n";
    
    $/ = "\n";
    
    while (<FILE1>)
    {
        $_ =~ s/\- \-/ \- /g;
        $_ =~ s/\-\-/\-/g;
        $_ =~ s/ \-([0-9])/ \- $1/g;
        $_ =~ s/([a-z])\- /$1 \- /g;
        $_ =~ s/[\t]+/ /g;
        $_ =~ s/[ ]+/ /g;
    
        if ($_ =~ /CHAP/)
        {
	    $chapnum = &get_chapnum($_);
        }
        elsif ($_ =~ /\-/)
        {
	    $parsed_input = &parse_input;
	    print FILE2 "$start";
	    print FILE2 "$parsed_input";
	    print FILE2 "$chap$chapnum\n";
	    print FILE2 "$year";
	    print FILE2 "$auth";
	    print FILE2 "$story";
	    print FILE2 "$book";
	    print FILE2 "$date";
	    print FILE2 "$place";
	    print FILE2 "$ed";
	    print FILE2 "$anth";
	    print FILE2 "$sect";
	    print FILE2 "$cit";
	    print FILE2 "$notes";
        }
        else
        {
	    print STDERR "\nERROR $_\n";
        }
    }
    print FILE2 "$end";
}
else
{
    exit;
}    

#################
## Subroutines ##
#################
sub check_for_file
{
    $count = 0;
    unless (-e $infile)
    {
	$count = 1;
    }
    return $count;
}   
sub tidy_skey
{
    local($skey) = $_[0];

    if ($skey =~ m|0|)
    {
        $skey =~ s/0/zero/;
    }
    if ($skey =~ m|1|)
    {
        $skey =~ s/1/one/;
    }
    if ($skey =~ m|2|)
    {
        $skey =~ s/2/two/;
    }
    if ($skey =~ m|3|)
    {
        $skey =~ s/3/three/;
    }
    if ($skey =~ m|4|)
    {
        $skey =~ s/4/four/;
    }
    if ($skey =~ m|5|)
    {
        $skey =~ s/5/five/;
    }
    if ($skey =~ m|6|)
    {
        $skey =~ s/6/six/;
    }
    if ($skey =~ m|7|)
    {
        $skey =~ s/7/seven/;
    }
    if ($skey =~ m|8|)
    {
        $skey =~ s/8/eight/;
    }
    if ($skey =~ m|9|)
    {
        $skey =~ s/9/nine/;
    }
    if ($skey =~ m|therim |)
    {
        $skey = "rim,the nom";
    }
    if ($skey =~ m|flyer|)
    {
        $skey =~ s/flyer/flier/;
    }
    if ($skey =~ m|orbit\,in |)
    {
        $skey = "inorbit phr";
    }
    if ($skey =~ m|thevoid |)
    {
        $skey = "void n";
    }
    if ($skey =~ m|ofspace$|)
    {
        $skey =~ s/(.*)(ofspace)/$2 $1/;
    }
    if ($skey =~ m|ofspace |)
    {
        $skey =~ s/(.*)(ofspace) (.*)/$2 $1/;
    }
    if ($skey =~ m|thebelt|)
    {
        $skey = "belt nom";
    }
    if ($skey =~ m|belt\,the|)
    {
        $skey = "belt nom";
    }
    if ($skey =~ m|stars\,the|)
    {
        $skey = "stars,the n";
    }
    if ($skey =~ m|thestars|)
    {
        $skey = "stars,the n";
    }
    if ($skey =~ m|gonova|)
    {
        $skey = "nova,go phr";
    }
    if ($skey =~ m|^cosmicray$| || $skey =~ m|^cosmicray |)
    {
        $skey = "cosmicrays";
    }
    if ($skey =~ m|^number [nom]+$|)
    {
        $skey = "number";
    }
    if ($skey =~ m|anachronism n|)
    {
        $skey = "anachronism";
    }
    if ($skey =~ m|exclamation|)
    {
        $skey = "exclamation";
    }
    if ($skey =~ m|\&..lig\;|)
    {
        $skey =~ s/\&(..)lig\;/$1/;
    }
    if ($skey =~ m|\&.acute\;|)
    {
        $skey =~ s|\&(.)acute\;|$1|g;
    }
    if ($skey =~ m|\&.grave\;|)
    {
        $skey =~ s|\&(.)grave\;|$1|g;
    }
    if ($skey =~ m|\&.circ\;|)
    {
        $skey =~ s|\&(.)circ\;|$1|g;
    }
    return $skey;
}

sub get_chapnum
{
    local($chapnum) = "";
    $_ =~ /CHAP (.*)/;
    $chapnum = $1;
    return $chapnum;
}
    
sub parse_input
{
    local($record) = $_;
    chop($record);

    if ($record =~ m|[0-9]/[0-9]|)
    {
	if ($record =~ / n \- /)
	{
	    $record =~ /(.*?) (n) \- ([0-9\-]+)\/([0-9]+)/;
	    $term = $1;
	    $pos = $2;
	    $page = $3;
	    $col = $4;
	    $skey = "$term\[JJJ\]$pos";
	    $skey =~ s/[ \'\-\.]+//g;
	    $skey =~ tr/A-Z/a-z/;
	}
	elsif ($record =~ / n[uc] \- /)
	{
	    $record =~ /(.*?) (n[uc]) \- ([0-9\-]+)\/([0-9]+)/;
	    $term = $1;
	    $pos = $2;
	    $page = $3;
	    $col = $4;
	    $skey = "$term\[JJJ\]$pos";
	    $skey =~ s/[ \'\-\.]+//g;
	    $skey =~ tr/A-Z/a-z/;
	}
	elsif ($record =~ / v[it] \- /)
	{
	    $record =~ /(.*?) (v[it]) \- ([0-9\-]+)\/([0-9]+)/;
	    $term = $1;
	    $pos = $2;
	    $page = $3;
	    $col = $4;
	    $skey = "$term\[JJJ\]$pos";
	    $skey =~ s/[ \'\-\.]+//g;
	    $skey =~ tr/A-Z/a-z/;
	}
	elsif (($record =~ / a \- /) || ($record =~ / adj \- /))
	{
	    $record =~ s/( a)dj( \- )/$1$2/gio;
	    $record =~ /(.*?) (a) \- ([0-9\-]+)\/([0-9]+)/;
	    $term = $1;
	    $pos = $2;
	    $page = $3;
	    $col = $4;
	    $skey = "$term\[JJJ\]$pos";
	    $skey =~ s/[ \'\-\.]+//g;
	    $skey =~ tr/A-Z/a-z/;
	}
	elsif ($record =~ / phr \- /)
	{
	    $record =~ /(.*?) (phr) \- ([0-9\-]+)\/([0-9]+)/;
	    $term = $1;
	    $pos = $2;
	    $page = $3;
	    $col = $4;
	    $skey = "$term\[JJJ\]$pos";
	    $skey =~ s/[ \'\-\.]+//g;
	    $skey =~ tr/A-Z/a-z/;
	}
	elsif (($record =~ / int \- /) || ($record =~ / interj \- /))
	{
	    $record =~ s/( int)erj( \- )/$1$2/gio;
	    $record =~ /(.*?) (int) \- ([0-9\-]+)\/([0-9]+)/;
	    $term = $1;
	    $pos = $2;
	    $page = $3;
	    $col = $4;
	    $skey = "$term\[JJJ\]$pos";
	    $skey =~ s/[ \'\-\.]+//g;
	    $skey =~ tr/A-Z/a-z/;
	}
	elsif ($record =~ / adv \- /)
	{
	    $record =~ /(.*?) (adv) \- ([0-9\-]+)\/([0-9]+)/;
	    $term = $1;
	    $pos = $2;
	    $page = $3;
	    $col = $4;
	    $skey = "$term\[JJJ\]$pos";
	    $skey =~ s/[ \'\-\.]+//g;
	    $skey =~ tr/A-Z/a-z/;
	}
	elsif ($record =~ / prep \- /)
	{
	    $record =~ /(.*?) (prep) \- ([0-9\-]+)\/([0-9]+)/;
	    $term = $1;
	    $pos = $2;
	    $page = $3;
	    $col = $4;
	    $skey = "$term\[JJJ\]$pos";
	    $skey =~ s/[ \'\-\.]+//g;
	    $skey =~ tr/A-Z/a-z/;
	}
	elsif ($record =~ / pron \- /)
	{
	    $record =~ /(.*?) (pron) \- ([0-9\-]+)\/([0-9]+)/;
	    $term = $1;
	    $pos = $2;
	    $page = $3;
	    $col = $4;
	    $skey = "$term\[JJJ\]$pos";
	    $skey =~ s/[ \'\-\.]+//g;
	    $skey =~ tr/A-Z/a-z/;
	}
	elsif ($record =~ / nom \- /)
	{
	    $record =~ /(.*?) (nom) \- ([0-9\-]+)\/([0-9]+)/;
	    $term = $1;
	    $pos = $2;
	    $page = $3;
	    $col = $4;
	    $skey = "$term\[JJJ\]$pos";
	    $skey =~ s/[ \'\-\.]+//g;
	    $skey =~ tr/A-Z/a-z/;
	}
	else
	{
	    $record =~ /(.*?) \- ([0-9\-]+)\/([0-9]+)/;
	    $term = $1;
	    $pos = "";
	    $page = $2;
	    $col = $3;
	    $skey = "$term\[JJJ\]$pos";
	    $skey =~ s/[ \'\-\.]+//g;
	    $skey =~ tr/A-Z/a-z/;
	    print STDERR "NO POS $term\n";
	}
	$skey =~ s/\[JJJ\]/ /i;
	$tidied_skey = &tidy_skey($skey);
	$ret_string = "\%sortkey\:$tidied_skey\n\%headword\:$term\n\%pos\:$pos\n\%page\:$page\n\%column\:$col\n";
    }
    else
    {
	if ($record =~ / n \- /)
	{
	    $record =~ /(.*?) (n) \- ([0-9\-]+)/;
	    $term = $1;
	    $pos = $2;
	    $page = $3;
	    $skey = "$term\[JJJ\]$pos";
	    $skey =~ s/[ \'\-\.]+//g;
	    $skey =~ tr/A-Z/a-z/;
	}
	elsif ($record =~ / n[uc] \- /)
	{
	    $record =~ /(.*?) (n[uc]) \- ([0-9\-]+)/;
	    $term = $1;
	    $pos = $2;
	    $page = $3;
	    $skey = "$term\[JJJ\]$pos";
	    $skey =~ s/[ \'\-\.]+//g;
	    $skey =~ tr/A-Z/a-z/;
	}
	elsif ($record =~ / v[it] \- /)
	{
	    $record =~ /(.*?) (v[it]) \- ([0-9\-]+)/;
	    $term = $1;
	    $pos = $2;
	    $page = $3;
	    $skey = "$term\[JJJ\]$pos";
	    $skey =~ s/[ \'\-\.]+//g;
	    $skey =~ tr/A-Z/a-z/;
	}
	elsif (($record =~ / a \- /) || ($record =~ / adj \- /))
	{
	    $record =~ s/( a)dj( \- )/$1$2/gio;
	    $record =~ /(.*?) (a) \- ([0-9\-]+)/;
	    $term = $1;
	    $pos = $2;
	    $page = $3;
	    $skey = "$term\[JJJ\]$pos";
	    $skey =~ s/[ \'\-\.]+//g;
	    $skey =~ tr/A-Z/a-z/;
	}
	elsif ($record =~ / phr \- /)
	{
	    $record =~ /(.*?) (phr) \- ([0-9\-]+)/;
	    $term = $1;
	    $pos = $2;
	    $page = $3;
	    $skey = "$term\[JJJ\]$pos";
	    $skey =~ s/[ \'\-\.]+//g;
	    $skey =~ tr/A-Z/a-z/;
	}
	elsif (($record =~ / int \- /) || ($record =~ / interj \- /))
	{
	    $record =~ s/( int)erj( \- )/$1$2/gio;
	    $record =~ /(.*?) (int) \- ([0-9\-]+)/;
	    $term = $1;
	    $pos = $2;
	    $page = $3;
	    $skey = "$term\[JJJ\]$pos";
	    $skey =~ s/[ \'\-\.]+//g;
	    $skey =~ tr/A-Z/a-z/;
	}
	elsif ($record =~ / adv \- /)
	{
	    $record =~ /(.*?) (adv) \- ([0-9\-]+)/;
	    $term = $1;
	    $pos = $2;
	    $page = $3;
	    $skey = "$term\[JJJ\]$pos";
	    $skey =~ s/[ \'\-\.]+//g;
	    $skey =~ tr/A-Z/a-z/;
	}
	elsif ($record =~ / prep \- /)
	{
	    $record =~ /(.*?) (prep) \- ([0-9\-]+)/;
	    $term = $1;
	    $pos = $2;
	    $page = $3;
	    $skey = "$term\[JJJ\]$pos";
	    $skey =~ s/[ \'\-\.]+//g;
	    $skey =~ tr/A-Z/a-z/;
	}
	elsif ($record =~ / pron \- /)
	{
	    $record =~ /(.*?) (pron) \- ([0-9\-]+)/;
	    $term = $1;
	    $pos = $2;
	    $page = $3;
	    $skey = "$term\[JJJ\]$pos";
	    $skey =~ s/[ \'\-\.]+//g;
	    $skey =~ tr/A-Z/a-z/;
	}
	elsif ($record =~ / nom \- /)
	{
	    $record =~ /(.*?) (nom) \- ([0-9\-]+)/;
	    $term = $1;
	    $pos = $2;
	    $page = $3;
	    $skey = "$term\[JJJ\]$pos";
	    $skey =~ s/[ \'\-\.]+//g;
	    $skey =~ tr/A-Z/a-z/;
	}
	else
	{
	    $record =~ /(.*?) \- ([0-9\-]+)/;
	    $term = $1;
	    $pos = "";
	    $page = $2;
	    $skey = "$term\[JJJ\]$pos";
	    $skey =~ s/[ \'\-\.]+//g;
	    $skey =~ tr/A-Z/a-z/;
	    print STDERR "NO POS $term\n";
	}
	$skey =~ s/\[JJJ\]/ /i;
	$tidied_skey = &tidy_skey($skey);
	$ret_string = "\%sortkey\:$tidied_skey\n\%headword\:$term\n\%pos\:$pos\n\%page\:$page\n\%column\:\n";
    }
    return $ret_string;
}

sub check_for_outfile
{
    if (-e $outfile)
    {
	print "The file $outfile already exists. Overwrite? (y/n)";
	$overwrite = <>;
	chop($overwrite);
	if ($overwrite =~ /[yY]/)
	{
	    open(FILE2, ">$outfile") || die "\"Cannot open file $outfile\"\n";
	}
	else
	{
	    print "Choose another output filename\n";
	    $outfile = <>;
	    chop($outfile);
	    &check_for_outfile;
	}
    }
    else
    {
	open(FILE2, ">$outfile") || die "\"Cannot open file $outfile\"\n";
    }
}