#!/usr/bin/perl -s

our ($n,$d,$t,$tok);    #-t=html for mkterminum; -tok nat-create tokenizer

use common::sense;
use utf8::all;
my $file = shift;
my $state = "0";
my $log=q{echo `date +'%F %T'` "#T:" $< "==> " $@ >> perfide-log };
my $dot=q{echo  $< "->" $@ ";" >> perfide.dot };
my $F;
my $makevars = "";
if($t){$makevars .= "TYPE='-t=$t'"}
if($tok){$makevars .= "TOK='-tokenize'"}
if($n){$F = *STDOUT;                                 }
else  {open( $F, "|-", "tee debug | make $makevars -f - $file "); }

while(<DATA>){ 
  my $tool='';
  if(/#tools/){$/=''; $state="tools"; next};
  if(s!^(\w+(?:-\w+)*)\s*:\s*(\w+)\s*→\s*(\w+)!.$2.$3:! ){  
          $tool=$1;
          s/#T\b/$tool/g;  
  }
  last if /__END__/;
  if($state eq "tools"){ s/(\s*)$/\n\t-$log\n\t-$dot$1/}
  print $F $_;
};

close $F;

__DATA__

SHELL = /bin/bash
.SUFFIXES: .tmx .tmxcl .atmx  .ptd .ucts .bws .pairs .tmxd .boot .aptd .pltmx .po .blocks
.PRECIOUS: %.tmx %.tmxcl %.atmx  %.ptd %.ucts %.bws %.pairs %.tmxd %.boot %.aptd %.pltmx %.po %.blocks

a := $(wildcard _*[a-zA-Z0-9-].tmx)
b := $(patsubst %.tmx,%.tmxcl,$(wildcard _*.tmx))

TYPE=
TOK=

#tools

tmxcleaner: tmx → tmxcl
	tmxclean -eq -len -junk -sw $< 
	tmxuniq  _cleaned_$< 
	- rm _cleaned_$< __tmxuniq_*
	mv _cleaned_$<._ $@

atmx2pltmx : atmx → pltmx
	atmx2pltmx $< 

mkaptd: pltmx → aptd
	nat-create -id=$@_ $(TOK) -tmx -csize=700000 $<
	touch $@

tmxpos-tagger : tmxcl → atmx
	tmx-POStagger -ner $< 
	mv pos_$< $@

tmx2ptd: tmxcl → ptd
	nat-create -id=$@_ $(TOK) -tmx -csize=700000 $<
	touch $@

ptd2ucts : ptd → ucts
	nat-ptd ucts $<_/source-target.dmp $<_/target-source.dmp | \
             awk -F "=" '$$1 != $$2 {print}' | sort -fV | \
             grep -v '[0-9;.!?"^]' > $@

ptd2bws: ptd → bws
	nat-ptd bws $<_/source-target.dmp $<_/target-source.dmp | \
             awk -F "=" '$$1 != $$2 {print}' | sort -u | sort -fV | \
             grep -Ev 'Langs: |[0-9;.!?"^]' > $@

mkterminum: pairs → tmxd
	mkterminum $(TYPE) -isutf8 -q0 -skipch $< 
	touch $@

tmxprepare: tmxd → tmx
#	iconv -f utf8 -t latin1 $<ir/$@  > $@
#	utf8restore -mkmiracles $<ir/$@  > $@
#
	cp $<ir/$@  $@
#        utf8restore -mkmiracles $<ir/$@  > $@
#        echo `date +'%F %T'` "#T:" $< "==> " $@ >> perfide-log

po2tmx: po → tmx
	perl -nE 'say /\b(\w\w|\w\w_\w\w)\b/' <<< $<
	po2tmx -s=$$a  $< 
	utf8restore -mkmiracles _1.aux  > $@

dir2pairs: boot → pairs
	mkterminum -isutf8 -q0 -skipch $</ $@

block2pairs: blocks → pairs
	mkterminum -isutf8 -q0 -skipch $< $@


__END__

=encoding utf8

=head1 NAME

perfidewf - Perl extension for 

=head1 SYNOPSIS

 perfidewf a.pt.fr.ptd

=head1 DESCRIPTION

=head2 EXPORT

=head1 AUTHOR

J.Joao Almeida, jj@di.uminho.pt

=head1 SEE ALSO

perl(1).

=cut