#!/usr/bin/env perl
# cuneiform.wrapper - Wrapper for "cuneiform" main program
# License:  Creative Commons Attribution-NonCommercial-ShareAlike 3.0
# Revision: 120627

# Note:  The license indicated above applies to this file.  It doesn't
# apply  to the  official  Cuneiform package,  or to any  file derived
# from that package.

#---------------------------------------------------------------------
#                         license information
#---------------------------------------------------------------------

# This section may not be modified except as approved by the author or
# licensor, or to make non-content changes such as adjustments to par-
# agraph formatting or white space.

# This  version of this software is  distributed  under  the following
# license:
#
#       Creative Commons Attribution-NonCommercial-ShareAlike 3.0

# You may use, modify, and redistribute this software without fees  or
# royalties, but only under the terms and conditions set forth  by the
# license.  In particular, copies and derived works cannot be used for
# commercial purposes.  Additionally, the license propagates to copies
# and derived works. Furthermore, you must provide attribution "in the
# manner  specified  by the  author or licensor".  The latter point is
# discussed below.

# The author [and licensor] hereby specifies that  attribution must be
# handled in the following manner:  a. If the software is interactive,
# any  About or Credits dialog boxes, windows, or output text provided
# by the  original version must be preserved and readily accessible to
# the end user at runtime.  b. If the software is non-interactive,  or
# if it does not  provide  About or Credits dialog boxes, windows,  or
# output text,  the  operating system and/or  desktop environment used
# must provide attribution that is  visible and/or  readily accessible
# to the end user at runtime.

# The following techniques  do not meet the  attribution requirements:
# Attribution through text  files, attribution  through  printed docu-
# mentation,  verbal  attribution, or postings on  external  web sites
# [i.e.,  web  sites  that are not an intrinsic local component of the
# operating system or  desktop environment used].  These  examples are
# provided for illustrative purposes only.

# It should be noted that trademarks are an additional issue.  If this
# software  uses any  trademarks,  trademark-related  restrictions may
# apply.

# This is not a  complete explanation of the  terms and conditions in-
# volved.  For more information, see the Creative Commons Attribution-
# NonCommercial-ShareAlike 3.0 license.

#---------------------------------------------------------------------
#                        standard module setup
#---------------------------------------------------------------------

require 5.8.1;
use strict;
use Carp;
use warnings;
use Cwd;
                                # Trap warnings
$SIG {__WARN__} = sub { die @_; };

#---------------------------------------------------------------------
#                        add LACSUB module[s]
#---------------------------------------------------------------------

use LACSUB::Filesystem
 ( @LACSUB::Filesystem::EXPORT_OK );

#---------------------------------------------------------------------
#                           basic constants
#---------------------------------------------------------------------

use constant ZERO  => 0;        # Zero
use constant ONE   => 1;        # One

use constant FALSE => 0;        # Boolean FALSE
use constant TRUE  => 1;        # Boolean TRUE

#---------------------------------------------------------------------
#                         program parameters
#---------------------------------------------------------------------

my $DIR_BASE = '__META_PREFIX__';
my $DIR_LIB  = "$DIR_BASE/lib";
my $DIR_PROG = "$DIR_BASE/program";

#---------------------------------------------------------------------
#                            main routine
#---------------------------------------------------------------------

sub Main
{
    my $CWD;                    # Current working directory
    my $cmd;                    # Shell-level command string
    my $language;               # Language code, "list", or undefined
    my $str;                    # Scratch

#---------------------------------------------------------------------
# Initial setup.

    select STDERR; $| = ONE;    # Set flush-on-write mode
    select STDOUT; $| = ONE;    # Ditto [note: this line must be last]

    $CWD =  getcwd();           # Current working directory
    $CWD =~ s@([^/])\z@$1/@;    # Make sure that it ends with a slash

                                # Set a required environment variable
    $ENV {CF_DATADIR} = "$DIR_PROG/datafiles";

#---------------------------------------------------------------------
# Print usage text, if necessary.

    if (!scalar (@ARGV))
    {
        print << 'END';
Usage: cuneiform file1.jpg /foo/file2.png ...
or     cuneiform --language=rus file1.jpg /foo/file2.png ...

The  specified arguments should be  filenames or  pathnames  for image
files. Both absolute and relative paths are allowed. Supported formats
include BMP, GIF, JPEG, PNM, PNG, and TIFF.

The image files should be  scanned copies of text pages.  If possible,
scans should be done at 300dpi.  Multi-column pages may work, but this
isn't guaranteed.

"cuneiform"  assumes that you've got  write access  to the  image-file
directories. For each input file NAME.png (or NAME.jpg, etc.), it cre-
ates a corresponding output file named NAME.cune. The output files are
text files  that contain  OCR'd text. Several input languages are sup-
ported. For a list of language codes, specify "--language=list". Note:
Russian output uses Code Page 1251 format.

Bug: Some images may crash the program.
END
        exit ONE;
    }

#---------------------------------------------------------------------
# Process the command line.

    for my $arg (@ARGV)
    {
        if ($arg =~ m@^-+lang(uage|)=(\w{3,})\z@i)
        {
            $language =  lc ($2);
            $language =~ s@^(help|info|list|show)\z@list@i;
            $language =~ s@^(.{3}).+\z@$1@ unless $language eq 'list';
        }
        elsif ($arg =~ m@^-@)
        {
            die "Error: Invalid option: $arg\n";
        }
    }

#---------------------------------------------------------------------
# Adjust LD_LIBRARY_PATH.

    $str =  $ENV {LD_LIBRARY_PATH};
    $str =  "" unless defined $str;
    $str =  ".:$DIR_LIB:$str";
    $str =~ s@^:+@@;
    $str =~ s@:+\z@@;
    $str =~ y/:/:/s;
    $ENV {LD_LIBRARY_PATH} = $str;

#---------------------------------------------------------------------
# Handle a special case.

    if (defined ($language) && ($language eq 'list'))
    {
        chdir $DIR_PROG ||
            die "Error: Can't enter directory: $!:\n$DIR_PROG\n";

        system "./cuneiform.bin -l";
        exit ZERO;
    }

#---------------------------------------------------------------------
# Process image-file arguments.

    for my $ifname (@ARGV)
    {
        next if $ifname =~ m@^-@;
        next unless -f $ifname;
        next unless -B $ifname;
        next unless $ifname =~
            m@\.(bmp|gif|jpe?g|pbm|pnm|png|tiff?)\z@i;
        print "Processing $ifname\n";

        if ($ifname !~ m@^/@)
        {
            $str    = $ifname;
            $ifname = "${CWD}$ifname";
            die "Unsupported filename: $str\n"
                unless &FixDotDot (-ref_path => \$ifname);
        }

        my $ofname = $ifname;
        die "Internal error\n" unless $ofname =~ s@\.\w+\z@.cune@i;

        chdir $DIR_PROG ||
            die "Error: Can't enter directory: $!:\n$DIR_PROG\n";

        open (OFD, ">$ofname") ||
            die "Error: Can't create output file: $!:\n$ofname\n";
        print OFD "\n";
        close (OFD) ||
            die "Error: Can't create output file: $!:\n$ofname\n";

        unlink ($ofname) ||
            die "Error: Can't delete output file: $!:\n$ofname\n";

        my $tempfile = "/tmp/cuneiform-$>-$$.png";
        my $cune_cmd;
        $cune_cmd =  "./cuneiform.bin -o \"$ofname\" $tempfile";
        $cune_cmd =~ s@(-o)\b@-l $language $1@ if defined ($language);

        system << "END";
convert "$ifname"       "$tempfile"
mogrify -type TrueColor "$tempfile"
$cune_cmd
END
        unlink $tempfile;
        chmod (0644, $ofname);

        if (!-f $ofname)
        {
            die "Error: cuneiform.bin failed for $ifname\n";
        }

        chdir $CWD ||
            die "Error: Can't return to directory: $!:\n$DIR_PROG\n";
    }

    undef;
}

#---------------------------------------------------------------------
#                            main program
#---------------------------------------------------------------------

&Main();                        # Call the main routine
exit ZERO;                      # Normal exit