mcu/tools/perl/Spreadsheet/WriteExcel/Workbook.pm

package Spreadsheet::WriteExcel::Workbook;

###############################################################################
#
# Workbook - A writer class for Excel Workbooks.
#
#
# Used in conjunction with Spreadsheet::WriteExcel
#
# Copyright 2000-2010, John McNamara, jmcnamara@cpan.org
#
# Documentation after __END__
#
use Exporter;
use strict;
use Carp;
use Spreadsheet::WriteExcel::BIFFwriter;
use Spreadsheet::WriteExcel::OLEwriter;
use Spreadsheet::WriteExcel::Worksheet;
use Spreadsheet::WriteExcel::Format;
use Spreadsheet::WriteExcel::Chart;
use Spreadsheet::WriteExcel::Properties ':property_sets';

use vars qw($VERSION @ISA);
@ISA = qw(Spreadsheet::WriteExcel::BIFFwriter Exporter);

$VERSION = '2.37';

###############################################################################
#
# new()
#
# Constructor. Creates a new Workbook object from a BIFFwriter object.
#
sub new {

    my $class       = shift;
    my $self        = Spreadsheet::WriteExcel::BIFFwriter->new();
    my $byte_order  = $self->{_byte_order};
    my $parser      = Spreadsheet::WriteExcel::Formula->new($byte_order);

    $self->{_filename}              = $_[0] || '';
    $self->{_parser}                = $parser;
    $self->{_tempdir}               = undef;
    $self->{_1904}                  = 0;
    $self->{_activesheet}           = 0;
    $self->{_firstsheet}            = 0;
    $self->{_selected}              = 0;
    $self->{_xf_index}              = 0;
    $self->{_fileclosed}            = 0;
    $self->{_biffsize}              = 0;
    $self->{_sheet_name}             = 'Sheet';
    $self->{_chart_name}             = 'Chart';
    $self->{_sheet_count}           = 0;
    $self->{_chart_count}           = 0;
    $self->{_url_format}            = '';
    $self->{_codepage}              = 0x04E4;
    $self->{_country}               = 1;
    $self->{_worksheets}            = [];
    $self->{_sheetnames}            = [];
    $self->{_formats}               = [];
    $self->{_palette}               = [];

    $self->{_using_tmpfile}         = 1;
    $self->{_filehandle}            = "";
    $self->{_temp_file}             = "";
    $self->{_internal_fh}           = 0;
    $self->{_fh_out}                = "";

    $self->{_str_total}             = 0;
    $self->{_str_unique}            = 0;
    $self->{_str_table}             = {};
    $self->{_str_array}             = [];
    $self->{_str_block_sizes}       = [];
    $self->{_extsst_offsets}        = [];
    $self->{_extsst_buckets}        = 0;
    $self->{_extsst_bucket_size}    = 0;

    $self->{_ext_ref_count}         = 0;
    $self->{_ext_refs}              = {};

    $self->{_mso_clusters}          = [];
    $self->{_mso_size}              = 0;

    $self->{_hideobj}               = 0;
    $self->{_compatibility}         = 0;

    $self->{_add_doc_properties}    = 0;
    $self->{_localtime}             = [localtime()];

    $self->{_defined_names}         = [];

    bless $self, $class;


    # Add the in-built style formats and the default cell format.
    $self->add_format(type => 1);                       #  0 Normal
    $self->add_format(type => 1);                       #  1 RowLevel 1
    $self->add_format(type => 1);                       #  2 RowLevel 2
    $self->add_format(type => 1);                       #  3 RowLevel 3
    $self->add_format(type => 1);                       #  4 RowLevel 4
    $self->add_format(type => 1);                       #  5 RowLevel 5
    $self->add_format(type => 1);                       #  6 RowLevel 6
    $self->add_format(type => 1);                       #  7 RowLevel 7
    $self->add_format(type => 1);                       #  8 ColLevel 1
    $self->add_format(type => 1);                       #  9 ColLevel 2
    $self->add_format(type => 1);                       # 10 ColLevel 3
    $self->add_format(type => 1);                       # 11 ColLevel 4
    $self->add_format(type => 1);                       # 12 ColLevel 5
    $self->add_format(type => 1);                       # 13 ColLevel 6
    $self->add_format(type => 1);                       # 14 ColLevel 7
    $self->add_format();                                # 15 Cell XF
    $self->add_format(type => 1, num_format => 0x2B);   # 16 Comma
    $self->add_format(type => 1, num_format => 0x29);   # 17 Comma[0]
    $self->add_format(type => 1, num_format => 0x2C);   # 18 Currency
    $self->add_format(type => 1, num_format => 0x2A);   # 19 Currency[0]
    $self->add_format(type => 1, num_format => 0x09);   # 20 Percent


    # Add the default format for hyperlinks
    $self->{_url_format} = $self->add_format(color => 'blue', underline => 1);


    # Check for a filename unless it is an existing filehandle
    if (not ref $self->{_filename} and $self->{_filename} eq '') {
        carp 'Filename required by Spreadsheet::WriteExcel->new()';
        return undef;
    }


    # Convert the filename to a filehandle to pass to the OLE writer when the
    # file is closed. If the filename is a reference it is assumed that it is
    # a valid filehandle.
    #
    if (not ref $self->{_filename}) {

        my $fh = FileHandle->new('>'. $self->{_filename});

        if (not defined $fh) {
            carp "Can't open " .
                  $self->{_filename} .
                  ". It may be in use or protected";
            return undef;
        }

        # binmode file whether platform requires it or not
        binmode($fh);
        $self->{_internal_fh} = 1;
        $self->{_fh_out}      = $fh;
    }
    else {
        $self->{_internal_fh} = 0;
        $self->{_fh_out}      = $self->{_filename};

    }


    # Set colour palette.
    $self->set_palette_xl97();

    # Load Encode if we can.
    require Encode if $] >= 5.008;

    $self->_initialize();
    $self->_get_checksum_method();
    return $self;
}


###############################################################################
#
# _initialize()
#
# Open a tmp file to store the majority of the Worksheet data. If this fails,
# for example due to write permissions, store the data in memory. This can be
# slow for large files.
#
# TODO: Move this and other methods shared with Worksheet up into BIFFwriter.
#
sub _initialize {

    my $self = shift;
    my $fh;
    my $tmp_dir;

    # The following code is complicated by Windows limitations. Porters can
    # choose a more direct method.



    # In the default case we use IO::File->new_tmpfile(). This may fail, in
    # particular with IIS on Windows, so we allow the user to specify a temp
    # directory via File::Temp.
    #
    if (defined $self->{_tempdir}) {

        # Delay loading File:Temp to reduce the module dependencies.
        eval { require File::Temp };
        die "The File::Temp module must be installed in order ".
            "to call set_tempdir().\n" if $@;


        # Trap but ignore File::Temp errors.
        eval { $fh = File::Temp::tempfile(DIR => $self->{_tempdir}) };

        # Store the failed tmp dir in case of errors.
        $tmp_dir = $self->{_tempdir} || File::Spec->tmpdir if not $fh;
    }
    else {

        $fh = IO::File->new_tmpfile();

        # Store the failed tmp dir in case of errors.
        $tmp_dir = "POSIX::tmpnam() directory" if not $fh;
    }


    # Check if the temp file creation was successful. Else store data in memory.
    if ($fh) {

        # binmode file whether platform requires it or not.
        binmode($fh);

        # Store filehandle
        $self->{_filehandle} = $fh;
    }
    else {

        # Set flag to store data in memory if XX::tempfile() failed.
        $self->{_using_tmpfile} = 0;

        if ($^W) {
            my $dir = $self->{_tempdir} || File::Spec->tmpdir();

            warn "Unable to create temp files in $tmp_dir. Data will be ".
                 "stored in memory. Refer to set_tempdir() in the ".
                 "Spreadsheet::WriteExcel documentation.\n" ;
        }
    }
}


###############################################################################
#
# _get_checksum_method.
#
# Check for modules available to calculate image checksum. Excel uses MD4 but
# MD5 will also work.
#
sub _get_checksum_method {

    my $self = shift;

    eval { require Digest::MD4};
    if (not $@) {
        $self->{_checksum_method} = 1;
        return;
    }


    eval { require Digest::Perl::MD4};
    if (not $@) {
        $self->{_checksum_method} = 2;
        return;
    }


    eval { require Digest::MD5};
    if (not $@) {
        $self->{_checksum_method} = 3;
        return;
    }

    # Default.
    $self->{_checksum_method} = 0;
}


###############################################################################
#
# _append(), overridden.
#
# Store Worksheet data in memory using the base class _append() or to a
# temporary file, the default.
#
sub _append {

    my $self = shift;
    my $data = '';

    if ($self->{_using_tmpfile}) {
        $data = join('', @_);

        # Add CONTINUE records if necessary
        $data = $self->_add_continue($data) if length($data) > $self->{_limit};

        # Protect print() from -l on the command line.
        local $\ = undef;

        print {$self->{_filehandle}} $data;
        $self->{_datasize} += length($data);
    }
    else {
        $data = $self->SUPER::_append(@_);
    }

    return $data;
}


###############################################################################
#
# get_data().
#
# Retrieves data from memory in one chunk, or from disk in $buffer
# sized chunks.
#
sub get_data {

    my $self   = shift;
    my $buffer = 4096;
    my $tmp;

    # Return data stored in memory
    if (defined $self->{_data}) {
        $tmp           = $self->{_data};
        $self->{_data} = undef;
        my $fh         = $self->{_filehandle};
        seek($fh, 0, 0) if $self->{_using_tmpfile};
        return $tmp;
    }

    # Return data stored on disk
    if ($self->{_using_tmpfile}) {
        return $tmp if read($self->{_filehandle}, $tmp, $buffer);
    }

    # No data to return
    return undef;
}


###############################################################################
#
# close()
#
# Calls finalization methods and explicitly close the OLEwriter file
# handle.
#
sub close {

    my $self = shift;

    return if $self->{_fileclosed}; # Prevent close() from being called twice.

    $self->{_fileclosed} = 1;

    return $self->_store_workbook();
}


###############################################################################
#
# DESTROY()
#
# Close the workbook if it hasn't already been explicitly closed.
#
sub DESTROY {

    my $self = shift;

    local ($@, $!, $^E, $?);

    $self->close() if not $self->{_fileclosed};
}


###############################################################################
#
# sheets(slice,...)
#
# An accessor for the _worksheets[] array
#
# Returns: an optionally sliced list of the worksheet objects in a workbook.
#
sub sheets {

    my $self = shift;

    if (@_) {
        # Return a slice of the array
        return @{$self->{_worksheets}}[@_];
    }
    else {
        # Return the entire list
        return @{$self->{_worksheets}};
    }
}


###############################################################################
#
# worksheets()
#
# An accessor for the _worksheets[] array.
# This method is now deprecated. Use the sheets() method instead.
#
# Returns: an array reference
#
sub worksheets {

    my $self = shift;

    return $self->{_worksheets};
}


###############################################################################
#
# add_worksheet($name, $encoding)
#
# Add a new worksheet to the Excel workbook.
#
# Returns: reference to a worksheet object
#
sub add_worksheet {

    my $self     = shift;
    my $index    = @{$self->{_worksheets}};

    my ($name, $encoding) = $self->_check_sheetname($_[0], $_[1]);


    # Porters take note, the following scheme of passing references to Workbook
    # data (in the \$self->{_foo} cases) instead of a reference to the Workbook
    # itself is a workaround to avoid circular references between Workbook and
    # Worksheet objects. Feel free to implement this in any way the suits your
    # language.
    #
    my @init_data = (
                         $name,
                         $index,
                         $encoding,
                        \$self->{_activesheet},
                        \$self->{_firstsheet},
                         $self->{_url_format},
                         $self->{_parser},
                         $self->{_tempdir},
                        \$self->{_str_total},
                        \$self->{_str_unique},
                        \$self->{_str_table},
                         $self->{_1904},
                         $self->{_compatibility},
                         undef, # Palette. Not used yet. See add_chart().
                    );

    my $worksheet = Spreadsheet::WriteExcel::Worksheet->new(@init_data);
    $self->{_worksheets}->[$index] = $worksheet;     # Store ref for iterator
    $self->{_sheetnames}->[$index] = $name;          # Store EXTERNSHEET names
    $self->{_parser}->set_ext_sheets($name, $index); # Store names in Formula.pm
    return $worksheet;
}

# Older method name for backwards compatibility.
*addworksheet = *add_worksheet;


###############################################################################
#
# add_chart(%args)
#
# Create a chart for embedding or as as new sheet.
#
#
sub add_chart {

    my $self     = shift;
    my %arg      = @_;
    my $name     = '';
    my $encoding = 0;
    my $index    = @{ $self->{_worksheets} };

    # Type must be specified so we can create the required chart instance.
    my $type = $arg{type};
    if ( !defined $type ) {
        croak "Must define chart type in add_chart()";
    }

    # Ensure that the chart defaults to non embedded.
    my $embedded = $arg{embedded} ||= 0;

    # Check the worksheet name for non-embedded charts.
    if ( !$embedded ) {
        ( $name, $encoding ) =
          $self->_check_sheetname( $arg{name}, $arg{name_encoding}, 1 );
    }

    my @init_data = (
                         $name,
                         $index,
                         $encoding,
                        \$self->{_activesheet},
                        \$self->{_firstsheet},
                         $self->{_url_format},
                         $self->{_parser},
                         $self->{_tempdir},
                        \$self->{_str_total},
                        \$self->{_str_unique},
                        \$self->{_str_table},
                         $self->{_1904},
                         $self->{_compatibility},
                         $self->{_palette},
                    );

    my $chart = Spreadsheet::WriteExcel::Chart->factory( $type, @init_data );

    # If the chart isn't embedded let the workbook control it.
    if ( !$embedded ) {
        $self->{_worksheets}->[$index] = $chart;    # Store ref for iterator
        $self->{_sheetnames}->[$index] = $name;     # Store EXTERNSHEET names
    }
    else {
        # Set index to 0 so that the activate() and set_first_sheet() methods
        # point back to the first worksheet if used for embedded charts.
        $chart->{_index} = 0;

        $chart->_set_embedded_config_data();
    }

    return $chart;
}


###############################################################################
#
# add_chart_ext($filename, $name)
#
# Add an externally created chart.
#
#
sub add_chart_ext {

    my $self     = shift;
    my $filename = $_[0];
    my $index    = @{$self->{_worksheets}};
    my $type     = 'external';

    my ($name, $encoding) = $self->_check_sheetname($_[1], $_[2]);


    my @init_data = (
                         $filename,
                         $name,
                         $index,
                         $encoding,
                        \$self->{_activesheet},
                        \$self->{_firstsheet},
                    );

    my $chart = Spreadsheet::WriteExcel::Chart->factory($type, @init_data);
    $self->{_worksheets}->[$index] = $chart;         # Store ref for iterator
    $self->{_sheetnames}->[$index] = $name;          # Store EXTERNSHEET names

    return $chart;
}


###############################################################################
#
# _check_sheetname($name, $encoding)
#
# Check for valid worksheet names. We check the length, if it contains any
# invalid characters and if the name is unique in the workbook.
#
sub _check_sheetname {

    my $self            = shift;
    my $name            = $_[0] || "";
    my $encoding        = $_[1] || 0;
    my $chart           = $_[2] || 0;
    my $limit           = $encoding ? 62 : 31;
    my $invalid_char    = qr([\[\]:*?/\\]);

    # Increment the Sheet/Chart number used for default sheet names below.
    if ( $chart ) {
        $self->{_chart_count}++;
    }
    else {
        $self->{_sheet_count}++;
    }

    # Supply default Sheet/Chart name if none has been defined.
    if ( $name eq "" ) {
        $encoding = 0;

        if ( $chart ) {
            $name = $self->{_chart_name} . $self->{_chart_count};
        }
        else {
            $name = $self->{_sheet_name} . $self->{_sheet_count};
        }
    }


    # Check that sheetname is <= 31 (1 or 2 byte chars). Excel limit.
    croak "Sheetname $name must be <= 31 chars" if length $name > $limit;

    # Check that Unicode sheetname has an even number of bytes
    croak 'Odd number of bytes in Unicode worksheet name:' . $name
          if $encoding == 1 and length($name) % 2;


    # Check that sheetname doesn't contain any invalid characters
    if ($encoding != 1 and $name =~ $invalid_char) {
        # Check ASCII names
        croak 'Invalid character []:*?/\\ in worksheet name: ' . $name;
    }
    else {
        # Extract any 8bit clean chars from the UTF16 name and validate them.
        for my $wchar ($name =~ /../sg) {
            my ($hi, $lo) = unpack "aa", $wchar;
            if ($hi eq "\0" and $lo =~ $invalid_char) {
                croak 'Invalid character []:*?/\\ in worksheet name: ' . $name;
            }
        }
    }


    # Handle utf8 strings in perl 5.8.
    if ($] >= 5.008) {
        require Encode;

        if (Encode::is_utf8($name)) {
            $name = Encode::encode("UTF-16BE", $name);
            $encoding = 1;
        }
    }


    # Check that the worksheet name doesn't already exist since this is a fatal
    # error in Excel 97. The check must also exclude case insensitive matches
    # since the names 'Sheet1' and 'sheet1' are equivalent. The tests also have
    # to take the encoding into account.
    #
    foreach my $worksheet (@{$self->{_worksheets}}) {
        my $name_a  = $name;
        my $encd_a  = $encoding;
        my $name_b  = $worksheet->{_name};
        my $encd_b  = $worksheet->{_encoding};
        my $error   = 0;

        if    ($encd_a == 0 and $encd_b == 0) {
            $error  = 1 if lc($name_a) eq lc($name_b);
        }
        elsif ($encd_a == 0 and $encd_b == 1) {
            $name_a = pack "n*", unpack "C*", $name_a;
            $error  = 1 if lc($name_a) eq lc($name_b);
        }
        elsif ($encd_a == 1 and $encd_b == 0) {
            $name_b = pack "n*", unpack "C*", $name_b;
            $error  = 1 if lc($name_a) eq lc($name_b);
        }
        elsif ($encd_a == 1 and $encd_b == 1) {
            # We can do a true case insensitive test with Perl 5.8 and utf8.
            if ($] >= 5.008) {
                $name_a = Encode::decode("UTF-16BE", $name_a);
                $name_b = Encode::decode("UTF-16BE", $name_b);
                $error  = 1 if lc($name_a) eq lc($name_b);
            }
            else {
            # We can't easily do a case insensitive test of the UTF16 names.
            # As a special case we check if all of the high bytes are nulls and
            # then do an ASCII style case insensitive test.

                # Strip out the high bytes (funkily).
                my $hi_a = grep {ord} $name_a =~ /(.)./sg;
                my $hi_b = grep {ord} $name_b =~ /(.)./sg;

                if ($hi_a or $hi_b) {
                    $error  = 1 if    $name_a  eq    $name_b;
                }
                else {
                    $error  = 1 if lc($name_a) eq lc($name_b);
                }
            }
        }

        # If any of the cases failed we throw the error here.
        if ($error) {
            croak "Worksheet name '$name', with case ignored, " .
                  "is already in use";
        }
    }

    return ($name,  $encoding);
}


###############################################################################
#
# add_format(%properties)
#
# Add a new format to the Excel workbook. This adds an XF record and
# a FONT record. Also, pass any properties to the Format::new().
#
sub add_format {

    my $self = shift;

    my $format = Spreadsheet::WriteExcel::Format->new($self->{_xf_index}, @_);

    $self->{_xf_index} += 1;
    push @{$self->{_formats}}, $format; # Store format reference

    return $format;
}

# Older method name for backwards compatibility.
*addformat = *add_format;


###############################################################################
#
# compatibility_mode()
#
# Set the compatibility mode.
#
# Excel doesn't require every possible Biff record to be present in a file.
# In particular if the indexing records INDEX, ROW and DBCELL aren't present
# it just ignores the fact and reads the cells anyway. This is also true of
# the EXTSST record. Gnumeric and OOo also take this approach. This allows
# WriteExcel to ignore these records in order to minimise the amount of data
# stored in memory. However, other third party applications that read Excel
# files often expect these records to be present. In "compatibility mode"
# WriteExcel writes these records and tries to be as close to an Excel
# generated file as possible.
#
# This requires additional data to be stored in memory until the file is
# about to be written. This incurs a memory and speed penalty and may not be
# suitable for very large files.
#
sub compatibility_mode {

    my $self      = shift;

    croak "compatibility_mode() must be called before add_worksheet()"
          if $self->sheets();

    if (defined($_[0])) {
        $self->{_compatibility} = $_[0];
    }
    else {
        $self->{_compatibility} = 1;
    }
}


###############################################################################
#
# set_1904()
#
# Set the date system: 0 = 1900 (the default), 1 = 1904
#
sub set_1904 {

    my $self      = shift;

    croak "set_1904() must be called before add_worksheet()"
          if $self->sheets();


    if (defined($_[0])) {
        $self->{_1904} = $_[0];
    }
    else {
        $self->{_1904} = 1;
    }
}


###############################################################################
#
# get_1904()
#
# Return the date system: 0 = 1900, 1 = 1904
#
sub get_1904 {

    my $self = shift;

    return $self->{_1904};
}


###############################################################################
#
# set_custom_color()
#
# Change the RGB components of the elements in the colour palette.
#
sub set_custom_color {

    my $self    = shift;


    # Match a HTML #xxyyzz style parameter
    if (defined $_[1] and $_[1] =~ /^#(\w\w)(\w\w)(\w\w)/ ) {
        @_ = ($_[0], hex $1, hex $2, hex $3);
    }


    my $index   = $_[0] || 0;
    my $red     = $_[1] || 0;
    my $green   = $_[2] || 0;
    my $blue    = $_[3] || 0;

    my $aref    = $self->{_palette};

    # Check that the colour index is the right range
    if ($index < 8 or $index > 64) {
        carp "Color index $index outside range: 8 <= index <= 64";
        return 0;
    }

    # Check that the colour components are in the right range
    if ( ($red   < 0 or $red   > 255) ||
         ($green < 0 or $green > 255) ||
         ($blue  < 0 or $blue  > 255) )
    {
        carp "Color component outside range: 0 <= color <= 255";
        return 0;
    }

    $index -=8; # Adjust colour index (wingless dragonfly)

    # Set the RGB value
    $aref->[$index] = [$red, $green, $blue, 0];

    return $index +8;
}


###############################################################################
#
# set_palette_xl97()
#
# Sets the colour palette to the Excel 97+ default.
#
sub set_palette_xl97 {

    my $self = shift;

    $self->{_palette} = [
                            [0x00, 0x00, 0x00, 0x00],   # 8
                            [0xff, 0xff, 0xff, 0x00],   # 9
                            [0xff, 0x00, 0x00, 0x00],   # 10
                            [0x00, 0xff, 0x00, 0x00],   # 11
                            [0x00, 0x00, 0xff, 0x00],   # 12
                            [0xff, 0xff, 0x00, 0x00],   # 13
                            [0xff, 0x00, 0xff, 0x00],   # 14
                            [0x00, 0xff, 0xff, 0x00],   # 15
                            [0x80, 0x00, 0x00, 0x00],   # 16
                            [0x00, 0x80, 0x00, 0x00],   # 17
                            [0x00, 0x00, 0x80, 0x00],   # 18
                            [0x80, 0x80, 0x00, 0x00],   # 19
                            [0x80, 0x00, 0x80, 0x00],   # 20
                            [0x00, 0x80, 0x80, 0x00],   # 21
                            [0xc0, 0xc0, 0xc0, 0x00],   # 22
                            [0x80, 0x80, 0x80, 0x00],   # 23
                            [0x99, 0x99, 0xff, 0x00],   # 24
                            [0x99, 0x33, 0x66, 0x00],   # 25
                            [0xff, 0xff, 0xcc, 0x00],   # 26
                            [0xcc, 0xff, 0xff, 0x00],   # 27
                            [0x66, 0x00, 0x66, 0x00],   # 28
                            [0xff, 0x80, 0x80, 0x00],   # 29
                            [0x00, 0x66, 0xcc, 0x00],   # 30
                            [0xcc, 0xcc, 0xff, 0x00],   # 31
                            [0x00, 0x00, 0x80, 0x00],   # 32
                            [0xff, 0x00, 0xff, 0x00],   # 33
                            [0xff, 0xff, 0x00, 0x00],   # 34
                            [0x00, 0xff, 0xff, 0x00],   # 35
                            [0x80, 0x00, 0x80, 0x00],   # 36
                            [0x80, 0x00, 0x00, 0x00],   # 37
                            [0x00, 0x80, 0x80, 0x00],   # 38
                            [0x00, 0x00, 0xff, 0x00],   # 39
                            [0x00, 0xcc, 0xff, 0x00],   # 40
                            [0xcc, 0xff, 0xff, 0x00],   # 41
                            [0xcc, 0xff, 0xcc, 0x00],   # 42
                            [0xff, 0xff, 0x99, 0x00],   # 43
                            [0x99, 0xcc, 0xff, 0x00],   # 44
                            [0xff, 0x99, 0xcc, 0x00],   # 45
                            [0xcc, 0x99, 0xff, 0x00],   # 46
                            [0xff, 0xcc, 0x99, 0x00],   # 47
                            [0x33, 0x66, 0xff, 0x00],   # 48
                            [0x33, 0xcc, 0xcc, 0x00],   # 49
                            [0x99, 0xcc, 0x00, 0x00],   # 50
                            [0xff, 0xcc, 0x00, 0x00],   # 51
                            [0xff, 0x99, 0x00, 0x00],   # 52
                            [0xff, 0x66, 0x00, 0x00],   # 53
                            [0x66, 0x66, 0x99, 0x00],   # 54
                            [0x96, 0x96, 0x96, 0x00],   # 55
                            [0x00, 0x33, 0x66, 0x00],   # 56
                            [0x33, 0x99, 0x66, 0x00],   # 57
                            [0x00, 0x33, 0x00, 0x00],   # 58
                            [0x33, 0x33, 0x00, 0x00],   # 59
                            [0x99, 0x33, 0x00, 0x00],   # 60
                            [0x99, 0x33, 0x66, 0x00],   # 61
                            [0x33, 0x33, 0x99, 0x00],   # 62
                            [0x33, 0x33, 0x33, 0x00],   # 63
                        ];

    return 0;
}


###############################################################################
#
# set_tempdir()
#
# Change the default temp directory used by _initialize() in Worksheet.pm.
#
sub set_tempdir {

    my $self = shift;

    # Windows workaround. See Worksheet::_initialize()
    my $dir  = shift || '';

    croak "$dir is not a valid directory"       if $dir ne '' and not -d $dir;
    croak "set_tempdir must be called before add_worksheet" if $self->sheets();

    $self->{_tempdir} = $dir ;
}


###############################################################################
#
# set_codepage()
#
# See also the _store_codepage method. This is used to store the code page, i.e.
# the character set used in the workbook.
#
sub set_codepage {

    my $self        = shift;

    my $codepage    = $_[0] || 1;
       $codepage    = 0x04E4 if $codepage == 1;
       $codepage    = 0x8000 if $codepage == 2;

    $self->{_codepage} = $codepage;
}


###############################################################################
#
# set_country()
#
# See also the _store_country method. This is used to store the country code.
# Some non-english versions of Excel may need this set to some value other
# than 1 = "United States". In general the country code is equal to the
# international dialling code.
#
sub set_country {

    my $self            = shift;

    $self->{_country}   = $_[0] || 1;
}







###############################################################################
#
# define_name()
#
# TODO.
#
sub define_name {

    my $self        = shift;
    my $name        = shift;
    my $formula     = shift;
    my $encoding    = shift || 0;
    my $sheet_index = 0;
    my @tokens;

    my $full_name   = $name;

    if ($name =~ /^(.*)!(.*)$/) {
        my $sheetname   = $1;
        $name           = $2;
        $sheet_index    = 1 + $self->{_parser}->_get_sheet_index($sheetname);
    }



    # Strip the = sign at the beginning of the formula string
    $formula    =~ s(^=)();

    # Parse the formula using the parser in Formula.pm
    my $parser  = $self->{_parser};

    # In order to raise formula errors from the point of view of the calling
    # program we use an eval block and re-raise the error from here.
    #
    eval { @tokens = $parser->parse_formula($formula) };

    if ($@) {
        $@ =~ s/\n$//;  # Strip the \n used in the Formula.pm die()
        croak $@;       # Re-raise the error
    }

    # Force 2d ranges to be a reference class.
    s/_ref3d/_ref3dR/     for @tokens;
    s/_range3d/_range3dR/ for @tokens;


    # Parse the tokens into a formula string.
    $formula = $parser->parse_tokens(@tokens);



    $full_name = lc $full_name;

    push @{$self->{_defined_names}},   {
                                            name        => $name,
                                            encoding    => $encoding,
                                            sheet_index => $sheet_index,
                                            formula     => $formula,
                                        };

    my $index = scalar @{$self->{_defined_names}};

    $parser->set_ext_name($name, $index);
}









###############################################################################
#
# set_properties()
#
# Set the document properties such as Title, Author etc. These are written to
# property sets in the OLE container.
#
sub set_properties {

    my $self    = shift;
    my %param;

    # Ignore if no args were passed.
    return -1 unless @_;


    # Allow the parameters to be passed as a hash or hash ref.
    if (ref $_[0] eq 'HASH') {
        %param = %{$_[0]};
    }
    else {
        %param = @_;
    }


    # List of valid input parameters.
    my %properties = (
                          codepage      => [0x0001, 'VT_I2'      ],
                          title         => [0x0002, 'VT_LPSTR'   ],
                          subject       => [0x0003, 'VT_LPSTR'   ],
                          author        => [0x0004, 'VT_LPSTR'   ],
                          keywords      => [0x0005, 'VT_LPSTR'   ],
                          comments      => [0x0006, 'VT_LPSTR'   ],
                          last_author   => [0x0008, 'VT_LPSTR'   ],
                          created       => [0x000C, 'VT_FILETIME'],
                          category      => [0x0002, 'VT_LPSTR'   ],
                          manager       => [0x000E, 'VT_LPSTR'   ],
                          company       => [0x000F, 'VT_LPSTR'   ],
                          utf8          => 1,
                      );

    # Check for valid input parameters.
    for my $parameter (keys %param) {
        if (not exists $properties{$parameter}) {
            carp "Unknown parameter '$parameter' in set_properties()";
            return -1;
        }
    }


    # Set the creation time unless specified by the user.
    if (!exists $param{created}){
        $param{created} = $self->{_localtime};
    }


    #
    # Create the SummaryInformation property set.
    #

    # Get the codepage of the strings in the property set.
    my @strings      = qw(title subject author keywords comments last_author);
    $param{codepage} = $self->_get_property_set_codepage(\%param,
                                                         \@strings);

    # Create an array of property set values.
    my @property_sets;

    for my $property (qw(codepage title    subject     author
                         keywords comments last_author created))
    {
        if (exists $param{$property} && defined $param{$property}) {
            push @property_sets, [
                                    $properties{$property}->[0],
                                    $properties{$property}->[1],
                                    $param{$property}
                                 ];
        }
    }

    # Pack the property sets.
    $self->{summary} = create_summary_property_set(\@property_sets);


    #
    # Create the DocSummaryInformation property set.
    #

    # Get the codepage of the strings in the property set.
    @strings         = qw(category manager company);
    $param{codepage} = $self->_get_property_set_codepage(\%param,
                                                         \@strings);

    # Create an array of property set values.
    @property_sets = ();

    for my $property (qw(codepage category manager company))
    {
        if (exists $param{$property} && defined $param{$property}) {
            push @property_sets, [
                                    $properties{$property}->[0],
                                    $properties{$property}->[1],
                                    $param{$property}
                                 ];
        }
    }

    # Pack the property sets.
    $self->{doc_summary} = create_doc_summary_property_set(\@property_sets);

    # Set a flag for when the files is written.
    $self->{_add_doc_properties} = 1;
}


###############################################################################
#
# _get_property_set_codepage()
#
# Get the character codepage used by the strings in a property set. If one of
# the strings used is utf8 then the codepage is marked as utf8. Otherwise
# Latin 1 is used (although in our case this is limited to 7bit ASCII).
#
sub _get_property_set_codepage {

    my $self        = shift;

    my $params      = $_[0];
    my $strings     = $_[1];

    # Allow for manually marked utf8 strings.
    return 0xFDE9 if defined $params->{utf8};

    # Check for utf8 strings in perl 5.8.
    if ($] >= 5.008) {
        require Encode;
        for my $string (@{$strings }) {
            next unless exists $params->{$string};
            return 0xFDE9 if Encode::is_utf8($params->{$string});
        }
    }

    return 0x04E4; # Default codepage, Latin 1.
}


###############################################################################
#
# _store_workbook()
#
# Assemble worksheets into a workbook and send the BIFF data to an OLE
# storage.
#
sub _store_workbook {

    my $self = shift;

    # Add a default worksheet if non have been added.
    $self->add_worksheet() if not @{$self->{_worksheets}};

    # Calculate size required for MSO records and update worksheets.
    $self->_calc_mso_sizes();

    # Ensure that at least one worksheet has been selected.
    if ($self->{_activesheet} == 0) {
        @{$self->{_worksheets}}[0]->{_selected} = 1;
        @{$self->{_worksheets}}[0]->{_hidden}   = 0;
    }

    # Calculate the number of selected sheet tabs and set the active sheet.
    foreach my $sheet (@{$self->{_worksheets}}) {
        $self->{_selected}++ if $sheet->{_selected};
        $sheet->{_active} = 1 if $sheet->{_index} == $self->{_activesheet};
    }

    # Add Workbook globals
    $self->_store_bof(0x0005);
    $self->_store_codepage();
    $self->_store_window1();
    $self->_store_hideobj();
    $self->_store_1904();
    $self->_store_all_fonts();
    $self->_store_all_num_formats();
    $self->_store_all_xfs();
    $self->_store_all_styles();
    $self->_store_palette();

    # Calculate the offsets required by the BOUNDSHEET records
    $self->_calc_sheet_offsets();

    # Add BOUNDSHEET records.
    foreach my $sheet (@{$self->{_worksheets}}) {
        $self->_store_boundsheet($sheet->{_name},
                                 $sheet->{_offset},
                                 $sheet->{_sheet_type},
                                 $sheet->{_hidden},
                                 $sheet->{_encoding});
    }

    # NOTE: If any records are added between here and EOF the
    # _calc_sheet_offsets() should be updated to include the new length.
    $self->_store_country();
    if ($self->{_ext_ref_count}) {
        $self->_store_supbook();
        $self->_store_externsheet();
        $self->_store_names();
    }
    $self->_add_mso_drawing_group();
    $self->_store_shared_strings();
    $self->_store_extsst();

    # End Workbook globals
    $self->_store_eof();

    # Store the workbook in an OLE container
    return $self->_store_OLE_file();
}


###############################################################################
#
# _store_OLE_file()
#
# Store the workbook in an OLE container using the default handler or using
# OLE::Storage_Lite if the workbook data is > ~ 7MB.
#
sub _store_OLE_file {

    my $self    = shift;
    my $maxsize = 7_087_104;

    if (!$self->{_add_doc_properties} && $self->{_biffsize} <= $maxsize) {
        # Write the OLE file using OLEwriter if data <= 7MB
        my $OLE  = Spreadsheet::WriteExcel::OLEwriter->new($self->{_fh_out});

        # Write the BIFF data without the OLE container for testing.
        $OLE->{_biff_only} = $self->{_biff_only};

        # Indicate that we created the filehandle and want to close it.
        $OLE->{_internal_fh} = $self->{_internal_fh};

        $OLE->set_size($self->{_biffsize});
        $OLE->write_header();

        while (my $tmp = $self->get_data()) {
            $OLE->write($tmp);
        }

        foreach my $worksheet (@{$self->{_worksheets}}) {
            while (my $tmp = $worksheet->get_data()) {
                $OLE->write($tmp);
            }
        }

        return $OLE->close();
    }
    else {
        # Write the OLE file using OLE::Storage_Lite if data > 7MB
        eval { require OLE::Storage_Lite };

        if (not $@) {

            # Protect print() from -l on the command line.
            local $\ = undef;

            my @streams;

            # Create the Workbook stream.
            my $stream   = pack 'v*', unpack 'C*', 'Workbook';
            my $workbook = OLE::Storage_Lite::PPS::File->newFile($stream);

            while (my $tmp = $self->get_data()) {
                $workbook->append($tmp);
            }

            foreach my $worksheet (@{$self->{_worksheets}}) {
                while (my $tmp = $worksheet->get_data()) {
                    $workbook->append($tmp);
                }
            }

            push @streams, $workbook;


            # Create the properties streams, if any.
            if ($self->{_add_doc_properties}) {
                my $stream;
                my $summary;

                $stream  = pack 'v*', unpack 'C*', "\5SummaryInformation";
                $summary = $self->{summary};
                $summary = OLE::Storage_Lite::PPS::File->new($stream, $summary);
                push @streams, $summary;

                $stream  = pack 'v*', unpack 'C*', "\5DocumentSummaryInformation";
                $summary = $self->{doc_summary};
                $summary = OLE::Storage_Lite::PPS::File->new($stream, $summary);
                push @streams, $summary;
            }

            # Create the OLE root document and add the substreams.
            my @localtime = @{ $self->{_localtime} };
            splice(@localtime, 6);

            my $ole_root = OLE::Storage_Lite::PPS::Root->new(\@localtime,
                                                             \@localtime,
                                                             \@streams);
            $ole_root->save($self->{_filename});


            # Close the filehandle if it was created internally.
            return CORE::close($self->{_fh_out}) if $self->{_internal_fh};
        }
        else {
            # File in greater than limit, set $! to "File too large"
            $! = 27; # Perl error code "File too large"

            croak "Maximum Spreadsheet::WriteExcel filesize, $maxsize bytes, ".
                  "exceeded. To create files bigger than this limit please "  .
                  "install OLE::Storage_Lite\n";

            # return 0;
        }
    }
}


###############################################################################
#
# _calc_sheet_offsets()
#
# Calculate Worksheet BOF offsets records for use in the BOUNDSHEET records.
#
sub _calc_sheet_offsets {

    my $self    = shift;
    my $BOF     = 12;
    my $EOF     = 4;
    my $offset  = $self->{_datasize};

    # Add the length of the COUNTRY record
    $offset += 8;

    # Add the length of the SST and associated CONTINUEs
    $offset += $self->_calculate_shared_string_sizes();

    # Add the length of the EXTSST record.
    $offset += $self->_calculate_extsst_size();

    # Add the length of the SUPBOOK, EXTERNSHEET and NAME records
    $offset += $self->_calculate_extern_sizes();

    # Add the length of the MSODRAWINGGROUP records including an extra 4 bytes
    # for any CONTINUE headers. See _add_mso_drawing_group_continue().
    my $mso_size = $self->{_mso_size};
    $mso_size += 4 * int(($mso_size -1) / $self->{_limit});
    $offset   += $mso_size ;

    foreach my $sheet (@{$self->{_worksheets}}) {
        $offset += $BOF + length($sheet->{_name});
    }

    $offset += $EOF;

    foreach my $sheet (@{$self->{_worksheets}}) {
        $sheet->{_offset} = $offset;
        $sheet->_close();
        $offset += $sheet->{_datasize};
    }

    $self->{_biffsize} = $offset;
}


###############################################################################
#
# _calc_mso_sizes()
#
# Calculate the MSODRAWINGGROUP sizes and the indexes of the Worksheet
# MSODRAWING records.
#
# In the following SPID is shape id, according to Escher nomenclature.
#
sub _calc_mso_sizes {

    my $self            = shift;

    my $mso_size        = 0;    # Size of the MSODRAWINGGROUP record
    my $start_spid      = 1024; # Initial spid for each sheet
    my $max_spid        = 1024; # spidMax
    my $num_clusters    = 1;    # cidcl
    my $shapes_saved    = 0;    # cspSaved
    my $drawings_saved  = 0;    # cdgSaved
    my @clusters        = ();


    $self->_process_images();

    # Add Bstore container size if there are images.
    $mso_size += 8 if @{$self->{_images_data}};


    # Iterate through the worksheets, calculate the MSODRAWINGGROUP parameters
    # and space required to store the record and the MSODRAWING parameters
    # required by each worksheet.
    #
    foreach my $sheet (@{$self->{_worksheets}}) {
        next unless $sheet->{_sheet_type} == 0x0000; # Ignore charts.

        my $num_images     = $sheet->{_num_images} || 0;
        my $image_mso_size = $sheet->{_image_mso_size} || 0;
        my $num_comments   = $sheet->_prepare_comments();
        my $num_charts     = $sheet->_prepare_charts();
        my $num_filters    = $sheet->{_filter_count};

        next unless $num_images + $num_comments + $num_charts +$num_filters;


        # Include 1 parent MSODRAWING shape, per sheet, in the shape count.
        my $num_shapes   += 1 + $num_images
                              + $num_comments
                              + $num_charts
                              + $num_filters;
           $shapes_saved += $num_shapes;
           $mso_size     += $image_mso_size;


        # Add a drawing object for each sheet with comments.
        $drawings_saved++;


        # For each sheet start the spids at the next 1024 interval.
        $max_spid   = 1024 * (1 + int(($max_spid -1)/1024));
        $start_spid = $max_spid;


        # Max spid for each sheet and eventually for the workbook.
        $max_spid  += $num_shapes;


        # Store the cluster ids
        for (my $i = $num_shapes; $i > 0; $i -= 1024) {
            $num_clusters  += 1;
            $mso_size      += 8;
            my $size        = $i > 1024 ? 1024 : $i;

            push @clusters, [$drawings_saved, $size];
        }


        # Pass calculated values back to the worksheet
        $sheet->{_object_ids} = [$start_spid, $drawings_saved,
                                  $num_shapes, $max_spid -1];
    }


    # Calculate the MSODRAWINGGROUP size if we have stored some shapes.
    $mso_size              += 86 if $mso_size; # Smallest size is 86+8=94


    $self->{_mso_size}      = $mso_size;
    $self->{_mso_clusters}  = [
                                $max_spid, $num_clusters, $shapes_saved,
                                $drawings_saved, [@clusters]
                              ];
}



###############################################################################
#
# _process_images()
#
# We need to process each image in each worksheet and extract information.
# Some of this information is stored and used in the Workbook and some is
# passed back into each Worksheet. The overall size for the image related
# BIFF structures in the Workbook is calculated here.
#
# MSO size =  8 bytes for bstore_container +
#            44 bytes for blip_store_entry +
#            25 bytes for blip
#          = 77 + image size.
#
sub _process_images {

    my $self = shift;

    my %images_seen;
    my @image_data;
    my @previous_images;
    my $image_id    = 1;
    my $images_size = 0;


    foreach my $sheet (@{$self->{_worksheets}}) {
        next unless $sheet->{_sheet_type} == 0x0000; # Ignore charts.
        next unless $sheet->_prepare_images();

        my $num_images      = 0;
        my $image_mso_size  = 0;


        for my $image_ref (@{$sheet->{_images_array}}) {
            my $filename = $image_ref->[2];
            $num_images++;

            #
            # For each Worksheet image we get a structure like this
            # [
            #   $row,
            #   $col,
            #   $name,
            #   $x_offset,
            #   $y_offset,
            #   $scale_x,
            #   $scale_y,
            # ]
            #
            # And we add additional information:
            #
            #   $image_id,
            #   $type,
            #   $width,
            #   $height;

            if (not exists $images_seen{$filename}) {
                # TODO should also match seen images based on checksum.

                # Open the image file and import the data.
                my $fh = FileHandle->new($filename);
                croak "Couldn't import $filename: $!" unless defined $fh;
                binmode $fh;

                # Slurp the file into a string and do some size calcs.
                my $data        = do {local $/; <$fh>};
                my $size        = length $data;
                my $checksum1   = $self->_image_checksum($data, $image_id);
                my $checksum2   = $checksum1;
                my $ref_count   = 1;


                # Process the image and extract dimensions.
                my ($type, $width, $height);

                # Test for PNGs...
                if    (unpack('x A3', $data) eq 'PNG') {
                    ($type, $width, $height) = $self->_process_png($data);
                }
                # Test for JFIF and Exif JPEGs...
                elsif ( (unpack('n', $data) == 0xFFD8) &&
                            ( (unpack('x6 A4', $data) eq 'JFIF') ||
                              (unpack('x6 A4', $data) eq 'Exif')
                            )
                      )
                {
                    ($type, $width, $height) = $self->_process_jpg($data, $filename);
                }
                # Test for BMPs...
                elsif (unpack('A2',   $data) eq 'BM') {
                    ($type, $width, $height) = $self->_process_bmp($data,
                                                                   $filename);
                    # The 14 byte header of the BMP is stripped off.
                    $data       = substr $data, 14;

                    # A checksum of the new image data is also required.
                    $checksum2  = $self->_image_checksum($data,
                                                         $image_id,
                                                         $image_id
                                                         );

                    # Adjust size -14 (header) + 16 (extra checksum).
                    $size += 2;
                }
                else {
                    croak "Unsupported image format for file: $filename\n";
                }


                # Push the new data back into the Worksheet array;
                push @$image_ref, $image_id, $type, $width, $height;

                # Also store new data for use in duplicate images.
                push @previous_images, [$image_id, $type, $width, $height];


                # Store information required by the Workbook.
                push @image_data, [$ref_count, $type, $data, $size,
                                   $checksum1, $checksum2];

                # Keep track of overall data size.
                $images_size       += $size +61; # Size for bstore container.
                $image_mso_size    += $size +69; # Size for dgg container.

                $images_seen{$filename} = $image_id++;
                $fh->close;
            }
            else {
                # We've processed this file already.
                my $index = $images_seen{$filename} -1;

                # Increase image reference count.
                $image_data[$index]->[0]++;

                # Add previously calculated data back onto the Worksheet array.
                # $image_id, $type, $width, $height
                my $a_ref = $sheet->{_images_array}->[$index];
                push @$image_ref, @{$previous_images[$index]};
            }
        }

        # Store information required by the Worksheet.
        $sheet->{_num_images}     = $num_images;
        $sheet->{_image_mso_size} = $image_mso_size;

    }


    # Store information required by the Workbook.
    $self->{_images_size} = $images_size;
    $self->{_images_data} = \@image_data; # Store the data for MSODRAWINGGROUP.

}


###############################################################################
#
# _image_checksum()
#
# Generate a checksum for the image using whichever module is available..The
# available modules are checked in _get_checksum_method(). Excel uses an MD4
# checksum but any other will do. In the event of no checksum module being
# available we simulate a checksum using the image index.
#
sub _image_checksum {

    my $self    = shift;

    my $data    = $_[0];
    my $index1  = $_[1];
    my $index2  = $_[2] || 0;

    if    ($self->{_checksum_method} == 1) {
        # Digest::MD4
        return Digest::MD4::md4_hex($data);
    }
    elsif ($self->{_checksum_method} == 2) {
        # Digest::Perl::MD4
        return Digest::Perl::MD4::md4_hex($data);
    }
    elsif ($self->{_checksum_method} == 3) {
        # Digest::MD5
        return Digest::MD5::md5_hex($data);
    }
    else {
        # Default
        return sprintf '%016X%016X', $index2, $index1;
    }
}


###############################################################################
#
# _process_png()
#
# Extract width and height information from a PNG file.
#
sub _process_png {

    my $self    = shift;

    my $type    = 6; # Excel Blip type (MSOBLIPTYPE).
    my $width   = unpack "N", substr $_[0], 16, 4;
    my $height  = unpack "N", substr $_[0], 20, 4;

    return ($type, $width, $height);
}


###############################################################################
#
# _process_bmp()
#
# Extract width and height information from a BMP file.
#
# Most of these checks came from the old Worksheet::_process_bitmap() method.
#
sub _process_bmp {

    my $self     = shift;
    my $data     = $_[0];
    my $filename = $_[1];
    my $type     = 7; # Excel Blip type (MSOBLIPTYPE).


    # Check that the file is big enough to be a bitmap.
    if (length $data <= 0x36) {
        croak "$filename doesn't contain enough data.";
    }


    # Read the bitmap width and height. Verify the sizes.
    my ($width, $height) = unpack "x18 V2", $data;

    if ($width > 0xFFFF) {
        croak "$filename: largest image width $width supported is 65k.";
    }

    if ($height > 0xFFFF) {
        croak "$filename: largest image height supported is 65k.";
    }

    # Read the bitmap planes and bpp data. Verify them.
    my ($planes, $bitcount) = unpack "x26 v2", $data;

    if ($bitcount != 24) {
        croak "$filename isn't a 24bit true color bitmap.";
    }

    if ($planes != 1) {
        croak "$filename: only 1 plane supported in bitmap image.";
    }


    # Read the bitmap compression. Verify compression.
    my $compression = unpack "x30 V", $data;

    if ($compression != 0) {
        croak "$filename: compression not supported in bitmap image.";
    }

    return ($type, $width, $height);
}


###############################################################################
#
# _process_jpg()
#
# Extract width and height information from a JPEG file.
#
sub _process_jpg {

    my $self     = shift;
    my $data     = $_[0];
    my $filename = $_[1];
    my $type     = 5; # Excel Blip type (MSOBLIPTYPE).
    my $width;
    my $height;

    my $offset = 2;
    my $data_length = length $data;

    # Search through the image data to find the 0xFFC0 marker. The height and
    # width are contained in the data for that sub element.
    while ($offset < $data_length) {

        my $marker  = unpack "n", substr $data, $offset,    2;
        my $length  = unpack "n", substr $data, $offset +2, 2;

        if ($marker == 0xFFC0 || $marker == 0xFFC2) {
            $height = unpack "n", substr $data, $offset +5, 2;
            $width  = unpack "n", substr $data, $offset +7, 2;
            last;
        }

        $offset = $offset + $length + 2;
        last if $marker == 0xFFDA;
    }

    if (not defined $height) {
        croak "$filename: no size data found in jpeg image.\n";
    }

    return ($type, $width, $height);
}


###############################################################################
#
# _store_all_fonts()
#
# Store the Excel FONT records.
#
sub _store_all_fonts {

    my $self    = shift;

    my $format  = $self->{_formats}->[15]; # The default cell format.
    my $font    = $format->get_font();

    # Fonts are 0-indexed. According to the SDK there is no index 4,
    for (0..3) {
        $self->_append($font);
    }


    # Add the default fonts for charts and comments. This aren't connected
    # to XF formats. Note, the font size, and some other properties of
    # chart fonts are set in the FBI record of the chart.
    my $tmp_format;

    # Index 5. Axis numbers.
    $tmp_format = Spreadsheet::WriteExcel::Format->new(
        undef,
        font_only => 1,
    );
    $self->_append( $tmp_format->get_font() );

    # Index 6. Series names.
    $tmp_format = Spreadsheet::WriteExcel::Format->new(
        undef,
        font_only => 1,
    );
    $self->_append( $tmp_format->get_font() );

    # Index 7. Title.
    $tmp_format = Spreadsheet::WriteExcel::Format->new(
        undef,
        font_only => 1,
        bold      => 1,
    );
    $self->_append( $tmp_format->get_font() );

    # Index 8. Axes.
    $tmp_format = Spreadsheet::WriteExcel::Format->new(
        undef,
        font_only => 1,
        bold      => 1,
    );
    $self->_append( $tmp_format->get_font() );

    # Index 9. Comments.
    $tmp_format = Spreadsheet::WriteExcel::Format->new(
        undef,
        font_only => 1,
        font      => 'Tahoma',
        size      => 8,
    );
    $self->_append( $tmp_format->get_font() );


    # Iterate through the XF objects and write a FONT record if it isn't the
    # same as the default FONT and if it hasn't already been used.
    #
    my %fonts;
    my $key;
    my $index = 10;                  # The first user defined FONT

    $key = $format->get_font_key(); # The default font for cell formats.
    $fonts{$key} = 0;               # Index of the default font

    # Fonts that are marked as '_font_only' are always stored. These are used
    # mainly for charts and may not have an associated XF record.

    foreach $format (@{$self->{_formats}}) {
        $key = $format->get_font_key();

        if (not $format->{_font_only} and exists $fonts{$key}) {
            # FONT has already been used
            $format->{_font_index} = $fonts{$key};
        }
        else {
            # Add a new FONT record

            if (not $format->{_font_only}) {
                $fonts{$key} = $index;
            }

            $format->{_font_index} = $index;
            $index++;
            $font = $format->get_font();
            $self->_append($font);
        }
    }
}


###############################################################################
#
# _store_all_num_formats()
#
# Store user defined numerical formats i.e. FORMAT records
#
sub _store_all_num_formats {

    my $self = shift;

    my %num_formats;
    my @num_formats;
    my $num_format;
    my $index = 164; # User defined FORMAT records start from 0xA4


    # Iterate through the XF objects and write a FORMAT record if it isn't a
    # built-in format type and if the FORMAT string hasn't already been used.
    #
    foreach my $format (@{$self->{_formats}}) {
        my $num_format = $format->{_num_format};
        my $encoding   = $format->{_num_format_enc};

        # Check if $num_format is an index to a built-in format.
        # Also check for a string of zeros, which is a valid format string
        # but would evaluate to zero.
        #
        if ($num_format !~ m/^0+\d/) {
            next if $num_format =~ m/^\d+$/; # built-in
        }

        if (exists($num_formats{$num_format})) {
            # FORMAT has already been used
            $format->{_num_format} = $num_formats{$num_format};
        }
        else{
            # Add a new FORMAT
            $num_formats{$num_format} = $index;
            $format->{_num_format}    = $index;
            $self->_store_num_format($num_format, $index, $encoding);
            $index++;
        }
    }
}


###############################################################################
#
# _store_all_xfs()
#
# Write all XF records.
#
sub _store_all_xfs {

    my $self = shift;

    foreach my $format (@{$self->{_formats}}) {
        my $xf = $format->get_xf();
        $self->_append($xf);
    }
}


###############################################################################
#
# _store_all_styles()
#
# Write all STYLE records.
#
sub _store_all_styles {

    my $self = shift;

    # Excel adds the built-in styles in alphabetical order.
    my @built_ins = (
        [0x03, 16], # Comma
        [0x06, 17], # Comma[0]
        [0x04, 18], # Currency
        [0x07, 19], # Currency[0]
        [0x00,  0], # Normal
        [0x05, 20], # Percent

        # We don't deal with these styles yet.
        #[0x08, 21], # Hyperlink
        #[0x02,  8], # ColLevel_n
        #[0x01,  1], # RowLevel_n
    );


    for my $aref (@built_ins) {
        my $type     = $aref->[0];
        my $xf_index = $aref->[1];

        $self->_store_style($type, $xf_index);
    }
}


###############################################################################
#
# _store_names()
#
# Write the NAME record to define the print area and the repeat rows and cols.
#
sub _store_names {

    my $self        = shift;
    my $index;
    my %ext_refs    = %{$self->{_ext_refs}};


    # Create the user defined names.
    for my $defined_name (@{$self->{_defined_names}}) {

        $self->_store_name(
            $defined_name->{name},
            $defined_name->{encoding},
            $defined_name->{sheet_index},
            $defined_name->{formula},
        );
    }

    # Sort the worksheets into alphabetical order by name. This is a
    # requirement for some non-English language Excel patch levels.
    my @worksheets = @{$self->{_worksheets}};
       @worksheets = sort { $a->{_name} cmp $b->{_name} } @worksheets;

    # Create the autofilter NAME records
    foreach my $worksheet (@worksheets) {
        $index  = $worksheet->{_index};
        my $key = "$index:$index";
        my $ref = $ext_refs{$key};

        # Write a Name record if Autofilter has been defined
        if ($worksheet->{_filter_count}) {
            $self->_store_name_short(
                $worksheet->{_index},
                0x0D, # NAME type = Filter Database
                $ref,
                $worksheet->{_filter_area}->[0],
                $worksheet->{_filter_area}->[1],
                $worksheet->{_filter_area}->[2],
                $worksheet->{_filter_area}->[3],
                1, # Hidden
            );
        }
    }

    # Create the print area NAME records
    foreach my $worksheet (@worksheets) {
        $index  = $worksheet->{_index};
        my $key = "$index:$index";
        my $ref = $ext_refs{$key};

        # Write a Name record if the print area has been defined
        if (defined $worksheet->{_print_rowmin}) {
            $self->_store_name_short(
                $worksheet->{_index},
                0x06, # NAME type = Print_Area
                $ref,
                $worksheet->{_print_rowmin},
                $worksheet->{_print_rowmax},
                $worksheet->{_print_colmin},
                $worksheet->{_print_colmax}
            );
        }
    }

    # Create the print title NAME records
    foreach my $worksheet (@worksheets) {
        $index  = $worksheet->{_index};

        my $rowmin = $worksheet->{_title_rowmin};
        my $rowmax = $worksheet->{_title_rowmax};
        my $colmin = $worksheet->{_title_colmin};
        my $colmax = $worksheet->{_title_colmax};
        my $key    = "$index:$index";
        my $ref    = $ext_refs{$key};

        # Determine if row + col, row, col or nothing has been defined
        # and write the appropriate record
        #
        if (defined $rowmin && defined $colmin) {
            # Row and column titles have been defined.
            # Row title has been defined.
            $self->_store_name_long(
                $worksheet->{_index},
                0x07, # NAME type = Print_Titles
                $ref,
                $rowmin,
                $rowmax,
                $colmin,
                $colmax
           );
        }
        elsif (defined $rowmin) {
            # Row title has been defined.
            $self->_store_name_short(
                $worksheet->{_index},
                0x07, # NAME type = Print_Titles
                $ref,
                $rowmin,
                $rowmax,
                0x00,
                0xff
            );
        }
        elsif (defined $colmin) {
            # Column title has been defined.
            $self->_store_name_short(
                $worksheet->{_index},
                0x07, # NAME type = Print_Titles
                $ref,
                0x0000,
                0xffff,
                $colmin,
                $colmax
            );
        }
        else {
            # Nothing left to do
        }
    }
}




###############################################################################
###############################################################################
#
# BIFF RECORDS
#


###############################################################################
#
# _store_window1()
#
# Write Excel BIFF WINDOW1 record.
#
sub _store_window1 {

    my $self      = shift;

    my $record    = 0x003D;                 # Record identifier
    my $length    = 0x0012;                 # Number of bytes to follow

    my $xWn       = 0x0000;                 # Horizontal position of window
    my $yWn       = 0x0000;                 # Vertical position of window
    my $dxWn      = 0x355C;                 # Width of window
    my $dyWn      = 0x30ED;                 # Height of window

    my $grbit     = 0x0038;                 # Option flags
    my $ctabsel   = $self->{_selected};     # Number of workbook tabs selected
    my $wTabRatio = 0x0258;                 # Tab to scrollbar ratio

    my $itabFirst = $self->{_firstsheet};   # 1st displayed worksheet
    my $itabCur   = $self->{_activesheet};  # Active worksheet

    my $header    = pack("vv",        $record, $length);
    my $data      = pack("vvvvvvvvv", $xWn, $yWn, $dxWn, $dyWn,
                                      $grbit,
                                      $itabCur, $itabFirst,
                                      $ctabsel, $wTabRatio);

    $self->_append($header, $data);
}


###############################################################################
#
# _store_boundsheet()
#
# Writes Excel BIFF BOUNDSHEET record.
#
sub _store_boundsheet {

    my $self      = shift;

    my $record    = 0x0085;               # Record identifier
    my $length    = 0x08 + length($_[0]); # Number of bytes to follow

    my $sheetname = $_[0];                # Worksheet name
    my $offset    = $_[1];                # Location of worksheet BOF
    my $type      = $_[2];                # Worksheet type
    my $hidden    = $_[3];                # Worksheet hidden flag
    my $encoding  = $_[4];                # Sheet name encoding
    my $cch       = length($sheetname);   # Length of sheet name

    my $grbit     = $type | $hidden;

    # Character length is num of chars not num of bytes
    $cch /= 2 if $encoding;

    # Change the UTF-16 name from BE to LE
    $sheetname = pack 'n*', unpack 'v*', $sheetname if $encoding;

    my $header    = pack("vv",   $record, $length);
    my $data      = pack("VvCC", $offset, $grbit, $cch, $encoding);

    $self->_append($header, $data, $sheetname);
}


###############################################################################
#
# _store_style()
#
# Write Excel BIFF STYLE records.
#
sub _store_style {

    my $self      = shift;

    my $record    = 0x0293; # Record identifier
    my $length    = 0x0004; # Bytes to follow

    my $type      = $_[0];  # Built-in style
    my $xf_index  = $_[1];  # Index to style XF
    my $level     = 0xff;   # Outline style level

    $xf_index    |= 0x8000; # Add flag to indicate built-in style.


    my $header    = pack("vv",  $record, $length);
    my $data      = pack("vCC", $xf_index, $type, $level);

    $self->_append($header, $data);
}


###############################################################################
#
# _store_num_format()
#
# Writes Excel FORMAT record for non "built-in" numerical formats.
#
sub _store_num_format {

    my $self      = shift;

    my $record    = 0x041E;         # Record identifier
    my $length;                     # Number of bytes to follow

    my $format    = $_[0];          # Custom format string
    my $ifmt      = $_[1];          # Format index code
    my $encoding  = $_[2];          # Char encoding for format string


    # Handle utf8 strings in perl 5.8.
    if ($] >= 5.008) {
        require Encode;

        if (Encode::is_utf8($format)) {
            $format = Encode::encode("UTF-16BE", $format);
            $encoding = 1;
        }
    }


    # Char length of format string
    my $cch = length $format;


    # Handle Unicode format strings.
    if ($encoding == 1) {
        croak "Uneven number of bytes in Unicode font name" if $cch % 2;
        $cch    /= 2 if $encoding;
        $format  = pack 'v*', unpack 'n*', $format;
    }


    # Special case to handle Euro symbol, 0x80, in non-Unicode strings.
    if ($encoding == 0 and $format =~ /\x80/) {
        $format   =  pack 'v*', unpack 'C*', $format;
        $format   =~ s/\x80\x00/\xAC\x20/g;
        $encoding =  1;
    }

    $length       = 0x05 + length $format;

    my $header    = pack("vv", $record, $length);
    my $data      = pack("vvC", $ifmt, $cch, $encoding);

    $self->_append($header, $data, $format);
}


###############################################################################
#
# _store_1904()
#
# Write Excel 1904 record to indicate the date system in use.
#
sub _store_1904 {

    my $self      = shift;

    my $record    = 0x0022;         # Record identifier
    my $length    = 0x0002;         # Bytes to follow

    my $f1904     = $self->{_1904}; # Flag for 1904 date system

    my $header    = pack("vv",  $record, $length);
    my $data      = pack("v", $f1904);

    $self->_append($header, $data);
}


###############################################################################
#
# _store_supbook()
#
# Write BIFF record SUPBOOK to indicate that the workbook contains external
# references, in our case, formula, print area and print title refs.
#
sub _store_supbook {

    my $self        = shift;

    my $record      = 0x01AE;                   # Record identifier
    my $length      = 0x0004;                   # Number of bytes to follow

    my $ctabs       = @{$self->{_worksheets}};  # Number of worksheets
    my $StVirtPath  = 0x0401;                   # Encoded workbook filename

    my $header      = pack("vv", $record, $length);
    my $data        = pack("vv", $ctabs, $StVirtPath);

    $self->_append($header, $data);
}


###############################################################################
#
# _store_externsheet()
#
# Writes the Excel BIFF EXTERNSHEET record. These references are used by
# formulas. TODO NAME record is required to define the print area and the
# repeat rows and columns.
#
sub _store_externsheet {

    my $self        = shift;

    my $record      = 0x0017;                   # Record identifier
    my $length;                                 # Number of bytes to follow


    # Get the external refs
    my %ext_refs = %{$self->{_ext_refs}};
    my @ext_refs = sort {$ext_refs{$a} <=> $ext_refs{$b}} keys %ext_refs;

    # Change the external refs from stringified "1:1" to [1, 1]
    foreach my $ref (@ext_refs) {
        $ref = [split /:/, $ref];
    }


    my $cxti        = scalar @ext_refs;         # Number of Excel XTI structures
    my $rgxti       = '';                       # Array of XTI structures

    # Write the XTI structs
    foreach my $ext_ref (@ext_refs) {
        $rgxti .= pack("vvv", 0, $ext_ref->[0], $ext_ref->[1])
    }


    my $data        = pack("v", $cxti) . $rgxti;
    my $header      = pack("vv", $record, length $data);

    $self->_append($header, $data);
}


###############################################################################
#
# _store_name()
#
#
# Store the NAME record used for storing the print area, repeat rows, repeat
# columns, autofilters and defined names.
#
# TODO. This is a more generic version that will replace _store_name_short()
#       and _store_name_long().
#
sub _store_name {

    my $self            = shift;

    my $record          = 0x0018;       # Record identifier
    my $length;                         # Number of bytes to follow

    my $name            = shift;
    my $encoding        = shift;
    my $sheet_index     = shift;
    my $formula         = shift;

    my $text_length     = length $name;
    my $formula_length  = length $formula;

    # UTF-16 string length is in characters not bytes.
    $text_length       /= 2 if $encoding;


    my $grbit           = 0x0000;       # Option flags
    my $shortcut        = 0x00;         # Keyboard shortcut
    my $ixals           = 0x0000;       # Unused index.
    my $menu_length     = 0x00;         # Length of cust menu text
    my $desc_length     = 0x00;         # Length of description text
    my $help_length     = 0x00;         # Length of help topic text
    my $status_length   = 0x00;         # Length of status bar text

    # Set grbit built-in flag and the hidden flag for autofilters.
    if ($text_length == 1) {
        $grbit = 0x0020 if ord $name == 0x06; # Print area
        $grbit = 0x0020 if ord $name == 0x07; # Print titles
        $grbit = 0x0021 if ord $name == 0x0D; # Autofilter
    }

    my $data            = pack "v", $grbit;
    $data              .= pack "C", $shortcut;
    $data              .= pack "C", $text_length;
    $data              .= pack "v", $formula_length;
    $data              .= pack "v", $ixals;
    $data              .= pack "v", $sheet_index;
    $data              .= pack "C", $menu_length;
    $data              .= pack "C", $desc_length;
    $data              .= pack "C", $help_length;
    $data              .= pack "C", $status_length;
    $data              .= pack "C", $encoding;
    $data              .= $name;
    $data              .= $formula;

    my $header          = pack "vv", $record, length $data;

    $self->_append($header, $data);
}


###############################################################################
#
# _store_name_short()
#
#
# Store the NAME record in the short format that is used for storing the print
# area, repeat rows only and repeat columns only.
#
sub _store_name_short {

    my $self            = shift;

    my $record          = 0x0018;       # Record identifier
    my $length          = 0x001b;       # Number of bytes to follow

    my $index           = shift;        # Sheet index
    my $type            = shift;
    my $ext_ref         = shift;        # TODO

    my $grbit           = 0x0020;       # Option flags
    my $chKey           = 0x00;         # Keyboard shortcut
    my $cch             = 0x01;         # Length of text name
    my $cce             = 0x000b;       # Length of text definition
    my $unknown01       = 0x0000;       #
    my $ixals           = $index +1;    # Sheet index
    my $unknown02       = 0x00;         #
    my $cchCustMenu     = 0x00;         # Length of cust menu text
    my $cchDescription  = 0x00;         # Length of description text
    my $cchHelptopic    = 0x00;         # Length of help topic text
    my $cchStatustext   = 0x00;         # Length of status bar text
    my $rgch            = $type;        # Built-in name type
    my $unknown03       = 0x3b;         #

    my $rowmin          = $_[0];        # Start row
    my $rowmax          = $_[1];        # End row
    my $colmin          = $_[2];        # Start column
    my $colmax          = $_[3];        # end column

    my $hidden          = $_[4];        # Name is hidden
    $grbit              = 0x0021 if $hidden;

    my $header          = pack("vv", $record, $length);
    my $data            = pack("v",  $grbit);
    $data              .= pack("C",  $chKey);
    $data              .= pack("C",  $cch);
    $data              .= pack("v",  $cce);
    $data              .= pack("v",  $unknown01);
    $data              .= pack("v",  $ixals);
    $data              .= pack("C",  $unknown02);
    $data              .= pack("C",  $cchCustMenu);
    $data              .= pack("C",  $cchDescription);
    $data              .= pack("C",  $cchHelptopic);
    $data              .= pack("C",  $cchStatustext);
    $data              .= pack("C",  $rgch);
    $data              .= pack("C",  $unknown03);
    $data              .= pack("v",  $ext_ref);

    $data              .= pack("v",  $rowmin);
    $data              .= pack("v",  $rowmax);
    $data              .= pack("v",  $colmin);
    $data              .= pack("v",  $colmax);

    $self->_append($header, $data);
}


###############################################################################
#
# _store_name_long()
#
#
# Store the NAME record in the long format that is used for storing the repeat
# rows and columns when both are specified. This share a lot of code with
# _store_name_short() but we use a separate method to keep the code clean.
# Code abstraction for reuse can be carried too far, and I should know. ;-)
#
sub _store_name_long {

    my $self            = shift;

    my $record          = 0x0018;       # Record identifier
    my $length          = 0x002a;       # Number of bytes to follow

    my $index           = shift;        # Sheet index
    my $type            = shift;
    my $ext_ref         = shift;        # TODO

    my $grbit           = 0x0020;       # Option flags
    my $chKey           = 0x00;         # Keyboard shortcut
    my $cch             = 0x01;         # Length of text name
    my $cce             = 0x001a;       # Length of text definition
    my $unknown01       = 0x0000;       #
    my $ixals           = $index +1;    # Sheet index
    my $unknown02       = 0x00;         #
    my $cchCustMenu     = 0x00;         # Length of cust menu text
    my $cchDescription  = 0x00;         # Length of description text
    my $cchHelptopic    = 0x00;         # Length of help topic text
    my $cchStatustext   = 0x00;         # Length of status bar text
    my $rgch            = $type;        # Built-in name type

    my $unknown03       = 0x29;
    my $unknown04       = 0x0017;
    my $unknown05       = 0x3b;

    my $rowmin          = $_[0];        # Start row
    my $rowmax          = $_[1];        # End row
    my $colmin          = $_[2];        # Start column
    my $colmax          = $_[3];        # end column


    my $header          = pack("vv", $record, $length);
    my $data            = pack("v",  $grbit);
    $data              .= pack("C",  $chKey);
    $data              .= pack("C",  $cch);
    $data              .= pack("v",  $cce);
    $data              .= pack("v",  $unknown01);
    $data              .= pack("v",  $ixals);
    $data              .= pack("C",  $unknown02);
    $data              .= pack("C",  $cchCustMenu);
    $data              .= pack("C",  $cchDescription);
    $data              .= pack("C",  $cchHelptopic);
    $data              .= pack("C",  $cchStatustext);
    $data              .= pack("C",  $rgch);

    # Column definition
    $data              .= pack("C",  $unknown03);
    $data              .= pack("v",  $unknown04);
    $data              .= pack("C",  $unknown05);
    $data              .= pack("v",  $ext_ref);
    $data              .= pack("v",  0x0000);
    $data              .= pack("v",  0xffff);
    $data              .= pack("v",  $colmin);
    $data              .= pack("v",  $colmax);

    # Row definition
    $data              .= pack("C",  $unknown05);
    $data              .= pack("v",  $ext_ref);
    $data              .= pack("v",  $rowmin);
    $data              .= pack("v",  $rowmax);
    $data              .= pack("v",  0x00);
    $data              .= pack("v",  0xff);
    # End of data
    $data              .= pack("C",  0x10);

    $self->_append($header, $data);
}


###############################################################################
#
# _store_palette()
#
# Stores the PALETTE biff record.
#
sub _store_palette {

    my $self            = shift;

    my $aref            = $self->{_palette};

    my $record          = 0x0092;           # Record identifier
    my $length          = 2 + 4 * @$aref;   # Number of bytes to follow
    my $ccv             =         @$aref;   # Number of RGB values to follow
    my $data;                               # The RGB data

    # Pack the RGB data
    $data .= pack "CCCC", @$_ for @$aref;

    my $header = pack("vvv",  $record, $length, $ccv);

    $self->_append($header, $data);
}


###############################################################################
#
# _store_codepage()
#
# Stores the CODEPAGE biff record.
#
sub _store_codepage {

    my $self            = shift;

    my $record          = 0x0042;               # Record identifier
    my $length          = 0x0002;               # Number of bytes to follow
    my $cv              = $self->{_codepage};   # The code page

    my $header          = pack("vv", $record, $length);
    my $data            = pack("v",  $cv);

    $self->_append($header, $data);
}


###############################################################################
#
# _store_country()
#
# Stores the COUNTRY biff record.
#
sub _store_country {

    my $self            = shift;

    my $record          = 0x008C;               # Record identifier
    my $length          = 0x0004;               # Number of bytes to follow
    my $country_default = $self->{_country};
    my $country_win_ini = $self->{_country};

    my $header          = pack("vv", $record, $length);
    my $data            = pack("vv", $country_default, $country_win_ini);

    $self->_append($header, $data);
}


###############################################################################
#
# _store_hideobj()
#
# Stores the HIDEOBJ biff record.
#
sub _store_hideobj {

    my $self            = shift;

    my $record          = 0x008D;               # Record identifier
    my $length          = 0x0002;               # Number of bytes to follow
    my $hide            = $self->{_hideobj};    # Option to hide objects

    my $header          = pack("vv", $record, $length);
    my $data            = pack("v",  $hide);

    $self->_append($header, $data);
}


###############################################################################
###############################################################################
###############################################################################



###############################################################################
#
# _calculate_extern_sizes()
#
# We need to calculate the space required by the SUPBOOK, EXTERNSHEET and NAME
# records so that it can be added to the BOUNDSHEET offsets.
#
sub _calculate_extern_sizes {

    my $self   = shift;


    my %ext_refs        = $self->{_parser}->get_ext_sheets();
    my $ext_ref_count   = scalar keys %ext_refs;
    my $length          = 0;
    my $index           = 0;


    if (@{$self->{_defined_names}}) {
        my $index   = 0;
        my $key     = "$index:$index";

        if (not exists $ext_refs{$key}) {
            $ext_refs{$key} = $ext_ref_count++;
        }
    }

    for my $defined_name (@{$self->{_defined_names}}) {
        $length += 19
                   + length($defined_name->{name})
                   + length($defined_name->{formula});
    }


    foreach my $worksheet (@{$self->{_worksheets}}) {

        my $rowmin      = $worksheet->{_title_rowmin};
        my $colmin      = $worksheet->{_title_colmin};
        my $filter      = $worksheet->{_filter_count};
        my $key         = "$index:$index";
        $index++;


        # Add area NAME records
        #
        if (defined $worksheet->{_print_rowmin}) {
            $ext_refs{$key} = $ext_ref_count++ if not exists $ext_refs{$key};

            $length += 31 ;
        }


        # Add title  NAME records
        #
        if (defined $rowmin and defined $colmin) {
            $ext_refs{$key} = $ext_ref_count++ if not exists $ext_refs{$key};

            $length += 46;
        }
        elsif (defined $rowmin or defined $colmin) {
            $ext_refs{$key} = $ext_ref_count++ if not exists $ext_refs{$key};

            $length += 31;
        }
        else {
            # TODO, may need this later.
        }


        # Add Autofilter  NAME records
        #
        if ($filter) {
            $ext_refs{$key} = $ext_ref_count++ if not exists $ext_refs{$key};

            $length += 31;
        }
    }


    # Update the ref counts.
    $self->{_ext_ref_count} = $ext_ref_count;
    $self->{_ext_refs}      = {%ext_refs};


    # If there are no external refs then we don't write, SUPBOOK, EXTERNSHEET
    # and NAME. Therefore the length is 0.

    return $length = 0 if $ext_ref_count == 0;


    # The SUPBOOK record is 8 bytes
    $length += 8;

    # The EXTERNSHEET record is 6 bytes + 6 bytes for each external ref
    $length += 6 * (1 + $ext_ref_count);

    return $length;
}


###############################################################################
#
# _calculate_shared_string_sizes()
#
# Handling of the SST continue blocks is complicated by the need to include an
# additional continuation byte depending on whether the string is split between
# blocks or whether it starts at the beginning of the block. (There are also
# additional complications that will arise later when/if Rich Strings are
# supported). As such we cannot use the simple CONTINUE mechanism provided by
# the _add_continue() method in BIFFwriter.pm. Thus we have to make two passes
# through the strings data. The first is to calculate the required block sizes
# and the second, in _store_shared_strings(), is to write the actual strings.
# The first pass through the data is also used to calculate the size of the SST
# and CONTINUE records for use in setting the BOUNDSHEET record offsets. The
# downside of this is that the same algorithm repeated in _store_shared_strings.
#
sub _calculate_shared_string_sizes {

    my $self    = shift;

    my @strings;
    $#strings = $self->{_str_unique} -1; # Pre-extend array

    while (my $key = each %{$self->{_str_table}}) {
        $strings[$self->{_str_table}->{$key}] = $key;
    }

    # The SST data could be very large, free some memory (maybe).
    $self->{_str_table} = undef;
    $self->{_str_array} = [@strings];


    # Iterate through the strings to calculate the CONTINUE block sizes.
    #
    # The SST blocks requires a specialised CONTINUE block, so we have to
    # ensure that the maximum data block size is less than the limit used by
    # _add_continue() in BIFFwriter.pm. For simplicity we use the same size
    # for the SST and CONTINUE records:
    #   8228 : Maximum Excel97 block size
    #     -4 : Length of block header
    #     -8 : Length of additional SST header information
    #     -8 : Arbitrary number to keep within _add_continue() limit
    # = 8208
    #
    my $continue_limit = 8208;
    my $block_length   = 0;
    my $written        = 0;
    my @block_sizes;
    my $continue       = 0;

    for my $string (@strings) {

        my $string_length = length $string;
        my $encoding      = unpack "xx C", $string;
        my $split_string  = 0;


        # Block length is the total length of the strings that will be
        # written out in a single SST or CONTINUE block.
        #
        $block_length += $string_length;


        # We can write the string if it doesn't cross a CONTINUE boundary
        if ($block_length < $continue_limit) {
            $written += $string_length;
            next;
        }


        # Deal with the cases where the next string to be written will exceed
        # the CONTINUE boundary. If the string is very long it may need to be
        # written in more than one CONTINUE record.
        #
        while ($block_length >= $continue_limit) {

            # We need to avoid the case where a string is continued in the first
            # n bytes that contain the string header information.
            #
            my $header_length   = 3; # Min string + header size -1
            my $space_remaining = $continue_limit -$written -$continue;


            # Unicode data should only be split on char (2 byte) boundaries.
            # Therefore, in some cases we need to reduce the amount of available
            # space by 1 byte to ensure the correct alignment.
            my $align = 0;

            # Only applies to Unicode strings
            if ($encoding == 1) {
                # Min string + header size -1
                $header_length = 4;

                if ($space_remaining > $header_length) {
                    # String contains 3 byte header => split on odd boundary
                    if (not $split_string and $space_remaining % 2 != 1) {
                        $space_remaining--;
                        $align = 1;
                    }
                    # Split section without header => split on even boundary
                    elsif ($split_string and $space_remaining % 2 == 1) {
                        $space_remaining--;
                        $align = 1;
                    }

                    $split_string = 1;
                }
            }


            if ($space_remaining > $header_length) {
                # Write as much as possible of the string in the current block
                $written      += $space_remaining;

                # Reduce the current block length by the amount written
                $block_length -= $continue_limit -$continue -$align;

                # Store the max size for this block
                push @block_sizes, $continue_limit -$align;

                # If the current string was split then the next CONTINUE block
                # should have the string continue flag (grbit) set unless the
                # split string fits exactly into the remaining space.
                #
                if ($block_length > 0) {
                    $continue = 1;
                }
                else {
                    $continue = 0;
                }

            }
            else {
                # Store the max size for this block
                push @block_sizes, $written +$continue;

                # Not enough space to start the string in the current block
                $block_length -= $continue_limit -$space_remaining -$continue;
                $continue = 0;

            }

            # If the string (or substr) is small enough we can write it in the
            # new CONTINUE block. Else, go through the loop again to write it in
            # one or more CONTINUE blocks
            #
            if ($block_length < $continue_limit) {
                $written = $block_length;
            }
            else {
                $written = 0;
            }
        }
    }

    # Store the max size for the last block unless it is empty
    push @block_sizes, $written +$continue if $written +$continue;


    $self->{_str_block_sizes} = [@block_sizes];


    # Calculate the total length of the SST and associated CONTINUEs (if any).
    # The SST record will have a length even if it contains no strings.
    # This length is required to set the offsets in the BOUNDSHEET records since
    # they must be written before the SST records
    #
    my $length  = 12;
    $length    +=     shift @block_sizes if    @block_sizes; # SST
    $length    += 4 + shift @block_sizes while @block_sizes; # CONTINUEs

    return $length;
}


###############################################################################
#
# _store_shared_strings()
#
# Write all of the workbooks strings into an indexed array.
#
# See the comments in _calculate_shared_string_sizes() for more information.
#
# We also use this routine to record the offsets required by the EXTSST table.
# In order to do this we first identify the first string in an EXTSST bucket
# and then store its global and local offset within the SST table. The offset
# occurs wherever the start of the bucket string is written out via append().
#
sub _store_shared_strings {

    my $self                = shift;

    my @strings = @{$self->{_str_array}};


    my $record              = 0x00FC;   # Record identifier
    my $length              = 0x0008;   # Number of bytes to follow
    my $total               = 0x0000;

    # Iterate through the strings to calculate the CONTINUE block sizes
    my $continue_limit = 8208;
    my $block_length   = 0;
    my $written        = 0;
    my $continue       = 0;

    # The SST and CONTINUE block sizes have been pre-calculated by
    # _calculate_shared_string_sizes()
    my @block_sizes    = @{$self->{_str_block_sizes}};


    # The SST record is required even if it contains no strings. Thus we will
    # always have a length
    #
    if (@block_sizes) {
        $length = 8 + shift @block_sizes;
    }
    else {
        # No strings
        $length = 8;
    }


    # Initialise variables used to track EXTSST bucket offsets.
    my $extsst_str_num  = -1;
    my $sst_block_start = $self->{_datasize};


    # Write the SST block header information
    my $header      = pack("vv", $record, $length);
    my $data        = pack("VV", $self->{_str_total}, $self->{_str_unique});
    $self->_append($header, $data);


    # Iterate through the strings and write them out
    for my $string (@strings) {

        my $string_length = length $string;
        my $encoding      = unpack "xx C", $string;
        my $split_string  = 0;
        my $bucket_string = 0; # Used to track EXTSST bucket offsets.


        # Check if the string is at the start of a EXTSST bucket.
        if (++$extsst_str_num % $self->{_extsst_bucket_size} == 0) {
            $bucket_string = 1;
        }


        # Block length is the total length of the strings that will be
        # written out in a single SST or CONTINUE block.
        #
        $block_length += $string_length;


        # We can write the string if it doesn't cross a CONTINUE boundary
        if ($block_length < $continue_limit) {

            # Store location of EXTSST bucket string.
            if ($bucket_string) {
                my $global_offset   = $self->{_datasize};
                my $local_offset    = $self->{_datasize} - $sst_block_start;

                push @{$self->{_extsst_offsets}}, [$global_offset, $local_offset];
                $bucket_string = 0;
            }

            $self->_append($string);
            $written += $string_length;
            next;
        }


        # Deal with the cases where the next string to be written will exceed
        # the CONTINUE boundary. If the string is very long it may need to be
        # written in more than one CONTINUE record.
        #
        while ($block_length >= $continue_limit) {

            # We need to avoid the case where a string is continued in the first
            # n bytes that contain the string header information.
            #
            my $header_length   = 3; # Min string + header size -1
            my $space_remaining = $continue_limit -$written -$continue;


            # Unicode data should only be split on char (2 byte) boundaries.
            # Therefore, in some cases we need to reduce the amount of available
            # space by 1 byte to ensure the correct alignment.
            my $align = 0;

            # Only applies to Unicode strings
            if ($encoding == 1) {
                # Min string + header size -1
                $header_length = 4;

                if ($space_remaining > $header_length) {
                    # String contains 3 byte header => split on odd boundary
                    if (not $split_string and $space_remaining % 2 != 1) {
                        $space_remaining--;
                        $align = 1;
                    }
                    # Split section without header => split on even boundary
                    elsif ($split_string and $space_remaining % 2 == 1) {
                        $space_remaining--;
                        $align = 1;
                    }

                    $split_string = 1;
                }
            }


            if ($space_remaining > $header_length) {
                # Write as much as possible of the string in the current block
                my $tmp = substr $string, 0, $space_remaining;

                # Store location of EXTSST bucket string.
                if ($bucket_string) {
                    my $global_offset   = $self->{_datasize};
                    my $local_offset    = $self->{_datasize} - $sst_block_start;

                    push @{$self->{_extsst_offsets}}, [$global_offset, $local_offset];
                    $bucket_string = 0;
                }

                $self->_append($tmp);


                # The remainder will be written in the next block(s)
                $string = substr $string, $space_remaining;

                # Reduce the current block length by the amount written
                $block_length -= $continue_limit -$continue -$align;

                # If the current string was split then the next CONTINUE block
                # should have the string continue flag (grbit) set unless the
                # split string fits exactly into the remaining space.
                #
                if ($block_length > 0) {
                    $continue = 1;
                }
                else {
                    $continue = 0;
                }
            }
            else {
                # Not enough space to start the string in the current block
                $block_length -= $continue_limit -$space_remaining -$continue;
                $continue = 0;
            }

            # Write the CONTINUE block header
            if (@block_sizes) {
                $sst_block_start= $self->{_datasize}; # Reset EXTSST offset.

                $record         = 0x003C;
                $length         = shift @block_sizes;

                $header         = pack("vv", $record, $length);
                $header        .= pack("C", $encoding) if $continue;

                $self->_append($header);
            }

            # If the string (or substr) is small enough we can write it in the
            # new CONTINUE block. Else, go through the loop again to write it in
            # one or more CONTINUE blocks
            #
            if ($block_length < $continue_limit) {

                # Store location of EXTSST bucket string.
                if ($bucket_string) {
                    my $global_offset   = $self->{_datasize};
                    my $local_offset    = $self->{_datasize} - $sst_block_start;

                    push @{$self->{_extsst_offsets}}, [$global_offset, $local_offset];

                    $bucket_string = 0;
                }
                $self->_append($string);

                $written = $block_length;
            }
            else {
                $written = 0;
            }
        }
    }
}


###############################################################################
#
# _calculate_extsst_size
#
# The number of buckets used in the EXTSST is between 0 and 128. The number of
# strings per bucket (bucket size) has a minimum value of 8 and a theoretical
# maximum of 2^16. For "number of strings" < 1024 there is a constant bucket
# size of 8. The following algorithm generates the same size/bucket ratio
# as Excel.
#
sub _calculate_extsst_size {

    my $self            = shift;

    my $unique_strings  = $self->{_str_unique};

    my $bucket_size;
    my $buckets;

    if ($unique_strings < 1024) {
        $bucket_size = 8;
    }
    else {
        $bucket_size = 1 + int($unique_strings / 128);
    }

    $buckets = int(($unique_strings + $bucket_size -1)  / $bucket_size);


    $self->{_extsst_buckets}        = $buckets ;
    $self->{_extsst_bucket_size}    = $bucket_size;


    return 6 + 8 * $buckets;
}


###############################################################################
#
# _store_extsst
#
# Write EXTSST table using the offsets calculated in _store_shared_strings().
#
sub _store_extsst {

    my $self = shift;

    my @offsets     = @{$self->{_extsst_offsets}};
    my $bucket_size = $self->{_extsst_bucket_size};

    my $record      = 0x00FF;             # Record identifier
    my $length      = 2 + 8 * @offsets;   # Bytes to follow

    my $header      = pack 'vv',   $record, $length;
    my $data        = pack 'v',    $bucket_size,;

    for my $offset (@offsets) {
       $data .= pack 'Vvv', $offset->[0], $offset->[1], 0;
    }

    $self->_append($header, $data);

}




#
# Methods related to comments and MSO objects.
#

###############################################################################
#
# _add_mso_drawing_group()
#
# Write the MSODRAWINGGROUP record that keeps track of the Escher drawing
# objects in the file such as images, comments and filters.
#
sub _add_mso_drawing_group {

    my $self    = shift;

    return unless $self->{_mso_size};

    my $record  = 0x00EB;               # Record identifier
    my $length  = 0x0000;               # Number of bytes to follow

    my $data    = $self->_store_mso_dgg_container();
       $data   .= $self->_store_mso_dgg(@{$self->{_mso_clusters}});
       $data   .= $self->_store_mso_bstore_container();
       $data   .= $self->_store_mso_images(@$_) for @{$self->{_images_data}};
       $data   .= $self->_store_mso_opt();
       $data   .= $self->_store_mso_split_menu_colors();

       $length  = length $data;
    my $header  = pack("vv", $record, $length);

    $self->_add_mso_drawing_group_continue($header . $data);

    return $header . $data; # For testing only.
}


###############################################################################
#
# _add_mso_drawing_group_continue()
#
# See first the Spreadsheet::WriteExcel::BIFFwriter::_add_continue() method.
#
# Add specialised CONTINUE headers to large MSODRAWINGGROUP data block.
# We use the Excel 97 max block size of 8228 - 4 bytes for the header = 8224.
#
# The structure depends on the size of the data block:
#
#     Case 1:  <=   8224 bytes      1 MSODRAWINGGROUP
#     Case 2:  <= 2*8224 bytes      1 MSODRAWINGGROUP + 1 CONTINUE
#     Case 3:  >  2*8224 bytes      2 MSODRAWINGGROUP + n CONTINUE
#
sub _add_mso_drawing_group_continue {

    my $self        = shift;

    my $data        = $_[0];
    my $limit       = 8228 -4;
    my $mso_group   = 0x00EB; # Record identifier
    my $continue    = 0x003C; # Record identifier
    my $block_count = 1;
    my $header;
    my $tmp;

    # Ignore the base class _add_continue() method.
    $self->{_ignore_continue} = 1;

    # Case 1 above. Just return the data as it is.
    if (length $data <= $limit) {
        $self->_append($data);
        return;
    }

    # Change length field of the first MSODRAWINGGROUP block. Case 2 and 3.
    $tmp = substr($data, 0, $limit +4, "");
    substr($tmp, 2, 2, pack("v", $limit));
    $self->_append($tmp);


    # Add MSODRAWINGGROUP and CONTINUE blocks for Case 3 above.
    while (length($data) > $limit) {
        if ($block_count == 1) {
            # Add extra MSODRAWINGGROUP block header.
            $header = pack("vv", $mso_group, $limit);
            $block_count++;
        }
        else {
            # Add normal CONTINUE header.
            $header = pack("vv", $continue, $limit);
        }

        $tmp = substr($data, 0, $limit, "");
        $self->_append($header, $tmp);
    }


    # Last CONTINUE block for remaining data. Case 2 and 3 above.
    $header = pack("vv", $continue, length($data));
    $self->_append($header, $data);


    # Turn the base class _add_continue() method back on.
    $self->{_ignore_continue} = 0;
}


###############################################################################
#
# _store_mso_dgg_container()
#
# Write the Escher DggContainer record that is part of MSODRAWINGGROUP.
#
sub _store_mso_dgg_container {

    my $self        = shift;

    my $type        = 0xF000;
    my $version     = 15;
    my $instance    = 0;
    my $data        = '';
    my $length      = $self->{_mso_size} -12; # -4 (biff header) -8 (for this).


    return $self->_add_mso_generic($type, $version, $instance, $data, $length);
}


###############################################################################
#
# _store_mso_dgg()
#
# Write the Escher Dgg record that is part of MSODRAWINGGROUP.
#
sub _store_mso_dgg {

    my $self            = shift;

    my $type            = 0xF006;
    my $version         = 0;
    my $instance        = 0;
    my $data            = '';
    my $length          = undef; # Calculate automatically.

    my $max_spid        = $_[0];
    my $num_clusters    = $_[1];
    my $shapes_saved    = $_[2];
    my $drawings_saved  = $_[3];
    my $clusters        = $_[4];

    $data               = pack "VVVV",  $max_spid,     $num_clusters,
                                        $shapes_saved, $drawings_saved;

    for my $aref (@$clusters) {
        my $drawing_id      = $aref->[0];
        my $shape_ids_used  = $aref->[1];

        $data              .= pack "VV",  $drawing_id, $shape_ids_used;
    }


    return $self->_add_mso_generic($type, $version, $instance, $data, $length);
}


###############################################################################
#
# _store_mso_bstore_container()
#
# Write the Escher BstoreContainer record that is part of MSODRAWINGGROUP.
#
sub _store_mso_bstore_container {

    my $self        = shift;

    return '' unless $self->{_images_size};

    my $type        = 0xF001;
    my $version     = 15;
    my $instance    = @{$self->{_images_data}}; # Number of images.
    my $data        = '';
    my $length      = $self->{_images_size} +8 *$instance;

    return $self->_add_mso_generic($type, $version, $instance, $data, $length);
}



###############################################################################
#
# _store_mso_images()
#
# Write the Escher BstoreContainer record that is part of MSODRAWINGGROUP.
#
sub _store_mso_images {

    my $self        = shift;

    my $ref_count   = $_[0];
    my $image_type  = $_[1];
    my $image       = $_[2];
    my $size        = $_[3];
    my $checksum1   = $_[4];
    my $checksum2   = $_[5];

    my $blip_store_entry =  $self->_store_mso_blip_store_entry($ref_count,
                                                               $image_type,
                                                               $size,
                                                               $checksum1);

    my $blip             =  $self->_store_mso_blip($image_type,
                                                   $image,
                                                   $size,
                                                   $checksum1,
                                                   $checksum2);

    return $blip_store_entry . $blip;
}



###############################################################################
#
# _store_mso_blip_store_entry()
#
# Write the Escher BlipStoreEntry record that is part of MSODRAWINGGROUP.
#
sub _store_mso_blip_store_entry {

    my $self        = shift;

    my $ref_count   = $_[0];
    my $image_type  = $_[1];
    my $size        = $_[2];
    my $checksum1   = $_[3];


    my $type        = 0xF007;
    my $version     = 2;
    my $instance    = $image_type;
    my $length      = $size +61;
    my $data        = pack('C',  $image_type)   # Win32
                    . pack('C',  $image_type)   # Mac
                    . pack('H*', $checksum1)    # Uid checksum
                    . pack('v',  0xFF)          # Tag
                    . pack('V',  $size +25)     # Next Blip size
                    . pack('V',  $ref_count)    # Image ref count
                    . pack('V',  0x00000000)    # File offset
                    . pack('C',  0x00)          # Usage
                    . pack('C',  0x00)          # Name length
                    . pack('C',  0x00)          # Unused
                    . pack('C',  0x00)          # Unused
                    ;

    return $self->_add_mso_generic($type, $version, $instance, $data, $length);
}


###############################################################################
#
# _store_mso_blip()
#
# Write the Escher Blip record that is part of MSODRAWINGGROUP.
#
sub _store_mso_blip {

    my $self        = shift;

    my $image_type  = $_[0];
    my $image_data  = $_[1];
    my $size        = $_[2];
    my $checksum1   = $_[3];
    my $checksum2   = $_[4];
    my $instance;

    $instance = 0x046A if $image_type == 5; # JPG
    $instance = 0x06E0 if $image_type == 6; # PNG
    $instance = 0x07A9 if $image_type == 7; # BMP

    # BMPs contain an extra checksum for the stripped data.
    if ( $image_type == 7) {
        $checksum1 = $checksum2 . $checksum1;
    }

    my $type        = 0xF018 + $image_type;
    my $version     = 0x0000;
    my $length      = $size +17;
    my $data        = pack('H*', $checksum1) # Uid checksum
                    . pack('C',  0xFF)       # Tag
                    . $image_data            # Image
                    ;

    return $self->_add_mso_generic($type, $version, $instance, $data, $length);
}



###############################################################################
#
# _store_mso_opt()
#
# Write the Escher Opt record that is part of MSODRAWINGGROUP.
#
sub _store_mso_opt {

    my $self        = shift;

    my $type        = 0xF00B;
    my $version     = 3;
    my $instance    = 3;
    my $data        = '';
    my $length      = 18;

    $data           = pack "H*", 'BF0008000800810109000008C0014000' .
                                 '0008';


    return $self->_add_mso_generic($type, $version, $instance, $data, $length);
}


###############################################################################
#
# _store_mso_split_menu_colors()
#
# Write the Escher SplitMenuColors record that is part of MSODRAWINGGROUP.
#
sub _store_mso_split_menu_colors {

    my $self        = shift;

    my $type        = 0xF11E;
    my $version     = 0;
    my $instance    = 4;
    my $data        = '';
    my $length      = 16;

    $data           = pack "H*", '0D0000080C00000817000008F7000010';

    return $self->_add_mso_generic($type, $version, $instance, $data, $length);
}


1;


__END__


=head1 NAME

Workbook - A writer class for Excel Workbooks.

=head1 SYNOPSIS

See the documentation for Spreadsheet::WriteExcel

=head1 DESCRIPTION

This module is used in conjunction with Spreadsheet::WriteExcel.

=head1 AUTHOR

John McNamara jmcnamara@cpan.org

=head1 COPYRIGHT

© MM-MMX, John McNamara.

All Rights Reserved. This module is free software. It may be used, redistributed and/or modified under the same terms as Perl itself.