1 |
dpavlin |
498 |
package WebPAC::Input::Excel; |
2 |
|
|
|
3 |
|
|
use warnings; |
4 |
|
|
use strict; |
5 |
|
|
|
6 |
|
|
use Spreadsheet::ParseExcel; |
7 |
|
|
use Spreadsheet::ParseExcel::Utility qw/int2col/; |
8 |
dpavlin |
728 |
use base qw/WebPAC::Common/; |
9 |
dpavlin |
498 |
|
10 |
|
|
=head1 NAME |
11 |
|
|
|
12 |
dpavlin |
894 |
WebPAC::Input::Excel - support for Microsoft Excel and compatibile files |
13 |
dpavlin |
498 |
|
14 |
|
|
=head1 VERSION |
15 |
|
|
|
16 |
dpavlin |
728 |
Version 0.04 |
17 |
dpavlin |
498 |
|
18 |
|
|
=cut |
19 |
|
|
|
20 |
dpavlin |
728 |
our $VERSION = '0.04'; |
21 |
dpavlin |
498 |
|
22 |
|
|
|
23 |
|
|
=head1 SYNOPSIS |
24 |
|
|
|
25 |
|
|
Open Microsoft Excell, or compatibile format (for e.g. from OpenOffice.org |
26 |
|
|
or Gnuemeric) in C<.xls> format. |
27 |
|
|
|
28 |
|
|
=head1 FUNCTIONS |
29 |
|
|
|
30 |
dpavlin |
728 |
=head2 new |
31 |
dpavlin |
498 |
|
32 |
|
|
Returns handle to database and size |
33 |
|
|
|
34 |
dpavlin |
728 |
my $excel = new WebPAC::Input::Excel( |
35 |
dpavlin |
498 |
path => '/path/to/workbook.xls' |
36 |
|
|
worksheet => 'name of sheet', |
37 |
dpavlin |
524 |
from => 42, |
38 |
|
|
to => 9999, |
39 |
dpavlin |
498 |
} |
40 |
|
|
|
41 |
|
|
C<worksheet> is case and white-space insensitive name of worksheet in Excel |
42 |
|
|
file to use. If not specified, it will use first worksheet in file. |
43 |
|
|
|
44 |
dpavlin |
524 |
C<from> and C<to> specify row numbers to start and finish import. |
45 |
|
|
|
46 |
dpavlin |
498 |
=cut |
47 |
|
|
|
48 |
dpavlin |
728 |
sub new { |
49 |
|
|
my $class = shift; |
50 |
|
|
my $self = {@_}; |
51 |
|
|
bless($self, $class); |
52 |
dpavlin |
498 |
|
53 |
|
|
my $log = $self->_get_logger(); |
54 |
|
|
|
55 |
dpavlin |
728 |
$log->logdie("can't open excel file $self->{path}: $!") unless (-r $self->{path} && -f $self->{path}); |
56 |
dpavlin |
498 |
|
57 |
dpavlin |
728 |
my $workbook = Spreadsheet::ParseExcel::Workbook->Parse($self->{path}); |
58 |
dpavlin |
498 |
|
59 |
dpavlin |
728 |
my $sheet; |
60 |
|
|
my $wanted_worksheet; |
61 |
dpavlin |
498 |
|
62 |
dpavlin |
728 |
if ($wanted_worksheet = $self->{worksheet}) { |
63 |
dpavlin |
498 |
my $name; |
64 |
|
|
do { |
65 |
|
|
$sheet = shift @{ $workbook->{Worksheet} }; |
66 |
dpavlin |
728 |
$log->logdie("can't find sheet '$wanted_worksheet' in $self->{path}\n") unless (defined($sheet)); |
67 |
dpavlin |
498 |
$name = $sheet->{Name}; |
68 |
|
|
$name =~ s/\s\s+/ /g; |
69 |
|
|
} until ($name =~ m/^\s*\Q$wanted_worksheet\E\s*$/i); |
70 |
|
|
|
71 |
|
|
} else { |
72 |
|
|
|
73 |
|
|
$sheet = shift @{ $workbook->{Worksheet} }; |
74 |
|
|
|
75 |
|
|
} |
76 |
|
|
|
77 |
dpavlin |
728 |
$self->{sheet} = $sheet; |
78 |
dpavlin |
498 |
|
79 |
dpavlin |
728 |
$self->{from} ||= $sheet->{MinRow}; |
80 |
|
|
$self->{to} ||= $sheet->{MaxRow}; |
81 |
dpavlin |
524 |
|
82 |
dpavlin |
728 |
my $size = $self->{to} - $self->{from}; |
83 |
|
|
$self->{size} = $size; |
84 |
|
|
|
85 |
|
|
$log->warn("opening Excel file '$self->{path}', using ", |
86 |
dpavlin |
498 |
$wanted_worksheet ? '' : 'first ', |
87 |
|
|
"worksheet: $sheet->{Name} [$size rows]" |
88 |
|
|
); |
89 |
|
|
|
90 |
dpavlin |
728 |
$self ? return $self : return undef; |
91 |
dpavlin |
498 |
} |
92 |
|
|
|
93 |
|
|
=head2 fetch_rec |
94 |
|
|
|
95 |
|
|
Return record with ID C<$mfn> from database |
96 |
|
|
|
97 |
dpavlin |
652 |
my $rec = $self->fetch_rec( $mfn ); |
98 |
dpavlin |
498 |
|
99 |
dpavlin |
1055 |
Columns are named C<A>, C<B> and so on... |
100 |
dpavlin |
498 |
|
101 |
|
|
=cut |
102 |
|
|
|
103 |
|
|
sub fetch_rec { |
104 |
|
|
my $self = shift; |
105 |
|
|
|
106 |
dpavlin |
652 |
my $mfn = shift; |
107 |
dpavlin |
498 |
|
108 |
|
|
my $log = $self->_get_logger(); |
109 |
|
|
|
110 |
dpavlin |
728 |
my $sheet = $self->{sheet}; |
111 |
dpavlin |
498 |
$log->logdie("can't find sheet hash") unless (defined($sheet)); |
112 |
|
|
$log->logdie("sheet hash isn't Spreadsheet::ParseExcel::Worksheet") unless ($sheet->isa('Spreadsheet::ParseExcel::Worksheet')); |
113 |
|
|
|
114 |
|
|
my $rec; |
115 |
|
|
|
116 |
dpavlin |
728 |
my $row = $self->{from} + $mfn - 1; |
117 |
dpavlin |
498 |
|
118 |
|
|
$log->debug("fetch_rec( $mfn ) row: $row cols: ",$sheet->{MinCol}," - ",$sheet->{MaxCol}); |
119 |
|
|
|
120 |
|
|
foreach my $col ( $sheet->{MinCol} ... $sheet->{MaxCol} ) { |
121 |
dpavlin |
1105 |
if (my $v = $sheet->{Cells}->[$row]->[$col]->{_Value}) { ## XXX _Value = formatted | Val = unformated ! |
122 |
dpavlin |
498 |
$rec->{ int2col($col) } = $v; |
123 |
|
|
} |
124 |
|
|
} |
125 |
|
|
|
126 |
|
|
# add mfn only to records with data |
127 |
dpavlin |
521 |
$rec->{'000'} = [ $mfn ] if ($rec); |
128 |
dpavlin |
498 |
|
129 |
|
|
return $rec; |
130 |
|
|
} |
131 |
|
|
|
132 |
dpavlin |
728 |
=head2 size |
133 |
|
|
|
134 |
|
|
Return number of records in database |
135 |
|
|
|
136 |
|
|
my $size = $isis->size; |
137 |
|
|
|
138 |
|
|
=cut |
139 |
|
|
|
140 |
|
|
sub size { |
141 |
|
|
my $self = shift; |
142 |
|
|
return $self->{size}; |
143 |
|
|
} |
144 |
dpavlin |
1100 |
|
145 |
|
|
sub default_encoding { 'UTF-16' } |
146 |
|
|
|
147 |
dpavlin |
498 |
=head1 AUTHOR |
148 |
|
|
|
149 |
|
|
Dobrica Pavlinusic, C<< <dpavlin@rot13.org> >> |
150 |
|
|
|
151 |
|
|
=head1 COPYRIGHT & LICENSE |
152 |
|
|
|
153 |
|
|
Copyright 2005-2006 Dobrica Pavlinusic, All Rights Reserved. |
154 |
|
|
|
155 |
|
|
This program is free software; you can redistribute it and/or modify it |
156 |
|
|
under the same terms as Perl itself. |
157 |
|
|
|
158 |
|
|
=cut |
159 |
|
|
|
160 |
|
|
1; # End of WebPAC::Input::Excel |