| line |
stmt |
bran |
cond |
sub |
pod |
time |
code |
|
1
|
|
|
|
|
|
|
package Spreadsheet::CSV; |
|
2
|
|
|
|
|
|
|
|
|
3
|
1
|
|
|
1
|
|
21018
|
use 5.006; |
|
|
1
|
|
|
|
|
3
|
|
|
|
1
|
|
|
|
|
54
|
|
|
4
|
1
|
|
|
1
|
|
6
|
use strict; |
|
|
1
|
|
|
|
|
2
|
|
|
|
1
|
|
|
|
|
45
|
|
|
5
|
1
|
|
|
1
|
|
5
|
use warnings FATAL => 'all'; |
|
|
1
|
|
|
|
|
7
|
|
|
|
1
|
|
|
|
|
54
|
|
|
6
|
1
|
|
|
1
|
|
6
|
use Fcntl(); |
|
|
1
|
|
|
|
|
1
|
|
|
|
1
|
|
|
|
|
16
|
|
|
7
|
1
|
|
|
1
|
|
2683
|
use Spreadsheet::ParseExcel(); |
|
|
1
|
|
|
|
|
65783
|
|
|
|
1
|
|
|
|
|
36
|
|
|
8
|
1
|
|
|
1
|
|
852
|
use Text::CSV_XS(); |
|
|
1
|
|
|
|
|
32214
|
|
|
|
1
|
|
|
|
|
66
|
|
|
9
|
1
|
|
|
1
|
|
1303
|
use Archive::Zip(); |
|
|
1
|
|
|
|
|
55365
|
|
|
|
1
|
|
|
|
|
23
|
|
|
10
|
1
|
|
|
1
|
|
192
|
use XML::Parser(); |
|
|
0
|
|
|
|
|
|
|
|
|
0
|
|
|
|
|
|
|
|
11
|
|
|
|
|
|
|
use English qw( -no_match_vars ); |
|
12
|
|
|
|
|
|
|
use Compress::Zlib(); |
|
13
|
|
|
|
|
|
|
use Carp(); |
|
14
|
|
|
|
|
|
|
use IO::File(); |
|
15
|
|
|
|
|
|
|
use charnames ':full'; |
|
16
|
|
|
|
|
|
|
|
|
17
|
|
|
|
|
|
|
sub _MAGIC_NUMBER_BUFFER_SIZE { return 2 } # for .zip and .gz files |
|
18
|
|
|
|
|
|
|
sub _GRANDPARENT_INDEX { return -2 } |
|
19
|
|
|
|
|
|
|
sub _PARENT_INDEX { return -1 } |
|
20
|
|
|
|
|
|
|
sub _EXCEL_COLUMN_RADIX { return 26 } |
|
21
|
|
|
|
|
|
|
sub _BUFFER_SIZE { return 4096 } |
|
22
|
|
|
|
|
|
|
|
|
23
|
|
|
|
|
|
|
our $VERSION = '0.20'; |
|
24
|
|
|
|
|
|
|
|
|
25
|
|
|
|
|
|
|
sub new { |
|
26
|
|
|
|
|
|
|
my ( $class, $params ) = @_; |
|
27
|
|
|
|
|
|
|
my $self = {}; |
|
28
|
|
|
|
|
|
|
if ( defined $params->{worksheet_name} ) { |
|
29
|
|
|
|
|
|
|
$self->{worksheet_name} = $params->{worksheet_name}; |
|
30
|
|
|
|
|
|
|
} |
|
31
|
|
|
|
|
|
|
elsif ( defined $params->{worksheet_number} ) { |
|
32
|
|
|
|
|
|
|
if ( $params->{worksheet_number} =~ /^\d+$/smx ) { |
|
33
|
|
|
|
|
|
|
$self->{worksheet_number} = $params->{worksheet_number}; |
|
34
|
|
|
|
|
|
|
} |
|
35
|
|
|
|
|
|
|
else { |
|
36
|
|
|
|
|
|
|
Carp::croak('worksheet_number parameter is not a number'); |
|
37
|
|
|
|
|
|
|
} |
|
38
|
|
|
|
|
|
|
} |
|
39
|
|
|
|
|
|
|
else { |
|
40
|
|
|
|
|
|
|
$self->{worksheet_number} = 1; |
|
41
|
|
|
|
|
|
|
} |
|
42
|
|
|
|
|
|
|
delete $params->{worksheet_number}; |
|
43
|
|
|
|
|
|
|
delete $params->{worksheet_name}; |
|
44
|
|
|
|
|
|
|
$self->{csv} = Text::CSV_XS->new( { binary => 1 } ); |
|
45
|
|
|
|
|
|
|
bless $self, $class; |
|
46
|
|
|
|
|
|
|
return $self; |
|
47
|
|
|
|
|
|
|
} |
|
48
|
|
|
|
|
|
|
|
|
49
|
|
|
|
|
|
|
sub getline { |
|
50
|
|
|
|
|
|
|
my ( $self, $handle ) = @_; |
|
51
|
|
|
|
|
|
|
if ( ( defined $self->{handle} ) && ( $self->{handle} eq $handle ) ) { |
|
52
|
|
|
|
|
|
|
} |
|
53
|
|
|
|
|
|
|
else { |
|
54
|
|
|
|
|
|
|
$self->{eof} = q[]; |
|
55
|
|
|
|
|
|
|
if ( $self->_setup_handle($handle) ) { |
|
56
|
|
|
|
|
|
|
} |
|
57
|
|
|
|
|
|
|
else { |
|
58
|
|
|
|
|
|
|
return; |
|
59
|
|
|
|
|
|
|
} |
|
60
|
|
|
|
|
|
|
} |
|
61
|
|
|
|
|
|
|
my $row = $self->{cells}->[ $self->{row_index}++ ]; |
|
62
|
|
|
|
|
|
|
if ( !defined $row ) { |
|
63
|
|
|
|
|
|
|
$self->{eof} = 1; |
|
64
|
|
|
|
|
|
|
} |
|
65
|
|
|
|
|
|
|
return $row; |
|
66
|
|
|
|
|
|
|
} |
|
67
|
|
|
|
|
|
|
|
|
68
|
|
|
|
|
|
|
sub eof { |
|
69
|
|
|
|
|
|
|
my ($self) = @_; |
|
70
|
|
|
|
|
|
|
return $self->{eof}; |
|
71
|
|
|
|
|
|
|
} |
|
72
|
|
|
|
|
|
|
|
|
73
|
|
|
|
|
|
|
sub error_diag { |
|
74
|
|
|
|
|
|
|
my ($self) = @_; |
|
75
|
|
|
|
|
|
|
return $self->{_ERROR_DIAG}; |
|
76
|
|
|
|
|
|
|
} |
|
77
|
|
|
|
|
|
|
|
|
78
|
|
|
|
|
|
|
sub _stuff_input_into_tmp_file { |
|
79
|
|
|
|
|
|
|
my ( $self, $input_handle ) = @_; |
|
80
|
|
|
|
|
|
|
seek $input_handle, 0, Fcntl::SEEK_SET() |
|
81
|
|
|
|
|
|
|
or |
|
82
|
|
|
|
|
|
|
Carp::croak("Failed to seek to start of filehandle:$EXTENDED_OS_ERROR"); |
|
83
|
|
|
|
|
|
|
my $handle = IO::File->new_tmpfile() |
|
84
|
|
|
|
|
|
|
or Carp::croak("Failed to create temporary file:$EXTENDED_OS_ERROR"); |
|
85
|
|
|
|
|
|
|
my $result; |
|
86
|
|
|
|
|
|
|
while ( $result = read $input_handle, my $buffer, _BUFFER_SIZE() ) { |
|
87
|
|
|
|
|
|
|
print {$handle} $buffer |
|
88
|
|
|
|
|
|
|
or |
|
89
|
|
|
|
|
|
|
Carp::croak("Failed to write to temporary file:$EXTENDED_OS_ERROR"); |
|
90
|
|
|
|
|
|
|
} |
|
91
|
|
|
|
|
|
|
defined $result |
|
92
|
|
|
|
|
|
|
or Carp::croak "Failed to read from input file:$EXTENDED_OS_ERROR"; |
|
93
|
|
|
|
|
|
|
seek $handle, 0, Fcntl::SEEK_SET() |
|
94
|
|
|
|
|
|
|
or |
|
95
|
|
|
|
|
|
|
Carp::croak("Failed to seek to start of filehandle:$EXTENDED_OS_ERROR"); |
|
96
|
|
|
|
|
|
|
seek $input_handle, 0, Fcntl::SEEK_SET() |
|
97
|
|
|
|
|
|
|
or |
|
98
|
|
|
|
|
|
|
Carp::croak("Failed to seek to start of filehandle:$EXTENDED_OS_ERROR"); |
|
99
|
|
|
|
|
|
|
return $handle; |
|
100
|
|
|
|
|
|
|
} |
|
101
|
|
|
|
|
|
|
|
|
102
|
|
|
|
|
|
|
sub _xls_parser { |
|
103
|
|
|
|
|
|
|
my ($self) = @_; |
|
104
|
|
|
|
|
|
|
my $parser = Spreadsheet::ParseExcel->new( |
|
105
|
|
|
|
|
|
|
CellHandler => sub { |
|
106
|
|
|
|
|
|
|
|
|
107
|
|
|
|
|
|
|
my ( $workbook, $sheet_index, $row, $col, $cell ) = @_; |
|
108
|
|
|
|
|
|
|
|
|
109
|
|
|
|
|
|
|
my $worksheet = $workbook->worksheet($sheet_index); |
|
110
|
|
|
|
|
|
|
my $process_worksheet = 0; |
|
111
|
|
|
|
|
|
|
if ( ( defined $self->{worksheet_name} ) |
|
112
|
|
|
|
|
|
|
&& ( $self->{worksheet_name} eq $worksheet->get_name() ) ) |
|
113
|
|
|
|
|
|
|
{ |
|
114
|
|
|
|
|
|
|
$self->{xls_worksheet_found} = 1; |
|
115
|
|
|
|
|
|
|
$process_worksheet = 1; |
|
116
|
|
|
|
|
|
|
} |
|
117
|
|
|
|
|
|
|
elsif (( $self->{worksheet_number} ) |
|
118
|
|
|
|
|
|
|
&& ( ( $self->{worksheet_number} - 1 ) == $sheet_index ) ) |
|
119
|
|
|
|
|
|
|
{ |
|
120
|
|
|
|
|
|
|
$self->{xls_worksheet_found} = 1; |
|
121
|
|
|
|
|
|
|
$process_worksheet = 1; |
|
122
|
|
|
|
|
|
|
} |
|
123
|
|
|
|
|
|
|
if ($process_worksheet) { |
|
124
|
|
|
|
|
|
|
$self->{cells}->[$row]->[$col] = $cell->{_Value}; |
|
125
|
|
|
|
|
|
|
$self->{cells}->[$row]->[$col] =~ |
|
126
|
|
|
|
|
|
|
s/\N{CARRIAGE RETURN}\N{LINE FEED}/\N{LINE FEED}/smxg; |
|
127
|
|
|
|
|
|
|
$self->{cells}->[$row]->[$col] =~ |
|
128
|
|
|
|
|
|
|
s/\N{LINE FEED}\N{CARRIAGE RETURN}/\N{LINE FEED}/smxg; |
|
129
|
|
|
|
|
|
|
} |
|
130
|
|
|
|
|
|
|
}, |
|
131
|
|
|
|
|
|
|
NotSetCell => 1, |
|
132
|
|
|
|
|
|
|
); |
|
133
|
|
|
|
|
|
|
return $parser; |
|
134
|
|
|
|
|
|
|
} |
|
135
|
|
|
|
|
|
|
|
|
136
|
|
|
|
|
|
|
sub _setup_handle { |
|
137
|
|
|
|
|
|
|
my ( $self, $input_handle ) = @_; |
|
138
|
|
|
|
|
|
|
my $magic_bytes = $self->_sniff_magic_bytes($input_handle); |
|
139
|
|
|
|
|
|
|
my $handle = $self->_stuff_input_into_tmp_file($input_handle); |
|
140
|
|
|
|
|
|
|
my $parser = $self->_xls_parser(); |
|
141
|
|
|
|
|
|
|
if ( $magic_bytes =~ /^PK/smx ) { |
|
142
|
|
|
|
|
|
|
if ( defined $self->_setup_zip($handle) ) { |
|
143
|
|
|
|
|
|
|
$self->{handle} = $input_handle; |
|
144
|
|
|
|
|
|
|
} |
|
145
|
|
|
|
|
|
|
else { |
|
146
|
|
|
|
|
|
|
return; |
|
147
|
|
|
|
|
|
|
} |
|
148
|
|
|
|
|
|
|
} |
|
149
|
|
|
|
|
|
|
elsif ( $magic_bytes =~ /^\037\213/smx ) { |
|
150
|
|
|
|
|
|
|
if ( $self->_setup_compress_zlib_spreadsheet($handle) ) { |
|
151
|
|
|
|
|
|
|
$self->{handle} = $input_handle; |
|
152
|
|
|
|
|
|
|
} |
|
153
|
|
|
|
|
|
|
else { |
|
154
|
|
|
|
|
|
|
return; |
|
155
|
|
|
|
|
|
|
} |
|
156
|
|
|
|
|
|
|
} |
|
157
|
|
|
|
|
|
|
elsif ( ( defined $parser ) && ( my $workbook = $parser->parse($handle) ) ) |
|
158
|
|
|
|
|
|
|
{ |
|
159
|
|
|
|
|
|
|
if ( $self->_setup_xls_spreadsheet($workbook) ) { |
|
160
|
|
|
|
|
|
|
$self->{handle} = $input_handle; |
|
161
|
|
|
|
|
|
|
} |
|
162
|
|
|
|
|
|
|
elsif ( !$self->{xls_worksheet_found} ) { |
|
163
|
|
|
|
|
|
|
$self->{_ERROR_DIAG} = 'ENOENT - Worksheet ' |
|
164
|
|
|
|
|
|
|
. ( |
|
165
|
|
|
|
|
|
|
defined $self->{worksheet_name} |
|
166
|
|
|
|
|
|
|
? $self->{worksheet_name} |
|
167
|
|
|
|
|
|
|
: $self->{worksheet_number} |
|
168
|
|
|
|
|
|
|
) . ' not found'; |
|
169
|
|
|
|
|
|
|
return; |
|
170
|
|
|
|
|
|
|
} |
|
171
|
|
|
|
|
|
|
else { |
|
172
|
|
|
|
|
|
|
return; |
|
173
|
|
|
|
|
|
|
} |
|
174
|
|
|
|
|
|
|
} |
|
175
|
|
|
|
|
|
|
else { |
|
176
|
|
|
|
|
|
|
$handle = $self->_stuff_input_into_tmp_file($input_handle); |
|
177
|
|
|
|
|
|
|
if (!$self->_check_file_utf8($input_handle)) { |
|
178
|
|
|
|
|
|
|
$self->{_ERROR_DIAG} = 'CSV - Failed to parse as CSV'; |
|
179
|
|
|
|
|
|
|
return; |
|
180
|
|
|
|
|
|
|
} |
|
181
|
|
|
|
|
|
|
binmode $handle, ':encoding(UTF-8)'; |
|
182
|
|
|
|
|
|
|
$self->{cells} = []; |
|
183
|
|
|
|
|
|
|
my $parsed_ok; |
|
184
|
|
|
|
|
|
|
eval { |
|
185
|
|
|
|
|
|
|
while ( my $row = $self->{csv}->getline($handle) ) { |
|
186
|
|
|
|
|
|
|
push @{ $self->{cells} }, $row; |
|
187
|
|
|
|
|
|
|
} |
|
188
|
|
|
|
|
|
|
$parsed_ok = 1; |
|
189
|
|
|
|
|
|
|
} or do { |
|
190
|
|
|
|
|
|
|
$self->{_ERROR_DIAG} = 'CSV - Failed to parse as CSV'; |
|
191
|
|
|
|
|
|
|
return; |
|
192
|
|
|
|
|
|
|
}; |
|
193
|
|
|
|
|
|
|
if ( ($parsed_ok) && ( $self->{csv}->eof() ) ) { |
|
194
|
|
|
|
|
|
|
$self->{content_type} = 'text/csv'; # according to RFC 4180 |
|
195
|
|
|
|
|
|
|
$self->{type} = 'csv'; |
|
196
|
|
|
|
|
|
|
$self->{row_index} = 0; |
|
197
|
|
|
|
|
|
|
$self->{handle} = $input_handle; |
|
198
|
|
|
|
|
|
|
} |
|
199
|
|
|
|
|
|
|
else { |
|
200
|
|
|
|
|
|
|
$self->{_ERROR_DIAG} = 'CSV - Failed to parse as CSV:' |
|
201
|
|
|
|
|
|
|
. ( 0 + $self->{csv}->error_diag() ); |
|
202
|
|
|
|
|
|
|
return; |
|
203
|
|
|
|
|
|
|
} |
|
204
|
|
|
|
|
|
|
} |
|
205
|
|
|
|
|
|
|
return $self; |
|
206
|
|
|
|
|
|
|
} |
|
207
|
|
|
|
|
|
|
|
|
208
|
|
|
|
|
|
|
sub _check_file_utf8 { |
|
209
|
|
|
|
|
|
|
my ($self, $handle) = @_; |
|
210
|
|
|
|
|
|
|
while(my $line = <$handle>) { |
|
211
|
|
|
|
|
|
|
if (!utf8::decode(my $text = $line)) { |
|
212
|
|
|
|
|
|
|
return; |
|
213
|
|
|
|
|
|
|
} |
|
214
|
|
|
|
|
|
|
} |
|
215
|
|
|
|
|
|
|
return 1; |
|
216
|
|
|
|
|
|
|
} |
|
217
|
|
|
|
|
|
|
|
|
218
|
|
|
|
|
|
|
sub _setup_zip { |
|
219
|
|
|
|
|
|
|
my ( $self, $handle ) = @_; |
|
220
|
|
|
|
|
|
|
my $zip = Archive::Zip->new(); |
|
221
|
|
|
|
|
|
|
Archive::Zip::setErrorHandler( sub { chomp $_[0]; die "$_[0]\n"; } ); |
|
222
|
|
|
|
|
|
|
my $result; |
|
223
|
|
|
|
|
|
|
my $zip_error = 'Corrupt ZIP file'; |
|
224
|
|
|
|
|
|
|
eval { $result = $zip->readFromFileHandle($handle); } |
|
225
|
|
|
|
|
|
|
or do { chomp $EVAL_ERROR; $zip_error = $EVAL_ERROR }; |
|
226
|
|
|
|
|
|
|
if ( ( defined $result ) && ( $result == Archive::Zip::AZ_OK() ) ) { |
|
227
|
|
|
|
|
|
|
} |
|
228
|
|
|
|
|
|
|
else { |
|
229
|
|
|
|
|
|
|
$self->{_ERROR_DIAG} = "ZIP - $zip_error"; |
|
230
|
|
|
|
|
|
|
return; |
|
231
|
|
|
|
|
|
|
} |
|
232
|
|
|
|
|
|
|
|
|
233
|
|
|
|
|
|
|
if ( $self->_parse_archive_zip_spreadsheet($zip) ) { |
|
234
|
|
|
|
|
|
|
return 1; |
|
235
|
|
|
|
|
|
|
} |
|
236
|
|
|
|
|
|
|
else { |
|
237
|
|
|
|
|
|
|
return; |
|
238
|
|
|
|
|
|
|
} |
|
239
|
|
|
|
|
|
|
} |
|
240
|
|
|
|
|
|
|
|
|
241
|
|
|
|
|
|
|
sub _sniff_magic_bytes { |
|
242
|
|
|
|
|
|
|
my ( $self, $handle ) = @_; |
|
243
|
|
|
|
|
|
|
seek $handle, 0, Fcntl::SEEK_SET() |
|
244
|
|
|
|
|
|
|
or |
|
245
|
|
|
|
|
|
|
Carp::croak("Failed to seek to start of filehandle:$EXTENDED_OS_ERROR"); |
|
246
|
|
|
|
|
|
|
defined read $handle, my $magic_bytes, _MAGIC_NUMBER_BUFFER_SIZE() |
|
247
|
|
|
|
|
|
|
or Carp::croak("Failed to read from filehandle:$EXTENDED_OS_ERROR"); |
|
248
|
|
|
|
|
|
|
seek $handle, 0, Fcntl::SEEK_SET() |
|
249
|
|
|
|
|
|
|
or |
|
250
|
|
|
|
|
|
|
Carp::croak("Failed to seek to start of filehandle:$EXTENDED_OS_ERROR"); |
|
251
|
|
|
|
|
|
|
return $magic_bytes; |
|
252
|
|
|
|
|
|
|
} |
|
253
|
|
|
|
|
|
|
|
|
254
|
|
|
|
|
|
|
sub _handle_workbook_type_zips { |
|
255
|
|
|
|
|
|
|
my ( $self, $zip ) = @_; |
|
256
|
|
|
|
|
|
|
my $shared_strings = $self->_xlsx_shared_strings($zip); |
|
257
|
|
|
|
|
|
|
if ( !defined $shared_strings ) { |
|
258
|
|
|
|
|
|
|
return; |
|
259
|
|
|
|
|
|
|
} |
|
260
|
|
|
|
|
|
|
my $worksheet_path = $self->_xlsx_worksheet_path($zip); |
|
261
|
|
|
|
|
|
|
if ( ( defined $worksheet_path ) |
|
262
|
|
|
|
|
|
|
&& ( my $worksheet = $zip->memberNamed($worksheet_path) ) ) |
|
263
|
|
|
|
|
|
|
{ |
|
264
|
|
|
|
|
|
|
$self->{'content_type'} = |
|
265
|
|
|
|
|
|
|
'application/vnd.openxmlformats-officedocument.spreadsheetml.sheet'; |
|
266
|
|
|
|
|
|
|
$self->{type} = 'xlsx'; |
|
267
|
|
|
|
|
|
|
$self->{zip} = $zip; |
|
268
|
|
|
|
|
|
|
$self->{row_index} = 0; |
|
269
|
|
|
|
|
|
|
my $content = $worksheet->contents(); |
|
270
|
|
|
|
|
|
|
my $cells = |
|
271
|
|
|
|
|
|
|
$self->_xlsx_cells( $content, $shared_strings, $worksheet_path ); |
|
272
|
|
|
|
|
|
|
if ( defined $cells ) { |
|
273
|
|
|
|
|
|
|
$self->{cells} = $cells; |
|
274
|
|
|
|
|
|
|
return 1; |
|
275
|
|
|
|
|
|
|
} |
|
276
|
|
|
|
|
|
|
else { |
|
277
|
|
|
|
|
|
|
return; |
|
278
|
|
|
|
|
|
|
} |
|
279
|
|
|
|
|
|
|
} |
|
280
|
|
|
|
|
|
|
elsif ( defined $worksheet_path ) { |
|
281
|
|
|
|
|
|
|
$self->{_ERROR_DIAG} = |
|
282
|
|
|
|
|
|
|
q[ZIP - Missing '] . $worksheet_path . q[' file in .xlsx file]; |
|
283
|
|
|
|
|
|
|
return; |
|
284
|
|
|
|
|
|
|
} |
|
285
|
|
|
|
|
|
|
else { |
|
286
|
|
|
|
|
|
|
return; |
|
287
|
|
|
|
|
|
|
} |
|
288
|
|
|
|
|
|
|
} |
|
289
|
|
|
|
|
|
|
|
|
290
|
|
|
|
|
|
|
sub _handle_mimetype_type_zips { |
|
291
|
|
|
|
|
|
|
my ( $self, $zip ) = @_; |
|
292
|
|
|
|
|
|
|
my $member = $zip->memberNamed('mimetype'); |
|
293
|
|
|
|
|
|
|
my $content_type; |
|
294
|
|
|
|
|
|
|
delete $self->{type}; |
|
295
|
|
|
|
|
|
|
if ( defined $member ) { |
|
296
|
|
|
|
|
|
|
$content_type = $member->contents(); |
|
297
|
|
|
|
|
|
|
if ( $content_type eq 'application/vnd.oasis.opendocument.spreadsheet' ) |
|
298
|
|
|
|
|
|
|
{ |
|
299
|
|
|
|
|
|
|
$self->{type} = 'ods'; |
|
300
|
|
|
|
|
|
|
} |
|
301
|
|
|
|
|
|
|
elsif ( $content_type eq 'application/vnd.sun.xml.calc' ) { |
|
302
|
|
|
|
|
|
|
$self->{type} = 'sxc'; |
|
303
|
|
|
|
|
|
|
} |
|
304
|
|
|
|
|
|
|
elsif ( $content_type eq 'application/x-kspread' ) { |
|
305
|
|
|
|
|
|
|
$self->{type} = 'ksp'; |
|
306
|
|
|
|
|
|
|
} |
|
307
|
|
|
|
|
|
|
} |
|
308
|
|
|
|
|
|
|
if ( ( $self->{type} ) && ( $self->{type} eq 'ksp' ) ) { |
|
309
|
|
|
|
|
|
|
$self->{content_type} = $content_type; |
|
310
|
|
|
|
|
|
|
my $maindoc_member = $zip->memberNamed('maindoc.xml'); |
|
311
|
|
|
|
|
|
|
if ( defined $maindoc_member ) { |
|
312
|
|
|
|
|
|
|
$self->{zip} = $zip; |
|
313
|
|
|
|
|
|
|
$self->{row_index} = 0; |
|
314
|
|
|
|
|
|
|
my $maindoc_data = $maindoc_member->contents(); |
|
315
|
|
|
|
|
|
|
my $cells = $self->_ksp_cells($maindoc_data); |
|
316
|
|
|
|
|
|
|
if ( defined $cells ) { |
|
317
|
|
|
|
|
|
|
$self->{cells} = $cells; |
|
318
|
|
|
|
|
|
|
return 1; |
|
319
|
|
|
|
|
|
|
} |
|
320
|
|
|
|
|
|
|
else { |
|
321
|
|
|
|
|
|
|
return; |
|
322
|
|
|
|
|
|
|
} |
|
323
|
|
|
|
|
|
|
} |
|
324
|
|
|
|
|
|
|
else { |
|
325
|
|
|
|
|
|
|
$self->{_ERROR_DIAG} = |
|
326
|
|
|
|
|
|
|
q[ZIP - Missing 'content.xml' file in .] |
|
327
|
|
|
|
|
|
|
. ( lc $self->{type} ) |
|
328
|
|
|
|
|
|
|
. q[ file]; |
|
329
|
|
|
|
|
|
|
return; |
|
330
|
|
|
|
|
|
|
} |
|
331
|
|
|
|
|
|
|
} |
|
332
|
|
|
|
|
|
|
elsif ( $self->{type} ) { |
|
333
|
|
|
|
|
|
|
$self->{content_type} = $content_type; |
|
334
|
|
|
|
|
|
|
my $content_member = $zip->memberNamed('content.xml'); |
|
335
|
|
|
|
|
|
|
if ( defined $content_member ) { |
|
336
|
|
|
|
|
|
|
$self->{zip} = $zip; |
|
337
|
|
|
|
|
|
|
$self->{row_index} = 0; |
|
338
|
|
|
|
|
|
|
my $content_data = $content_member->contents(); |
|
339
|
|
|
|
|
|
|
my $cells = $self->_ods_cells($content_data); |
|
340
|
|
|
|
|
|
|
if ( defined $cells ) { |
|
341
|
|
|
|
|
|
|
$self->{cells} = $cells; |
|
342
|
|
|
|
|
|
|
return 1; |
|
343
|
|
|
|
|
|
|
} |
|
344
|
|
|
|
|
|
|
else { |
|
345
|
|
|
|
|
|
|
return; |
|
346
|
|
|
|
|
|
|
} |
|
347
|
|
|
|
|
|
|
} |
|
348
|
|
|
|
|
|
|
else { |
|
349
|
|
|
|
|
|
|
$self->{_ERROR_DIAG} = |
|
350
|
|
|
|
|
|
|
q[ZIP - Missing 'content.xml' file in .] |
|
351
|
|
|
|
|
|
|
. ( lc $self->{type} ) |
|
352
|
|
|
|
|
|
|
. q[ file]; |
|
353
|
|
|
|
|
|
|
return; |
|
354
|
|
|
|
|
|
|
} |
|
355
|
|
|
|
|
|
|
} |
|
356
|
|
|
|
|
|
|
else { |
|
357
|
|
|
|
|
|
|
$self->{_ERROR_DIAG} = |
|
358
|
|
|
|
|
|
|
q[ZIP - mimetype file does not contain any known MIME Types in OpenOffice document]; |
|
359
|
|
|
|
|
|
|
return; |
|
360
|
|
|
|
|
|
|
} |
|
361
|
|
|
|
|
|
|
} |
|
362
|
|
|
|
|
|
|
|
|
363
|
|
|
|
|
|
|
sub _parse_archive_zip_spreadsheet { |
|
364
|
|
|
|
|
|
|
my ( $self, $zip ) = @_; |
|
365
|
|
|
|
|
|
|
if ( $zip->memberNamed('xl/workbook.xml') ) { |
|
366
|
|
|
|
|
|
|
if ( !defined $self->_handle_workbook_type_zips($zip) ) { |
|
367
|
|
|
|
|
|
|
return; |
|
368
|
|
|
|
|
|
|
} |
|
369
|
|
|
|
|
|
|
} |
|
370
|
|
|
|
|
|
|
elsif ( $zip->memberNamed('mimetype') ) { |
|
371
|
|
|
|
|
|
|
if ( !defined $self->_handle_mimetype_type_zips($zip) ) { |
|
372
|
|
|
|
|
|
|
return; |
|
373
|
|
|
|
|
|
|
} |
|
374
|
|
|
|
|
|
|
} |
|
375
|
|
|
|
|
|
|
else { |
|
376
|
|
|
|
|
|
|
$self->{_ERROR_DIAG} = |
|
377
|
|
|
|
|
|
|
q[ZIP - Missing any identifiable spreadsheet file in ZIP archive]; |
|
378
|
|
|
|
|
|
|
return; |
|
379
|
|
|
|
|
|
|
} |
|
380
|
|
|
|
|
|
|
return 1; |
|
381
|
|
|
|
|
|
|
} |
|
382
|
|
|
|
|
|
|
|
|
383
|
|
|
|
|
|
|
sub _setup_compress_zlib_spreadsheet { |
|
384
|
|
|
|
|
|
|
my ( $self, $handle ) = @_; |
|
385
|
|
|
|
|
|
|
seek $handle, 0, Fcntl::SEEK_SET() |
|
386
|
|
|
|
|
|
|
or |
|
387
|
|
|
|
|
|
|
Carp::croak("Failed to seek to start of filehandle:$EXTENDED_OS_ERROR"); |
|
388
|
|
|
|
|
|
|
|
|
389
|
|
|
|
|
|
|
my $contents; |
|
390
|
|
|
|
|
|
|
my $gzip_error = 'Corrupt GZIP'; |
|
391
|
|
|
|
|
|
|
my $result = 0; |
|
392
|
|
|
|
|
|
|
eval { |
|
393
|
|
|
|
|
|
|
my $gz = Compress::Zlib::gzopen( $handle, 'rb' ) |
|
394
|
|
|
|
|
|
|
or die "Cannot open handle: $Compress::Zlib::gzerrno\n"; |
|
395
|
|
|
|
|
|
|
|
|
396
|
|
|
|
|
|
|
while ( $gz->gzread( my $uncompressed ) > 0 ) { |
|
397
|
|
|
|
|
|
|
$contents .= $uncompressed; |
|
398
|
|
|
|
|
|
|
} |
|
399
|
|
|
|
|
|
|
if ( $gz->gzerror() != Compress::Zlib::Z_STREAM_END() ) { |
|
400
|
|
|
|
|
|
|
die "Failed to read from handle:$Compress::Zlib::gzerrno\n"; |
|
401
|
|
|
|
|
|
|
} |
|
402
|
|
|
|
|
|
|
$gz->gzclose() |
|
403
|
|
|
|
|
|
|
and die "Failed to close compression:$Compress::Zlib::gzerrno\n"; |
|
404
|
|
|
|
|
|
|
$result = 1; |
|
405
|
|
|
|
|
|
|
} or do { |
|
406
|
|
|
|
|
|
|
chomp $EVAL_ERROR; |
|
407
|
|
|
|
|
|
|
if ($EVAL_ERROR) { |
|
408
|
|
|
|
|
|
|
$gzip_error = $EVAL_ERROR; |
|
409
|
|
|
|
|
|
|
} |
|
410
|
|
|
|
|
|
|
}; |
|
411
|
|
|
|
|
|
|
if ( $result == 0 ) { |
|
412
|
|
|
|
|
|
|
$self->{_ERROR_DIAG} = q[GZIP - ] . $gzip_error; |
|
413
|
|
|
|
|
|
|
return; |
|
414
|
|
|
|
|
|
|
} |
|
415
|
|
|
|
|
|
|
$self->{content_type} = 'application/x-gnumeric'; |
|
416
|
|
|
|
|
|
|
$self->{type} = 'gnumeric'; |
|
417
|
|
|
|
|
|
|
$self->{row_index} = 0; |
|
418
|
|
|
|
|
|
|
my $cells = $self->_gnumeric_cells($contents); |
|
419
|
|
|
|
|
|
|
if ( defined $cells ) { |
|
420
|
|
|
|
|
|
|
$self->{cells} = $cells; |
|
421
|
|
|
|
|
|
|
} |
|
422
|
|
|
|
|
|
|
else { |
|
423
|
|
|
|
|
|
|
return; |
|
424
|
|
|
|
|
|
|
} |
|
425
|
|
|
|
|
|
|
return 1; |
|
426
|
|
|
|
|
|
|
} |
|
427
|
|
|
|
|
|
|
|
|
428
|
|
|
|
|
|
|
sub _setup_xls_spreadsheet { |
|
429
|
|
|
|
|
|
|
my ( $self, $workbook ) = @_; |
|
430
|
|
|
|
|
|
|
my $worksheet; |
|
431
|
|
|
|
|
|
|
if ( defined $self->{worksheet_name} ) { |
|
432
|
|
|
|
|
|
|
$worksheet = $workbook->worksheet( $self->{worksheet_name} ); |
|
433
|
|
|
|
|
|
|
} |
|
434
|
|
|
|
|
|
|
elsif ( $self->{worksheet_number} ) { |
|
435
|
|
|
|
|
|
|
$worksheet = $workbook->worksheet( $self->{worksheet_number} - 1 ); |
|
436
|
|
|
|
|
|
|
} |
|
437
|
|
|
|
|
|
|
if ( !defined $worksheet ) { |
|
438
|
|
|
|
|
|
|
return; |
|
439
|
|
|
|
|
|
|
} |
|
440
|
|
|
|
|
|
|
$self->{content_type} = 'application/vnd.ms-excel'; |
|
441
|
|
|
|
|
|
|
$self->{type} = 'xls'; |
|
442
|
|
|
|
|
|
|
$self->{row_index} = 0; |
|
443
|
|
|
|
|
|
|
return 1; |
|
444
|
|
|
|
|
|
|
} |
|
445
|
|
|
|
|
|
|
|
|
446
|
|
|
|
|
|
|
sub _process_worksheet { |
|
447
|
|
|
|
|
|
|
my ( $self, $current_worksheet_name, $current_worksheet_index ) = @_; |
|
448
|
|
|
|
|
|
|
my $process_worksheet = 0; |
|
449
|
|
|
|
|
|
|
if ( ( defined $self->{worksheet_name} ) |
|
450
|
|
|
|
|
|
|
&& ( $self->{worksheet_name} eq $current_worksheet_name ) ) |
|
451
|
|
|
|
|
|
|
{ |
|
452
|
|
|
|
|
|
|
$process_worksheet = 1; |
|
453
|
|
|
|
|
|
|
} |
|
454
|
|
|
|
|
|
|
elsif (( $self->{worksheet_number} ) |
|
455
|
|
|
|
|
|
|
&& ( ( $self->{worksheet_number} - 1 ) == $current_worksheet_index ) ) |
|
456
|
|
|
|
|
|
|
{ |
|
457
|
|
|
|
|
|
|
$process_worksheet = 1; |
|
458
|
|
|
|
|
|
|
} |
|
459
|
|
|
|
|
|
|
return $process_worksheet; |
|
460
|
|
|
|
|
|
|
} |
|
461
|
|
|
|
|
|
|
|
|
462
|
|
|
|
|
|
|
sub _ksp_cells { |
|
463
|
|
|
|
|
|
|
my ( $self, $maindoc_content ) = @_; |
|
464
|
|
|
|
|
|
|
my $cells; |
|
465
|
|
|
|
|
|
|
my $current_worksheet_name; |
|
466
|
|
|
|
|
|
|
my $parameter_stack; |
|
467
|
|
|
|
|
|
|
my $element_stack; |
|
468
|
|
|
|
|
|
|
my $worksheet_count; |
|
469
|
|
|
|
|
|
|
my $process_worksheet = 0; |
|
470
|
|
|
|
|
|
|
my $xml = XML::Parser->new( |
|
471
|
|
|
|
|
|
|
Handlers => { |
|
472
|
|
|
|
|
|
|
Entity => sub { |
|
473
|
|
|
|
|
|
|
die |
|
474
|
|
|
|
|
|
|
"XML Entities have been detected and rejected in the XML, due to security concerns\n"; |
|
475
|
|
|
|
|
|
|
}, |
|
476
|
|
|
|
|
|
|
Start => sub { |
|
477
|
|
|
|
|
|
|
my ( $expat, $element_name, %element_parameters ) = @_; |
|
478
|
|
|
|
|
|
|
push @{$element_stack}, $element_name; |
|
479
|
|
|
|
|
|
|
push @{$parameter_stack}, \%element_parameters; |
|
480
|
|
|
|
|
|
|
if ( $element_name eq 'table' ) { |
|
481
|
|
|
|
|
|
|
if ( defined $worksheet_count ) { |
|
482
|
|
|
|
|
|
|
$worksheet_count += 1; |
|
483
|
|
|
|
|
|
|
} |
|
484
|
|
|
|
|
|
|
else { |
|
485
|
|
|
|
|
|
|
$worksheet_count = 0; |
|
486
|
|
|
|
|
|
|
} |
|
487
|
|
|
|
|
|
|
$process_worksheet = |
|
488
|
|
|
|
|
|
|
$self->_process_worksheet( $element_parameters{name}, |
|
489
|
|
|
|
|
|
|
$worksheet_count ); |
|
490
|
|
|
|
|
|
|
} |
|
491
|
|
|
|
|
|
|
}, |
|
492
|
|
|
|
|
|
|
End => sub { |
|
493
|
|
|
|
|
|
|
my ( $expat, $element_name ) = @_; |
|
494
|
|
|
|
|
|
|
pop @{$parameter_stack}; |
|
495
|
|
|
|
|
|
|
if ( $element_name ne pop @{$element_stack} ) { |
|
496
|
|
|
|
|
|
|
Carp::croak( |
|
497
|
|
|
|
|
|
|
'Internal confusion in processing XML elements'); |
|
498
|
|
|
|
|
|
|
} |
|
499
|
|
|
|
|
|
|
}, |
|
500
|
|
|
|
|
|
|
Char => sub { |
|
501
|
|
|
|
|
|
|
my ( $expat, $content ) = @_; |
|
502
|
|
|
|
|
|
|
if ($process_worksheet) { |
|
503
|
|
|
|
|
|
|
if ( ( defined $element_stack->[ _GRANDPARENT_INDEX() ] ) |
|
504
|
|
|
|
|
|
|
&& ( |
|
505
|
|
|
|
|
|
|
$element_stack->[ _GRANDPARENT_INDEX() ] eq 'cell' ) |
|
506
|
|
|
|
|
|
|
&& ( $element_stack->[ _PARENT_INDEX() ] eq 'text' ) ) |
|
507
|
|
|
|
|
|
|
{ |
|
508
|
|
|
|
|
|
|
$cells->[ $parameter_stack->[ _GRANDPARENT_INDEX() ] |
|
509
|
|
|
|
|
|
|
->{row} - 1 ] |
|
510
|
|
|
|
|
|
|
->[ $parameter_stack->[ _GRANDPARENT_INDEX() ] |
|
511
|
|
|
|
|
|
|
->{column} - 1 ] .= $content; |
|
512
|
|
|
|
|
|
|
} |
|
513
|
|
|
|
|
|
|
} |
|
514
|
|
|
|
|
|
|
} |
|
515
|
|
|
|
|
|
|
} |
|
516
|
|
|
|
|
|
|
); |
|
517
|
|
|
|
|
|
|
eval { $xml->parse($maindoc_content); } or do { |
|
518
|
|
|
|
|
|
|
chomp $EVAL_ERROR; |
|
519
|
|
|
|
|
|
|
$EVAL_ERROR =~ s/^\s*//smx; |
|
520
|
|
|
|
|
|
|
$self->{_ERROR_DIAG} = "XML - Invalid XML in maindoc.xml:$EVAL_ERROR"; |
|
521
|
|
|
|
|
|
|
return; |
|
522
|
|
|
|
|
|
|
}; |
|
523
|
|
|
|
|
|
|
if ( defined $cells ) { |
|
524
|
|
|
|
|
|
|
return $cells; |
|
525
|
|
|
|
|
|
|
} |
|
526
|
|
|
|
|
|
|
else { |
|
527
|
|
|
|
|
|
|
$self->{_ERROR_DIAG} = 'ENOENT - Worksheet ' |
|
528
|
|
|
|
|
|
|
. ( |
|
529
|
|
|
|
|
|
|
defined $self->{worksheet_name} |
|
530
|
|
|
|
|
|
|
? $self->{worksheet_name} |
|
531
|
|
|
|
|
|
|
: $self->{worksheet_number} |
|
532
|
|
|
|
|
|
|
) . ' not found'; |
|
533
|
|
|
|
|
|
|
return; |
|
534
|
|
|
|
|
|
|
} |
|
535
|
|
|
|
|
|
|
} |
|
536
|
|
|
|
|
|
|
|
|
537
|
|
|
|
|
|
|
sub _ods_cells { |
|
538
|
|
|
|
|
|
|
my ( $self, $ods_content ) = @_; |
|
539
|
|
|
|
|
|
|
my $cells; |
|
540
|
|
|
|
|
|
|
my $current_row; |
|
541
|
|
|
|
|
|
|
my $current_column_number; |
|
542
|
|
|
|
|
|
|
my $element_stack; |
|
543
|
|
|
|
|
|
|
my $process_worksheet = 0; |
|
544
|
|
|
|
|
|
|
my $worksheet_count; |
|
545
|
|
|
|
|
|
|
my $xml = XML::Parser->new( |
|
546
|
|
|
|
|
|
|
Handlers => { |
|
547
|
|
|
|
|
|
|
Entity => sub { |
|
548
|
|
|
|
|
|
|
die |
|
549
|
|
|
|
|
|
|
"XML Entities have been detected and rejected in the XML, due to security concerns\n"; |
|
550
|
|
|
|
|
|
|
}, |
|
551
|
|
|
|
|
|
|
Start => sub { |
|
552
|
|
|
|
|
|
|
my ( $expat, $element_name, %element_parameters ) = @_; |
|
553
|
|
|
|
|
|
|
push @{$element_stack}, $element_name; |
|
554
|
|
|
|
|
|
|
my ( $prefix, $suffix ) = split /:/smx, $element_name; |
|
555
|
|
|
|
|
|
|
if ( $prefix eq 'table' ) { |
|
556
|
|
|
|
|
|
|
if ( $suffix eq 'table-row' ) { |
|
557
|
|
|
|
|
|
|
$current_row = []; |
|
558
|
|
|
|
|
|
|
$current_column_number = 0 - 1; |
|
559
|
|
|
|
|
|
|
} |
|
560
|
|
|
|
|
|
|
elsif ( $suffix eq 'table' ) { |
|
561
|
|
|
|
|
|
|
if ( defined $worksheet_count ) { |
|
562
|
|
|
|
|
|
|
$worksheet_count += 1; |
|
563
|
|
|
|
|
|
|
} |
|
564
|
|
|
|
|
|
|
else { |
|
565
|
|
|
|
|
|
|
$worksheet_count = 0; |
|
566
|
|
|
|
|
|
|
} |
|
567
|
|
|
|
|
|
|
$process_worksheet = |
|
568
|
|
|
|
|
|
|
$self->_process_worksheet( |
|
569
|
|
|
|
|
|
|
$element_parameters{'table:name'}, |
|
570
|
|
|
|
|
|
|
$worksheet_count ); |
|
571
|
|
|
|
|
|
|
} |
|
572
|
|
|
|
|
|
|
elsif ( $suffix eq 'table-cell' ) { |
|
573
|
|
|
|
|
|
|
$current_column_number += 1; |
|
574
|
|
|
|
|
|
|
} |
|
575
|
|
|
|
|
|
|
} |
|
576
|
|
|
|
|
|
|
elsif ( $prefix eq 'text' ) { |
|
577
|
|
|
|
|
|
|
if ( $suffix eq 'p' ) { |
|
578
|
|
|
|
|
|
|
if ( ( defined $current_row ) |
|
579
|
|
|
|
|
|
|
&& ( |
|
580
|
|
|
|
|
|
|
defined $current_row->[$current_column_number] ) |
|
581
|
|
|
|
|
|
|
) |
|
582
|
|
|
|
|
|
|
{ |
|
583
|
|
|
|
|
|
|
$current_row->[$current_column_number] .= "\n"; |
|
584
|
|
|
|
|
|
|
} |
|
585
|
|
|
|
|
|
|
} |
|
586
|
|
|
|
|
|
|
elsif ( $suffix eq 's' ) { |
|
587
|
|
|
|
|
|
|
if ( ( defined $current_row ) |
|
588
|
|
|
|
|
|
|
&& ( |
|
589
|
|
|
|
|
|
|
defined $current_row->[$current_column_number] ) |
|
590
|
|
|
|
|
|
|
) |
|
591
|
|
|
|
|
|
|
{ |
|
592
|
|
|
|
|
|
|
$current_row->[$current_column_number] .= q[ ]; |
|
593
|
|
|
|
|
|
|
} |
|
594
|
|
|
|
|
|
|
} |
|
595
|
|
|
|
|
|
|
} |
|
596
|
|
|
|
|
|
|
}, |
|
597
|
|
|
|
|
|
|
End => sub { |
|
598
|
|
|
|
|
|
|
my ( $expat, $element_name ) = @_; |
|
599
|
|
|
|
|
|
|
if ( $element_name ne pop @{$element_stack} ) { |
|
600
|
|
|
|
|
|
|
Carp::croak( |
|
601
|
|
|
|
|
|
|
'Internal confusion in processing XML elements'); |
|
602
|
|
|
|
|
|
|
} |
|
603
|
|
|
|
|
|
|
if ( $element_name eq 'table:table-row' ) { |
|
604
|
|
|
|
|
|
|
if ( @{$current_row} ) { |
|
605
|
|
|
|
|
|
|
push @{$cells}, $current_row; |
|
606
|
|
|
|
|
|
|
} |
|
607
|
|
|
|
|
|
|
} |
|
608
|
|
|
|
|
|
|
elsif ( $element_name eq 'table:table' ) { |
|
609
|
|
|
|
|
|
|
$process_worksheet = 0; |
|
610
|
|
|
|
|
|
|
} |
|
611
|
|
|
|
|
|
|
}, |
|
612
|
|
|
|
|
|
|
Char => sub { |
|
613
|
|
|
|
|
|
|
my ( $expat, $content ) = @_; |
|
614
|
|
|
|
|
|
|
if ($process_worksheet) { |
|
615
|
|
|
|
|
|
|
if ( $element_stack->[_PARENT_INDEX] eq 'text:p' ) { |
|
616
|
|
|
|
|
|
|
if ( $element_stack->[ _GRANDPARENT_INDEX() ] eq |
|
617
|
|
|
|
|
|
|
'table:table-cell' ) |
|
618
|
|
|
|
|
|
|
{ |
|
619
|
|
|
|
|
|
|
if ( |
|
620
|
|
|
|
|
|
|
defined $current_row->[$current_column_number] ) |
|
621
|
|
|
|
|
|
|
{ |
|
622
|
|
|
|
|
|
|
$current_row->[$current_column_number] .= |
|
623
|
|
|
|
|
|
|
$content; |
|
624
|
|
|
|
|
|
|
} |
|
625
|
|
|
|
|
|
|
else { |
|
626
|
|
|
|
|
|
|
$current_row->[$current_column_number] = |
|
627
|
|
|
|
|
|
|
$content; |
|
628
|
|
|
|
|
|
|
} |
|
629
|
|
|
|
|
|
|
} |
|
630
|
|
|
|
|
|
|
} |
|
631
|
|
|
|
|
|
|
} |
|
632
|
|
|
|
|
|
|
} |
|
633
|
|
|
|
|
|
|
} |
|
634
|
|
|
|
|
|
|
); |
|
635
|
|
|
|
|
|
|
eval { $xml->parse($ods_content); } or do { |
|
636
|
|
|
|
|
|
|
chomp $EVAL_ERROR; |
|
637
|
|
|
|
|
|
|
$EVAL_ERROR =~ s/^\s*//smx; |
|
638
|
|
|
|
|
|
|
$self->{_ERROR_DIAG} = "XML - Invalid XML in content.xml:$EVAL_ERROR"; |
|
639
|
|
|
|
|
|
|
return; |
|
640
|
|
|
|
|
|
|
}; |
|
641
|
|
|
|
|
|
|
if ( defined $cells ) { |
|
642
|
|
|
|
|
|
|
return $cells; |
|
643
|
|
|
|
|
|
|
} |
|
644
|
|
|
|
|
|
|
else { |
|
645
|
|
|
|
|
|
|
$self->{_ERROR_DIAG} = 'ENOENT - Worksheet ' |
|
646
|
|
|
|
|
|
|
. ( |
|
647
|
|
|
|
|
|
|
defined $self->{worksheet_name} |
|
648
|
|
|
|
|
|
|
? $self->{worksheet_name} |
|
649
|
|
|
|
|
|
|
: $self->{worksheet_number} |
|
650
|
|
|
|
|
|
|
) . ' not found'; |
|
651
|
|
|
|
|
|
|
return; |
|
652
|
|
|
|
|
|
|
} |
|
653
|
|
|
|
|
|
|
} |
|
654
|
|
|
|
|
|
|
|
|
655
|
|
|
|
|
|
|
sub _gnumeric_cells { |
|
656
|
|
|
|
|
|
|
my ( $self, $gnumeric_content ) = @_; |
|
657
|
|
|
|
|
|
|
my $cells; |
|
658
|
|
|
|
|
|
|
my $current_row_number; |
|
659
|
|
|
|
|
|
|
my $current_column_number; |
|
660
|
|
|
|
|
|
|
my $current_worksheet_name; |
|
661
|
|
|
|
|
|
|
my $current_sheet_cells; |
|
662
|
|
|
|
|
|
|
my $element_stack; |
|
663
|
|
|
|
|
|
|
my $process_worksheet = 0; |
|
664
|
|
|
|
|
|
|
my $worksheet_count; |
|
665
|
|
|
|
|
|
|
my $xml = XML::Parser->new( |
|
666
|
|
|
|
|
|
|
Handlers => { |
|
667
|
|
|
|
|
|
|
Entity => sub { |
|
668
|
|
|
|
|
|
|
die |
|
669
|
|
|
|
|
|
|
"XML Entities have been detected and rejected in the XML, due to security concerns\n"; |
|
670
|
|
|
|
|
|
|
}, |
|
671
|
|
|
|
|
|
|
Start => sub { |
|
672
|
|
|
|
|
|
|
my ( $expat, $element_name, %element_parameters ) = @_; |
|
673
|
|
|
|
|
|
|
push @{$element_stack}, $element_name; |
|
674
|
|
|
|
|
|
|
if ( $element_name eq 'gnm:Cell' ) { |
|
675
|
|
|
|
|
|
|
$current_row_number = $element_parameters{Row}; |
|
676
|
|
|
|
|
|
|
$current_column_number = $element_parameters{Col}; |
|
677
|
|
|
|
|
|
|
} |
|
678
|
|
|
|
|
|
|
elsif ( $element_name eq 'gnm:Sheet' ) { |
|
679
|
|
|
|
|
|
|
$current_sheet_cells = []; |
|
680
|
|
|
|
|
|
|
} |
|
681
|
|
|
|
|
|
|
}, |
|
682
|
|
|
|
|
|
|
End => sub { |
|
683
|
|
|
|
|
|
|
my ( $expat, $element_name ) = @_; |
|
684
|
|
|
|
|
|
|
if ( $element_name ne pop @{$element_stack} ) { |
|
685
|
|
|
|
|
|
|
Carp::croak( |
|
686
|
|
|
|
|
|
|
'Internal confusion in processing XML elements'); |
|
687
|
|
|
|
|
|
|
} |
|
688
|
|
|
|
|
|
|
elsif ( $element_name eq 'gnm:Sheet' ) { |
|
689
|
|
|
|
|
|
|
if ( defined $worksheet_count ) { |
|
690
|
|
|
|
|
|
|
$worksheet_count += 1; |
|
691
|
|
|
|
|
|
|
} |
|
692
|
|
|
|
|
|
|
else { |
|
693
|
|
|
|
|
|
|
$worksheet_count = 0; |
|
694
|
|
|
|
|
|
|
} |
|
695
|
|
|
|
|
|
|
if ( |
|
696
|
|
|
|
|
|
|
$self->_process_worksheet( |
|
697
|
|
|
|
|
|
|
$current_worksheet_name, $worksheet_count |
|
698
|
|
|
|
|
|
|
) |
|
699
|
|
|
|
|
|
|
) |
|
700
|
|
|
|
|
|
|
{ |
|
701
|
|
|
|
|
|
|
$cells = $current_sheet_cells; |
|
702
|
|
|
|
|
|
|
} |
|
703
|
|
|
|
|
|
|
} |
|
704
|
|
|
|
|
|
|
}, |
|
705
|
|
|
|
|
|
|
Char => sub { |
|
706
|
|
|
|
|
|
|
my ( $expat, $content ) = @_; |
|
707
|
|
|
|
|
|
|
if ( $element_stack->[ _PARENT_INDEX() ] eq 'gnm:Name' ) { |
|
708
|
|
|
|
|
|
|
if ( $element_stack->[ _GRANDPARENT_INDEX() ] eq |
|
709
|
|
|
|
|
|
|
'gnm:Sheet' ) |
|
710
|
|
|
|
|
|
|
{ |
|
711
|
|
|
|
|
|
|
$current_worksheet_name = $content; |
|
712
|
|
|
|
|
|
|
} |
|
713
|
|
|
|
|
|
|
} |
|
714
|
|
|
|
|
|
|
if ( $element_stack->[ _PARENT_INDEX() ] eq 'gnm:Cell' ) { |
|
715
|
|
|
|
|
|
|
if ( $content eq "\N{LINE FEED}" ) { |
|
716
|
|
|
|
|
|
|
$current_sheet_cells->[$current_row_number] |
|
717
|
|
|
|
|
|
|
->[$current_column_number] .= |
|
718
|
|
|
|
|
|
|
$expat->original_string(); |
|
719
|
|
|
|
|
|
|
} |
|
720
|
|
|
|
|
|
|
else { |
|
721
|
|
|
|
|
|
|
$current_sheet_cells->[$current_row_number] |
|
722
|
|
|
|
|
|
|
->[$current_column_number] .= $content; |
|
723
|
|
|
|
|
|
|
} |
|
724
|
|
|
|
|
|
|
} |
|
725
|
|
|
|
|
|
|
} |
|
726
|
|
|
|
|
|
|
} |
|
727
|
|
|
|
|
|
|
); |
|
728
|
|
|
|
|
|
|
eval { $xml->parse($gnumeric_content); } or do { |
|
729
|
|
|
|
|
|
|
chomp $EVAL_ERROR; |
|
730
|
|
|
|
|
|
|
$EVAL_ERROR =~ s/^\s*//smx; |
|
731
|
|
|
|
|
|
|
$self->{_ERROR_DIAG} = |
|
732
|
|
|
|
|
|
|
"XML - Invalid XML in gzipped gnumeric file:$EVAL_ERROR"; |
|
733
|
|
|
|
|
|
|
return; |
|
734
|
|
|
|
|
|
|
}; |
|
735
|
|
|
|
|
|
|
if ( defined $cells ) { |
|
736
|
|
|
|
|
|
|
return $cells; |
|
737
|
|
|
|
|
|
|
} |
|
738
|
|
|
|
|
|
|
else { |
|
739
|
|
|
|
|
|
|
$self->{_ERROR_DIAG} = 'ENOENT - Worksheet ' |
|
740
|
|
|
|
|
|
|
. ( |
|
741
|
|
|
|
|
|
|
defined $self->{worksheet_name} |
|
742
|
|
|
|
|
|
|
? $self->{worksheet_name} |
|
743
|
|
|
|
|
|
|
: $self->{worksheet_number} |
|
744
|
|
|
|
|
|
|
) . ' not found'; |
|
745
|
|
|
|
|
|
|
return; |
|
746
|
|
|
|
|
|
|
} |
|
747
|
|
|
|
|
|
|
} |
|
748
|
|
|
|
|
|
|
|
|
749
|
|
|
|
|
|
|
sub _xlsx_shared_strings { |
|
750
|
|
|
|
|
|
|
my ( $self, $zip ) = @_; |
|
751
|
|
|
|
|
|
|
my $member = $zip->memberNamed('xl/sharedStrings.xml'); |
|
752
|
|
|
|
|
|
|
if ( !$member ) { |
|
753
|
|
|
|
|
|
|
$self->{_ERROR_DIAG} = q[ZIP - Missing 'xl/sharedStrings.xml' file]; |
|
754
|
|
|
|
|
|
|
return; |
|
755
|
|
|
|
|
|
|
} |
|
756
|
|
|
|
|
|
|
my $shared_string_content = $member->contents(); |
|
757
|
|
|
|
|
|
|
my $element_stack; |
|
758
|
|
|
|
|
|
|
my $current_index; |
|
759
|
|
|
|
|
|
|
my $shared_strings = {}; |
|
760
|
|
|
|
|
|
|
my $xml = XML::Parser->new( |
|
761
|
|
|
|
|
|
|
Handlers => { |
|
762
|
|
|
|
|
|
|
Entity => sub { |
|
763
|
|
|
|
|
|
|
die |
|
764
|
|
|
|
|
|
|
"XML Entities have been detected and rejected in the XML, due to security concerns\n"; |
|
765
|
|
|
|
|
|
|
}, |
|
766
|
|
|
|
|
|
|
Start => sub { |
|
767
|
|
|
|
|
|
|
my ( $expat, $element_name, %element_parameters ) = @_; |
|
768
|
|
|
|
|
|
|
push @{$element_stack}, $element_name; |
|
769
|
|
|
|
|
|
|
if ( $element_name eq 'sst' ) { |
|
770
|
|
|
|
|
|
|
} |
|
771
|
|
|
|
|
|
|
elsif ( $element_name eq 'si' ) { |
|
772
|
|
|
|
|
|
|
if ( defined $current_index ) { |
|
773
|
|
|
|
|
|
|
$current_index += 1; |
|
774
|
|
|
|
|
|
|
} |
|
775
|
|
|
|
|
|
|
else { |
|
776
|
|
|
|
|
|
|
$current_index = 0; |
|
777
|
|
|
|
|
|
|
} |
|
778
|
|
|
|
|
|
|
} |
|
779
|
|
|
|
|
|
|
elsif ( $element_name eq 't' ) { |
|
780
|
|
|
|
|
|
|
} |
|
781
|
|
|
|
|
|
|
}, |
|
782
|
|
|
|
|
|
|
End => sub { |
|
783
|
|
|
|
|
|
|
my ( $expat, $element_name ) = @_; |
|
784
|
|
|
|
|
|
|
if ( $element_name ne pop @{$element_stack} ) { |
|
785
|
|
|
|
|
|
|
Carp::croak( |
|
786
|
|
|
|
|
|
|
'Internal confusion in processing XML elements'); |
|
787
|
|
|
|
|
|
|
} |
|
788
|
|
|
|
|
|
|
}, |
|
789
|
|
|
|
|
|
|
Char => sub { |
|
790
|
|
|
|
|
|
|
my ( $expat, $content ) = @_; |
|
791
|
|
|
|
|
|
|
$content =~ s/_x000D_//smxg; |
|
792
|
|
|
|
|
|
|
if ( defined $shared_strings->{$current_index} ) { |
|
793
|
|
|
|
|
|
|
$shared_strings->{$current_index} .= $content; |
|
794
|
|
|
|
|
|
|
} |
|
795
|
|
|
|
|
|
|
else { |
|
796
|
|
|
|
|
|
|
$shared_strings->{$current_index} = $content; |
|
797
|
|
|
|
|
|
|
} |
|
798
|
|
|
|
|
|
|
} |
|
799
|
|
|
|
|
|
|
} |
|
800
|
|
|
|
|
|
|
); |
|
801
|
|
|
|
|
|
|
eval { $xml->parse($shared_string_content); } or do { |
|
802
|
|
|
|
|
|
|
chomp $EVAL_ERROR; |
|
803
|
|
|
|
|
|
|
$EVAL_ERROR =~ s/^\s*//smx; |
|
804
|
|
|
|
|
|
|
$self->{_ERROR_DIAG} = |
|
805
|
|
|
|
|
|
|
"XML - Invalid XML in sharedStrings.xml:$EVAL_ERROR"; |
|
806
|
|
|
|
|
|
|
return; |
|
807
|
|
|
|
|
|
|
}; |
|
808
|
|
|
|
|
|
|
return $shared_strings; |
|
809
|
|
|
|
|
|
|
} |
|
810
|
|
|
|
|
|
|
|
|
811
|
|
|
|
|
|
|
sub _xlsx_worksheet_path { |
|
812
|
|
|
|
|
|
|
my ( $self, $zip ) = @_; |
|
813
|
|
|
|
|
|
|
my $member = $zip->memberNamed('xl/workbook.xml'); |
|
814
|
|
|
|
|
|
|
if ( !$member ) { |
|
815
|
|
|
|
|
|
|
$self->{_ERROR_DIAG} = q[ZIP - Missing 'xl/workbook.xml' file]; |
|
816
|
|
|
|
|
|
|
return; |
|
817
|
|
|
|
|
|
|
} |
|
818
|
|
|
|
|
|
|
my $content = $member->contents(); |
|
819
|
|
|
|
|
|
|
my $worksheet_number; |
|
820
|
|
|
|
|
|
|
my $sheets = []; |
|
821
|
|
|
|
|
|
|
my $worksheet_count; |
|
822
|
|
|
|
|
|
|
my $xml = XML::Parser->new( |
|
823
|
|
|
|
|
|
|
Handlers => { |
|
824
|
|
|
|
|
|
|
Entity => sub { |
|
825
|
|
|
|
|
|
|
die |
|
826
|
|
|
|
|
|
|
"XML Entities have been detected and rejected in the XML, due to security concerns\n"; |
|
827
|
|
|
|
|
|
|
}, |
|
828
|
|
|
|
|
|
|
Start => sub { |
|
829
|
|
|
|
|
|
|
my ( $expat, $element_name, %element_parameters ) = @_; |
|
830
|
|
|
|
|
|
|
if ( $element_name eq 'sheet' ) { |
|
831
|
|
|
|
|
|
|
push @{$sheets}, \%element_parameters; |
|
832
|
|
|
|
|
|
|
if ( !defined $worksheet_number ) { |
|
833
|
|
|
|
|
|
|
if ( defined $worksheet_count ) { |
|
834
|
|
|
|
|
|
|
$worksheet_count += 1; |
|
835
|
|
|
|
|
|
|
} |
|
836
|
|
|
|
|
|
|
else { |
|
837
|
|
|
|
|
|
|
$worksheet_count = 0; |
|
838
|
|
|
|
|
|
|
} |
|
839
|
|
|
|
|
|
|
if ( |
|
840
|
|
|
|
|
|
|
( defined $self->{worksheet_name} ) |
|
841
|
|
|
|
|
|
|
&& ( defined $element_parameters{name} ) |
|
842
|
|
|
|
|
|
|
&& ( $self->{worksheet_name} eq |
|
843
|
|
|
|
|
|
|
$element_parameters{name} ) |
|
844
|
|
|
|
|
|
|
) |
|
845
|
|
|
|
|
|
|
{ |
|
846
|
|
|
|
|
|
|
$worksheet_number = $worksheet_count + 1; |
|
847
|
|
|
|
|
|
|
} |
|
848
|
|
|
|
|
|
|
elsif ( |
|
849
|
|
|
|
|
|
|
( $self->{worksheet_number} ) |
|
850
|
|
|
|
|
|
|
&& ( ( $self->{worksheet_number} - 1 ) == |
|
851
|
|
|
|
|
|
|
$worksheet_count ) |
|
852
|
|
|
|
|
|
|
) |
|
853
|
|
|
|
|
|
|
{ |
|
854
|
|
|
|
|
|
|
$worksheet_number = $worksheet_count + 1; |
|
855
|
|
|
|
|
|
|
} |
|
856
|
|
|
|
|
|
|
} |
|
857
|
|
|
|
|
|
|
} |
|
858
|
|
|
|
|
|
|
} |
|
859
|
|
|
|
|
|
|
} |
|
860
|
|
|
|
|
|
|
); |
|
861
|
|
|
|
|
|
|
eval { $xml->parse($content); } or do { |
|
862
|
|
|
|
|
|
|
chomp $EVAL_ERROR; |
|
863
|
|
|
|
|
|
|
$EVAL_ERROR =~ s/^\s*//smx; |
|
864
|
|
|
|
|
|
|
$self->{_ERROR_DIAG} = |
|
865
|
|
|
|
|
|
|
"XML - Invalid XML in xl/workbook.xml:$EVAL_ERROR"; |
|
866
|
|
|
|
|
|
|
return; |
|
867
|
|
|
|
|
|
|
}; |
|
868
|
|
|
|
|
|
|
if ( !defined $worksheet_number ) { |
|
869
|
|
|
|
|
|
|
$self->{_ERROR_DIAG} = 'ENOENT - Worksheet ' |
|
870
|
|
|
|
|
|
|
. ( |
|
871
|
|
|
|
|
|
|
defined $self->{worksheet_name} |
|
872
|
|
|
|
|
|
|
? $self->{worksheet_name} |
|
873
|
|
|
|
|
|
|
: $self->{worksheet_number} |
|
874
|
|
|
|
|
|
|
) . ' not found'; |
|
875
|
|
|
|
|
|
|
return; |
|
876
|
|
|
|
|
|
|
} |
|
877
|
|
|
|
|
|
|
return 'xl/worksheets/sheet' . $worksheet_number . '.xml'; |
|
878
|
|
|
|
|
|
|
} |
|
879
|
|
|
|
|
|
|
|
|
880
|
|
|
|
|
|
|
sub _xlsx_cells { |
|
881
|
|
|
|
|
|
|
my ( $self, $xlsx_content, $shared_strings, $worksheet_path ) = @_; |
|
882
|
|
|
|
|
|
|
my $cells = []; |
|
883
|
|
|
|
|
|
|
my $current_worksheet_name; |
|
884
|
|
|
|
|
|
|
my $current_sheet_cells = []; |
|
885
|
|
|
|
|
|
|
my $parameter_stack; |
|
886
|
|
|
|
|
|
|
my $element_stack; |
|
887
|
|
|
|
|
|
|
my $process_worksheet = 0; |
|
888
|
|
|
|
|
|
|
my $xml = XML::Parser->new( |
|
889
|
|
|
|
|
|
|
Handlers => { |
|
890
|
|
|
|
|
|
|
Entity => sub { |
|
891
|
|
|
|
|
|
|
die |
|
892
|
|
|
|
|
|
|
"XML Entities have been detected and rejected in the XML, due to security concerns\n"; |
|
893
|
|
|
|
|
|
|
}, |
|
894
|
|
|
|
|
|
|
Start => sub { |
|
895
|
|
|
|
|
|
|
my ( $expat, $element_name, %element_parameters ) = @_; |
|
896
|
|
|
|
|
|
|
push @{$element_stack}, $element_name; |
|
897
|
|
|
|
|
|
|
push @{$parameter_stack}, \%element_parameters; |
|
898
|
|
|
|
|
|
|
}, |
|
899
|
|
|
|
|
|
|
End => sub { |
|
900
|
|
|
|
|
|
|
my ( $expat, $element_name ) = @_; |
|
901
|
|
|
|
|
|
|
pop @{$parameter_stack}; |
|
902
|
|
|
|
|
|
|
if ( $element_name ne pop @{$element_stack} ) { |
|
903
|
|
|
|
|
|
|
Carp::croak( |
|
904
|
|
|
|
|
|
|
'Internal confusion in processing XML elements'); |
|
905
|
|
|
|
|
|
|
} |
|
906
|
|
|
|
|
|
|
}, |
|
907
|
|
|
|
|
|
|
Char => sub { |
|
908
|
|
|
|
|
|
|
my ( $expat, $content ) = @_; |
|
909
|
|
|
|
|
|
|
if ( ( $element_stack->[ _GRANDPARENT_INDEX() ] eq 'c' ) |
|
910
|
|
|
|
|
|
|
&& ( $element_stack->[ _PARENT_INDEX() ] eq 'v' ) ) |
|
911
|
|
|
|
|
|
|
{ |
|
912
|
|
|
|
|
|
|
my $cell_reference = |
|
913
|
|
|
|
|
|
|
$parameter_stack->[ _GRANDPARENT_INDEX() ]->{r}; |
|
914
|
|
|
|
|
|
|
if ( $cell_reference =~ /^([[:upper:]]+)(\d+)$/smx ) { |
|
915
|
|
|
|
|
|
|
my ( $column_designator, $row_number ) = ( $1, $2 - 1 ); |
|
916
|
|
|
|
|
|
|
my $column_number = 0; |
|
917
|
|
|
|
|
|
|
my $letter_index = 0; |
|
918
|
|
|
|
|
|
|
foreach |
|
919
|
|
|
|
|
|
|
my $letter ( reverse split //smx, $column_designator ) |
|
920
|
|
|
|
|
|
|
{ |
|
921
|
|
|
|
|
|
|
$column_number += |
|
922
|
|
|
|
|
|
|
( ( ord uc $letter ) - ( ord 'A' ) ) * |
|
923
|
|
|
|
|
|
|
( _EXCEL_COLUMN_RADIX()**$letter_index ); |
|
924
|
|
|
|
|
|
|
$letter_index += 1; |
|
925
|
|
|
|
|
|
|
} |
|
926
|
|
|
|
|
|
|
if ( |
|
927
|
|
|
|
|
|
|
( |
|
928
|
|
|
|
|
|
|
defined |
|
929
|
|
|
|
|
|
|
$parameter_stack->[ _GRANDPARENT_INDEX() ]->{t} |
|
930
|
|
|
|
|
|
|
) |
|
931
|
|
|
|
|
|
|
&& ( $parameter_stack->[ _GRANDPARENT_INDEX() ]->{t} |
|
932
|
|
|
|
|
|
|
eq 's' ) |
|
933
|
|
|
|
|
|
|
) |
|
934
|
|
|
|
|
|
|
{ |
|
935
|
|
|
|
|
|
|
$cells->[$row_number]->[$column_number] = |
|
936
|
|
|
|
|
|
|
$shared_strings->{$content}; |
|
937
|
|
|
|
|
|
|
} |
|
938
|
|
|
|
|
|
|
else { |
|
939
|
|
|
|
|
|
|
$cells->[$row_number]->[$column_number] = $content; |
|
940
|
|
|
|
|
|
|
} |
|
941
|
|
|
|
|
|
|
} |
|
942
|
|
|
|
|
|
|
else { |
|
943
|
|
|
|
|
|
|
Carp::croak('Unable to determine cell reference'); |
|
944
|
|
|
|
|
|
|
} |
|
945
|
|
|
|
|
|
|
} |
|
946
|
|
|
|
|
|
|
} |
|
947
|
|
|
|
|
|
|
} |
|
948
|
|
|
|
|
|
|
); |
|
949
|
|
|
|
|
|
|
eval { $xml->parse($xlsx_content); } or do { |
|
950
|
|
|
|
|
|
|
chomp $EVAL_ERROR; |
|
951
|
|
|
|
|
|
|
$self->{_ERROR_DIAG} = |
|
952
|
|
|
|
|
|
|
"XML - Invalid XML in $worksheet_path:$EVAL_ERROR"; |
|
953
|
|
|
|
|
|
|
return; |
|
954
|
|
|
|
|
|
|
}; |
|
955
|
|
|
|
|
|
|
return $cells; |
|
956
|
|
|
|
|
|
|
} |
|
957
|
|
|
|
|
|
|
|
|
958
|
|
|
|
|
|
|
sub content_type { |
|
959
|
|
|
|
|
|
|
my ($self) = @_; |
|
960
|
|
|
|
|
|
|
return $self->{content_type}; |
|
961
|
|
|
|
|
|
|
} |
|
962
|
|
|
|
|
|
|
|
|
963
|
|
|
|
|
|
|
sub suffix { |
|
964
|
|
|
|
|
|
|
my ($self) = @_; |
|
965
|
|
|
|
|
|
|
return $self->{type}; |
|
966
|
|
|
|
|
|
|
} |
|
967
|
|
|
|
|
|
|
|
|
968
|
|
|
|
|
|
|
1; |
|
969
|
|
|
|
|
|
|
__END__ |