| line |
stmt |
bran |
cond |
sub |
pod |
time |
code |
|
1
|
|
|
|
|
|
|
# |
|
2
|
|
|
|
|
|
|
# $Id: Tape.pm,v 1.7 2005/09/01 08:19:27 patrick Exp $ |
|
3
|
|
|
|
|
|
|
# |
|
4
|
|
|
|
|
|
|
|
|
5
|
|
|
|
|
|
|
=head1 NAME |
|
6
|
|
|
|
|
|
|
|
|
7
|
|
|
|
|
|
|
XML::Tape - module for the manipulation of XMLtape archives |
|
8
|
|
|
|
|
|
|
|
|
9
|
|
|
|
|
|
|
=head1 SYNOPSIS |
|
10
|
|
|
|
|
|
|
|
|
11
|
|
|
|
|
|
|
use XML::Tape qw(:all); |
|
12
|
|
|
|
|
|
|
|
|
13
|
|
|
|
|
|
|
$tape = tapeopen('tape.xml','w'); |
|
14
|
|
|
|
|
|
|
$tape->add_record("info:archive_id/1", "2005-05-31", $xml_record); |
|
15
|
|
|
|
|
|
|
$tape->tapeclose(); |
|
16
|
|
|
|
|
|
|
|
|
17
|
|
|
|
|
|
|
$tape = tapeopen('tape.xml','r'); |
|
18
|
|
|
|
|
|
|
while ($record = $tape->get_record()) { |
|
19
|
|
|
|
|
|
|
printf "id: %s\n" , $record->getIdentifier; |
|
20
|
|
|
|
|
|
|
printf "date: %s\n" , $record->getDate; |
|
21
|
|
|
|
|
|
|
printf "xml: %s\n", $record->getRecord; |
|
22
|
|
|
|
|
|
|
} |
|
23
|
|
|
|
|
|
|
$tape->tapeclose(); |
|
24
|
|
|
|
|
|
|
|
|
25
|
|
|
|
|
|
|
=head1 DESCRIPTION |
|
26
|
|
|
|
|
|
|
|
|
27
|
|
|
|
|
|
|
The XMLtape provides a write-once/read-many XML wrapper for a |
|
28
|
|
|
|
|
|
|
collection of XML documents. The wrapper provides an easy |
|
29
|
|
|
|
|
|
|
storage format for big collections of XML files which can be |
|
30
|
|
|
|
|
|
|
processed with off the shelf tools and validated against a |
|
31
|
|
|
|
|
|
|
schema. The XMLtape is typically used in digital preservation |
|
32
|
|
|
|
|
|
|
projects. |
|
33
|
|
|
|
|
|
|
|
|
34
|
|
|
|
|
|
|
=cut |
|
35
|
|
|
|
|
|
|
package XML::Tape; |
|
36
|
1
|
|
|
1
|
|
79824
|
use strict; |
|
|
1
|
|
|
|
|
2
|
|
|
|
1
|
|
|
|
|
55
|
|
|
37
|
|
|
|
|
|
|
require Exporter; |
|
38
|
1
|
|
|
1
|
|
5
|
use vars qw($VERSION); |
|
|
1
|
|
|
|
|
1
|
|
|
|
1
|
|
|
|
|
684
|
|
|
39
|
|
|
|
|
|
|
|
|
40
|
|
|
|
|
|
|
( $VERSION ) = '$Revision: 1.7 $ ' =~ /\$Revision:\s+([^\s]+)/;; |
|
41
|
|
|
|
|
|
|
|
|
42
|
|
|
|
|
|
|
@XML::Tape::ISA = qw(Exporter); |
|
43
|
|
|
|
|
|
|
@XML::Tape::EXPORT_OK = qw(tapeopen); |
|
44
|
|
|
|
|
|
|
%XML::Tape::EXPORT_TAGS = (all => [qw(tapeopen)]); |
|
45
|
|
|
|
|
|
|
$XML::Tape::SCHEMA_LOCATION = 'http://purl.lanl.gov/aDORe/schemas/2005-08/XMLtape.xsd'; |
|
46
|
|
|
|
|
|
|
|
|
47
|
|
|
|
|
|
|
=head1 FUNCTIONS |
|
48
|
|
|
|
|
|
|
|
|
49
|
|
|
|
|
|
|
=over 4 |
|
50
|
|
|
|
|
|
|
|
|
51
|
|
|
|
|
|
|
=item tapeopen($filename, $mode, [, @admin]) |
|
52
|
|
|
|
|
|
|
|
|
53
|
|
|
|
|
|
|
Filename is the location of an XMLtape file or an opened |
|
54
|
|
|
|
|
|
|
IO::Handle. |
|
55
|
|
|
|
|
|
|
When mode is 'r' this function opens a XMLtape for reading. |
|
56
|
|
|
|
|
|
|
When mode is 'w' this function creates a new XMLtape on disk. |
|
57
|
|
|
|
|
|
|
Optionally an array of strings can be provided which contain in |
|
58
|
|
|
|
|
|
|
XML format metadata about the XMLtape. E.g. |
|
59
|
|
|
|
|
|
|
|
|
60
|
|
|
|
|
|
|
tapeopen( |
|
61
|
|
|
|
|
|
|
"tape.xml", |
|
62
|
|
|
|
|
|
|
"w" |
|
63
|
|
|
|
|
|
|
"2005-05-31" |
|
64
|
|
|
|
|
|
|
); |
|
65
|
|
|
|
|
|
|
|
|
66
|
|
|
|
|
|
|
Returns a XMLtape instance on success or undef on error. |
|
67
|
|
|
|
|
|
|
|
|
68
|
|
|
|
|
|
|
=cut |
|
69
|
|
|
|
|
|
|
sub tapeopen { |
|
70
|
2
|
|
|
2
|
|
2289
|
my ($filename, $mode, @admin) = @_; |
|
71
|
|
|
|
|
|
|
|
|
72
|
2
|
50
|
33
|
|
|
19
|
die "usage: tapeopen(\$filename, \$mode, [\@admin])" unless ($filename && $mode =~ /^r|w$/); |
|
73
|
|
|
|
|
|
|
|
|
74
|
2
|
100
|
|
|
|
7
|
if ($mode eq 'w') { |
|
75
|
1
|
|
|
|
|
9
|
return new XML::Tape::Writer($filename,@admin); |
|
76
|
|
|
|
|
|
|
} |
|
77
|
|
|
|
|
|
|
else { |
|
78
|
1
|
|
|
|
|
207
|
my $identifier = new XML::Tape::Identifier; |
|
79
|
0
|
|
|
|
|
0
|
my $namespace = $identifier->identify($filename); |
|
80
|
0
|
0
|
|
|
|
0
|
if ($namespace eq 'http://library.lanl.gov/2005-01/STB-RL/tape/') { |
|
|
|
0
|
|
|
|
|
|
|
81
|
0
|
|
|
|
|
0
|
return new XML::Tape::Reader::v2005_01($filename); |
|
82
|
|
|
|
|
|
|
} |
|
83
|
|
|
|
|
|
|
elsif ($namespace eq 'http://library.lanl.gov/2005-08/aDORe/XMLtape/') { |
|
84
|
0
|
|
|
|
|
0
|
return new XML::Tape::Reader($filename); |
|
85
|
|
|
|
|
|
|
} |
|
86
|
|
|
|
|
|
|
else { |
|
87
|
0
|
|
|
|
|
0
|
die "unknown tape version $namespace"; |
|
88
|
|
|
|
|
|
|
} |
|
89
|
|
|
|
|
|
|
} |
|
90
|
|
|
|
|
|
|
|
|
91
|
0
|
|
|
|
|
0
|
return undef; |
|
92
|
|
|
|
|
|
|
} |
|
93
|
|
|
|
|
|
|
|
|
94
|
|
|
|
|
|
|
package XML::Tape::Writer; |
|
95
|
1
|
|
|
1
|
|
1453
|
use IO::File; |
|
|
1
|
|
|
|
|
1816
|
|
|
|
1
|
|
|
|
|
1012
|
|
|
96
|
|
|
|
|
|
|
|
|
97
|
|
|
|
|
|
|
sub new { |
|
98
|
1
|
|
|
1
|
|
3
|
my ($pkg, $filename,@admin) = @_; |
|
99
|
1
|
|
|
|
|
2
|
my $fh; |
|
100
|
|
|
|
|
|
|
|
|
101
|
1
|
50
|
33
|
|
|
5
|
if (ref $filename && $filename->isa('Tie::Handle')) { |
|
102
|
0
|
|
|
|
|
0
|
$fh = $filename; |
|
103
|
|
|
|
|
|
|
} |
|
104
|
|
|
|
|
|
|
else { |
|
105
|
1
|
|
|
|
|
10
|
$fh = new IO::File; |
|
106
|
1
|
50
|
|
|
|
46
|
$fh->open("> $filename") || return undef; |
|
107
|
|
|
|
|
|
|
} |
|
108
|
|
|
|
|
|
|
|
|
109
|
1
|
|
|
|
|
115
|
my $obj = bless { |
|
110
|
|
|
|
|
|
|
fh => $fh , |
|
111
|
|
|
|
|
|
|
init => 0, |
|
112
|
|
|
|
|
|
|
recnum => 0 , |
|
113
|
|
|
|
|
|
|
} , $pkg; |
|
114
|
1
|
50
|
|
|
|
5
|
$obj->add_admin(@admin) if (@admin > 0); |
|
115
|
1
|
|
|
|
|
6
|
return $obj; |
|
116
|
|
|
|
|
|
|
} |
|
117
|
|
|
|
|
|
|
|
|
118
|
|
|
|
|
|
|
sub init { |
|
119
|
1
|
|
|
1
|
|
2
|
my ($this) = shift; |
|
120
|
1
|
|
|
|
|
2
|
my $fh = $this->{fh}; |
|
121
|
1
|
50
|
|
|
|
4
|
die "init: not allowed at this stage" unless $this->{init} == 0; |
|
122
|
1
|
|
|
|
|
14
|
print $fh "\n"; |
|
123
|
1
|
|
|
|
|
4
|
print $fh ""; |
|
124
|
1
|
|
|
|
|
3
|
$this->{init}++; |
|
125
|
|
|
|
|
|
|
} |
|
126
|
|
|
|
|
|
|
|
|
127
|
|
|
|
|
|
|
sub add_admin { |
|
128
|
0
|
|
|
0
|
|
0
|
my ($this,@admin) = @_; |
|
129
|
0
|
|
|
|
|
0
|
my $fh = $this->{fh}; |
|
130
|
0
|
0
|
|
|
|
0
|
$this->init() unless ($this->{init}); |
|
131
|
0
|
0
|
|
|
|
0
|
die "add_admin: not allowed at this stage" unless $this->{recnum} == 0; |
|
132
|
0
|
|
|
|
|
0
|
foreach (@admin) { |
|
133
|
0
|
|
|
|
|
0
|
printf $fh "%s", $_; |
|
134
|
|
|
|
|
|
|
} |
|
135
|
|
|
|
|
|
|
} |
|
136
|
|
|
|
|
|
|
|
|
137
|
|
|
|
|
|
|
=item $tape->add_record($identifier, $date, $record [, @admin]) |
|
138
|
|
|
|
|
|
|
|
|
139
|
|
|
|
|
|
|
Add a XML document to the XMLtape with identifier $identifier, date stamp |
|
140
|
|
|
|
|
|
|
$date and XML string representation $record. Optionally |
|
141
|
|
|
|
|
|
|
an array of strings can be provided which contain in |
|
142
|
|
|
|
|
|
|
XML format metadata about the record. |
|
143
|
|
|
|
|
|
|
|
|
144
|
|
|
|
|
|
|
Returns true on success undef on error. |
|
145
|
|
|
|
|
|
|
|
|
146
|
|
|
|
|
|
|
=cut |
|
147
|
|
|
|
|
|
|
sub add_record { |
|
148
|
6
|
|
|
6
|
|
25
|
my ($this, $identifier, $date, $record, @admin) = @_; |
|
149
|
6
|
|
|
|
|
7
|
my $fh = $this->{fh}; |
|
150
|
6
|
100
|
|
|
|
16
|
$this->init() unless ($this->{init}); |
|
151
|
|
|
|
|
|
|
|
|
152
|
6
|
|
|
|
|
8
|
print $fh ""; |
|
153
|
6
|
|
|
|
|
7
|
print $fh ""; |
|
154
|
6
|
|
|
|
|
10
|
print $fh "" , &escape($identifier) , ""; |
|
155
|
6
|
|
|
|
|
12
|
print $fh "" , &escape($date) , ""; |
|
156
|
6
|
|
|
|
|
13
|
foreach my $admin (@admin) { |
|
157
|
0
|
|
|
|
|
0
|
print $fh "" , $admin , ""; |
|
158
|
|
|
|
|
|
|
} |
|
159
|
6
|
|
|
|
|
9
|
print $fh ""; |
|
160
|
6
|
|
|
|
|
8
|
print $fh "" , $record , ""; |
|
161
|
6
|
|
|
|
|
8
|
print $fh ""; |
|
162
|
|
|
|
|
|
|
|
|
163
|
6
|
|
|
|
|
10
|
$this->{recnum}++; |
|
164
|
|
|
|
|
|
|
|
|
165
|
6
|
|
|
|
|
11
|
return 1; |
|
166
|
|
|
|
|
|
|
} |
|
167
|
|
|
|
|
|
|
|
|
168
|
|
|
|
|
|
|
=item $tape->tapeclose |
|
169
|
|
|
|
|
|
|
|
|
170
|
|
|
|
|
|
|
Closes the XMLtape. |
|
171
|
|
|
|
|
|
|
|
|
172
|
|
|
|
|
|
|
Returns true on success undef on error. |
|
173
|
|
|
|
|
|
|
|
|
174
|
|
|
|
|
|
|
=cut |
|
175
|
|
|
|
|
|
|
sub tapeclose { |
|
176
|
1
|
|
|
1
|
|
257
|
my ($this) = shift; |
|
177
|
1
|
|
|
|
|
2
|
my $fh = $this->{fh}; |
|
178
|
1
|
50
|
|
|
|
6
|
$this->init() unless ($this->{init}); |
|
179
|
1
|
|
|
|
|
2
|
print $fh ""; |
|
180
|
1
|
|
|
|
|
8
|
$fh->close; |
|
181
|
|
|
|
|
|
|
} |
|
182
|
|
|
|
|
|
|
|
|
183
|
|
|
|
|
|
|
sub escape { |
|
184
|
12
|
|
|
12
|
|
12
|
my $str = shift; |
|
185
|
12
|
|
|
|
|
14
|
$str =~ s/&/&/g; |
|
186
|
12
|
|
|
|
|
13
|
$str =~ s/</g; |
|
187
|
12
|
|
|
|
|
13
|
$str =~ s/>/>/g; |
|
188
|
12
|
|
|
|
|
17
|
$str =~ s/'/'/g; |
|
189
|
12
|
|
|
|
|
12
|
$str =~ s/"/"/g; |
|
190
|
12
|
|
|
|
|
19
|
return $str; |
|
191
|
|
|
|
|
|
|
} |
|
192
|
|
|
|
|
|
|
|
|
193
|
|
|
|
|
|
|
package XML::Tape::Identifier; |
|
194
|
1
|
|
|
1
|
|
662
|
use XML::Parser; |
|
|
0
|
|
|
|
|
0
|
|
|
|
0
|
|
|
|
|
0
|
|
|
195
|
|
|
|
|
|
|
|
|
196
|
|
|
|
|
|
|
sub new { |
|
197
|
|
|
|
|
|
|
my $pkg = shift; |
|
198
|
|
|
|
|
|
|
return bless {} , $pkg; |
|
199
|
|
|
|
|
|
|
} |
|
200
|
|
|
|
|
|
|
|
|
201
|
|
|
|
|
|
|
sub identify { |
|
202
|
|
|
|
|
|
|
my ($this,$filename) = @_; |
|
203
|
|
|
|
|
|
|
my $parser = new XML::Parser(Namespaces => 1, |
|
204
|
|
|
|
|
|
|
Handlers => { |
|
205
|
|
|
|
|
|
|
Start => sub { $this->handle_start(@_); }, |
|
206
|
|
|
|
|
|
|
}); |
|
207
|
|
|
|
|
|
|
eval { |
|
208
|
|
|
|
|
|
|
$parser->parsefile($filename); |
|
209
|
|
|
|
|
|
|
}; |
|
210
|
|
|
|
|
|
|
return $this->{namespace}; |
|
211
|
|
|
|
|
|
|
} |
|
212
|
|
|
|
|
|
|
|
|
213
|
|
|
|
|
|
|
sub handle_start { |
|
214
|
|
|
|
|
|
|
my ($this, $xp, $elem, %attr) = @_; |
|
215
|
|
|
|
|
|
|
$this->{namespace} = $xp->namespace($elem); |
|
216
|
|
|
|
|
|
|
die "ok"; |
|
217
|
|
|
|
|
|
|
} |
|
218
|
|
|
|
|
|
|
|
|
219
|
|
|
|
|
|
|
package XML::Tape::Reader; |
|
220
|
|
|
|
|
|
|
# NS version http://library.lanl.gov/2005-08/aDORe/XMLtape/ |
|
221
|
|
|
|
|
|
|
use XML::Parser; |
|
222
|
|
|
|
|
|
|
use IO::File; |
|
223
|
|
|
|
|
|
|
|
|
224
|
|
|
|
|
|
|
$XML::Tape::Reader::BUFF_SIZE = 1024; |
|
225
|
|
|
|
|
|
|
|
|
226
|
|
|
|
|
|
|
sub new { |
|
227
|
|
|
|
|
|
|
my ($pkg, $filename,%options) = @_; |
|
228
|
|
|
|
|
|
|
my $obj = bless {} , $pkg; |
|
229
|
|
|
|
|
|
|
my $fh; |
|
230
|
|
|
|
|
|
|
|
|
231
|
|
|
|
|
|
|
if (ref $filename && $filename->isa('Tie::Handle')) { |
|
232
|
|
|
|
|
|
|
$fh = $filename; |
|
233
|
|
|
|
|
|
|
} |
|
234
|
|
|
|
|
|
|
else { |
|
235
|
|
|
|
|
|
|
$fh = new IO::File; |
|
236
|
|
|
|
|
|
|
$fh->open("< $filename") || return undef; |
|
237
|
|
|
|
|
|
|
} |
|
238
|
|
|
|
|
|
|
|
|
239
|
|
|
|
|
|
|
$obj->{fh} = $fh; # XML file handle |
|
240
|
|
|
|
|
|
|
$obj->{records} = []; # Temporary storage for XML::Tape::Record |
|
241
|
|
|
|
|
|
|
$obj->{admins} = []; # Temporary storage for XML::Tape::Admin |
|
242
|
|
|
|
|
|
|
$obj->{curr} = undef; # Current record to be read |
|
243
|
|
|
|
|
|
|
$obj->{parse_init} = 0; # Flag to indicate if we started reading XML |
|
244
|
|
|
|
|
|
|
$obj->{parse_done} = 0; # Flag to indicate if we still reading XML |
|
245
|
|
|
|
|
|
|
$obj->{parser} = undef; # XML::Parser |
|
246
|
|
|
|
|
|
|
$obj->{parsernb} = undef; # XML::Parser::ExpatNB |
|
247
|
|
|
|
|
|
|
$obj->{nav} = {}; # Hash to navigate in the XML record |
|
248
|
|
|
|
|
|
|
|
|
249
|
|
|
|
|
|
|
return $obj; |
|
250
|
|
|
|
|
|
|
} |
|
251
|
|
|
|
|
|
|
|
|
252
|
|
|
|
|
|
|
=item $tape->get_admin() |
|
253
|
|
|
|
|
|
|
|
|
254
|
|
|
|
|
|
|
Reads one XMLtape admin section. Returns an instance of XML::Tape::Admin on success |
|
255
|
|
|
|
|
|
|
or undef when no more XMLtape admin sections are available. |
|
256
|
|
|
|
|
|
|
|
|
257
|
|
|
|
|
|
|
=cut |
|
258
|
|
|
|
|
|
|
sub get_admin { |
|
259
|
|
|
|
|
|
|
my ($this) = shift; |
|
260
|
|
|
|
|
|
|
|
|
261
|
|
|
|
|
|
|
$this->parse() until ( ( scalar @{$this->{records}} ) || ( $this->{parse_done} ) ); |
|
262
|
|
|
|
|
|
|
|
|
263
|
|
|
|
|
|
|
return shift( @{$this->{admins}} ); |
|
264
|
|
|
|
|
|
|
} |
|
265
|
|
|
|
|
|
|
|
|
266
|
|
|
|
|
|
|
=item $tape->get_record() |
|
267
|
|
|
|
|
|
|
|
|
268
|
|
|
|
|
|
|
Reads one XMLtape record section. Returns an instance of XML::Tape::Record on success |
|
269
|
|
|
|
|
|
|
or undef when no more records are available. |
|
270
|
|
|
|
|
|
|
|
|
271
|
|
|
|
|
|
|
=cut |
|
272
|
|
|
|
|
|
|
sub get_record { |
|
273
|
|
|
|
|
|
|
my ($this) = shift; |
|
274
|
|
|
|
|
|
|
|
|
275
|
|
|
|
|
|
|
# Parse the XML until we read a new record or the parse is done... |
|
276
|
|
|
|
|
|
|
$this->parse() until ( ( scalar @{$this->{records}} ) || ( $this->{parse_done} ) ); |
|
277
|
|
|
|
|
|
|
|
|
278
|
|
|
|
|
|
|
return shift( @{$this->{records}} ); |
|
279
|
|
|
|
|
|
|
} |
|
280
|
|
|
|
|
|
|
|
|
281
|
|
|
|
|
|
|
sub tapeclose { |
|
282
|
|
|
|
|
|
|
my ($this) = shift; |
|
283
|
|
|
|
|
|
|
$this->{fh}->close; |
|
284
|
|
|
|
|
|
|
} |
|
285
|
|
|
|
|
|
|
|
|
286
|
|
|
|
|
|
|
sub parse_init { |
|
287
|
|
|
|
|
|
|
my ($this) = shift; |
|
288
|
|
|
|
|
|
|
|
|
289
|
|
|
|
|
|
|
$this->{parser} = new XML::Parser( Handlers => { |
|
290
|
|
|
|
|
|
|
Start => sub { $this->handle_start(@_); }, |
|
291
|
|
|
|
|
|
|
Char => sub { $this->handle_char(@_); }, |
|
292
|
|
|
|
|
|
|
Comment => sub { $this->handle_comment(@_); }, |
|
293
|
|
|
|
|
|
|
Proc => sub { $this->handle_proc(@_); }, |
|
294
|
|
|
|
|
|
|
CdataStart => sub { $this->handle_cdata_start(@_); }, |
|
295
|
|
|
|
|
|
|
CdataEnd => sub { $this->handle_cdata_end(@_); }, |
|
296
|
|
|
|
|
|
|
End => sub { $this->handle_end(@_); }, |
|
297
|
|
|
|
|
|
|
Final => sub { $this->handle_final(@_); }, |
|
298
|
|
|
|
|
|
|
}); |
|
299
|
|
|
|
|
|
|
|
|
300
|
|
|
|
|
|
|
$this->{parsernb} = $this->{parser}->parse_start(); |
|
301
|
|
|
|
|
|
|
|
|
302
|
|
|
|
|
|
|
return undef unless $this->{parsernb}; |
|
303
|
|
|
|
|
|
|
|
|
304
|
|
|
|
|
|
|
$this->{parse_init} = 1; |
|
305
|
|
|
|
|
|
|
|
|
306
|
|
|
|
|
|
|
return 1; |
|
307
|
|
|
|
|
|
|
} |
|
308
|
|
|
|
|
|
|
|
|
309
|
|
|
|
|
|
|
sub parse { |
|
310
|
|
|
|
|
|
|
my ($this) = shift; |
|
311
|
|
|
|
|
|
|
|
|
312
|
|
|
|
|
|
|
unless ($this->{parse_init}) { |
|
313
|
|
|
|
|
|
|
$this->parse_init() || return undef; |
|
314
|
|
|
|
|
|
|
} |
|
315
|
|
|
|
|
|
|
|
|
316
|
|
|
|
|
|
|
if (defined $this->{fh}) { |
|
317
|
|
|
|
|
|
|
my $buffer; |
|
318
|
|
|
|
|
|
|
|
|
319
|
|
|
|
|
|
|
# Read a chunk of XML... |
|
320
|
|
|
|
|
|
|
read($this->{fh}, $buffer, $XML::Tape::Reader::BUFF_SIZE); |
|
321
|
|
|
|
|
|
|
|
|
322
|
|
|
|
|
|
|
# If the buffer isn't empty then, parse it |
|
323
|
|
|
|
|
|
|
# otherwise we reached the end of the file... |
|
324
|
|
|
|
|
|
|
if (length $buffer) { |
|
325
|
|
|
|
|
|
|
$this->{parsernb}->parse_more($buffer); |
|
326
|
|
|
|
|
|
|
} |
|
327
|
|
|
|
|
|
|
else { |
|
328
|
|
|
|
|
|
|
$this->{parsernb}->parse_done(); |
|
329
|
|
|
|
|
|
|
$this->{parse_done} = 1; |
|
330
|
|
|
|
|
|
|
} |
|
331
|
|
|
|
|
|
|
} |
|
332
|
|
|
|
|
|
|
} |
|
333
|
|
|
|
|
|
|
|
|
334
|
|
|
|
|
|
|
sub handle_start { |
|
335
|
|
|
|
|
|
|
my ($this, $xp, $elem, %attr) = @_; |
|
336
|
|
|
|
|
|
|
|
|
337
|
|
|
|
|
|
|
if (0) {} |
|
338
|
|
|
|
|
|
|
elsif ($this->{nav}->{in_record}) { |
|
339
|
|
|
|
|
|
|
$this->{curr}->addRecordXML($xp->original_string); |
|
340
|
|
|
|
|
|
|
} |
|
341
|
|
|
|
|
|
|
elsif ($this->{nav}->{in_record_admin}) { |
|
342
|
|
|
|
|
|
|
$this->{curr}->addAdminXML($xp->original_string); |
|
343
|
|
|
|
|
|
|
} |
|
344
|
|
|
|
|
|
|
elsif ($this->{nav}->{in_tape_admin}) { |
|
345
|
|
|
|
|
|
|
$this->{curr}->addAdminXML($xp->original_string); |
|
346
|
|
|
|
|
|
|
} |
|
347
|
|
|
|
|
|
|
|
|
348
|
|
|
|
|
|
|
if (0) {} |
|
349
|
|
|
|
|
|
|
elsif ($xp->depth == 1 && $elem =~ /^(\w+:)?tapeAdmin$/) { |
|
350
|
|
|
|
|
|
|
$this->{nav}->{in_tape_admin} = 1; |
|
351
|
|
|
|
|
|
|
$this->{curr} = XML::Tape::Admin->new(); |
|
352
|
|
|
|
|
|
|
} |
|
353
|
|
|
|
|
|
|
elsif ($xp->depth == 1 && $elem =~ /^(\w+:)?tapeRecord$/) { |
|
354
|
|
|
|
|
|
|
$this->{curr} = XML::Tape::Record->new(); |
|
355
|
|
|
|
|
|
|
} |
|
356
|
|
|
|
|
|
|
elsif ($xp->depth == 2 && $elem =~ /^(\w+:)?tapeRecordAdmin$/) { |
|
357
|
|
|
|
|
|
|
$this->{nav}->{in_tape_record_admin} = 1; |
|
358
|
|
|
|
|
|
|
} |
|
359
|
|
|
|
|
|
|
elsif ($this->{nav}->{in_tape_record_admin} == 1 && $elem =~ /^(\w+:)?identifier$/) { |
|
360
|
|
|
|
|
|
|
$this->{nav}->{in_record_identifier} = 1; |
|
361
|
|
|
|
|
|
|
} |
|
362
|
|
|
|
|
|
|
elsif ($this->{nav}->{in_tape_record_admin} == 1 && $elem =~ /^(\w+:)?date$/) { |
|
363
|
|
|
|
|
|
|
$this->{nav}->{in_record_date} = 1; |
|
364
|
|
|
|
|
|
|
} |
|
365
|
|
|
|
|
|
|
elsif ($this->{nav}->{in_tape_record_admin} == 1 && $elem =~ /^(\w+:)?recordAdmin$/) { |
|
366
|
|
|
|
|
|
|
$this->{nav}->{in_record_admin} = 1; |
|
367
|
|
|
|
|
|
|
$this->{curr}->pushAdmin(); |
|
368
|
|
|
|
|
|
|
} |
|
369
|
|
|
|
|
|
|
elsif ($xp->depth == 2 && $elem =~ /^(\w+:)?record$/) { |
|
370
|
|
|
|
|
|
|
$this->{nav}->{in_record} = 1; |
|
371
|
|
|
|
|
|
|
$this->{curr}->setStartByte($xp->current_byte + length $xp->original_string); |
|
372
|
|
|
|
|
|
|
} |
|
373
|
|
|
|
|
|
|
} |
|
374
|
|
|
|
|
|
|
|
|
375
|
|
|
|
|
|
|
sub handle_end { |
|
376
|
|
|
|
|
|
|
my ($this, $xp, $elem, %attr) = @_; |
|
377
|
|
|
|
|
|
|
|
|
378
|
|
|
|
|
|
|
if (0) {} |
|
379
|
|
|
|
|
|
|
elsif ($xp->depth == 1 && $elem =~ /^(\w+:)?tapeAdmin$/) { |
|
380
|
|
|
|
|
|
|
$this->{nav}->{in_tape_admin} = 0; |
|
381
|
|
|
|
|
|
|
push(@{$this->{admins}}, $this->{curr}); |
|
382
|
|
|
|
|
|
|
} |
|
383
|
|
|
|
|
|
|
elsif ($xp->depth == 1 && $elem =~ /^(\w+:)?tapeRecord$/) { |
|
384
|
|
|
|
|
|
|
push(@{$this->{records}}, $this->{curr}); |
|
385
|
|
|
|
|
|
|
} |
|
386
|
|
|
|
|
|
|
elsif ($xp->depth == 2 && $elem =~ /^(\w+:)?tapeRecordAdmin$/) { |
|
387
|
|
|
|
|
|
|
$this->{nav}->{in_tape_record_admin} = 0; |
|
388
|
|
|
|
|
|
|
} |
|
389
|
|
|
|
|
|
|
elsif ($this->{nav}->{in_tape_record_admin} == 1 && $elem =~ /^(\w+:)?identifier$/) { |
|
390
|
|
|
|
|
|
|
$this->{nav}->{in_record_identifier} = 0; |
|
391
|
|
|
|
|
|
|
} |
|
392
|
|
|
|
|
|
|
elsif ($this->{nav}->{in_tape_record_admin} == 1 && $elem =~ /^(\w+:)?date$/) { |
|
393
|
|
|
|
|
|
|
$this->{nav}->{in_record_date} = 0; |
|
394
|
|
|
|
|
|
|
} |
|
395
|
|
|
|
|
|
|
elsif ($this->{nav}->{in_tape_record_admin} == 1 && $elem =~ /^(\w+:)?recordAdmin$/) { |
|
396
|
|
|
|
|
|
|
$this->{nav}->{in_record_admin} = 0; |
|
397
|
|
|
|
|
|
|
} |
|
398
|
|
|
|
|
|
|
elsif ($xp->depth == 2 && $elem =~ /^(\w+:)?record$/) { |
|
399
|
|
|
|
|
|
|
$this->{nav}->{in_record} = 0; |
|
400
|
|
|
|
|
|
|
$this->{curr}->setEndByte($xp->current_byte); |
|
401
|
|
|
|
|
|
|
} |
|
402
|
|
|
|
|
|
|
|
|
403
|
|
|
|
|
|
|
if (0) {} |
|
404
|
|
|
|
|
|
|
elsif ($this->{nav}->{in_record}) { |
|
405
|
|
|
|
|
|
|
$this->{curr}->addRecordXML($xp->original_string); |
|
406
|
|
|
|
|
|
|
} |
|
407
|
|
|
|
|
|
|
elsif ($this->{nav}->{in_record_admin}) { |
|
408
|
|
|
|
|
|
|
$this->{curr}->addAdminXML($xp->original_string); |
|
409
|
|
|
|
|
|
|
} |
|
410
|
|
|
|
|
|
|
elsif ($this->{nav}->{in_tape_admin}) { |
|
411
|
|
|
|
|
|
|
$this->{curr}->addAdminXML($xp->original_string); |
|
412
|
|
|
|
|
|
|
} |
|
413
|
|
|
|
|
|
|
} |
|
414
|
|
|
|
|
|
|
|
|
415
|
|
|
|
|
|
|
sub handle_char { |
|
416
|
|
|
|
|
|
|
my ($this, $xp, $data) = @_; |
|
417
|
|
|
|
|
|
|
|
|
418
|
|
|
|
|
|
|
if (0) {} |
|
419
|
|
|
|
|
|
|
elsif ($this->{nav}->{in_tape_admin}) { |
|
420
|
|
|
|
|
|
|
$this->{curr}->addAdminXML($xp->original_string); |
|
421
|
|
|
|
|
|
|
} |
|
422
|
|
|
|
|
|
|
elsif ($this->{nav}->{in_record}) { |
|
423
|
|
|
|
|
|
|
$this->{curr}->addRecordXML($xp->original_string); |
|
424
|
|
|
|
|
|
|
} |
|
425
|
|
|
|
|
|
|
elsif ($this->{nav}->{in_record_identifier}) { |
|
426
|
|
|
|
|
|
|
$this->{curr}->addIdentifier($data); |
|
427
|
|
|
|
|
|
|
} |
|
428
|
|
|
|
|
|
|
elsif ($this->{nav}->{in_record_date}) { |
|
429
|
|
|
|
|
|
|
$this->{curr}->addDate($data); |
|
430
|
|
|
|
|
|
|
} |
|
431
|
|
|
|
|
|
|
} |
|
432
|
|
|
|
|
|
|
|
|
433
|
|
|
|
|
|
|
sub handle_comment { |
|
434
|
|
|
|
|
|
|
my ($this, $xp, $data) = @_; |
|
435
|
|
|
|
|
|
|
|
|
436
|
|
|
|
|
|
|
if (0) {} |
|
437
|
|
|
|
|
|
|
elsif ($this->{nav}->{in_tape_admin}) { |
|
438
|
|
|
|
|
|
|
$this->{curr}->addAdminXML($xp->original_string); |
|
439
|
|
|
|
|
|
|
} |
|
440
|
|
|
|
|
|
|
elsif ($this->{nav}->{in_record}) { |
|
441
|
|
|
|
|
|
|
$this->{curr}->addRecordXML($xp->original_string); |
|
442
|
|
|
|
|
|
|
} |
|
443
|
|
|
|
|
|
|
} |
|
444
|
|
|
|
|
|
|
|
|
445
|
|
|
|
|
|
|
sub handle_proc { |
|
446
|
|
|
|
|
|
|
my ($this, $xp) = @_; |
|
447
|
|
|
|
|
|
|
|
|
448
|
|
|
|
|
|
|
if (0) {} |
|
449
|
|
|
|
|
|
|
elsif ($this->{nav}->{in_tape_admin}) { |
|
450
|
|
|
|
|
|
|
$this->{curr}->addAdminXML($xp->original_string); |
|
451
|
|
|
|
|
|
|
} |
|
452
|
|
|
|
|
|
|
elsif ($this->{nav}->{in_record}) { |
|
453
|
|
|
|
|
|
|
$this->{curr}->addRecordXML($xp->original_string); |
|
454
|
|
|
|
|
|
|
} |
|
455
|
|
|
|
|
|
|
} |
|
456
|
|
|
|
|
|
|
|
|
457
|
|
|
|
|
|
|
sub handle_cdata_start { |
|
458
|
|
|
|
|
|
|
my ($this, $xp) = @_; |
|
459
|
|
|
|
|
|
|
|
|
460
|
|
|
|
|
|
|
if (0) {} |
|
461
|
|
|
|
|
|
|
elsif ($this->{nav}->{in_tape_admin}) { |
|
462
|
|
|
|
|
|
|
$this->{curr}->addAdminXML($xp->original_string); |
|
463
|
|
|
|
|
|
|
} |
|
464
|
|
|
|
|
|
|
elsif ($this->{nav}->{in_record}) { |
|
465
|
|
|
|
|
|
|
$this->{curr}->addRecordXML($xp->original_string); |
|
466
|
|
|
|
|
|
|
} |
|
467
|
|
|
|
|
|
|
} |
|
468
|
|
|
|
|
|
|
|
|
469
|
|
|
|
|
|
|
sub handle_cdata_end { |
|
470
|
|
|
|
|
|
|
my ($this, $xp) = @_; |
|
471
|
|
|
|
|
|
|
|
|
472
|
|
|
|
|
|
|
if (0) {} |
|
473
|
|
|
|
|
|
|
elsif ($this->{nav}->{in_tape_admin}) { |
|
474
|
|
|
|
|
|
|
$this->{curr}->addAdminXML($xp->original_string); |
|
475
|
|
|
|
|
|
|
} |
|
476
|
|
|
|
|
|
|
elsif ($this->{nav}->{in_record}) { |
|
477
|
|
|
|
|
|
|
$this->{curr}->addRecordXML($xp->original_string); |
|
478
|
|
|
|
|
|
|
} |
|
479
|
|
|
|
|
|
|
} |
|
480
|
|
|
|
|
|
|
|
|
481
|
|
|
|
|
|
|
sub handle_final { |
|
482
|
|
|
|
|
|
|
return 1; |
|
483
|
|
|
|
|
|
|
} |
|
484
|
|
|
|
|
|
|
|
|
485
|
|
|
|
|
|
|
package XML::Tape::Reader::v2005_01; |
|
486
|
|
|
|
|
|
|
# Old tape reader keeping for backwards compatibility reasons; |
|
487
|
|
|
|
|
|
|
# NS version http://library.lanl.gov/2005-01/STB-RL/tape/ |
|
488
|
|
|
|
|
|
|
use XML::Parser; |
|
489
|
|
|
|
|
|
|
use IO::File; |
|
490
|
|
|
|
|
|
|
|
|
491
|
|
|
|
|
|
|
$XML::Tape::Reader::BUFF_SIZE = 1024; |
|
492
|
|
|
|
|
|
|
|
|
493
|
|
|
|
|
|
|
sub new { |
|
494
|
|
|
|
|
|
|
my ($pkg, $filename,%options) = @_; |
|
495
|
|
|
|
|
|
|
my $obj = bless {} , $pkg; |
|
496
|
|
|
|
|
|
|
my $fh; |
|
497
|
|
|
|
|
|
|
|
|
498
|
|
|
|
|
|
|
if (ref $filename && $filename->isa('Tie::Handle')) { |
|
499
|
|
|
|
|
|
|
$fh = $filename; |
|
500
|
|
|
|
|
|
|
} |
|
501
|
|
|
|
|
|
|
else { |
|
502
|
|
|
|
|
|
|
$fh = new IO::File; |
|
503
|
|
|
|
|
|
|
$fh->open("< $filename") || return undef; |
|
504
|
|
|
|
|
|
|
} |
|
505
|
|
|
|
|
|
|
|
|
506
|
|
|
|
|
|
|
$obj->{fh} = $fh; # XML file handle |
|
507
|
|
|
|
|
|
|
$obj->{records} = []; # Temporary storage for XML::Tape::Record |
|
508
|
|
|
|
|
|
|
$obj->{admins} = []; # Temporary storage for XML::Tape::Admin |
|
509
|
|
|
|
|
|
|
$obj->{curr} = undef; # Current record to be read |
|
510
|
|
|
|
|
|
|
$obj->{parse_init} = 0; # Flag to indicate if we started reading XML |
|
511
|
|
|
|
|
|
|
$obj->{parse_done} = 0; # Flag to indicate if we still reading XML |
|
512
|
|
|
|
|
|
|
$obj->{parser} = undef; # XML::Parser |
|
513
|
|
|
|
|
|
|
$obj->{parsernb} = undef; # XML::Parser::ExpatNB |
|
514
|
|
|
|
|
|
|
$obj->{nav} = {}; # Hash to navigate in the XML record |
|
515
|
|
|
|
|
|
|
|
|
516
|
|
|
|
|
|
|
return $obj; |
|
517
|
|
|
|
|
|
|
} |
|
518
|
|
|
|
|
|
|
|
|
519
|
|
|
|
|
|
|
sub get_admin { |
|
520
|
|
|
|
|
|
|
my ($this) = shift; |
|
521
|
|
|
|
|
|
|
|
|
522
|
|
|
|
|
|
|
$this->parse() until ( ( scalar @{$this->{records}} ) || ( $this->{parse_done} ) ); |
|
523
|
|
|
|
|
|
|
|
|
524
|
|
|
|
|
|
|
return shift( @{$this->{admins}} ); |
|
525
|
|
|
|
|
|
|
} |
|
526
|
|
|
|
|
|
|
|
|
527
|
|
|
|
|
|
|
sub get_record { |
|
528
|
|
|
|
|
|
|
my ($this) = shift; |
|
529
|
|
|
|
|
|
|
|
|
530
|
|
|
|
|
|
|
# Parse the XML until we read a new record or the parse is done... |
|
531
|
|
|
|
|
|
|
$this->parse() until ( ( scalar @{$this->{records}} ) || ( $this->{parse_done} ) ); |
|
532
|
|
|
|
|
|
|
|
|
533
|
|
|
|
|
|
|
return shift( @{$this->{records}} ); |
|
534
|
|
|
|
|
|
|
} |
|
535
|
|
|
|
|
|
|
|
|
536
|
|
|
|
|
|
|
sub tapeclose { |
|
537
|
|
|
|
|
|
|
my ($this) = shift; |
|
538
|
|
|
|
|
|
|
$this->{fh}->close; |
|
539
|
|
|
|
|
|
|
} |
|
540
|
|
|
|
|
|
|
|
|
541
|
|
|
|
|
|
|
sub parse_init { |
|
542
|
|
|
|
|
|
|
my ($this) = shift; |
|
543
|
|
|
|
|
|
|
|
|
544
|
|
|
|
|
|
|
$this->{parser} = new XML::Parser( Handlers => { |
|
545
|
|
|
|
|
|
|
Start => sub { $this->handle_start(@_); }, |
|
546
|
|
|
|
|
|
|
Char => sub { $this->handle_char(@_); }, |
|
547
|
|
|
|
|
|
|
Comment => sub { $this->handle_comment(@_); }, |
|
548
|
|
|
|
|
|
|
Proc => sub { $this->handle_proc(@_); }, |
|
549
|
|
|
|
|
|
|
CdataStart => sub { $this->handle_cdata_start(@_); }, |
|
550
|
|
|
|
|
|
|
CdataEnd => sub { $this->handle_cdata_end(@_); }, |
|
551
|
|
|
|
|
|
|
End => sub { $this->handle_end(@_); }, |
|
552
|
|
|
|
|
|
|
Final => sub { $this->handle_final(@_); }, |
|
553
|
|
|
|
|
|
|
}); |
|
554
|
|
|
|
|
|
|
|
|
555
|
|
|
|
|
|
|
$this->{parsernb} = $this->{parser}->parse_start(); |
|
556
|
|
|
|
|
|
|
|
|
557
|
|
|
|
|
|
|
return undef unless $this->{parsernb}; |
|
558
|
|
|
|
|
|
|
|
|
559
|
|
|
|
|
|
|
$this->{parse_init} = 1; |
|
560
|
|
|
|
|
|
|
|
|
561
|
|
|
|
|
|
|
return 1; |
|
562
|
|
|
|
|
|
|
} |
|
563
|
|
|
|
|
|
|
|
|
564
|
|
|
|
|
|
|
sub parse { |
|
565
|
|
|
|
|
|
|
my ($this) = shift; |
|
566
|
|
|
|
|
|
|
|
|
567
|
|
|
|
|
|
|
unless ($this->{parse_init}) { |
|
568
|
|
|
|
|
|
|
$this->parse_init() || return undef; |
|
569
|
|
|
|
|
|
|
} |
|
570
|
|
|
|
|
|
|
|
|
571
|
|
|
|
|
|
|
if (defined $this->{fh}) { |
|
572
|
|
|
|
|
|
|
my $buffer; |
|
573
|
|
|
|
|
|
|
|
|
574
|
|
|
|
|
|
|
# Read a chunk of XML... |
|
575
|
|
|
|
|
|
|
read($this->{fh}, $buffer, $XML::Tape::Reader::BUFF_SIZE); |
|
576
|
|
|
|
|
|
|
|
|
577
|
|
|
|
|
|
|
# If the buffer isn't empty then, parse it |
|
578
|
|
|
|
|
|
|
# otherwise we reached the end of the file... |
|
579
|
|
|
|
|
|
|
if (length $buffer) { |
|
580
|
|
|
|
|
|
|
$this->{parsernb}->parse_more($buffer); |
|
581
|
|
|
|
|
|
|
} |
|
582
|
|
|
|
|
|
|
else { |
|
583
|
|
|
|
|
|
|
$this->{parsernb}->parse_done(); |
|
584
|
|
|
|
|
|
|
$this->{parse_done} = 1; |
|
585
|
|
|
|
|
|
|
} |
|
586
|
|
|
|
|
|
|
} |
|
587
|
|
|
|
|
|
|
} |
|
588
|
|
|
|
|
|
|
|
|
589
|
|
|
|
|
|
|
sub handle_start { |
|
590
|
|
|
|
|
|
|
my ($this, $xp, $elem, %attr) = @_; |
|
591
|
|
|
|
|
|
|
|
|
592
|
|
|
|
|
|
|
if (0) {} |
|
593
|
|
|
|
|
|
|
elsif ($this->{nav}->{in_record}) { |
|
594
|
|
|
|
|
|
|
$this->{curr}->addRecordXML($xp->original_string); |
|
595
|
|
|
|
|
|
|
} |
|
596
|
|
|
|
|
|
|
elsif ($this->{nav}->{in_record_admin}) { |
|
597
|
|
|
|
|
|
|
$this->{curr}->addAdminXML($xp->original_string); |
|
598
|
|
|
|
|
|
|
} |
|
599
|
|
|
|
|
|
|
elsif ($this->{nav}->{in_tape_admin}) { |
|
600
|
|
|
|
|
|
|
$this->{curr}->addAdminXML($xp->original_string); |
|
601
|
|
|
|
|
|
|
} |
|
602
|
|
|
|
|
|
|
|
|
603
|
|
|
|
|
|
|
if (0) {} |
|
604
|
|
|
|
|
|
|
elsif ($xp->depth == 1 && $elem =~ /^(\w+:)?tape-admin$/) { |
|
605
|
|
|
|
|
|
|
$this->{nav}->{in_tape_admin} = 1; |
|
606
|
|
|
|
|
|
|
$this->{curr} = XML::Tape::Admin->new(); |
|
607
|
|
|
|
|
|
|
} |
|
608
|
|
|
|
|
|
|
elsif ($xp->depth == 1 && $elem =~ /^(\w+:)?tape-record$/) { |
|
609
|
|
|
|
|
|
|
$this->{curr} = XML::Tape::Record->new(); |
|
610
|
|
|
|
|
|
|
} |
|
611
|
|
|
|
|
|
|
elsif ($xp->depth == 2 && $elem =~ /^(\w+:)?tape-record-admin$/) { |
|
612
|
|
|
|
|
|
|
$this->{nav}->{in_tape_record_admin} = 1; |
|
613
|
|
|
|
|
|
|
} |
|
614
|
|
|
|
|
|
|
elsif ($this->{nav}->{in_tape_record_admin} == 1 && $elem =~ /^(\w+:)?identifier$/) { |
|
615
|
|
|
|
|
|
|
$this->{nav}->{in_record_identifier} = 1; |
|
616
|
|
|
|
|
|
|
} |
|
617
|
|
|
|
|
|
|
elsif ($this->{nav}->{in_tape_record_admin} == 1 && $elem =~ /^(\w+:)?date$/) { |
|
618
|
|
|
|
|
|
|
$this->{nav}->{in_record_date} = 1; |
|
619
|
|
|
|
|
|
|
} |
|
620
|
|
|
|
|
|
|
elsif ($this->{nav}->{in_tape_record_admin} == 1 && $elem =~ /^(\w+:)?record-admin$/) { |
|
621
|
|
|
|
|
|
|
$this->{nav}->{in_record_admin} = 1; |
|
622
|
|
|
|
|
|
|
$this->{curr}->pushAdmin(); |
|
623
|
|
|
|
|
|
|
} |
|
624
|
|
|
|
|
|
|
elsif ($xp->depth == 2 && $elem =~ /^(\w+:)?record$/) { |
|
625
|
|
|
|
|
|
|
$this->{nav}->{in_record} = 1; |
|
626
|
|
|
|
|
|
|
$this->{curr}->setStartByte($xp->current_byte + length $xp->original_string); |
|
627
|
|
|
|
|
|
|
} |
|
628
|
|
|
|
|
|
|
} |
|
629
|
|
|
|
|
|
|
|
|
630
|
|
|
|
|
|
|
sub handle_end { |
|
631
|
|
|
|
|
|
|
my ($this, $xp, $elem, %attr) = @_; |
|
632
|
|
|
|
|
|
|
|
|
633
|
|
|
|
|
|
|
if (0) {} |
|
634
|
|
|
|
|
|
|
elsif ($xp->depth == 1 && $elem =~ /^(\w+:)?tape-admin$/) { |
|
635
|
|
|
|
|
|
|
$this->{nav}->{in_tape_admin} = 0; |
|
636
|
|
|
|
|
|
|
push(@{$this->{admins}}, $this->{curr}); |
|
637
|
|
|
|
|
|
|
} |
|
638
|
|
|
|
|
|
|
elsif ($xp->depth == 1 && $elem =~ /^(\w+:)?tape-record$/) { |
|
639
|
|
|
|
|
|
|
push(@{$this->{records}}, $this->{curr}); |
|
640
|
|
|
|
|
|
|
} |
|
641
|
|
|
|
|
|
|
elsif ($xp->depth == 2 && $elem =~ /^(\w+:)?tape-record-admin$/) { |
|
642
|
|
|
|
|
|
|
$this->{nav}->{in_tape_record_admin} = 0; |
|
643
|
|
|
|
|
|
|
} |
|
644
|
|
|
|
|
|
|
elsif ($this->{nav}->{in_tape_record_admin} == 1 && $elem =~ /^(\w+:)?identifier$/) { |
|
645
|
|
|
|
|
|
|
$this->{nav}->{in_record_identifier} = 0; |
|
646
|
|
|
|
|
|
|
} |
|
647
|
|
|
|
|
|
|
elsif ($this->{nav}->{in_tape_record_admin} == 1 && $elem =~ /^(\w+:)?date$/) { |
|
648
|
|
|
|
|
|
|
$this->{nav}->{in_record_date} = 0; |
|
649
|
|
|
|
|
|
|
} |
|
650
|
|
|
|
|
|
|
elsif ($this->{nav}->{in_tape_record_admin} == 1 && $elem =~ /^(\w+:)?record-admin$/) { |
|
651
|
|
|
|
|
|
|
$this->{nav}->{in_record_admin} = 0; |
|
652
|
|
|
|
|
|
|
} |
|
653
|
|
|
|
|
|
|
elsif ($xp->depth == 2 && $elem =~ /^(\w+:)?record$/) { |
|
654
|
|
|
|
|
|
|
$this->{nav}->{in_record} = 0; |
|
655
|
|
|
|
|
|
|
$this->{curr}->setEndByte($xp->current_byte); |
|
656
|
|
|
|
|
|
|
} |
|
657
|
|
|
|
|
|
|
|
|
658
|
|
|
|
|
|
|
if (0) {} |
|
659
|
|
|
|
|
|
|
elsif ($this->{nav}->{in_record}) { |
|
660
|
|
|
|
|
|
|
$this->{curr}->addRecordXML($xp->original_string); |
|
661
|
|
|
|
|
|
|
} |
|
662
|
|
|
|
|
|
|
elsif ($this->{nav}->{in_record_admin}) { |
|
663
|
|
|
|
|
|
|
$this->{curr}->addAdminXML($xp->original_string); |
|
664
|
|
|
|
|
|
|
} |
|
665
|
|
|
|
|
|
|
elsif ($this->{nav}->{in_tape_admin}) { |
|
666
|
|
|
|
|
|
|
$this->{curr}->addAdminXML($xp->original_string); |
|
667
|
|
|
|
|
|
|
} |
|
668
|
|
|
|
|
|
|
} |
|
669
|
|
|
|
|
|
|
|
|
670
|
|
|
|
|
|
|
sub handle_char { |
|
671
|
|
|
|
|
|
|
my ($this, $xp, $data) = @_; |
|
672
|
|
|
|
|
|
|
|
|
673
|
|
|
|
|
|
|
if (0) {} |
|
674
|
|
|
|
|
|
|
elsif ($this->{nav}->{in_tape_admin}) { |
|
675
|
|
|
|
|
|
|
$this->{curr}->addAdminXML($xp->original_string); |
|
676
|
|
|
|
|
|
|
} |
|
677
|
|
|
|
|
|
|
elsif ($this->{nav}->{in_record}) { |
|
678
|
|
|
|
|
|
|
$this->{curr}->addRecordXML($xp->original_string); |
|
679
|
|
|
|
|
|
|
} |
|
680
|
|
|
|
|
|
|
elsif ($this->{nav}->{in_record_identifier}) { |
|
681
|
|
|
|
|
|
|
$this->{curr}->addIdentifier($data); |
|
682
|
|
|
|
|
|
|
} |
|
683
|
|
|
|
|
|
|
elsif ($this->{nav}->{in_record_date}) { |
|
684
|
|
|
|
|
|
|
$this->{curr}->addDate($data); |
|
685
|
|
|
|
|
|
|
} |
|
686
|
|
|
|
|
|
|
} |
|
687
|
|
|
|
|
|
|
|
|
688
|
|
|
|
|
|
|
sub handle_comment { |
|
689
|
|
|
|
|
|
|
my ($this, $xp, $data) = @_; |
|
690
|
|
|
|
|
|
|
|
|
691
|
|
|
|
|
|
|
if (0) {} |
|
692
|
|
|
|
|
|
|
elsif ($this->{nav}->{in_tape_admin}) { |
|
693
|
|
|
|
|
|
|
$this->{curr}->addAdminXML($xp->original_string); |
|
694
|
|
|
|
|
|
|
} |
|
695
|
|
|
|
|
|
|
elsif ($this->{nav}->{in_record}) { |
|
696
|
|
|
|
|
|
|
$this->{curr}->addRecordXML($xp->original_string); |
|
697
|
|
|
|
|
|
|
} |
|
698
|
|
|
|
|
|
|
} |
|
699
|
|
|
|
|
|
|
|
|
700
|
|
|
|
|
|
|
sub handle_proc { |
|
701
|
|
|
|
|
|
|
my ($this, $xp) = @_; |
|
702
|
|
|
|
|
|
|
|
|
703
|
|
|
|
|
|
|
if (0) {} |
|
704
|
|
|
|
|
|
|
elsif ($this->{nav}->{in_tape_admin}) { |
|
705
|
|
|
|
|
|
|
$this->{curr}->addAdminXML($xp->original_string); |
|
706
|
|
|
|
|
|
|
} |
|
707
|
|
|
|
|
|
|
elsif ($this->{nav}->{in_record}) { |
|
708
|
|
|
|
|
|
|
$this->{curr}->addRecordXML($xp->original_string); |
|
709
|
|
|
|
|
|
|
} |
|
710
|
|
|
|
|
|
|
} |
|
711
|
|
|
|
|
|
|
|
|
712
|
|
|
|
|
|
|
sub handle_cdata_start { |
|
713
|
|
|
|
|
|
|
my ($this, $xp) = @_; |
|
714
|
|
|
|
|
|
|
|
|
715
|
|
|
|
|
|
|
if (0) {} |
|
716
|
|
|
|
|
|
|
elsif ($this->{nav}->{in_tape_admin}) { |
|
717
|
|
|
|
|
|
|
$this->{curr}->addAdminXML($xp->original_string); |
|
718
|
|
|
|
|
|
|
} |
|
719
|
|
|
|
|
|
|
elsif ($this->{nav}->{in_record}) { |
|
720
|
|
|
|
|
|
|
$this->{curr}->addRecordXML($xp->original_string); |
|
721
|
|
|
|
|
|
|
} |
|
722
|
|
|
|
|
|
|
} |
|
723
|
|
|
|
|
|
|
|
|
724
|
|
|
|
|
|
|
sub handle_cdata_end { |
|
725
|
|
|
|
|
|
|
my ($this, $xp) = @_; |
|
726
|
|
|
|
|
|
|
|
|
727
|
|
|
|
|
|
|
if (0) {} |
|
728
|
|
|
|
|
|
|
elsif ($this->{nav}->{in_tape_admin}) { |
|
729
|
|
|
|
|
|
|
$this->{curr}->addAdminXML($xp->original_string); |
|
730
|
|
|
|
|
|
|
} |
|
731
|
|
|
|
|
|
|
elsif ($this->{nav}->{in_record}) { |
|
732
|
|
|
|
|
|
|
$this->{curr}->addRecordXML($xp->original_string); |
|
733
|
|
|
|
|
|
|
} |
|
734
|
|
|
|
|
|
|
} |
|
735
|
|
|
|
|
|
|
|
|
736
|
|
|
|
|
|
|
sub handle_final { |
|
737
|
|
|
|
|
|
|
return 1; |
|
738
|
|
|
|
|
|
|
} |
|
739
|
|
|
|
|
|
|
|
|
740
|
|
|
|
|
|
|
|
|
741
|
|
|
|
|
|
|
=back |
|
742
|
|
|
|
|
|
|
|
|
743
|
|
|
|
|
|
|
=head1 XML::Tape::Admin METHODS |
|
744
|
|
|
|
|
|
|
|
|
745
|
|
|
|
|
|
|
=over 4 |
|
746
|
|
|
|
|
|
|
|
|
747
|
|
|
|
|
|
|
=item $admin->getRecord() |
|
748
|
|
|
|
|
|
|
|
|
749
|
|
|
|
|
|
|
Returns a XML string representation of a XMLtape administrative record. |
|
750
|
|
|
|
|
|
|
|
|
751
|
|
|
|
|
|
|
=back |
|
752
|
|
|
|
|
|
|
|
|
753
|
|
|
|
|
|
|
=cut |
|
754
|
|
|
|
|
|
|
package XML::Tape::Admin; |
|
755
|
|
|
|
|
|
|
|
|
756
|
|
|
|
|
|
|
sub new { |
|
757
|
|
|
|
|
|
|
my ($pkg) = shift; |
|
758
|
|
|
|
|
|
|
return bless { |
|
759
|
|
|
|
|
|
|
adminXML => undef , |
|
760
|
|
|
|
|
|
|
} , $pkg; |
|
761
|
|
|
|
|
|
|
} |
|
762
|
|
|
|
|
|
|
|
|
763
|
|
|
|
|
|
|
sub addAdminXML { |
|
764
|
|
|
|
|
|
|
my ($this, $str) = @_; |
|
765
|
|
|
|
|
|
|
$this->{adminXML} .= $str; |
|
766
|
|
|
|
|
|
|
} |
|
767
|
|
|
|
|
|
|
|
|
768
|
|
|
|
|
|
|
sub getRecord { |
|
769
|
|
|
|
|
|
|
my ($this) = @_; |
|
770
|
|
|
|
|
|
|
return $this->{adminXML}; |
|
771
|
|
|
|
|
|
|
} |
|
772
|
|
|
|
|
|
|
|
|
773
|
|
|
|
|
|
|
=head1 XML::Tape::Record METHODS |
|
774
|
|
|
|
|
|
|
|
|
775
|
|
|
|
|
|
|
=over 4 |
|
776
|
|
|
|
|
|
|
|
|
777
|
|
|
|
|
|
|
=item $record->getIdentifier() |
|
778
|
|
|
|
|
|
|
|
|
779
|
|
|
|
|
|
|
Returns the record identifier as string. |
|
780
|
|
|
|
|
|
|
|
|
781
|
|
|
|
|
|
|
=item $record->getDate() |
|
782
|
|
|
|
|
|
|
|
|
783
|
|
|
|
|
|
|
Returns the record datestamp as string. |
|
784
|
|
|
|
|
|
|
|
|
785
|
|
|
|
|
|
|
=item $record->getAdmin() |
|
786
|
|
|
|
|
|
|
|
|
787
|
|
|
|
|
|
|
Returns an ARRAY of administrative records |
|
788
|
|
|
|
|
|
|
|
|
789
|
|
|
|
|
|
|
=item $record->getRecord() |
|
790
|
|
|
|
|
|
|
|
|
791
|
|
|
|
|
|
|
Returns a XML string representation of a XMLtape record. |
|
792
|
|
|
|
|
|
|
|
|
793
|
|
|
|
|
|
|
=item $record->getStartByte() |
|
794
|
|
|
|
|
|
|
|
|
795
|
|
|
|
|
|
|
Returns the start byte position of the record in the XMLtape |
|
796
|
|
|
|
|
|
|
|
|
797
|
|
|
|
|
|
|
=item $record->getEndByte() |
|
798
|
|
|
|
|
|
|
|
|
799
|
|
|
|
|
|
|
Returns the end byte positorion of the record in the XMLtape |
|
800
|
|
|
|
|
|
|
|
|
801
|
|
|
|
|
|
|
=back |
|
802
|
|
|
|
|
|
|
|
|
803
|
|
|
|
|
|
|
=cut |
|
804
|
|
|
|
|
|
|
package XML::Tape::Record; |
|
805
|
|
|
|
|
|
|
|
|
806
|
|
|
|
|
|
|
sub new { |
|
807
|
|
|
|
|
|
|
my ($pkg) = shift; |
|
808
|
|
|
|
|
|
|
return bless { |
|
809
|
|
|
|
|
|
|
startByte => 0 , |
|
810
|
|
|
|
|
|
|
endByte => 0 , |
|
811
|
|
|
|
|
|
|
recordXML => undef, |
|
812
|
|
|
|
|
|
|
identifier => undef, |
|
813
|
|
|
|
|
|
|
date => undef, |
|
814
|
|
|
|
|
|
|
admin => [], |
|
815
|
|
|
|
|
|
|
} , $pkg; |
|
816
|
|
|
|
|
|
|
} |
|
817
|
|
|
|
|
|
|
|
|
818
|
|
|
|
|
|
|
sub setStartByte { |
|
819
|
|
|
|
|
|
|
my ($this,$num) = @_; |
|
820
|
|
|
|
|
|
|
$this->{startByte} = $num; |
|
821
|
|
|
|
|
|
|
} |
|
822
|
|
|
|
|
|
|
|
|
823
|
|
|
|
|
|
|
sub getStartByte { |
|
824
|
|
|
|
|
|
|
my ($this) = @_; |
|
825
|
|
|
|
|
|
|
return $this->{startByte}; |
|
826
|
|
|
|
|
|
|
} |
|
827
|
|
|
|
|
|
|
|
|
828
|
|
|
|
|
|
|
sub setEndByte { |
|
829
|
|
|
|
|
|
|
my ($this,$num) = @_; |
|
830
|
|
|
|
|
|
|
$this->{endByte} = $num; |
|
831
|
|
|
|
|
|
|
} |
|
832
|
|
|
|
|
|
|
|
|
833
|
|
|
|
|
|
|
sub getEndByte { |
|
834
|
|
|
|
|
|
|
my ($this) = @_; |
|
835
|
|
|
|
|
|
|
return $this->{endByte}; |
|
836
|
|
|
|
|
|
|
} |
|
837
|
|
|
|
|
|
|
|
|
838
|
|
|
|
|
|
|
sub addRecordXML { |
|
839
|
|
|
|
|
|
|
my ($this, $str) = @_; |
|
840
|
|
|
|
|
|
|
$this->{recordXML} .= $str; |
|
841
|
|
|
|
|
|
|
} |
|
842
|
|
|
|
|
|
|
|
|
843
|
|
|
|
|
|
|
sub getRecord { |
|
844
|
|
|
|
|
|
|
my ($this) = @_; |
|
845
|
|
|
|
|
|
|
return $this->{recordXML}; |
|
846
|
|
|
|
|
|
|
} |
|
847
|
|
|
|
|
|
|
|
|
848
|
|
|
|
|
|
|
sub addIdentifier { |
|
849
|
|
|
|
|
|
|
my ($this, $str) = @_; |
|
850
|
|
|
|
|
|
|
$this->{identifier} .= $str; |
|
851
|
|
|
|
|
|
|
} |
|
852
|
|
|
|
|
|
|
|
|
853
|
|
|
|
|
|
|
sub getIdentifier { |
|
854
|
|
|
|
|
|
|
my ($this) = @_; |
|
855
|
|
|
|
|
|
|
return $this->{identifier}; |
|
856
|
|
|
|
|
|
|
} |
|
857
|
|
|
|
|
|
|
|
|
858
|
|
|
|
|
|
|
sub addDate { |
|
859
|
|
|
|
|
|
|
my ($this, $str) = @_; |
|
860
|
|
|
|
|
|
|
$this->{date} .= $str; |
|
861
|
|
|
|
|
|
|
} |
|
862
|
|
|
|
|
|
|
|
|
863
|
|
|
|
|
|
|
sub getDate { |
|
864
|
|
|
|
|
|
|
my ($this) = @_; |
|
865
|
|
|
|
|
|
|
return $this->{date}; |
|
866
|
|
|
|
|
|
|
} |
|
867
|
|
|
|
|
|
|
|
|
868
|
|
|
|
|
|
|
sub pushAdmin { |
|
869
|
|
|
|
|
|
|
my ($this) = @_; |
|
870
|
|
|
|
|
|
|
push(@{$this->{admin}},''); |
|
871
|
|
|
|
|
|
|
} |
|
872
|
|
|
|
|
|
|
|
|
873
|
|
|
|
|
|
|
sub addAdminXML { |
|
874
|
|
|
|
|
|
|
my ($this, $xml) = @_; |
|
875
|
|
|
|
|
|
|
my $num = @{$this->{admin}}; |
|
876
|
|
|
|
|
|
|
$this->{admin}->[$num-1] .= $xml; |
|
877
|
|
|
|
|
|
|
} |
|
878
|
|
|
|
|
|
|
|
|
879
|
|
|
|
|
|
|
sub getAdmin { |
|
880
|
|
|
|
|
|
|
my ($this) = @_; |
|
881
|
|
|
|
|
|
|
return $this->{admin}; |
|
882
|
|
|
|
|
|
|
} |
|
883
|
|
|
|
|
|
|
|
|
884
|
|
|
|
|
|
|
1; |
|
885
|
|
|
|
|
|
|
|
|
886
|
|
|
|
|
|
|
=head1 FURTHER INFORMATION |
|
887
|
|
|
|
|
|
|
|
|
888
|
|
|
|
|
|
|
'File-based storage of Digital Objects and constituent datastreams: XMLtapes and Internet Archive ARC files' |
|
889
|
|
|
|
|
|
|
http://arxiv.org/abs/cs.DL/0503016 |
|
890
|
|
|
|
|
|
|
|
|
891
|
|
|
|
|
|
|
'The multi-faceted use of the OAI-PMH in the LANL Repository' |
|
892
|
|
|
|
|
|
|
http://yar.sourceforge.net/jcdl2004-submitted-draft.pdf |
|
893
|
|
|
|
|
|
|
|
|
894
|
|
|
|
|
|
|
=head1 BUGS |
|
895
|
|
|
|
|
|
|
|
|
896
|
|
|
|
|
|
|
UTF-8 encoding is mandatory. |
|
897
|
|
|
|
|
|
|
Doesn't check for UTF-8 encoding. |
|
898
|
|
|
|
|
|
|
|
|
899
|
|
|
|
|
|
|
=head1 CREDITS |
|
900
|
|
|
|
|
|
|
|
|
901
|
|
|
|
|
|
|
XMLtape archives were developed by the Digital Library Research & Prototyping |
|
902
|
|
|
|
|
|
|
team at Los Alamos National Laboratory. |
|
903
|
|
|
|
|
|
|
|
|
904
|
|
|
|
|
|
|
XML parsing in the module was inspired by Robert Hanson's XML::RAX module. |
|
905
|
|
|
|
|
|
|
|
|
906
|
|
|
|
|
|
|
=head1 SEE ALSO |
|
907
|
|
|
|
|
|
|
|
|
908
|
|
|
|
|
|
|
L |
|
909
|
|
|
|
|
|
|
|
|
910
|
|
|
|
|
|
|
In bin/oaitape you'll find an example of a OAI-PMH interface on XML::Tape |
|
911
|
|
|
|
|
|
|
|
|
912
|
|
|
|
|
|
|
=head1 AUTHOR |
|
913
|
|
|
|
|
|
|
|
|
914
|
|
|
|
|
|
|
Patrick Hochstenbach |
|
915
|
|
|
|
|
|
|
|
|
916
|
|
|
|
|
|
|
=cut |
|
917
|
|
|
|
|
|
|
|
|
918
|
|
|
|
|
|
|
1; |