1
|
#!/usr/bin/perl
|
2
|
|
3
|
# This script is INCOMPLETE -- see TODO markers throughout
|
4
|
|
5
|
# This script loops through a list of metadata documents and associated data
|
6
|
# files, and for each one uploads the data file to metacat, puts the accession
|
7
|
# number in the metadata, then inserts the metadata to metacat.
|
8
|
# Matt Jones February 14, 2008
|
9
|
|
10
|
use Metacat;
|
11
|
use XML::DOM;
|
12
|
use strict;
|
13
|
|
14
|
############################################################################
|
15
|
#
|
16
|
# MAIN program block
|
17
|
#
|
18
|
############################################################################
|
19
|
|
20
|
# check that the correct number or parameters are passed from the commandline
|
21
|
if (($#ARGV +1) != 1) {die "Usage: %./insert_fgdc.pl <metacat_url> \n\n";}
|
22
|
# Get the URL to the metacat server from the command line options
|
23
|
my ($url) = @ARGV;
|
24
|
my $dn = 'uid=jones,o=NCEAS,dc=ecoinformatics,dc=org';
|
25
|
my $password = 'foobar';
|
26
|
|
27
|
# Open a metacat connection and login
|
28
|
my $metacat = openMetacatConnection($url);
|
29
|
loginToMetacat($metacat, $dn, $password);
|
30
|
|
31
|
# Read in the list of FGDC documents
|
32
|
my $dataDir = "/tmp/data";
|
33
|
# datalist key = data filename, value = metadata filename
|
34
|
my $datalist = getDataList($dataDir);
|
35
|
my $datafilename;
|
36
|
|
37
|
# Loop through the list of datafiles
|
38
|
foreach $datafilename (keys %$datalist) {
|
39
|
print("Processing... $datafilename $$datalist{$datafilename}\n");
|
40
|
|
41
|
# Upload the data file into Metacat, catching accession # errors
|
42
|
my $dataId = uploadData($metacat, $datafilename);
|
43
|
|
44
|
# Reference the accession # URL for the data file in the FGDC metadata
|
45
|
my $metadata = createDataReference($dataId, $$datalist{$datafilename});
|
46
|
|
47
|
# Insert the metadata file into Metacat, catching accession # errors
|
48
|
my $metadataId = insertMetadata($metacat, $metadata );
|
49
|
|
50
|
# Set access control rules for the metadata and data files
|
51
|
setAccess($dataId);
|
52
|
setAccess($metadataId);
|
53
|
}
|
54
|
|
55
|
exit(0);
|
56
|
|
57
|
############################################################################
|
58
|
#
|
59
|
# SUBROUTINES
|
60
|
#
|
61
|
############################################################################
|
62
|
|
63
|
#
|
64
|
# Create a connection to the metacat server
|
65
|
#
|
66
|
sub openMetacatConnection {
|
67
|
my $url = shift;
|
68
|
|
69
|
my $metacat = Metacat->new();
|
70
|
if ($metacat) {
|
71
|
$metacat->set_options( metacatUrl => $url );
|
72
|
} else {
|
73
|
die("Could not open connection to Metacat url: $url\n");
|
74
|
}
|
75
|
return $metacat;
|
76
|
}
|
77
|
|
78
|
#
|
79
|
# Login to the metacat server
|
80
|
#
|
81
|
sub loginToMetacat {
|
82
|
my $metacat = shift;
|
83
|
my $dn = shift;
|
84
|
my $password = shift;
|
85
|
|
86
|
my $response1 = $metacat->login($dn, $password);
|
87
|
if (! $response1) {
|
88
|
print $metacat->getMessage();
|
89
|
print "Failed during login: metacat.\n";
|
90
|
} else {
|
91
|
print "Connected to metacat\n";
|
92
|
}
|
93
|
}
|
94
|
|
95
|
#
|
96
|
# TODO
|
97
|
# Get a hashed list of datafile names and their associated metadata file names
|
98
|
#
|
99
|
sub getDataList {
|
100
|
my $dataDir = shift;
|
101
|
|
102
|
my %dataList = ("datafile_a"=>"metadatafile_ma",
|
103
|
"datafile_b"=>"metadatafile_mb");
|
104
|
|
105
|
return \%dataList;
|
106
|
}
|
107
|
|
108
|
#
|
109
|
# TODO
|
110
|
# Upload a data file to metacat, returning the identifier used
|
111
|
#
|
112
|
sub uploadData {
|
113
|
my $metacat = shift;
|
114
|
my $datafilename = shift;
|
115
|
|
116
|
# TODO: Create a unique ID
|
117
|
my $identifier;
|
118
|
|
119
|
# TODO: Read in the data file from disk
|
120
|
# my $data;
|
121
|
|
122
|
# TODO: Do the metacat insertion as a MIME insertion
|
123
|
#my $response = $metacat->upload($identifier, $data);
|
124
|
|
125
|
# TODO: Check the insertion succeeded, if not possibly try again with new id
|
126
|
my $message = $metacat->getMessage();
|
127
|
|
128
|
return $identifier;
|
129
|
|
130
|
return $identifier;
|
131
|
}
|
132
|
|
133
|
#
|
134
|
# TODO
|
135
|
# Insert a reference to a data file into the FGDC metadata document
|
136
|
#
|
137
|
sub createDataReference {
|
138
|
my $metacat = shift;
|
139
|
my $dataId = shift;
|
140
|
my $metadatafilename = shift;
|
141
|
|
142
|
my $metadata;
|
143
|
|
144
|
return $metadata;
|
145
|
}
|
146
|
|
147
|
#
|
148
|
# TODO
|
149
|
# Insert the metadata document into the Metacat database
|
150
|
#
|
151
|
sub insertMetadata {
|
152
|
my $metacat = shift;
|
153
|
my $metadatafilename = shift;
|
154
|
|
155
|
# TODO: Create a unique ID
|
156
|
my $identifier;
|
157
|
|
158
|
# TODO: Read in the metadata file from disk
|
159
|
# my $metadata;
|
160
|
|
161
|
# Do the metacat insertion
|
162
|
# my $response = $metacat->insert($identifier, $metadata);
|
163
|
|
164
|
# TODO: Check the insertion succeeded, if not possibly try again with new id
|
165
|
my $message = $metacat->getMessage();
|
166
|
|
167
|
return $identifier;
|
168
|
}
|
169
|
|
170
|
#
|
171
|
# TODO
|
172
|
# Set access control permissions on the file identified
|
173
|
#
|
174
|
sub setAccess {
|
175
|
my $metacat = shift;
|
176
|
my $identifier;
|
177
|
|
178
|
}
|