| 1 |
2563
|
harris
|
#!/usr/bin/perl
|
| 2 |
|
|
|
| 3 |
|
|
# check that the correct number or parameters are issued
|
| 4 |
|
|
if (($#ARGV +1) != 3) {die "Usage: %./get_eml.pl <metacat_url> <login> <passwd>\n\n";}
|
| 5 |
|
|
|
| 6 |
|
|
($url, $username, $password) = @ARGV; #get the input file and output file names
|
| 7 |
|
|
|
| 8 |
|
|
|
| 9 |
|
|
use Metacat;
|
| 10 |
|
|
use XML::DOM;
|
| 11 |
|
|
|
| 12 |
|
|
my $metacat = Metacat->new();
|
| 13 |
|
|
|
| 14 |
|
|
|
| 15 |
|
|
if ($metacat) {
|
| 16 |
|
|
$metacat->set_options( metacatUrl => $url );
|
| 17 |
|
|
} else {
|
| 18 |
|
|
#print "Failed during metacat creation.";
|
| 19 |
|
|
$error = 1;
|
| 20 |
|
|
exit();
|
| 21 |
|
|
}
|
| 22 |
|
|
|
| 23 |
|
|
# Login to metacat
|
| 24 |
|
|
# //print "Logging in to metacat..........\n";
|
| 25 |
|
|
my $response1 = $metacat->login($username, $password);
|
| 26 |
|
|
if (! $response1) {
|
| 27 |
|
|
# //print $metacat->getMessage();
|
| 28 |
|
|
# //print "Failed during login: metacat.\n";
|
| 29 |
|
|
$error = 2;
|
| 30 |
|
|
} else {
|
| 31 |
|
|
# //print "Logged in.\n";
|
| 32 |
|
|
}
|
| 33 |
|
|
|
| 34 |
|
|
|
| 35 |
|
|
my $query = "<?xml version=\"1.0\" ?> <pathquery version=\"1.2\"> <querytitle>Untitled-Search-2</querytitle> <returndoctype>-//ecoinformatics.org//eml-dataset-2.0.0beta6//EN</returndoctype> <returndoctype>-//NCEAS//eml-dataset-2.0//EN</returndoctype> <returndoctype>eml://ecoinformatics.org/eml-2.0.0</returndoctype> <returnfield>dataset/title</returnfield> <returnfield>individualName/surName</returnfield> <returnfield>keyword</returnfield><returnfield>westBoundingCoordinate</returnfield><returnfield>eastBoundingCoordinate</returnfield><returnfield>northBoundingCoordinate</returnfield><returnfield>southBoundingCoordinate</returnfield><returnfield>westbc</returnfield><returnfield>eastbc</returnfield><returnfield>northbc</returnfield><returnfield>southbc</returnfield><querygroup operator=\"INTERSECT\"><querygroup operator=\"UNION\"><querygroup operator=\"UNION\"><queryterm searchmode=\"contains\" casesensitive=\"false\"><value>%25</value></queryterm></querygroup></querygroup></querygroup></pathquery>";
|
| 36 |
|
|
|
| 37 |
|
|
my $response = $metacat->squery($query);
|
| 38 |
|
|
|
| 39 |
|
|
my $mesg =$metacat->getMessage();
|
| 40 |
|
|
|
| 41 |
|
|
#print "\n\ngetting the message ... \n";
|
| 42 |
|
|
#print $mesg;
|
| 43 |
|
|
|
| 44 |
|
|
if($mesg eq ""){
|
| 45 |
|
|
# print ("Too much time is reply back from metacat...");
|
| 46 |
|
|
exit();
|
| 47 |
|
|
}
|
| 48 |
|
|
|
| 49 |
|
|
my $parser = new XML::DOM::Parser;
|
| 50 |
|
|
my $node;
|
| 51 |
|
|
my $name;
|
| 52 |
|
|
my $doc = $parser->parse($mesg);
|
| 53 |
|
|
my $nodes = $doc->getElementsByTagName("docid");
|
| 54 |
|
|
|
| 55 |
|
|
$numberNodes = $nodes->getLength;
|
| 56 |
|
|
|
| 57 |
|
|
|
| 58 |
|
|
for (my $loop_index =0; $loop_index < $numberNodes; $loop_index++)
|
| 59 |
|
|
{
|
| 60 |
|
|
$node = $nodes->item($loop_index);
|
| 61 |
2638
|
harris
|
$name = trimwhitespace($node->getFirstChild()->getNodeValue());
|
| 62 |
2563
|
harris
|
|
| 63 |
|
|
$node = $node->getParentNode();
|
| 64 |
|
|
my $tempnodes = $node->getElementsByTagName("param");
|
| 65 |
|
|
my $tempnumberNodes = $tempnodes->getLength;
|
| 66 |
|
|
|
| 67 |
|
|
my $title = "";
|
| 68 |
|
|
my $keyword = "";
|
| 69 |
|
|
my $cName = "";
|
| 70 |
|
|
my $eBC = "";
|
| 71 |
|
|
my $wBC = "";
|
| 72 |
|
|
my $nBC = "";
|
| 73 |
|
|
my $sBC = "";
|
| 74 |
|
|
|
| 75 |
|
|
for (my $loop =0; $loop < $tempnumberNodes; $loop++)
|
| 76 |
|
|
{
|
| 77 |
|
|
my $tempnode = $tempnodes->item($loop);
|
| 78 |
|
|
my $paramname = $tempnode->getAttributeNode("name")->getValue();
|
| 79 |
|
|
if($paramname eq "dataset/title"){
|
| 80 |
2638
|
harris
|
$title = trimwhitespace($tempnode->getFirstChild()->getNodeValue());
|
| 81 |
2563
|
harris
|
}
|
| 82 |
|
|
if($paramname eq "keyword"){
|
| 83 |
2638
|
harris
|
$keyword = trimwhitespace($keyword.",".$tempnode->getFirstChild()->getNodeValue());
|
| 84 |
2563
|
harris
|
}
|
| 85 |
|
|
if($paramname eq "individualName/surName"){
|
| 86 |
2638
|
harris
|
$cName = trimwhitespace($cName.",".$tempnode->getFirstChild()->getNodeValue());
|
| 87 |
2563
|
harris
|
}
|
| 88 |
|
|
if($paramname eq "eastBoundingCoordinate"){
|
| 89 |
2638
|
harris
|
$eBC = trimwhitespace($tempnode->getFirstChild()->getNodeValue());
|
| 90 |
2563
|
harris
|
}
|
| 91 |
|
|
if($paramname eq "eastbc"){
|
| 92 |
2638
|
harris
|
$eBC = trimwhitespace($tempnode->getFirstChild()->getNodeValue());
|
| 93 |
2563
|
harris
|
}
|
| 94 |
|
|
if($paramname eq "westBoundingCoordinate"){
|
| 95 |
2638
|
harris
|
$wBC = trimwhitespace($tempnode->getFirstChild()->getNodeValue());
|
| 96 |
2563
|
harris
|
}
|
| 97 |
|
|
if($paramname eq "westbc"){
|
| 98 |
2638
|
harris
|
$wBC = trimwhitespace($tempnode->getFirstChild()->getNodeValue());
|
| 99 |
2563
|
harris
|
}
|
| 100 |
|
|
if($paramname eq "northBoundingCoordinate"){
|
| 101 |
2638
|
harris
|
$nBC = trimwhitespace($tempnode->getFirstChild()->getNodeValue());
|
| 102 |
2563
|
harris
|
}
|
| 103 |
|
|
if($paramname eq "northbc"){
|
| 104 |
2638
|
harris
|
$nBC = trimwhitespace($tempnode->getFirstChild()->getNodeValue());
|
| 105 |
2563
|
harris
|
}
|
| 106 |
|
|
if($paramname eq "southBoundingCoordinate"){
|
| 107 |
2638
|
harris
|
$sBC = trimwhitespace($tempnode->getFirstChild()->getNodeValue());
|
| 108 |
2563
|
harris
|
}
|
| 109 |
|
|
if($paramname eq "southbc"){
|
| 110 |
2638
|
harris
|
$sBC = trimwhitespace($tempnode->getFirstChild()->getNodeValue());
|
| 111 |
2563
|
harris
|
}
|
| 112 |
|
|
}
|
| 113 |
|
|
|
| 114 |
|
|
if($keyword ne "" ){
|
| 115 |
|
|
$keyword = substr ($keyword, 1);
|
| 116 |
|
|
}
|
| 117 |
|
|
if($cName ne "" ){
|
| 118 |
|
|
$cName = substr ($cName, 1);
|
| 119 |
|
|
}
|
| 120 |
|
|
|
| 121 |
|
|
# print "$name, $title, ($cName), ($keyword), ($eBC,$wBC,$nBC,$sBC), $url?action=read&docid=$name&qformat=knb, \n";
|
| 122 |
2638
|
harris
|
# print trimwhitespace($eBC) + " \n";
|
| 123 |
2563
|
harris
|
if ($eBC ne "" && $nBC ne "" && $eBC ne "0" && $nBC ne "0") {
|
| 124 |
2638
|
harris
|
print "$name $wBC $eBC $sBC $nBC $url?action=read&docid=$name&qformat=knb \n";
|
| 125 |
2563
|
harris
|
}
|
| 126 |
|
|
}
|
| 127 |
|
|
|
| 128 |
2638
|
harris
|
|
| 129 |
|
|
# Remove whitespace from the start and end of the string
|
| 130 |
|
|
sub trimwhitespace($)
|
| 131 |
|
|
{
|
| 132 |
|
|
my $string = shift;
|
| 133 |
|
|
$string =~ s/^\s+//;
|
| 134 |
|
|
$string =~ s/\s+$//;
|
| 135 |
|
|
return $string;
|
| 136 |
|
|
}
|
| 137 |
2563
|
harris
|
#print $numberNodes;
|