| 1 |
3814
|
tao
|
package edu.ucsb.nceas.metacat;
|
| 2 |
|
|
|
| 3 |
3816
|
tao
|
import java.sql.Statement;
|
| 4 |
|
|
|
| 5 |
|
|
import org.apache.log4j.Logger;
|
| 6 |
|
|
|
| 7 |
5015
|
daigle
|
import edu.ucsb.nceas.metacat.database.DBConnection;
|
| 8 |
|
|
import edu.ucsb.nceas.metacat.database.DBConnectionPool;
|
| 9 |
5030
|
daigle
|
import edu.ucsb.nceas.metacat.properties.PropertyService;
|
| 10 |
3816
|
tao
|
|
| 11 |
3814
|
tao
|
/**
|
| 12 |
|
|
* Before Metacat 1.8.1 release, Metacat uses the eml201 schema with the tag
|
| 13 |
|
|
* RELEASE_EML_2_0_1_UPDATE_5. Unfortunately, this tag points at wrong version
|
| 14 |
|
|
* of eml-resource.xsd. In this schema, the element "references" has an attribute named
|
| 15 |
|
|
* "system" and the attribute has a default value "document". Metacat will add
|
| 16 |
|
|
* the attribute system="document" to "references" element even the orginal eml didn't have it
|
| 17 |
|
|
* (this is another bug and see bug 1601), so this causes metacat generated some invalid eml201
|
| 18 |
|
|
* documents. This class provides a path to fix the existed invalid eml201 documents. It will
|
| 19 |
|
|
* remove the attribute system="document" of the element "references" in xml_nodes and xml_index
|
| 20 |
|
|
* tables.
|
| 21 |
|
|
* @author tao
|
| 22 |
|
|
*
|
| 23 |
|
|
*/
|
| 24 |
3827
|
tao
|
public class EML201DocumentCorrector
|
| 25 |
3814
|
tao
|
{
|
| 26 |
3816
|
tao
|
private Logger logMetacat = Logger.getLogger(EML201DocumentCorrector.class);
|
| 27 |
|
|
|
| 28 |
3814
|
tao
|
/**
|
| 29 |
|
|
* Default constructor
|
| 30 |
|
|
*
|
| 31 |
|
|
*/
|
| 32 |
|
|
public EML201DocumentCorrector()
|
| 33 |
|
|
{
|
| 34 |
|
|
|
| 35 |
|
|
}
|
| 36 |
|
|
|
| 37 |
|
|
/**
|
| 38 |
|
|
* It will remove the records - attribute system="document" of element "refrence"
|
| 39 |
3816
|
tao
|
* in both xml_nodes and xml_index table. Since xml_index has a foreign key (nodeid)which
|
| 40 |
|
|
* references nodeid in xml_nodes table, we should delete records in xml_index table first.
|
| 41 |
3814
|
tao
|
*/
|
| 42 |
3858
|
tao
|
public boolean run()
|
| 43 |
3814
|
tao
|
{
|
| 44 |
3816
|
tao
|
DBConnection dbconn = null;
|
| 45 |
3858
|
tao
|
boolean success = false;
|
| 46 |
3816
|
tao
|
int serialNumber = 0;
|
| 47 |
|
|
try
|
| 48 |
|
|
{
|
| 49 |
|
|
|
| 50 |
|
|
//checkout the dbconnection
|
| 51 |
3864
|
tao
|
dbconn = DBConnectionPool.getDBConnection("EML201DocumentCorrector.run");
|
| 52 |
3816
|
tao
|
serialNumber = dbconn.getCheckOutSerialNumber();
|
| 53 |
|
|
Statement deletingStatement = dbconn.createStatement();
|
| 54 |
|
|
|
| 55 |
|
|
// delete the records in xml_index table
|
| 56 |
|
|
String deletingIndex = generateXML_IndexDeletingSQL();
|
| 57 |
|
|
deletingStatement.execute(deletingIndex);
|
| 58 |
|
|
|
| 59 |
|
|
// delete the records in xml_nodes table
|
| 60 |
|
|
String deletingNode = generateXML_NodeDeletingSQL();
|
| 61 |
|
|
deletingStatement.execute(deletingNode);
|
| 62 |
|
|
|
| 63 |
3827
|
tao
|
// delete the records in xml_nodes_revisions table
|
| 64 |
|
|
String deletingNodeRevision = generateXML_Node_RevisionsDeletingSQL();
|
| 65 |
|
|
deletingStatement.execute(deletingNodeRevision);
|
| 66 |
|
|
|
| 67 |
3816
|
tao
|
//close statement and connection
|
| 68 |
|
|
deletingStatement.close();
|
| 69 |
3864
|
tao
|
//dbconn.close();
|
| 70 |
3858
|
tao
|
success = true;
|
| 71 |
3816
|
tao
|
}
|
| 72 |
|
|
catch (Exception ee)
|
| 73 |
|
|
{
|
| 74 |
3864
|
tao
|
logMetacat.error("EML201DocumentCorrector.run: "
|
| 75 |
3816
|
tao
|
+ ee.getMessage());
|
| 76 |
|
|
ee.printStackTrace();
|
| 77 |
|
|
}
|
| 78 |
|
|
finally
|
| 79 |
|
|
{
|
| 80 |
|
|
DBConnectionPool.returnDBConnection(dbconn, serialNumber);
|
| 81 |
|
|
} //finally
|
| 82 |
3858
|
tao
|
return success;
|
| 83 |
3814
|
tao
|
}
|
| 84 |
|
|
|
| 85 |
|
|
/*
|
| 86 |
3816
|
tao
|
* Generate the sql command to delete the records in xml_node table.
|
| 87 |
|
|
* Since it is leaf node, so we can just delete it without any other side-effect.
|
| 88 |
3814
|
tao
|
*/
|
| 89 |
|
|
private String generateXML_NodeDeletingSQL()
|
| 90 |
|
|
{
|
| 91 |
3922
|
tao
|
String sql ="delete from xml_nodes where nodetype='ATTRIBUTE' and nodename='system' "+
|
| 92 |
3814
|
tao
|
"and parentnodeid in (select nodeid from xml_nodes where nodetype='ELEMENT' and nodename='references') and docid in "+
|
| 93 |
3816
|
tao
|
"(select docid from xml_documents where doctype ='eml://ecoinformatics.org/eml-2.0.1')";
|
| 94 |
3814
|
tao
|
return sql;
|
| 95 |
|
|
}
|
| 96 |
|
|
|
| 97 |
|
|
/*
|
| 98 |
3827
|
tao
|
* Generate the sql command to delete the records in xml_node table.
|
| 99 |
|
|
* Since it is leaf node, so we can just delete it without any other side-effect.
|
| 100 |
|
|
*/
|
| 101 |
|
|
private String generateXML_Node_RevisionsDeletingSQL()
|
| 102 |
|
|
{
|
| 103 |
3922
|
tao
|
String sql ="delete from xml_nodes_revisions where nodetype='ATTRIBUTE' and nodename='system' "+
|
| 104 |
3917
|
tao
|
"and parentnodeid in (select nodeid from xml_nodes_revisions where nodetype='ELEMENT' and nodename='references') and docid in "+
|
| 105 |
|
|
"(select docid from xml_revisions where doctype ='eml://ecoinformatics.org/eml-2.0.1')";
|
| 106 |
3827
|
tao
|
return sql;
|
| 107 |
|
|
}
|
| 108 |
|
|
|
| 109 |
|
|
/*
|
| 110 |
3814
|
tao
|
* Generate the sql command to delete the records in xml_nidex table;
|
| 111 |
|
|
*/
|
| 112 |
|
|
private String generateXML_IndexDeletingSQL()
|
| 113 |
|
|
{
|
| 114 |
3818
|
tao
|
String sql ="delete from xml_index where doctype ='eml://ecoinformatics.org/eml-2.0.1' AND nodeid in "+
|
| 115 |
|
|
"(select nodeid from xml_index where path ='references/@system')";
|
| 116 |
3814
|
tao
|
return sql;
|
| 117 |
|
|
}
|
| 118 |
3816
|
tao
|
|
| 119 |
|
|
/**
|
| 120 |
|
|
* Runs the job to correct eml201 documents - deleting extral nodes in
|
| 121 |
|
|
* @param argus
|
| 122 |
|
|
* @throws Exception
|
| 123 |
|
|
*/
|
| 124 |
5030
|
daigle
|
public static void main(String[] args) throws Exception
|
| 125 |
3816
|
tao
|
{
|
| 126 |
|
|
|
| 127 |
|
|
//initialize options and connection pool
|
| 128 |
5030
|
daigle
|
PropertyService.getInstance(args[0]);
|
| 129 |
3816
|
tao
|
DBConnectionPool connPool = DBConnectionPool.getInstance();
|
| 130 |
4124
|
daigle
|
|
| 131 |
3816
|
tao
|
// run the thread
|
| 132 |
|
|
EML201DocumentCorrector correct = new EML201DocumentCorrector();
|
| 133 |
3827
|
tao
|
//Thread thread = new Thread(correct);
|
| 134 |
|
|
//thread.start();
|
| 135 |
|
|
correct.run();
|
| 136 |
3816
|
tao
|
}
|
| 137 |
3814
|
tao
|
}
|