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 |
5170
|
daigle
|
logMetacat.debug("EML201DocumentCorrector.run - deleting the records in xml_index table with sql: " + deletingIndex);
|
58 |
3816
|
tao
|
deletingStatement.execute(deletingIndex);
|
59 |
|
|
|
60 |
|
|
// delete the records in xml_nodes table
|
61 |
|
|
String deletingNode = generateXML_NodeDeletingSQL();
|
62 |
5170
|
daigle
|
logMetacat.debug("EML201DocumentCorrector.run - deleting the records in xml_nodes table with sql: " + deletingNode);
|
63 |
3816
|
tao
|
deletingStatement.execute(deletingNode);
|
64 |
|
|
|
65 |
3827
|
tao
|
// delete the records in xml_nodes_revisions table
|
66 |
|
|
String deletingNodeRevision = generateXML_Node_RevisionsDeletingSQL();
|
67 |
5170
|
daigle
|
logMetacat.debug("EML201DocumentCorrector.run - deleting the records in xml_nodes_revisions table with sql: " + deletingNodeRevision);
|
68 |
3827
|
tao
|
deletingStatement.execute(deletingNodeRevision);
|
69 |
|
|
|
70 |
3816
|
tao
|
//close statement and connection
|
71 |
|
|
deletingStatement.close();
|
72 |
3864
|
tao
|
//dbconn.close();
|
73 |
3858
|
tao
|
success = true;
|
74 |
3816
|
tao
|
}
|
75 |
|
|
catch (Exception ee)
|
76 |
|
|
{
|
77 |
3864
|
tao
|
logMetacat.error("EML201DocumentCorrector.run: "
|
78 |
3816
|
tao
|
+ ee.getMessage());
|
79 |
|
|
ee.printStackTrace();
|
80 |
|
|
}
|
81 |
|
|
finally
|
82 |
|
|
{
|
83 |
|
|
DBConnectionPool.returnDBConnection(dbconn, serialNumber);
|
84 |
|
|
} //finally
|
85 |
3858
|
tao
|
return success;
|
86 |
3814
|
tao
|
}
|
87 |
|
|
|
88 |
|
|
/*
|
89 |
3816
|
tao
|
* Generate the sql command to delete the records in xml_node table.
|
90 |
|
|
* Since it is leaf node, so we can just delete it without any other side-effect.
|
91 |
3814
|
tao
|
*/
|
92 |
|
|
private String generateXML_NodeDeletingSQL()
|
93 |
|
|
{
|
94 |
3922
|
tao
|
String sql ="delete from xml_nodes where nodetype='ATTRIBUTE' and nodename='system' "+
|
95 |
3814
|
tao
|
"and parentnodeid in (select nodeid from xml_nodes where nodetype='ELEMENT' and nodename='references') and docid in "+
|
96 |
3816
|
tao
|
"(select docid from xml_documents where doctype ='eml://ecoinformatics.org/eml-2.0.1')";
|
97 |
3814
|
tao
|
return sql;
|
98 |
|
|
}
|
99 |
|
|
|
100 |
|
|
/*
|
101 |
3827
|
tao
|
* Generate the sql command to delete the records in xml_node table.
|
102 |
|
|
* Since it is leaf node, so we can just delete it without any other side-effect.
|
103 |
|
|
*/
|
104 |
|
|
private String generateXML_Node_RevisionsDeletingSQL()
|
105 |
|
|
{
|
106 |
3922
|
tao
|
String sql ="delete from xml_nodes_revisions where nodetype='ATTRIBUTE' and nodename='system' "+
|
107 |
3917
|
tao
|
"and parentnodeid in (select nodeid from xml_nodes_revisions where nodetype='ELEMENT' and nodename='references') and docid in "+
|
108 |
|
|
"(select docid from xml_revisions where doctype ='eml://ecoinformatics.org/eml-2.0.1')";
|
109 |
3827
|
tao
|
return sql;
|
110 |
|
|
}
|
111 |
|
|
|
112 |
|
|
/*
|
113 |
3814
|
tao
|
* Generate the sql command to delete the records in xml_nidex table;
|
114 |
|
|
*/
|
115 |
|
|
private String generateXML_IndexDeletingSQL()
|
116 |
|
|
{
|
117 |
3818
|
tao
|
String sql ="delete from xml_index where doctype ='eml://ecoinformatics.org/eml-2.0.1' AND nodeid in "+
|
118 |
|
|
"(select nodeid from xml_index where path ='references/@system')";
|
119 |
3814
|
tao
|
return sql;
|
120 |
|
|
}
|
121 |
3816
|
tao
|
|
122 |
|
|
/**
|
123 |
|
|
* Runs the job to correct eml201 documents - deleting extral nodes in
|
124 |
|
|
* @param argus
|
125 |
|
|
* @throws Exception
|
126 |
|
|
*/
|
127 |
5030
|
daigle
|
public static void main(String[] args) throws Exception
|
128 |
3816
|
tao
|
{
|
129 |
|
|
|
130 |
|
|
//initialize options and connection pool
|
131 |
5030
|
daigle
|
PropertyService.getInstance(args[0]);
|
132 |
3816
|
tao
|
DBConnectionPool connPool = DBConnectionPool.getInstance();
|
133 |
4124
|
daigle
|
|
134 |
3816
|
tao
|
// run the thread
|
135 |
|
|
EML201DocumentCorrector correct = new EML201DocumentCorrector();
|
136 |
3827
|
tao
|
//Thread thread = new Thread(correct);
|
137 |
|
|
//thread.start();
|
138 |
|
|
correct.run();
|
139 |
3816
|
tao
|
}
|
140 |
3814
|
tao
|
}
|