|
1
|
|
|
package br.ufrj.ppgi.greco.kettle; |
|
2
|
|
|
|
|
3
|
|
|
import org.pentaho.di.core.exception.KettleException; |
|
4
|
|
|
import org.pentaho.di.core.exception.KettleStepException; |
|
5
|
|
|
import org.pentaho.di.core.row.RowDataUtil; |
|
6
|
|
|
import org.pentaho.di.core.row.RowMetaInterface; |
|
7
|
|
|
import org.pentaho.di.trans.Trans; |
|
8
|
|
|
import org.pentaho.di.trans.TransMeta; |
|
9
|
|
|
import org.pentaho.di.trans.step.BaseStep; |
|
10
|
|
|
import org.pentaho.di.trans.step.StepDataInterface; |
|
11
|
|
|
import org.pentaho.di.trans.step.StepInterface; |
|
12
|
|
|
import org.pentaho.di.trans.step.StepMeta; |
|
13
|
|
|
import org.pentaho.di.trans.step.StepMetaInterface; |
|
14
|
|
|
|
|
15
|
|
|
import br.ufrj.ppgi.greco.kettle.plugin.tools.datatable.DataTable; |
|
16
|
|
|
|
|
17
|
|
|
public class TurtleGeneratorStep extends BaseStep implements StepInterface { |
|
18
|
|
|
|
|
19
|
|
|
int j = 0; |
|
20
|
|
|
|
|
21
|
|
|
public TurtleGeneratorStep(StepMeta stepMeta, StepDataInterface stepDataInterface, int copyNr, TransMeta transMeta, |
|
22
|
|
|
Trans trans) { |
|
23
|
|
|
super(stepMeta, stepDataInterface, copyNr, transMeta, trans); |
|
24
|
|
|
} |
|
25
|
|
|
|
|
26
|
|
|
@Override |
|
27
|
|
|
public boolean init(StepMetaInterface smi, StepDataInterface sdi) { |
|
28
|
|
|
if (super.init(smi, sdi)) { |
|
29
|
|
|
return true; |
|
30
|
|
|
} else { |
|
31
|
|
|
return false; |
|
32
|
|
|
} |
|
33
|
|
|
} |
|
34
|
|
|
|
|
35
|
|
|
@Override |
|
36
|
|
|
public void dispose(StepMetaInterface smi, StepDataInterface sdi) { |
|
37
|
|
|
super.dispose(smi, sdi); |
|
38
|
|
|
} |
|
39
|
|
|
|
|
40
|
|
|
/** |
|
41
|
|
|
* Metodo chamado para cada linha que entra no step |
|
42
|
|
|
*/ |
|
43
|
|
|
public boolean processRow(StepMetaInterface smi, StepDataInterface sdi) throws KettleException { |
|
44
|
|
|
|
|
45
|
|
|
TurtleGeneratorStepMeta meta = (TurtleGeneratorStepMeta) smi; |
|
46
|
|
|
TurtleGeneratorStepData data = (TurtleGeneratorStepData) sdi; |
|
47
|
|
|
|
|
48
|
|
|
// Obtem linha do fluxo de entrada e termina caso nao haja mais entrada |
|
49
|
|
|
Object[] row = getRow(); |
|
50
|
|
|
if (row == null) { |
|
51
|
|
|
setOutputDone(); |
|
52
|
|
|
return false; |
|
53
|
|
|
} |
|
54
|
|
|
|
|
55
|
|
|
if (first) { |
|
56
|
|
|
first = false; |
|
57
|
|
|
|
|
58
|
|
|
RowMetaInterface rowMeta = getInputRowMeta(); |
|
59
|
|
|
data.outputRowMeta = rowMeta.clone(); |
|
60
|
|
|
|
|
61
|
|
|
// Adiciona os metadados do step atual |
|
62
|
|
|
meta.getFields(data.outputRowMeta, getStepname(), null, null, this); |
|
63
|
|
|
|
|
64
|
|
|
// Prefixos |
|
65
|
|
|
String prefix = meta.getPrefixes().toString(); |
|
66
|
|
|
prefix = prefix.replace("[", ""); |
|
67
|
|
|
prefix = prefix.replace(", ", " <"); |
|
68
|
|
|
prefix = prefix.replace("]", ">. "); |
|
69
|
|
|
prefix = prefix.replace("<@", "@"); |
|
70
|
|
|
prefix = prefix.replace(">. >.", ">."); |
|
71
|
|
|
prefix = prefix.replace(" <>. ", ""); |
|
72
|
|
|
prefix = prefix.replace(". ", "." + System.getProperty("line.separator")); |
|
73
|
|
|
|
|
74
|
|
|
putOutRow(row, meta, data, prefix); |
|
75
|
|
|
putOutRow(row, meta, data, ""); |
|
76
|
|
|
putOutRow(row, meta, data, "#"); |
|
77
|
|
|
putOutRow(row, meta, data, "# PROPERTIES DEFINITION"); |
|
78
|
|
|
putOutRow(row, meta, data, "#"); |
|
79
|
|
|
putOutRow(row, meta, data, ""); |
|
80
|
|
|
|
|
81
|
|
|
DataTable<String> table = meta.getMapTable(); |
|
82
|
|
|
for (int i = 0; i < table.size(); i++) { |
|
83
|
|
|
String dimensionField = table.getValue(i, |
|
84
|
|
|
TurtleGeneratorStepMeta.Field.MAP_TABLE_DIMENSIONS_FIELD_NAME.name()); |
|
85
|
|
|
String labelDimensao = table.getValue(i, TurtleGeneratorStepMeta.Field.MAP_TABLE_LABELS_FIELD_NAME.name()); |
|
86
|
|
|
String dimensionURIType = table.getValue(i, |
|
87
|
|
|
TurtleGeneratorStepMeta.Field.MAP_TABLE_URI_TYPE_FIELD_NAME.name()); |
|
88
|
|
|
if (dimensionField != null) { |
|
89
|
|
|
putOutRow(row, meta, data, "exProp:" + removeSignals(dimensionField).toLowerCase() + " owl:sameAs <" + dimensionURIType + ">;"); |
|
90
|
|
|
putOutRow(row, meta, data, " rdfs:label \"" + labelDimensao + "\"@en ."); |
|
91
|
|
|
putOutRow(row, meta, data, ""); |
|
92
|
|
|
} |
|
93
|
|
|
} |
|
94
|
|
|
|
|
95
|
|
|
if (TurtleGeneratorStepMeta.Field.MAP_TABLE_HIERARCHY_FIELD_NAME.name() != null) { |
|
96
|
|
|
|
|
97
|
|
|
putOutRow(row, meta, data, "#"); |
|
98
|
|
|
putOutRow(row, meta, data, "# HIERARCHIES"); |
|
99
|
|
|
putOutRow(row, meta, data, "#"); |
|
100
|
|
|
putOutRow(row, meta, data, ""); |
|
101
|
|
|
|
|
102
|
|
|
table = meta.getMapTable2(); |
|
103
|
|
|
for (int i = 0; i < table.size(); i++) { |
|
104
|
|
|
String hierarquia = table.getValue(i, TurtleGeneratorStepMeta.Field.MAP_TABLE_HIERARCHY_FIELD_NAME.name()); |
|
105
|
|
|
String hierarquiaDe = table.getValue(i, |
|
106
|
|
|
TurtleGeneratorStepMeta.Field.MAP_TABLE_HIERARCHY_DE_FIELD_NAME.name()); |
|
107
|
|
|
String hierarquiaLabel = table.getValue(i, |
|
108
|
|
|
TurtleGeneratorStepMeta.Field.MAP_TABLE_HIERARCHY_LABEL_FIELD_NAME.name()); |
|
109
|
|
|
String hierarquiaPara = table.getValue(i, |
|
110
|
|
|
TurtleGeneratorStepMeta.Field.MAP_TABLE_HIERARCHY_PARA_FIELD_NAME.name()); |
|
111
|
|
|
if (hierarquia != null && hierarquiaDe != null && hierarquiaPara != null) { |
|
112
|
|
|
putOutRow(row, meta, data, "ex:" + removeSignals(hierarquiaDe).toLowerCase() + " a ex:" |
|
113
|
|
|
+ removeSignals(hierarquia).toLowerCase() + " ;"); |
|
114
|
|
|
putOutRow(row, meta, data, " obo:part_of ex:" + removeSignals(hierarquiaPara) + ";"); |
|
115
|
|
|
putOutRow(row, meta, data, " rdfs:label \"" + hierarquiaLabel + "\"@en ."); |
|
116
|
|
|
putOutRow(row, meta, data, ""); |
|
117
|
|
|
} |
|
118
|
|
|
} |
|
119
|
|
|
} |
|
120
|
|
|
} |
|
121
|
|
|
|
|
122
|
|
|
DataTable<String> table = meta.getMapTable(); |
|
123
|
|
|
String line = ""; |
|
124
|
|
|
|
|
125
|
|
|
for (int i = 0; i < table.size(); i++) { |
|
126
|
|
|
String dimensionField = table.getValue(i, TurtleGeneratorStepMeta.Field.MAP_TABLE_DIMENSIONS_FIELD_NAME.name()); |
|
127
|
|
|
String dimension = getInputRowMeta().getString(row, dimensionField, ""); |
|
128
|
|
|
if (!dimension.startsWith("http")) { |
|
129
|
|
|
if (i == 0) { |
|
130
|
|
|
line = " exProp:" + removeSignals(dimensionField).toLowerCase() + " " + "\"" + dimension + "\";"; |
|
131
|
|
|
} else |
|
132
|
|
|
line = " exProp:" + removeSignals(dimensionField).toLowerCase() + " " + "\"" + dimension + "\";"; |
|
133
|
|
|
} else { |
|
134
|
|
|
line = " exProp:" + removeSignals(dimensionField).toLowerCase() + " " + "<" + dimension + ">;"; |
|
135
|
|
|
} |
|
136
|
|
|
if (i == 0) { |
|
137
|
|
|
line = "ex:obj" + j + line; |
|
138
|
|
|
} |
|
139
|
|
|
putOutRow(row, meta, data, line); |
|
140
|
|
|
} |
|
141
|
|
|
|
|
142
|
|
|
putOutRow(row, meta, data, " rdfs:label \"" + meta.getunity() + "\" ."); |
|
143
|
|
|
putOutRow(row, meta, data, ""); |
|
144
|
|
|
|
|
145
|
|
|
j++; |
|
146
|
|
|
|
|
147
|
|
|
return true; |
|
148
|
|
|
} |
|
149
|
|
|
|
|
150
|
|
|
private void putOutRow(Object[] inputRow, TurtleGeneratorStepMeta meta, TurtleGeneratorStepData data, String turtle) |
|
151
|
|
|
throws KettleStepException { |
|
152
|
|
|
|
|
153
|
|
|
Object[] outputRow = null; |
|
154
|
|
|
|
|
155
|
|
|
outputRow = new Object[0]; |
|
156
|
|
|
|
|
157
|
|
|
outputRow = RowDataUtil.addValueData(outputRow, outputRow.length, turtle); |
|
158
|
|
|
// Coloca linha no fluxo |
|
159
|
|
|
putRow(data.outputRowMeta, outputRow); |
|
160
|
|
|
} |
|
161
|
|
|
|
|
162
|
|
|
private static String removeSignals(String value) { |
|
163
|
|
|
if (value != null) { |
|
164
|
|
|
return value.replaceAll("á", "a").replaceAll("à", "a").replaceAll("ä", "a").replaceAll("ã", "a") |
|
165
|
|
|
.replaceAll("ú", "u").replaceAll("ù", "u").replaceAll("é", "e").replaceAll("è", "e") |
|
166
|
|
|
.replaceAll("ó", "o").replaceAll("ò", "o").replaceAll("ú", "u").replaceAll("ç", "c") |
|
167
|
|
|
.replaceAll("í", "i").replaceAll("ì", "i").replaceAll(" ", "").trim(); |
|
168
|
|
|
} else { |
|
169
|
|
|
return ""; |
|
170
|
|
|
} |
|
171
|
|
|
} |
|
172
|
|
|
|
|
173
|
|
|
} |
|
174
|
|
|
|