|
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 DataCubeStep extends BaseStep implements StepInterface { |
|
18
|
|
|
|
|
19
|
|
|
int i = 0; |
|
20
|
|
|
public static final String OBJ = "exProp:%s ex:%s;"; |
|
21
|
|
|
|
|
22
|
|
|
public DataCubeStep(StepMeta stepMeta, StepDataInterface stepDataInterface, int copyNr, TransMeta transMeta, |
|
23
|
|
|
Trans trans) { |
|
24
|
|
|
super(stepMeta, stepDataInterface, copyNr, transMeta, trans); |
|
25
|
|
|
} |
|
26
|
|
|
|
|
27
|
|
|
@Override |
|
28
|
|
|
public boolean init(StepMetaInterface smi, StepDataInterface sdi) { |
|
29
|
|
|
if (super.init(smi, sdi)) { |
|
30
|
|
|
return true; |
|
31
|
|
|
} else |
|
32
|
|
|
return false; |
|
33
|
|
|
} |
|
34
|
|
|
|
|
35
|
|
|
@Override |
|
36
|
|
|
public void dispose(StepMetaInterface smi, StepDataInterface sdi) { |
|
37
|
|
|
super.dispose(smi, sdi); |
|
38
|
|
|
} |
|
39
|
|
|
|
|
40
|
|
|
private String formatUri(String uri) { |
|
41
|
|
|
return "<" + uri + ">"; |
|
42
|
|
|
} |
|
43
|
|
|
|
|
44
|
|
|
private String formatPrefix(String prefix) { |
|
45
|
|
|
if (!prefix.equals("@base")){ |
|
46
|
|
|
return "@prefix " + prefix + ":"; |
|
47
|
|
|
}else { |
|
48
|
|
|
return prefix; |
|
49
|
|
|
} |
|
50
|
|
|
} |
|
51
|
|
|
|
|
52
|
|
|
/** |
|
53
|
|
|
* Método chamado para cada linha que entra no step |
|
54
|
|
|
*/ |
|
55
|
|
|
public boolean processRow(StepMetaInterface smi, StepDataInterface sdi) throws KettleException { |
|
56
|
|
|
|
|
57
|
|
|
DataCubeStepMeta meta = (DataCubeStepMeta) smi; |
|
58
|
|
|
DataCubeStepData data = (DataCubeStepData) sdi; |
|
59
|
|
|
|
|
60
|
|
|
Object[] row = getRow(); |
|
61
|
|
|
if (row == null) { |
|
62
|
|
|
setOutputDone(); |
|
63
|
|
|
return false; |
|
64
|
|
|
} |
|
65
|
|
|
|
|
66
|
|
|
if (first) { |
|
67
|
|
|
first = false; |
|
68
|
|
|
|
|
69
|
|
|
RowMetaInterface rowMeta = getInputRowMeta(); |
|
70
|
|
|
data.outputRowMeta = rowMeta.clone(); |
|
71
|
|
|
|
|
72
|
|
|
// Adiciona os metadados do step atual |
|
73
|
|
|
meta.getFields(data.outputRowMeta, getStepname(), null, null, this); |
|
74
|
|
|
|
|
75
|
|
|
DataTable<String> vocabularyTable = meta.getVocabularyTable(); |
|
76
|
|
|
for (int k = 0; k < vocabularyTable.size(); k++) { |
|
77
|
|
|
String prefix = vocabularyTable.getValue(k, DataCubeStepMeta.Field.VOCABULARY_TABLE_PREFIX.name()); |
|
78
|
|
|
String uri = vocabularyTable.getValue(k, DataCubeStepMeta.Field.VOCABULARY_TABLE_URI.name()); |
|
79
|
|
|
putOutRow(row, meta, data, this.formatPrefix(prefix) + " " + this.formatUri(uri) + " ."); |
|
80
|
|
|
} |
|
81
|
|
|
|
|
82
|
|
|
putOutRow(row, meta, data, "", |
|
83
|
|
|
"<http://purl.org/linked-data/cube> a owl:Ontology ;", |
|
84
|
|
|
"rdfs:label \"Example DataCube Knowledge Base\" ;", |
|
85
|
|
|
"dc:description \"This knowledgebase contains one Data Structure Definition with one Data Set. This Data Set has a couple of Components and Observations.\" .", |
|
86
|
|
|
"", |
|
87
|
|
|
"# Data Structure Definitions", |
|
88
|
|
|
"", |
|
89
|
|
|
"ex:dsd a cube:DataStructureDefinition ;", |
|
90
|
|
|
" rdfs:label \"A Data Structure Definition\"@en ;", |
|
91
|
|
|
" rdfs:comment \"Defines the structure of a DataSet or slice.\" ;"); |
|
92
|
|
|
|
|
93
|
|
|
DataTable<String> dimensionTable = meta.getDimensionTable(); |
|
94
|
|
|
for (int k = 0; k < dimensionTable.size(); k++) { |
|
95
|
|
|
String dimURI = dimensionTable.getValue(k, DataCubeStepMeta.Field.DIMENSION_TABLE_URI.name()); |
|
96
|
|
|
if (k == 0) { |
|
97
|
|
|
putOutRow(row, meta, data, " cube:component " + this.formatUri(dimURI) + ","); |
|
98
|
|
|
} else if (k == dimensionTable.size() - 1) { |
|
99
|
|
|
putOutRow(row, meta, data, " " + this.formatUri(dimURI) + " ."); |
|
100
|
|
|
} else { |
|
101
|
|
|
putOutRow(row, meta, data, " " + this.formatUri(dimURI) + ","); |
|
102
|
|
|
} |
|
103
|
|
|
} |
|
104
|
|
|
|
|
105
|
|
|
putOutRow(row, meta, data, "", "# Component Specifications", ""); |
|
106
|
|
|
|
|
107
|
|
|
for (int k = 0; k < dimensionTable.size(); k++) { |
|
108
|
|
|
String label = dimensionTable.getValue(k, DataCubeStepMeta.Field.DIMENSION_TABLE_LABEL.name()); |
|
109
|
|
|
String dimURI = dimensionTable.getValue(k, DataCubeStepMeta.Field.DIMENSION_TABLE_URI.name()); |
|
110
|
|
|
String dimName = dimensionTable.getValue(k, DataCubeStepMeta.Field.DIMENSION_TABLE_NAME.name()); |
|
111
|
|
|
putOutRow(row, meta, data, |
|
112
|
|
|
this.formatUri(dimURI) + " a cube:ComponentSpecification ;", |
|
113
|
|
|
" rdfs:label \"" + label + "\" ;", |
|
114
|
|
|
" cube: cube:dimension exProp:" + dimName + " .", ""); |
|
115
|
|
|
} |
|
116
|
|
|
|
|
117
|
|
|
putOutRow(row, meta, data, "# Data Set", |
|
118
|
|
|
"rdfs:label \"A DataSet\"^^<http://www.w3.org/2001/XMLSchema#string> ;", |
|
119
|
|
|
"rdfs:comment \"Represents a collection of observations and conforming to some common dimensional structure.\" ;", |
|
120
|
|
|
"cube:structure ex:dsd .", "", "# Dimensions, Unit and Measure"); |
|
121
|
|
|
|
|
122
|
|
|
for (int k = 0; k < dimensionTable.size(); k++) { |
|
123
|
|
|
String type = dimensionTable.getValue(k, DataCubeStepMeta.Field.DIMENSION_TABLE_TYPE.name()); |
|
124
|
|
|
String label = dimensionTable.getValue(k, DataCubeStepMeta.Field.DIMENSION_TABLE_LABEL.name()); |
|
125
|
|
|
String dimName = dimensionTable.getValue(k, DataCubeStepMeta.Field.DIMENSION_TABLE_NAME.name()); |
|
126
|
|
|
if (type.trim().equals("")) { // it's not a measure |
|
127
|
|
|
putOutRow(row, meta, data, |
|
128
|
|
|
"exProp:" + dimName + " a cube:DimensionProperty ;", |
|
129
|
|
|
" rdfs:label \"" + label + "\"@en .", ""); |
|
130
|
|
|
} else { |
|
131
|
|
|
putOutRow(row, meta, data, "exProp:unit a cube:AttributeProperty ;", |
|
132
|
|
|
" exProp:" + dimName + " a cube:MeasureProperty ;", |
|
133
|
|
|
" rdfs:label \"" + label + "\"@en .", ""); |
|
134
|
|
|
} |
|
135
|
|
|
} |
|
136
|
|
|
|
|
137
|
|
|
} |
|
138
|
|
|
|
|
139
|
|
|
putOutRow(row, meta, data, "ex:ob" + i + " a cube:Observation;", |
|
140
|
|
|
" cube:dataSet ex:dataset;"); |
|
141
|
|
|
|
|
142
|
|
|
DataTable<String> dimensionTable = meta.getDimensionTable(); |
|
143
|
|
|
for (int k = 0; k < dimensionTable.size(); k++) { |
|
144
|
|
|
String dimName = dimensionTable.getValue(k, DataCubeStepMeta.Field.DIMENSION_TABLE_NAME.name()); |
|
145
|
|
|
String type = dimensionTable.getValue(k, DataCubeStepMeta.Field.DIMENSION_TABLE_TYPE.name()); |
|
146
|
|
|
String label = dimensionTable.getValue(k, DataCubeStepMeta.Field.DIMENSION_TABLE_LABEL.name()); |
|
147
|
|
|
if (type.trim().equals("")) { |
|
148
|
|
|
putOutRow(row, meta, data, " " + String.format(OBJ, dimName, getInputRowMeta().getString(row, dimName, ""))); |
|
149
|
|
|
} else { |
|
150
|
|
|
putOutRow(row, meta, data, |
|
151
|
|
|
" exProp:unit \"" + label + "\";", |
|
152
|
|
|
" exProp:value \"" + getInputRowMeta().getString(row, dimName, "") + "\"^^" + type + ";"); |
|
153
|
|
|
} |
|
154
|
|
|
} |
|
155
|
|
|
|
|
156
|
|
|
putOutRow(row, meta, data, " rdfs:label \"\".", ""); |
|
157
|
|
|
|
|
158
|
|
|
i++; |
|
159
|
|
|
return true; |
|
160
|
|
|
} |
|
161
|
|
|
|
|
162
|
|
|
private void putOutRow(Object[] inputRow, DataCubeStepMeta meta, DataCubeStepData data, String... lines) |
|
163
|
|
|
throws KettleStepException { |
|
164
|
|
|
for (String arg : lines) { |
|
165
|
|
|
Object[] outputRow = null; |
|
166
|
|
|
outputRow = meta.isKeepInputFields() ? inputRow : new Object[0]; |
|
167
|
|
|
outputRow = RowDataUtil.addValueData(outputRow, outputRow.length, arg); |
|
168
|
|
|
putRow(data.outputRowMeta, outputRow); |
|
169
|
|
|
} |
|
170
|
|
|
} |
|
171
|
|
|
|
|
172
|
|
|
} |
johncurcio /
ETL4LODPlus
