|
1
|
|
|
package br.ufrj.ppgi.greco.kettle; |
|
2
|
|
|
|
|
3
|
|
|
import java.io.File; |
|
4
|
|
|
import java.nio.file.Paths; |
|
5
|
|
|
import java.util.Collection; |
|
6
|
|
|
import java.util.Iterator; |
|
7
|
|
|
|
|
8
|
|
|
import org.apache.jena.riot.Lang; |
|
9
|
|
|
import org.apache.jena.riot.RDFLanguages; |
|
10
|
|
|
import org.pentaho.di.core.exception.KettleException; |
|
11
|
|
|
import org.pentaho.di.core.exception.KettleStepException; |
|
12
|
|
|
import org.pentaho.di.core.row.RowDataUtil; |
|
13
|
|
|
import org.pentaho.di.core.row.RowMeta; |
|
14
|
|
|
import org.pentaho.di.core.row.RowMetaInterface; |
|
15
|
|
|
import org.pentaho.di.trans.Trans; |
|
16
|
|
|
import org.pentaho.di.trans.TransMeta; |
|
17
|
|
|
import org.pentaho.di.trans.step.BaseStep; |
|
18
|
|
|
import org.pentaho.di.trans.step.StepDataInterface; |
|
19
|
|
|
import org.pentaho.di.trans.step.StepInterface; |
|
20
|
|
|
import org.pentaho.di.trans.step.StepMeta; |
|
21
|
|
|
import org.pentaho.di.trans.step.StepMetaInterface; |
|
22
|
|
|
|
|
23
|
|
|
import org.apache.jena.ontology.OntClass; |
|
24
|
|
|
import org.apache.jena.ontology.OntModel; |
|
25
|
|
|
import org.apache.jena.ontology.OntModelSpec; |
|
26
|
|
|
import org.apache.jena.ontology.OntProperty; |
|
27
|
|
|
import org.apache.jena.rdf.model.ModelFactory; |
|
28
|
|
|
|
|
29
|
|
|
import br.ufrj.ppgi.greco.kettle.plugin.tools.datatable.DataTable; |
|
30
|
|
|
|
|
31
|
|
|
public class OwlInputStep extends BaseStep implements StepInterface { |
|
32
|
|
|
boolean repetir = true; |
|
33
|
|
|
public OntModel model = ModelFactory.createOntologyModel(OntModelSpec.OWL_MEM); |
|
34
|
|
|
|
|
35
|
|
|
public OwlInputStep(StepMeta s, StepDataInterface stepDataInterface, int c, TransMeta t, Trans dis) { |
|
36
|
|
|
super(s, stepDataInterface, c, t, dis); |
|
37
|
|
|
} |
|
38
|
|
|
|
|
39
|
|
|
public boolean init(StepMetaInterface smi, StepDataInterface sdi) { |
|
40
|
|
|
return super.init(smi, sdi); |
|
41
|
|
|
} |
|
42
|
|
|
|
|
43
|
|
|
public boolean processRow(StepMetaInterface smi, StepDataInterface sdi) throws KettleException { |
|
44
|
|
|
OwlInputStepMeta meta = (OwlInputStepMeta) smi; |
|
45
|
|
|
OwlInputStepData data = (OwlInputStepData) sdi; |
|
46
|
|
|
|
|
47
|
|
|
Object[] row = getRow(); |
|
48
|
|
|
|
|
49
|
|
|
if (first) { |
|
50
|
|
|
first = false; |
|
51
|
|
|
RowMetaInterface rowMeta = getInputRowMeta(row != null); |
|
52
|
|
|
data.outputRowMeta = rowMeta.clone(); |
|
53
|
|
|
meta.getFields(data.outputRowMeta, getStepname(), null, null, this); |
|
54
|
|
|
} |
|
55
|
|
|
|
|
56
|
|
|
DataTable<String> table = meta.getVocabTable(); |
|
57
|
|
|
boolean isTableEmpty = table.getValue(0, OwlInputStepMeta.Field.VOCAB_TABLE_URI.name()).isEmpty(); |
|
58
|
|
|
|
|
59
|
|
|
if (!isTableEmpty){ |
|
60
|
|
|
for (int k = 0; k < table.size(); k++) { |
|
61
|
|
|
putOutRow(row, meta, data, |
|
62
|
|
|
table.getValue(k, OwlInputStepMeta.Field.VOCAB_TABLE_PREFIX.name()), |
|
63
|
|
|
table.getValue(k, OwlInputStepMeta.Field.VOCAB_TABLE_URI.name()), |
|
64
|
|
|
table.getValue(k, OwlInputStepMeta.Field.VOCAB_TABLE_PROPERTY.name()), |
|
65
|
|
|
table.getValue(k, OwlInputStepMeta.Field.VOCAB_TABLE_TYPE.name())); |
|
66
|
|
|
} |
|
67
|
|
|
} else { |
|
68
|
|
|
table = meta.getMapTable(); |
|
69
|
|
|
for (int k = 0; k < table.size(); k++) { |
|
70
|
|
|
String owlFile = getOwlFile(table.getValue(k, OwlInputStepMeta.Field.MAP_TABLE_ONTOLOGY_URI.name())); |
|
71
|
|
|
try { |
|
72
|
|
|
model = ModelFactory.createOntologyModel(OntModelSpec.OWL_MEM); |
|
73
|
|
|
this.logBasic("Attempting to read " + owlFile + " as RDF/XML"); |
|
74
|
|
|
model.read(owlFile); |
|
75
|
|
|
this.logBasic(owlFile + " has been read successfully"); |
|
76
|
|
|
} catch (Exception eox) { |
|
77
|
|
|
this.logBasic("Error reading " + owlFile + " as RDF/XML: " + eox.getMessage()); |
|
78
|
|
|
Collection<Lang> registeredLanguages = RDFLanguages.getRegisteredLanguages(); |
|
79
|
|
|
for (Lang c : registeredLanguages) { |
|
80
|
|
|
try { |
|
81
|
|
|
this.logBasic("Trying to read the file as... " + c.getName()); |
|
82
|
|
|
model.read(owlFile, c.getName()); |
|
83
|
|
|
this.logBasic(owlFile + " has been read successfully"); |
|
84
|
|
|
break; |
|
85
|
|
|
} catch (Exception e) { |
|
86
|
|
|
this.logBasic("File could not be read as " + c.getName() + ": " + e.getMessage()); |
|
87
|
|
|
} |
|
88
|
|
|
} |
|
89
|
|
|
} |
|
90
|
|
|
|
|
91
|
|
|
if (!model.isEmpty()) { |
|
92
|
|
|
String ontoField = table.getValue(k, OwlInputStepMeta.Field.MAP_TABLE_ONTOLOGY_NAME.name()); |
|
93
|
|
|
for (Iterator<OntClass> i = model.listClasses(); i.hasNext();) { |
|
94
|
|
|
OntClass cls = i.next(); |
|
95
|
|
|
if (cls.getLocalName() != null) { |
|
96
|
|
|
putOutRow(row, meta, data, |
|
97
|
|
|
ontoField.trim(), cls.getURI(), "rdf:type", "rdfs:class"); |
|
98
|
|
|
} |
|
99
|
|
|
} |
|
100
|
|
|
|
|
101
|
|
|
for (Iterator<OntProperty> j = model.listAllOntProperties(); j.hasNext();) { |
|
102
|
|
|
OntProperty proper = j.next(); |
|
103
|
|
|
if (proper.getLocalName() != null) { |
|
104
|
|
|
putOutRow(row, meta, data, |
|
105
|
|
|
ontoField.trim(), proper.getURI(), "rdf:type", "rdfs:property"); |
|
106
|
|
|
} |
|
107
|
|
|
} |
|
108
|
|
|
} |
|
109
|
|
|
} |
|
110
|
|
|
} |
|
111
|
|
|
|
|
112
|
|
|
if (row == null){ |
|
113
|
|
|
setOutputDone(); |
|
114
|
|
|
return false; |
|
115
|
|
|
} |
|
116
|
|
|
|
|
117
|
|
|
return true; |
|
118
|
|
|
} |
|
119
|
|
|
|
|
120
|
|
|
/** |
|
121
|
|
|
* Verifies whether src is an actual file or and URI. |
|
122
|
|
|
* If the String src is a file, return it's valid path |
|
123
|
|
|
* else return the URI |
|
124
|
|
|
* |
|
125
|
|
|
* @param src the original file path |
|
126
|
|
|
* @return a valid file path or an URI |
|
127
|
|
|
*/ |
|
128
|
|
|
private String getOwlFile(String src) { |
|
129
|
|
|
File file = new File(src); |
|
130
|
|
|
if (!file.isDirectory()){ |
|
131
|
|
|
file = file.getParentFile(); |
|
132
|
|
|
} |
|
133
|
|
|
if (file.exists()){ |
|
134
|
|
|
return Paths.get(src).toUri().toString(); |
|
135
|
|
|
}else { |
|
136
|
|
|
return src; |
|
137
|
|
|
} |
|
138
|
|
|
} |
|
139
|
|
|
|
|
140
|
|
|
public void dispose(StepMetaInterface smi, StepDataInterface sdi) { |
|
141
|
|
|
super.dispose(smi, sdi); |
|
142
|
|
|
} |
|
143
|
|
|
|
|
144
|
|
|
private void putOutRow(Object[] inputRow, OwlInputStepMeta meta, OwlInputStepData data, String... var) throws KettleStepException { |
|
145
|
|
|
|
|
146
|
|
|
int outputRowPos = 0; |
|
147
|
|
|
Object[] outputRow = null; |
|
148
|
|
|
|
|
149
|
|
|
if (meta.isKeepInputFields()) { |
|
150
|
|
|
outputRow = inputRow; |
|
151
|
|
|
outputRowPos = getInputRowMeta().size(); |
|
152
|
|
|
} else { |
|
153
|
|
|
outputRow = new Object[4]; |
|
154
|
|
|
} |
|
155
|
|
|
|
|
156
|
|
|
if (outputRow != null){ |
|
157
|
|
|
for (String arg : var) { |
|
158
|
|
|
outputRow = RowDataUtil.addValueData(outputRow, outputRowPos++, arg); |
|
159
|
|
|
} |
|
160
|
|
|
|
|
161
|
|
|
putRow(data.outputRowMeta, outputRow); |
|
162
|
|
|
} |
|
163
|
|
|
} |
|
164
|
|
|
|
|
165
|
|
|
private RowMetaInterface getInputRowMeta(boolean hasInputRow) { |
|
166
|
|
|
|
|
167
|
|
|
RowMetaInterface rowMeta = null; |
|
168
|
|
|
if (hasInputRow) |
|
169
|
|
|
rowMeta = getInputRowMeta(); |
|
170
|
|
|
else |
|
171
|
|
|
rowMeta = new RowMeta(); |
|
172
|
|
|
|
|
173
|
|
|
return rowMeta; |
|
174
|
|
|
} |
|
175
|
|
|
|
|
176
|
|
|
} |
|
177
|
|
|
|
johncurcio /
ETL4LODPlus
