|
1
|
package eu.dnetlib.data.collective.transformation.engine;
|
|
2
|
|
|
3
|
import java.io.StringReader;
|
|
4
|
import java.util.HashMap;
|
|
5
|
import java.util.LinkedList;
|
|
6
|
import java.util.List;
|
|
7
|
import java.util.Map;
|
|
8
|
|
|
9
|
import org.apache.commons.logging.Log;
|
|
10
|
import org.apache.commons.logging.LogFactory;
|
|
11
|
import org.dom4j.Document;
|
|
12
|
import org.dom4j.DocumentException;
|
|
13
|
import org.dom4j.DocumentHelper;
|
|
14
|
import org.dom4j.Node;
|
|
15
|
import org.dom4j.XPath;
|
|
16
|
import org.dom4j.io.SAXReader;
|
|
17
|
|
|
18
|
import eu.dnetlib.data.collective.transformation.core.xsl.ext.TransformationFunctionProxy;
|
|
19
|
import eu.dnetlib.data.collective.transformation.engine.functions.Convert;
|
|
20
|
import eu.dnetlib.data.collective.transformation.engine.functions.Extract;
|
|
21
|
import eu.dnetlib.data.collective.transformation.engine.functions.ProcessingException;
|
|
22
|
import eu.dnetlib.data.collective.transformation.engine.functions.RegularExpression;
|
|
23
|
import eu.dnetlib.data.collective.transformation.engine.functions.RetrieveValue;
|
|
24
|
import eu.dnetlib.data.collective.transformation.rulelanguage.IRule;
|
|
25
|
import eu.dnetlib.data.collective.transformation.rulelanguage.Rules;
|
|
26
|
import eu.dnetlib.data.collective.transformation.rulelanguage.util.FunctionCall;
|
|
27
|
|
|
28
|
/**
|
|
29
|
* @author jochen
|
|
30
|
*
|
|
31
|
*/
|
|
32
|
public class PreProcessor {
|
|
33
|
|
|
34
|
@SuppressWarnings("unused")
|
|
35
|
private static final Log log = LogFactory.getLog(PreProcessor.class);
|
|
36
|
private Convert convertFunction;
|
|
37
|
private Extract extractFunction;
|
|
38
|
private RetrieveValue retrieveFunction;
|
|
39
|
private RegularExpression regExprFunction;
|
|
40
|
private TransformationFunctionProxy functionProxy;
|
|
41
|
private SAXReader reader = new SAXReader();
|
|
42
|
private Map<String, String> nsMap = new HashMap<String, String>();
|
|
43
|
|
|
44
|
/**
|
|
45
|
* pre-process output values from object records using a function call
|
|
46
|
* @param aFunctionCall function call object
|
|
47
|
* @param aObjectRecords list of object records
|
|
48
|
* @param aNamespaceMap map of namespace prefixes and uris
|
|
49
|
*/
|
|
50
|
public void preprocess(
|
|
51
|
FunctionCall aFunctionCall,
|
|
52
|
List<String> aObjectRecords,
|
|
53
|
Map<String, String> aNamespaceMap,
|
|
54
|
Map<String, String> aStaticResults,
|
|
55
|
Map<String, String> aJobProperties,
|
|
56
|
Map<String, IRule> aVarRules){
|
|
57
|
this.nsMap = aNamespaceMap;
|
|
58
|
FunctionResults functionResults = new FunctionResults();
|
|
59
|
|
|
60
|
try {
|
|
61
|
if (aFunctionCall.getExternalFunctionName().equals("extract")){
|
|
62
|
String featureName = aFunctionCall.getParameters().get(Extract.paramNameFeature);
|
|
63
|
functionResults.addAll(extractFunction.execute(aObjectRecords, featureName));
|
|
64
|
}else{
|
|
65
|
for (String objRecord: aObjectRecords){
|
|
66
|
String result = null;
|
|
67
|
if (aFunctionCall.getExternalFunctionName().equals("convert")){
|
|
68
|
if (aFunctionCall.isStatic() && aStaticResults.containsKey(aFunctionCall.getUuid())){
|
|
69
|
functionResults.add(aStaticResults.get(aFunctionCall.getUuid()));
|
|
70
|
}else{
|
|
71
|
String vocabName = aFunctionCall.getParameters().get(Convert.paramVocabularyName);
|
|
72
|
String fieldExpr = aFunctionCall.getParameters().get(Convert.paramFieldValue);
|
|
73
|
List<String> recordValues = getValuesFromRecord(objRecord, fieldExpr);
|
|
74
|
if (aFunctionCall.isStatic())
|
|
75
|
aStaticResults.put(aFunctionCall.getUuid(), convertFunction.executeSingleValue(vocabName, recordValues));
|
|
76
|
else
|
|
77
|
functionResults.add(convertFunction.executeAllValues(vocabName, recordValues));
|
|
78
|
}
|
|
79
|
}else if (aFunctionCall.getExternalFunctionName().equals("getValue")){
|
|
80
|
if (aFunctionCall.isStatic() && aStaticResults.containsKey(aFunctionCall.getUuid()))
|
|
81
|
functionResults.add(aStaticResults.get(aFunctionCall.getUuid()));
|
|
82
|
else{
|
|
83
|
String functionName = aFunctionCall.getParameters().get(RetrieveValue.paramFunctionName);
|
|
84
|
result = retrieveFunction.executeSingleValue(functionName, aFunctionCall.getArguments(), objRecord, nsMap);
|
|
85
|
functionResults.add(result);
|
|
86
|
if (aFunctionCall.isStatic())
|
|
87
|
aStaticResults.put(aFunctionCall.getUuid(), result);
|
|
88
|
}
|
|
89
|
}else if (aFunctionCall.getExternalFunctionName().equals("regExpr")){
|
|
90
|
// TODO
|
|
91
|
if (aFunctionCall.isStatic() && aStaticResults.containsKey(aFunctionCall.getUuid())){
|
|
92
|
//log.debug("static functioncal; static result exist to compute regexpr: " + aFunctionCall.getParameters().get(RegularExpression.paramRegularExpr));
|
|
93
|
// functionResults.add(aStaticResults.get(aFunctionCall.getUuid()));
|
|
94
|
}else{
|
|
95
|
// log.debug("static functioncal to compute regexpr: " + aFunctionCall.getParameters().get(RegularExpression.paramRegularExpr));
|
|
96
|
String regularExpression = aFunctionCall.getParameters().get(RegularExpression.paramRegularExpr); //.replaceAll("'", "");
|
|
97
|
String expression1 = aFunctionCall.getParameters().get(RegularExpression.paramExpr1);
|
|
98
|
List<String> recordValues = null;
|
|
99
|
// distinguish xpath-expr, jobConst, var
|
|
100
|
// log.debug("expression1: " + expression1);
|
|
101
|
|
|
102
|
if (aJobProperties.containsKey(expression1)){
|
|
103
|
recordValues = new LinkedList<String>();
|
|
104
|
recordValues.add(aJobProperties.get(expression1));
|
|
105
|
}else{
|
|
106
|
recordValues = getValuesFromRecord(objRecord, expression1);
|
|
107
|
}
|
|
108
|
|
|
109
|
String expression2 = aFunctionCall.getParameters().get(RegularExpression.paramExpr2);
|
|
110
|
String replacement = "";
|
|
111
|
if (aJobProperties.containsKey(expression2)){
|
|
112
|
replacement = aJobProperties.get(expression2);
|
|
113
|
}else if (aVarRules.containsKey(expression2)){
|
|
114
|
Rules varRule = (Rules)aVarRules.get(expression2);
|
|
115
|
replacement = varRule.getConstant().replace("'", ""); // currently limited to constant rules.
|
|
116
|
}else {
|
|
117
|
replacement = getValuesFromRecord(objRecord, expression2).get(0); // get the first available value
|
|
118
|
}
|
|
119
|
List<String> regExprResults = new LinkedList<String>();
|
|
120
|
for (String fieldValue: recordValues){
|
|
121
|
try {
|
|
122
|
int lastSlash = regularExpression.lastIndexOf("/");
|
|
123
|
String trailingOptions = regularExpression.substring(lastSlash);
|
|
124
|
int replacementSlash = regularExpression.substring(0, lastSlash).lastIndexOf("/");
|
|
125
|
String replacementFromExpression = regularExpression.substring(replacementSlash + 1, lastSlash);
|
|
126
|
String newRegExpr = regularExpression.substring(0, replacementSlash + 1) + replacement + replacementFromExpression + trailingOptions; // ???
|
|
127
|
result = regExprFunction.executeSingleValue(newRegExpr, fieldValue, replacement);
|
|
128
|
regExprResults.add(result);
|
|
129
|
} catch (ProcessingException e) {
|
|
130
|
throw new IllegalStateException(e);
|
|
131
|
}
|
|
132
|
// regExprResults.add(regExprFunction.executeSingleValue(regularExpression, fieldValue, expression2));
|
|
133
|
}
|
|
134
|
functionResults.add(regExprResults);
|
|
135
|
// assuming 1 result only
|
|
136
|
if (aFunctionCall.isStatic()){
|
|
137
|
aStaticResults.put(aFunctionCall.getUuid(), result);
|
|
138
|
}
|
|
139
|
|
|
140
|
// unsupported
|
|
141
|
// if (aFunctionCall.isStatic()){
|
|
142
|
// aStaticResults.put(aFunctionCall.getUuid(), result);
|
|
143
|
// }
|
|
144
|
}
|
|
145
|
}
|
|
146
|
}
|
|
147
|
}
|
|
148
|
} catch (ProcessingException e) {
|
|
149
|
throw new IllegalStateException(e);
|
|
150
|
} catch (DocumentException e) {
|
|
151
|
throw new IllegalStateException(e);
|
|
152
|
}
|
|
153
|
functionProxy.setResults(aFunctionCall.getUuid(), functionResults);
|
|
154
|
}
|
|
155
|
|
|
156
|
public void setFunctionProxy(TransformationFunctionProxy functionProxy) {
|
|
157
|
this.functionProxy = functionProxy;
|
|
158
|
}
|
|
159
|
|
|
160
|
public TransformationFunctionProxy getFunctionProxy() {
|
|
161
|
return functionProxy;
|
|
162
|
}
|
|
163
|
|
|
164
|
public void setConvertFunction(Convert convertFunction) {
|
|
165
|
this.convertFunction = convertFunction;
|
|
166
|
}
|
|
167
|
|
|
168
|
public Convert getConvertFunction() {
|
|
169
|
return convertFunction;
|
|
170
|
}
|
|
171
|
|
|
172
|
/**
|
|
173
|
* @param retrieveFunction the retrieveFunction to set
|
|
174
|
*/
|
|
175
|
public void setRetrieveFunction(RetrieveValue retrieveFunction) {
|
|
176
|
this.retrieveFunction = retrieveFunction;
|
|
177
|
}
|
|
178
|
|
|
179
|
/**
|
|
180
|
* @return the retrieveFunction
|
|
181
|
*/
|
|
182
|
public RetrieveValue getRetrieveFunction() {
|
|
183
|
return retrieveFunction;
|
|
184
|
}
|
|
185
|
|
|
186
|
/**
|
|
187
|
* @return the regExprFunction
|
|
188
|
*/
|
|
189
|
public RegularExpression getRegExprFunction() {
|
|
190
|
return regExprFunction;
|
|
191
|
}
|
|
192
|
|
|
193
|
/**
|
|
194
|
* @param regExprFunction the regExprFunction to set
|
|
195
|
*/
|
|
196
|
public void setRegExprFunction(RegularExpression regExprFunction) {
|
|
197
|
this.regExprFunction = regExprFunction;
|
|
198
|
}
|
|
199
|
|
|
200
|
/**
|
|
201
|
* @param extractFunction the extractFunction to set
|
|
202
|
*/
|
|
203
|
public void setExtractFunction(Extract extractFunction) {
|
|
204
|
this.extractFunction = extractFunction;
|
|
205
|
}
|
|
206
|
|
|
207
|
/**
|
|
208
|
* @return the extractFunction
|
|
209
|
*/
|
|
210
|
public Extract getExtractFunction() {
|
|
211
|
return extractFunction;
|
|
212
|
}
|
|
213
|
|
|
214
|
/**
|
|
215
|
* evaluate given XPath Expr applied on a record and return the values as a list of strings
|
|
216
|
* @param record
|
|
217
|
* @param xpathExpr
|
|
218
|
* @return list of strings
|
|
219
|
* @throws DocumentException
|
|
220
|
*/
|
|
221
|
@SuppressWarnings("unchecked")
|
|
222
|
private List<String> getValuesFromRecord(String record, String xpathExpr) throws DocumentException{
|
|
223
|
List<String> values = new LinkedList<String>();
|
|
224
|
Document doc = reader.read(new StringReader(record));
|
|
225
|
XPath xpath = DocumentHelper.createXPath(xpathExpr);
|
|
226
|
xpath.setNamespaceURIs(nsMap);
|
|
227
|
Object context = xpath.evaluate(doc);
|
|
228
|
if (context instanceof String)
|
|
229
|
values.add((String)context);
|
|
230
|
else if (context instanceof List)
|
|
231
|
for (Node node: (List<Node>)context)
|
|
232
|
values.add(node.getText());
|
|
233
|
else if (context instanceof Node)
|
|
234
|
values.add( ((Node)context).getText());
|
|
235
|
else if (context instanceof Number)
|
|
236
|
values.add( ((Number)context).intValue() + "");
|
|
237
|
return values;
|
|
238
|
}
|
|
239
|
|
|
240
|
}
|