241 lines
9.2 KiB
Java
241 lines
9.2 KiB
Java
package eu.dnetlib.data.collective.transformation.engine;
|
|
|
|
import java.io.StringReader;
|
|
import java.util.HashMap;
|
|
import java.util.LinkedList;
|
|
import java.util.List;
|
|
import java.util.Map;
|
|
|
|
import org.apache.commons.logging.Log;
|
|
import org.apache.commons.logging.LogFactory;
|
|
import org.dom4j.Document;
|
|
import org.dom4j.DocumentException;
|
|
import org.dom4j.DocumentHelper;
|
|
import org.dom4j.Node;
|
|
import org.dom4j.XPath;
|
|
import org.dom4j.io.SAXReader;
|
|
|
|
import eu.dnetlib.data.collective.transformation.core.xsl.ext.TransformationFunctionProxy;
|
|
import eu.dnetlib.data.collective.transformation.engine.functions.Convert;
|
|
import eu.dnetlib.data.collective.transformation.engine.functions.Extract;
|
|
import eu.dnetlib.data.collective.transformation.engine.functions.ProcessingException;
|
|
import eu.dnetlib.data.collective.transformation.engine.functions.RegularExpression;
|
|
import eu.dnetlib.data.collective.transformation.engine.functions.RetrieveValue;
|
|
import eu.dnetlib.data.collective.transformation.rulelanguage.IRule;
|
|
import eu.dnetlib.data.collective.transformation.rulelanguage.Rules;
|
|
import eu.dnetlib.data.collective.transformation.rulelanguage.util.FunctionCall;
|
|
|
|
/**
|
|
* @author jochen
|
|
*
|
|
*/
|
|
public class PreProcessor {
|
|
|
|
@SuppressWarnings("unused")
|
|
private static final Log log = LogFactory.getLog(PreProcessor.class);
|
|
private Convert convertFunction;
|
|
private Extract extractFunction;
|
|
private RetrieveValue retrieveFunction;
|
|
private RegularExpression regExprFunction;
|
|
private TransformationFunctionProxy functionProxy;
|
|
private SAXReader reader = new SAXReader();
|
|
private Map<String, String> nsMap = new HashMap<String, String>();
|
|
|
|
/**
|
|
* pre-process output values from object records using a function call
|
|
* @param aFunctionCall function call object
|
|
* @param aObjectRecords list of object records
|
|
* @param aNamespaceMap map of namespace prefixes and uris
|
|
*/
|
|
public void preprocess(
|
|
FunctionCall aFunctionCall,
|
|
List<String> aObjectRecords,
|
|
Map<String, String> aNamespaceMap,
|
|
Map<String, String> aStaticResults,
|
|
Map<String, String> aJobProperties,
|
|
Map<String, IRule> aVarRules){
|
|
this.nsMap = aNamespaceMap;
|
|
FunctionResults functionResults = new FunctionResults();
|
|
|
|
try {
|
|
if (aFunctionCall.getExternalFunctionName().equals("extract")){
|
|
String featureName = aFunctionCall.getParameters().get(Extract.paramNameFeature);
|
|
functionResults.addAll(extractFunction.execute(aObjectRecords, featureName));
|
|
}else{
|
|
for (String objRecord: aObjectRecords){
|
|
String result = null;
|
|
if (aFunctionCall.getExternalFunctionName().equals("convert")){
|
|
if (aFunctionCall.isStatic() && aStaticResults.containsKey(aFunctionCall.getUuid())){
|
|
functionResults.add(aStaticResults.get(aFunctionCall.getUuid()));
|
|
}else{
|
|
String vocabName = aFunctionCall.getParameters().get(Convert.paramVocabularyName);
|
|
String fieldExpr = aFunctionCall.getParameters().get(Convert.paramFieldValue);
|
|
List<String> recordValues = getValuesFromRecord(objRecord, fieldExpr);
|
|
if (aFunctionCall.isStatic())
|
|
aStaticResults.put(aFunctionCall.getUuid(), convertFunction.executeSingleValue(vocabName, recordValues));
|
|
else
|
|
functionResults.add(convertFunction.executeAllValues(vocabName, recordValues));
|
|
}
|
|
}else if (aFunctionCall.getExternalFunctionName().equals("getValue")){
|
|
if (aFunctionCall.isStatic() && aStaticResults.containsKey(aFunctionCall.getUuid()))
|
|
functionResults.add(aStaticResults.get(aFunctionCall.getUuid()));
|
|
else{
|
|
String functionName = aFunctionCall.getParameters().get(RetrieveValue.paramFunctionName);
|
|
result = retrieveFunction.executeSingleValue(functionName, aFunctionCall.getArguments(), objRecord, nsMap);
|
|
functionResults.add(result);
|
|
if (aFunctionCall.isStatic())
|
|
aStaticResults.put(aFunctionCall.getUuid(), result);
|
|
}
|
|
}else if (aFunctionCall.getExternalFunctionName().equals("regExpr")){
|
|
// TODO
|
|
if (aFunctionCall.isStatic() && aStaticResults.containsKey(aFunctionCall.getUuid())){
|
|
//log.debug("static functioncal; static result exist to compute regexpr: " + aFunctionCall.getParameters().get(RegularExpression.paramRegularExpr));
|
|
// functionResults.add(aStaticResults.get(aFunctionCall.getUuid()));
|
|
}else{
|
|
// log.debug("static functioncal to compute regexpr: " + aFunctionCall.getParameters().get(RegularExpression.paramRegularExpr));
|
|
String regularExpression = aFunctionCall.getParameters().get(RegularExpression.paramRegularExpr); //.replaceAll("'", "");
|
|
String expression1 = aFunctionCall.getParameters().get(RegularExpression.paramExpr1);
|
|
List<String> recordValues = null;
|
|
// distinguish xpath-expr, jobConst, var
|
|
// log.debug("expression1: " + expression1);
|
|
|
|
if (aJobProperties.containsKey(expression1)){
|
|
recordValues = new LinkedList<String>();
|
|
recordValues.add(aJobProperties.get(expression1));
|
|
}else{
|
|
recordValues = getValuesFromRecord(objRecord, expression1);
|
|
}
|
|
|
|
String expression2 = aFunctionCall.getParameters().get(RegularExpression.paramExpr2);
|
|
String replacement = "";
|
|
if (aJobProperties.containsKey(expression2)){
|
|
replacement = aJobProperties.get(expression2);
|
|
}else if (aVarRules.containsKey(expression2)){
|
|
Rules varRule = (Rules)aVarRules.get(expression2);
|
|
replacement = varRule.getConstant().replace("'", ""); // currently limited to constant rules.
|
|
}else {
|
|
replacement = getValuesFromRecord(objRecord, expression2).get(0); // get the first available value
|
|
}
|
|
List<String> regExprResults = new LinkedList<String>();
|
|
for (String fieldValue: recordValues){
|
|
try {
|
|
int lastSlash = regularExpression.lastIndexOf("/");
|
|
String trailingOptions = regularExpression.substring(lastSlash);
|
|
int replacementSlash = regularExpression.substring(0, lastSlash).lastIndexOf("/");
|
|
String replacementFromExpression = regularExpression.substring(replacementSlash + 1, lastSlash);
|
|
String newRegExpr = regularExpression.substring(0, replacementSlash + 1) + replacement + replacementFromExpression + trailingOptions; // ???
|
|
result = regExprFunction.executeSingleValue(newRegExpr, fieldValue, replacement);
|
|
regExprResults.add(result);
|
|
} catch (ProcessingException e) {
|
|
throw new IllegalStateException(e);
|
|
}
|
|
// regExprResults.add(regExprFunction.executeSingleValue(regularExpression, fieldValue, expression2));
|
|
}
|
|
functionResults.add(regExprResults);
|
|
// assuming 1 result only
|
|
if (aFunctionCall.isStatic()){
|
|
aStaticResults.put(aFunctionCall.getUuid(), result);
|
|
}
|
|
|
|
// unsupported
|
|
// if (aFunctionCall.isStatic()){
|
|
// aStaticResults.put(aFunctionCall.getUuid(), result);
|
|
// }
|
|
}
|
|
}
|
|
}
|
|
}
|
|
} catch (ProcessingException e) {
|
|
throw new IllegalStateException(e);
|
|
} catch (DocumentException e) {
|
|
throw new IllegalStateException(e);
|
|
}
|
|
functionProxy.setResults(aFunctionCall.getUuid(), functionResults);
|
|
}
|
|
|
|
public void setFunctionProxy(TransformationFunctionProxy functionProxy) {
|
|
this.functionProxy = functionProxy;
|
|
}
|
|
|
|
public TransformationFunctionProxy getFunctionProxy() {
|
|
return functionProxy;
|
|
}
|
|
|
|
public void setConvertFunction(Convert convertFunction) {
|
|
this.convertFunction = convertFunction;
|
|
}
|
|
|
|
public Convert getConvertFunction() {
|
|
return convertFunction;
|
|
}
|
|
|
|
/**
|
|
* @param retrieveFunction the retrieveFunction to set
|
|
*/
|
|
public void setRetrieveFunction(RetrieveValue retrieveFunction) {
|
|
this.retrieveFunction = retrieveFunction;
|
|
}
|
|
|
|
/**
|
|
* @return the retrieveFunction
|
|
*/
|
|
public RetrieveValue getRetrieveFunction() {
|
|
return retrieveFunction;
|
|
}
|
|
|
|
/**
|
|
* @return the regExprFunction
|
|
*/
|
|
public RegularExpression getRegExprFunction() {
|
|
return regExprFunction;
|
|
}
|
|
|
|
/**
|
|
* @param regExprFunction the regExprFunction to set
|
|
*/
|
|
public void setRegExprFunction(RegularExpression regExprFunction) {
|
|
this.regExprFunction = regExprFunction;
|
|
}
|
|
|
|
/**
|
|
* @param extractFunction the extractFunction to set
|
|
*/
|
|
public void setExtractFunction(Extract extractFunction) {
|
|
this.extractFunction = extractFunction;
|
|
}
|
|
|
|
/**
|
|
* @return the extractFunction
|
|
*/
|
|
public Extract getExtractFunction() {
|
|
return extractFunction;
|
|
}
|
|
|
|
/**
|
|
* evaluate given XPath Expr applied on a record and return the values as a list of strings
|
|
* @param record
|
|
* @param xpathExpr
|
|
* @return list of strings
|
|
* @throws DocumentException
|
|
*/
|
|
@SuppressWarnings("unchecked")
|
|
private List<String> getValuesFromRecord(String record, String xpathExpr) throws DocumentException{
|
|
List<String> values = new LinkedList<String>();
|
|
Document doc = reader.read(new StringReader(record));
|
|
XPath xpath = DocumentHelper.createXPath(xpathExpr);
|
|
xpath.setNamespaceURIs(nsMap);
|
|
Object context = xpath.evaluate(doc);
|
|
if (context instanceof String)
|
|
values.add((String)context);
|
|
else if (context instanceof List)
|
|
for (Node node: (List<Node>)context)
|
|
values.add(node.getText());
|
|
else if (context instanceof Node)
|
|
values.add( ((Node)context).getText());
|
|
else if (context instanceof Number)
|
|
values.add( ((Number)context).intValue() + "");
|
|
return values;
|
|
}
|
|
|
|
}
|