Changeset 6311

Timestamp:

06/19/15 13:58:05 (9 years ago)

Author:

Twan Goosen

Message:

added support for derived value facets
refs #770

Location:

vlo/trunk/vlo-importer/src/main/java/eu/clarin/cmdi/vlo/importer

Files:

• CMDIParserVTDXML.java (modified) (2 diffs)
• FacetConceptMapping.java (modified) (14 diffs)
• FacetConfiguration.java (modified) (6 diffs)
• FacetMappingFactory.java (modified) (1 diff)

vlo/trunk/vlo-importer/src/main/java/eu/clarin/cmdi/vlo/importer/CMDIParserVTDXML.java

@@ -268 +268 @@
         setNameSpace(ap);
         ap.selectXPath(pattern);
-        
+
         boolean matchedPattern = false;
         int index = ap.evalXPath();
@@ -287 +287 @@
             }
 
-            final List<String> valueList = postProcess(config.getName(), value);
-            insertFacetValues(config.getName(), valueList, cmdiData, languageCode, allowMultipleValues, config.isCaseInsensitive());
+            final List<String> values = postProcess(config.getName(), value);
+            insertFacetValues(config.getName(), values, cmdiData, languageCode, allowMultipleValues, config.isCaseInsensitive());
+
+            // insert post-processed values into derived facet(s) if configured
+            for (String derivedFacet : config.getDerivedFacets()) {
+                final List<String> derivedValues = new ArrayList<String>();
+                for (String postProcessedValue : values) {
+                    derivedValues.addAll(postProcess(derivedFacet, postProcessedValue));
+                }
+                insertFacetValues(derivedFacet, derivedValues, cmdiData, languageCode, allowMultipleValues, config.isCaseInsensitive());
+            }
+
             index = ap.evalXPath();
 

vlo/trunk/vlo-importer/src/main/java/eu/clarin/cmdi/vlo/importer/FacetConceptMapping.java

@@ -15 +15 @@
 import javax.xml.bind.annotation.XmlRootElement;
 
-
 /**
  * Corresponds to the facet concepts file.
- * 
- * This class holds the mapping of facet name -> facetConcepts/patterns
- * A facetConcept is a ISOcat conceptLink e.g.: http://www.isocat.org/datcat/DC-2544
- * the conceptLink will be analysed and translated into a valid Xpath expression to extract data out of the metadata. 
- * Valid xpath expression e.g. /c:CMD/c:Header/c:MdSelfLink/text(), the 'c' namespace will be mapped to http://www.clarin.eu/cmd/ in the parser.
- * A pattern is an xpath expression used directly on the metadata. Use patterns only when a conceptLink does not suffice. 
+ *
+ * This class holds the mapping of facet name -> facetConcepts/patterns A
+ * facetConcept is a ISOcat conceptLink e.g.:
+ * http://www.isocat.org/datcat/DC-2544 the conceptLink will be analysed and
+ * translated into a valid Xpath expression to extract data out of the metadata.
+ * Valid xpath expression e.g. /c:CMD/c:Header/c:MdSelfLink/text(), the 'c'
+ * namespace will be mapped to http://www.clarin.eu/cmd/ in the parser. A
+ * pattern is an xpath expression used directly on the metadata. Use patterns
+ * only when a conceptLink does not suffice.
  *
  */
@@ -42 +44 @@
         this.facetConcepts = facetConcepts;
     }
-    
+
     public Map<String, FacetConcept> getFacetConceptMap() {
-        Map<String, FacetConcept> facetConceptMap = new HashMap<String, FacetConcept>();
-        for(FacetConcept facet : getFacetConcepts())
-                facetConceptMap.put(facet.getName(), facet);
-        
-        return facetConceptMap;
-    }
-    
+        Map<String, FacetConcept> facetConceptMap = new HashMap<String, FacetConcept>();
+        for (FacetConcept facet : getFacetConcepts()) {
+            facetConceptMap.put(facet.getName(), facet);
+        }
+
+        return facetConceptMap;
+    }
+
     public void check() {
         for (FacetConcept facetConcept : getFacetConcepts()) {
@@ -56 +59 @@
                 AcceptableContext acceptableContext = facetConcept.getAcceptableContext();
                 RejectableContext rejectableContext = facetConcept.getRejectableContext();
-                if (acceptableContext.includeAny() && rejectableContext.includeAny())
-                    LOG.error("Error: any context is both acceptable and rejectable for facet '"+facetConcept.getName()+"'");
-                if (acceptableContext.includeEmpty() && rejectableContext.includeEmpty())
-                    LOG.error("Error: empty context is both acceptable and rejectable for facet '"+facetConcept.getName()+"'");
+                if (acceptableContext.includeAny() && rejectableContext.includeAny()) {
+                    LOG.error("Error: any context is both acceptable and rejectable for facet '" + facetConcept.getName() + "'");
+                }
+                if (acceptableContext.includeEmpty() && rejectableContext.includeEmpty()) {
+                    LOG.error("Error: empty context is both acceptable and rejectable for facet '" + facetConcept.getName() + "'");
+                }
             }
         }
@@ -67 +72 @@
     @XmlRootElement(name = "facetConcept")
     public static class FacetConcept {
+
         @XmlAttribute
         private String name;
 
         /**
-         * Values will be stored lowercase by default, set isCaseInsensitive to true if you want to keep the case of the value
+         * Values will be stored lowercase by default, set isCaseInsensitive to
+         * true if you want to keep the case of the value
          */
         @XmlAttribute
@@ -77 +84 @@
 
         /**
-         * By default multiple values that are found for a matching pattern will be stored. For some facets this leads to too much values
-         * with little value for instance for "subject". Set allowMultipleValues to false will only store the first found value.
+         * By default multiple values that are found for a matching pattern will
+         * be stored. For some facets this leads to too much values with little
+         * value for instance for "subject". Set allowMultipleValues to false
+         * will only store the first found value.
          */
         @XmlAttribute
         private boolean allowMultipleValues = true;
-        
+
         @XmlAttribute
         private String description = "";
-        
+
         @XmlAttribute
         private String definition = "";
@@ -100 +109 @@
         @XmlElement(name = "pattern")
         private List<String> patterns = new ArrayList<String>();
-        
+
         @XmlElement(name = "blacklistPattern")
         private List<String> blacklistPatterns = new ArrayList<String>();
-        
+
+        @XmlElement(name = "derivedFacet")
+        private List<String> derivedFacets = new ArrayList<String>();
+
         public void setConcepts(List<String> concepts) {
             this.concepts = concepts;
@@ -119 +131 @@
             return acceptableContext;
         }
-        
+
         public boolean hasAcceptableContext() {
-            return (acceptableContext!=null);
+            return (acceptableContext != null);
         }
 
@@ -131 +143 @@
             return rejectableContext;
         }
-        
+
         public boolean hasRejectableContext() {
-            return (rejectableContext!=null);
-        }
-        
+            return (rejectableContext != null);
+        }
+
         public boolean hasContext() {
             return (hasAcceptableContext() || hasRejectableContext());
@@ -163 +175 @@
             return allowMultipleValues;
         }
-        
+
         public void setDescription(String description) {
-                this.description = description;
-        }
-        
+            this.description = description;
+        }
+
         public String getDescription() {
-                return this.description;
-        }
-        
+            return this.description;
+        }
+
         public void setDefinition(String definition) {
-                this.definition = definition;
-        }
-        
+            this.definition = definition;
+        }
+
         public String getDefinition() {
-                return this.definition;
+            return this.definition;
         }
 
@@ -187 +199 @@
             return patterns;
         }
-        
+
         public void setBlacklistPatterns(List<String> blacklistPatterns) {
             this.blacklistPatterns = blacklistPatterns;
@@ -194 +206 @@
         public List<String> getBlacklistPatterns() {
             return blacklistPatterns;
+        }
+
+        public List<String> getDerivedFacets() {
+            return derivedFacets;
+        }
+
+        public void setDerivedFacets(List<String> derivedFacets) {
+            this.derivedFacets = derivedFacets;
         }
 
@@ -202 +222 @@
 
     }
-    
+
     @XmlAccessorType(XmlAccessType.FIELD)
     @XmlRootElement(name = "acceptableContext")
@@ -239 +259 @@
             return includeEmpty;
         }
-        
+
     }
 
@@ -278 +298 @@
             return includeEmpty;
         }
-        
+
     }
 

vlo/trunk/vlo-importer/src/main/java/eu/clarin/cmdi/vlo/importer/FacetConfiguration.java

@@ -6 +6 @@
 
 /**
- * Once created contains the information about the facets and such.
- * Just a container for some information, doesn't do processing.
+ * Once created contains the information about the facets and such. Just a
+ * container for some information, doesn't do processing.
  */
-
 public class FacetConfiguration {
 
     private String name;
-    private boolean caseInsensitive= false;
+    private boolean caseInsensitive = false;
     private List<String> patterns = new ArrayList<String>();
     private List<String> fallbackPatterns = new ArrayList<String>();
+    private List<String> derivedFacets = new ArrayList<String>();
     private boolean allowMultipleValues = true;
 
@@ -29 +29 @@
         this.patterns = patterns;
     }
-    
+
     public void setFallbackPatterns(List<String> fallbackPatterns) {
         this.fallbackPatterns = fallbackPatterns;
@@ -37 +37 @@
         this.patterns = Collections.singletonList(pattern);
     }
-    
+
     public void setFallbackPattern(String fallbackPattern) {
         this.fallbackPatterns = Collections.singletonList(fallbackPattern);
@@ -48 +48 @@
         return patterns;
     }
-    
+
     public List<String> getFallbackPatterns() {
         return fallbackPatterns;
@@ -63 +63 @@
     @Override
     public String toString() {
-        return "name="+name+", pattern="+patterns;
+        return "name=" + name + ", pattern=" + patterns;
     }
 
@@ -73 +73 @@
         this.allowMultipleValues = allowMultipleValues;
     }
+
+    public List<String> getDerivedFacets() {
+        return derivedFacets;
+    }
+
+    public void setDerivedFacets(List<String> derivedFacets) {
+        this.derivedFacets = derivedFacets;
+    }
+
 }

vlo/trunk/vlo-importer/src/main/java/eu/clarin/cmdi/vlo/importer/FacetMappingFactory.java

@@ -179 +179 @@
                 config.setPatterns(new ArrayList<String>(linkedHashSet));
                 config.setFallbackPatterns(facetConcept.getPatterns());
+                config.setDerivedFacets(facetConcept.getDerivedFacets());
 
                 if (!config.getPatterns().isEmpty() || !config.getFallbackPatterns().isEmpty()) {