Merge pull request #316 from dice-group/dev

Release 1.8.1

docs/user_manual/configuration_file/index.md

@@ -65,6 +65,27 @@ Optional properties can be set to segment the requested dataset.
 * The graph of the endpoint can be specified directly ofter the `ENDPOINT` tag using the `GRAPH` tag.
 * The limits of the query can be set with the `MINOFFSET` and `MAXOFFSET` tags directly after the `PAGESIZE` tag. The resulting query will ask about the statements in the interval [`MINOFFSET`, `MAXOFFSET`]. Note that `MINOFFSET` must be smaller than `MAXOFFSET`! If both `SOURCE` and `TARGET` are restricted, a warning is generated.
 
+Please note that LIMES does not allow namespace IRIs to be used in the `PROPERTY`, `RESTRICTION`, and `OPTIONAL_PROPERTY` tag. Please use namespace prefixes and reference the namespace IRI using a prefix. Example: Do not use
+```
+<SOURCE>
+	...
+    <PROPERTY>http://xmlns.com/foaf/0.1#name</PROPERTY>
+	...
+</SOURCE>
+```
+instead use the following
+```
+<PREFIX>
+    <NAMESPACE>http://xmlns.com/foaf/0.1/</NAMESPACE>
+    <LABEL>foaf</LABEL>
+</PREFIX>
+<SOURCE>
+	...
+    <PROPERTY>foaf:name</PROPERTY>
+	...
+</SOURCE>
+```
+
 ### Preprocessing Functions
 #### Simple
 

docs/user_manual/running_limes.md

@@ -136,7 +136,8 @@ The Prefixes component consists of two parts:
 
 The Data source and data target consists of the two similar components, which include three input fields:
 * *Sparql endpoint/Local file*: One of two options can be chosen. Sparql endpoint means that the user will select the endpont from the list. Local file means that the file should be provided as an endpoint.
-* *Endpoint*: A dropdown list of available endpoints. Moreover, the user can try to search for the endpoint, typing it in the input field or write your own endpoint. After clicking on the endpoint from the list or writing it by hand and press the Enter, the user will get the list of restriction classes according to this endpoint.
+* *Endpoint*: A dropdown list of available endpoints. Moreover, the user can try to search for the endpoint, typing it in the input field or write your own endpoint. After clicking on the endpoint from the list or writing it by hand and press the Enter, the user will get the list of restriction classes according to this endpoint. 
+ In case you want to upload a configuration file as well as a data source and a target file, you must upload the configuration file prior to uploading the data source and target files.
 * *Restriction*: Contains of three parameters splitted by space (?s rdf:type some:Type). The third parameter will be changed automatically after changing the restriction class. 
 * *Restriction class*: A dropdown list of restriction classes according to the endpoint. You can start typing the name of the class and the list will be filtered automatically. After choosing the restriction class, you will get all the properties related to this class.
 

limes-core/pom.xml

@@ -16,7 +16,7 @@
     <parent>
         <groupId>org.aksw.limes</groupId>
         <artifactId>limes-full</artifactId>
-        <version>1.8.0-SNAPSHOT</version>
+        <version>1.8.1-SNAPSHOT</version>
     </parent>
 
     <modelVersion>4.0.0</modelVersion>

limes-core/src/main/java/org/aksw/limes/core/io/config/reader/AConfigurationReader.java

@@ -18,6 +18,9 @@
 package org.aksw.limes.core.io.config.reader;
 
 import org.aksw.limes.core.io.config.Configuration;
+import org.aksw.limes.core.io.config.KBInfo;
+
+import java.util.*;
 
 /**
  * @author Mohamed Sherif ([email protected])
@@ -48,5 +51,78 @@ public void setConfiguration(Configuration configuration) {
         this.configuration = configuration;
     }
 
+    /**
+     * This method replaces any URIs used in the kbInfo with their prefixes
+     * @param info
+     */
+    public static void replaceURIsWithPrefixes(KBInfo info) {
+        Map<String, String> prefixes = info.getPrefixes();
+        HashMap<String, String> rev = new HashMap<>();
+        for(Map.Entry<String, String> entry : prefixes.entrySet()) {
+            rev.put(entry.getValue(), entry.getKey());
+        }
+        info.setProperties(replaceURIsWithPrefixes(info.getProperties(), rev));
+        info.setOptionalProperties(replaceURIsWithPrefixes(info.getOptionalProperties(), rev));
+        info.setRestrictions(replaceURIsWithPrefixes(info.getRestrictions(), rev));
+        info.setFunctions(replaceURIsWithPrefixes(info.getFunctions(), rev));
+    }
+
+    private static ArrayList<String> replaceURIsWithPrefixes(Collection<String> props, HashMap<String, String> rev) {
+        ArrayList<String> replacements = new ArrayList<>();
+        for (String property : props) {
+            String originalProp = property;
+            for (Map.Entry<String, String> prefixEntry : rev.entrySet()) {
+                if(property.contains(prefixEntry.getKey())){
+                    property = property.replace(prefixEntry.getKey(), prefixEntry.getValue() + ":");
+                }
+            }
+            replacements.add(property);
+
+            if(property.contains("://")){
+                throw new IllegalArgumentException("LIMES does not support using URIs in the properties, optionalProperties, restrictions, or functions in the configuration file. " +
+                        "Please define a prefix and use the prefix for the following URI: " + originalProp);
+            }
+        }
+        return replacements;
+    }
+
+    private static LinkedHashMap<String, Map<String, String>> replaceURIsWithPrefixes(Map<String, Map<String, String>> funcs, HashMap<String, String> rev) {
+        LinkedHashMap<String, Map<String, String>> replacements = new LinkedHashMap<>();
+        for (Map.Entry<String, Map<String, String>> entry : funcs.entrySet()) {
+            String property = entry.getKey();
+            String originalProp = property;
+
+            //Replace key of function
+            for (Map.Entry<String, String> prefixEntry : rev.entrySet()) {
+                if(property.contains(prefixEntry.getKey())){
+                    property = property.replace(prefixEntry.getKey(), prefixEntry.getValue() + ":");
+                }
+            }
+            if(property.contains("://")){
+                throw new IllegalArgumentException("LIMES does not support using namespace IRIs in the properties, optionalProperties, restrictions, or functions in the configuration file. " +
+                        "Please define a prefix and use the prefix for the namespace of the following IRI: " + originalProp);
+            }
+
+            //Replace value map
+            Map<String, String> intermediateReplacement = new HashMap<>();
+            for (Map.Entry<String, String> stringEntry : entry.getValue().entrySet()) {
+                String subKey = stringEntry.getKey();
+                String origSubKey = subKey;
+                for (Map.Entry<String, String> prefixEntry : rev.entrySet()) {
+                    subKey = subKey.replace(prefixEntry.getKey(), prefixEntry.getValue() + ":");
+                }
+                intermediateReplacement.put(subKey, stringEntry.getValue());
+                if(subKey.contains("://")){
+                    throw new IllegalArgumentException("LIMES does not support using namespace IRIs in the properties, optionalProperties, restrictions, or functions in the configuration file. " +
+                            "Please define a prefix and use the prefix for the namespace of the following IRI: " + origSubKey);
+                }
+
+            }
+
+            replacements.put(property, intermediateReplacement);
+        }
+        return replacements;
+    }
+
 
 }

limes-core/src/main/java/org/aksw/limes/core/io/config/reader/rdf/RDFConfigurationReader.java

@@ -166,6 +166,15 @@ public Configuration read(Model configurationModel) {
             configuration.setOutputFormat(output.toString());
         }
 
+        // 10. Check for invalid config because of URIs without prefixes
+        if(configuration.getSourceInfo() != null){
+            replaceURIsWithPrefixes(configuration.getSourceInfo());
+        }
+        if(configuration.getTargetInfo() != null){
+            replaceURIsWithPrefixes(configuration.getTargetInfo());
+        }
+
+
         return configuration;
     }
 

limes-core/src/main/java/org/aksw/limes/core/io/config/reader/xml/XMLConfigurationReader.java

@@ -513,6 +513,15 @@ public InputSource resolveEntity(String publicId, String systemId) throws SAXExc
                     children = list.item(0).getChildNodes();
                     configuration.setOutputFormat(getText(list.item(0)));
                 }
+
+                // 10. Check for invalid config because of URIs without prefixes
+                if(configuration.getSourceInfo() != null){
+                    replaceURIsWithPrefixes(configuration.getSourceInfo());
+                }
+                if(configuration.getTargetInfo() != null){
+                    replaceURIsWithPrefixes(configuration.getTargetInfo());
+                }
+
             }
         } catch (Exception e) {
             logger.warn(e.getMessage());

limes-core/src/main/java/org/aksw/limes/core/io/preprocessing/PreprocessingFunctionFactory.java

@@ -44,10 +44,10 @@ public class PreprocessingFunctionFactory {
     public static final String REMOVE_NON_ALPHANUMERIC = "regularalphabet";
     public static final String URI_AS_STRING = "uriasstring";
     public static final String SPLIT = "split";
-    public static final String TO_WKT_POINT = "toWktPoint";
+    public static final String TO_WKT_POINT = "towktpoint"; //toWktPoint
 
     public static PreprocessingFunctionType getPreprocessingType(String expression){
-        switch(expression.trim()){
+        switch(expression.trim().toLowerCase()){
             case(CLEAN_IRI):
                 return PreprocessingFunctionType.CLEAN_IRI;
             case(CLEAN_NUMBER):

limes-core/src/main/java/org/aksw/limes/core/io/query/SparqlQueryModule.java

@@ -258,6 +258,7 @@ else if (q[ql].contains("^")) {
         if (kb.getRestrictions().size() > 0) {
             query = query + "}";
         }
+
         logger.info("Query issued is \n" + query);
         return query;
     }

limes-core/src/test/java/org/aksw/limes/core/io/config/reader/xml/AConfigurationReaderTest.java

@@ -0,0 +1,137 @@
+package org.aksw.limes.core.io.config.reader.xml;
+
+import org.aksw.limes.core.io.config.Configuration;
+import org.aksw.limes.core.io.config.KBInfo;
+import org.aksw.limes.core.io.config.reader.AConfigurationReader;
+import org.junit.Before;
+import org.junit.Test;
+
+import static org.junit.Assert.assertTrue;
+import static org.junit.Assert.assertThrows;
+
+import java.util.*;
+
+public class AConfigurationReaderTest {
+    Map<String, String> prefixes;
+    LinkedHashMap<String, Map<String, String>> functions;
+    KBInfo sourceInfo, targetInfo;
+    Configuration testConf;
+
+    @Before
+    public void init() {
+        prefixes = new HashMap<>();
+        prefixes.put("geos", "http://www.opengis.net/ont/geosparql#");
+        prefixes.put("lgdo", "http://linkedgeodata.org/ontology/");
+        prefixes.put("geom", "http://geovocab.org/geometry#");
+        prefixes.put("rdfs", "http://www.w3.org/2000/01/rdf-schema#");
+        prefixes.put("limes", "http://limes.sf.net/ontology/");
+
+        functions = new LinkedHashMap<>();
+        Map<String, String> f = new LinkedHashMap<>();
+        f.put("polygon", null);
+        functions.put("geom:geometry/geos:asWKT", f);
+
+        sourceInfo = new KBInfo(
+                "linkedgeodata",                                                  //String id
+                "http://linkedgeodata.org/sparql",                                //String endpoint
+                null,                                                             //String graph
+                "?x",                                                             //String var
+                new ArrayList<>(Arrays.asList("geom:geometry/geos:asWKT")), //List<String> properties
+                new ArrayList<>(),                                          //List<String> optionalProperties
+                new ArrayList<>(Arrays.asList("?x a lgdo:RelayBox")),       //ArrayList<String> restrictions
+                functions,                                                        //LinkedHashMap<String, Map<String, String>> functions
+                prefixes,                                                         //Map<String, String> prefixes
+                2000,                                                             //int pageSize
+                "sparql",                                                         //String type
+                -1,                                                               //int minOffset
+                -1                                                                //int maxoffset
+        );
+
+        targetInfo = new KBInfo(
+                "linkedgeodata",                                                  //String id
+                "http://linkedgeodata.org/sparql",                                //String endpoint
+                null,                                                             //String graph
+                "?y",                                                             //String var
+                new ArrayList<>(Arrays.asList("geom:geometry/geos:asWKT")), //List<String> properties
+                new ArrayList<>(),                                          //List<String> optionalProperties
+                new ArrayList<>(Arrays.asList("?y a lgdo:RelayBox")),       //ArrayList<String> restrictions
+                functions,                                                        //LinkedHashMap<String, Map<String, String>> functions
+                prefixes,                                                         //Map<String, String> prefixes
+                2000,                                                             //int pageSize
+                "sparql",                                                         //String type
+                -1,                                                               //int minOffset
+                -1                                                                //int maxoffset
+        );
+
+        testConf = new Configuration();
+        testConf.setPrefixes(prefixes);
+        testConf.setSourceInfo(sourceInfo);
+        testConf.setTargetInfo(targetInfo);
+
+    }
+
+    @Test
+    public void testReplaceWithPrefixProp() {
+        testConf.getSourceInfo().addProperty("http://www.opengis.net/ont/geosparql#test");
+        AConfigurationReader.replaceURIsWithPrefixes(testConf.getSourceInfo());
+        assertTrue(testConf.getSourceInfo().getProperties().contains("geos:test"));
+    }
+
+    @Test
+    public void testReplaceWithPrefixOptionalProp() {
+        testConf.getSourceInfo().addOptionalProperty("http://www.opengis.net/ont/geosparql#test");
+        AConfigurationReader.replaceURIsWithPrefixes(testConf.getSourceInfo());
+        assertTrue(testConf.getSourceInfo().getOptionalProperties().contains("geos:test"));
+    }
+
+    @Test
+    public void testReplaceWithPrefixRestriction() {
+        testConf.getSourceInfo().addRestriction("http://www.opengis.net/ont/geosparql#test");
+        AConfigurationReader.replaceURIsWithPrefixes(testConf.getSourceInfo());
+        assertTrue(testConf.getSourceInfo().getRestrictions().contains("geos:test"));
+    }
+
+    @Test
+    public void testReplaceWithPrefixFunc() {
+        LinkedHashMap<String, Map<String, String>> x = testConf.getSourceInfo().getFunctions();
+        HashMap<String, String> y = new HashMap<>();
+        y.put("http://www.opengis.net/ont/geosparql#test", "toLower");
+        x.put("http://www.opengis.net/ont/geosparql#test", y);
+        testConf.getSourceInfo().setFunctions(x);
+        AConfigurationReader.replaceURIsWithPrefixes(testConf.getSourceInfo());
+        assertTrue(testConf.getSourceInfo().getFunctions().containsKey("geos:test"));
+    }
+
+    @Test
+    public void testReplaceWithPrefixPropError() {
+        testConf.getPrefixes().remove("geos");
+        testConf.getSourceInfo().addProperty("http://www.opengis.net/ont/geosparql#test");
+        assertThrows(IllegalArgumentException.class, () -> AConfigurationReader.replaceURIsWithPrefixes(testConf.getSourceInfo()));
+    }
+
+    @Test
+    public void testReplaceWithPrefixOptionalPropError() {
+        testConf.getPrefixes().remove("geos");
+        testConf.getSourceInfo().addOptionalProperty("http://www.opengis.net/ont/geosparql#test");
+        assertThrows(IllegalArgumentException.class, () -> AConfigurationReader.replaceURIsWithPrefixes(testConf.getSourceInfo()));
+    }
+
+    @Test
+    public void testReplaceWithPrefixRestrictionError() {
+        testConf.getPrefixes().remove("geos");
+        testConf.getSourceInfo().addRestriction("http://www.opengis.net/ont/geosparql#test");
+        assertThrows(IllegalArgumentException.class, () -> AConfigurationReader.replaceURIsWithPrefixes(testConf.getSourceInfo()));
+    }
+
+    @Test
+    public void testReplaceWithPrefixFuncError() {
+        testConf.getPrefixes().remove("geos");
+        LinkedHashMap<String, Map<String, String>> x = testConf.getSourceInfo().getFunctions();
+        HashMap<String, String> y = new HashMap<>();
+        y.put("http://www.opengis.net/ont/geosparql#test", "toLower");
+        x.put("http://www.opengis.net/ont/geosparql#test", y);
+        testConf.getSourceInfo().setFunctions(x);
+        assertThrows(IllegalArgumentException.class, () -> AConfigurationReader.replaceURIsWithPrefixes(testConf.getSourceInfo()));
+    }
+
+}

limes-debian-cli/pom.xml

@@ -19,7 +19,7 @@
 	<parent>
 		<groupId>org.aksw.limes</groupId>
 		<artifactId>limes-full</artifactId>
-		<version>1.8.0-SNAPSHOT</version>
+		<version>1.8.1-SNAPSHOT</version>
 	</parent>
 
 	<artifactId>limes-debian-cli</artifactId>

pom.xml

@@ -26,7 +26,7 @@
 
     <groupId>org.aksw.limes</groupId>
     <artifactId>limes-full</artifactId>
-    <version>1.8.0-SNAPSHOT</version>
+    <version>1.8.1-SNAPSHOT</version>
     <name>LIMES</name>
     <description>LIMES – Link Discovery Framework for Metric Spaces.</description>
     <url>https://aksw.org/Projects/LIMES</url>