Kbss cvut/termit UI#587 remember manual annotations

src/main/java/cz/cvut/kbss/termit/exception/SelectorGenerationException.java

@@ -0,0 +1,11 @@
+package cz.cvut.kbss.termit.exception;
+
+/**
+ * Indicates that the application was unable to generate a selector.
+ */
+public class SelectorGenerationException extends AnnotationGenerationException {
+
+    public SelectorGenerationException(String message) {
+        super(message);
+    }
+}

src/main/java/cz/cvut/kbss/termit/model/assignment/DefinitionalOccurrenceTarget.java

@@ -19,8 +19,12 @@
 
 import cz.cvut.kbss.jopa.model.annotations.OWLClass;
 import cz.cvut.kbss.termit.model.AbstractTerm;
+import cz.cvut.kbss.termit.model.util.Copyable;
+import cz.cvut.kbss.termit.util.Utils;
 import cz.cvut.kbss.termit.util.Vocabulary;
 
+import java.util.stream.Collectors;
+
 /**
  * Target representing the definition of a term.
  * <p>
@@ -36,6 +40,14 @@ public DefinitionalOccurrenceTarget(AbstractTerm source) {
         super(source);
     }
 
+    @Override
+    public DefinitionalOccurrenceTarget copy() {
+        final DefinitionalOccurrenceTarget copy = new DefinitionalOccurrenceTarget();
+        copy.setSource(getSource());
+        copy.setSelectors(Utils.emptyIfNull(getSelectors()).stream().map(Copyable::copy).collect(Collectors.toSet()));
+        return copy;
+    }
+
     @Override
     public String toString() {
         return "DefinitionalOccurrenceTarget{" + super.toString() + '}';

src/main/java/cz/cvut/kbss/termit/model/assignment/FileOccurrenceTarget.java

@@ -19,8 +19,12 @@
 
 import cz.cvut.kbss.jopa.model.annotations.OWLClass;
 import cz.cvut.kbss.termit.model.resource.File;
+import cz.cvut.kbss.termit.model.util.Copyable;
+import cz.cvut.kbss.termit.util.Utils;
 import cz.cvut.kbss.termit.util.Vocabulary;
 
+import java.util.stream.Collectors;
+
 /**
  * Target representing the content of a file.
  * <p>
@@ -36,6 +40,14 @@ public FileOccurrenceTarget(File source) {
         super(source);
     }
 
+    @Override
+    public FileOccurrenceTarget copy() {
+        final FileOccurrenceTarget copy = new FileOccurrenceTarget();
+        copy.setSource(getSource());
+        copy.setSelectors(Utils.emptyIfNull(getSelectors()).stream().map(Copyable::copy).collect(Collectors.toSet()));
+        return copy;
+    }
+
     @Override
     public String toString() {
         return "FileOccurrenceTarget{" + super.toString() + '}';

src/main/java/cz/cvut/kbss/termit/model/assignment/OccurrenceTarget.java

@@ -26,6 +26,7 @@
 import cz.cvut.kbss.termit.model.AbstractEntity;
 import cz.cvut.kbss.termit.model.Asset;
 import cz.cvut.kbss.termit.model.selector.Selector;
+import cz.cvut.kbss.termit.model.util.Copyable;
 import cz.cvut.kbss.termit.util.Vocabulary;
 import jakarta.validation.constraints.NotEmpty;
 import jakarta.validation.constraints.NotNull;
@@ -36,7 +37,7 @@
 
 @JsonTypeInfo(use = JsonTypeInfo.Id.CLASS, property = "javaClass")
 @OWLClass(iri = Vocabulary.s_c_cil_vyskytu)
-public abstract class OccurrenceTarget extends AbstractEntity {
+public abstract class OccurrenceTarget extends AbstractEntity implements Copyable<OccurrenceTarget> {
 
     @NotNull
     @ParticipationConstraints(nonEmpty = true)

src/main/java/cz/cvut/kbss/termit/model/assignment/TermDefinitionSource.java

@@ -34,4 +34,9 @@ public TermDefinitionSource() {
     public TermDefinitionSource(URI term, FileOccurrenceTarget target) {
         super(term, target);
     }
+
+    @Override
+    public TermDefinitionSource copy() {
+        return new TermDefinitionSource(getTerm(), getTarget().copy());
+    }
 }

src/main/java/cz/cvut/kbss/termit/model/assignment/TermDefinitionalOccurrence.java

@@ -44,4 +44,9 @@ public DefinitionalOccurrenceTarget getTarget() {
     public void setTarget(DefinitionalOccurrenceTarget target) {
         this.target = target;
     }
+
+    @Override
+    public TermDefinitionalOccurrence copy() {
+        return new TermDefinitionalOccurrence(getTerm(), getTarget().copy());
+    }
 }

src/main/java/cz/cvut/kbss/termit/model/assignment/TermFileOccurrence.java

@@ -44,4 +44,9 @@ public FileOccurrenceTarget getTarget() {
     public void setTarget(FileOccurrenceTarget target) {
         this.target = target;
     }
+
+    @Override
+    public TermFileOccurrence copy() {
+        return new TermFileOccurrence(getTerm(), getTarget().copy());
+    }
 }

src/main/java/cz/cvut/kbss/termit/model/assignment/TermOccurrence.java

@@ -24,10 +24,13 @@
 import cz.cvut.kbss.jopa.model.annotations.OWLDataProperty;
 import cz.cvut.kbss.jopa.model.annotations.OWLObjectProperty;
 import cz.cvut.kbss.jopa.model.annotations.ParticipationConstraints;
+import cz.cvut.kbss.jopa.model.annotations.Transient;
 import cz.cvut.kbss.jopa.model.annotations.Types;
 import cz.cvut.kbss.jopa.vocabulary.DC;
 import cz.cvut.kbss.termit.model.AbstractEntity;
+import cz.cvut.kbss.termit.model.util.Copyable;
 import cz.cvut.kbss.termit.model.util.HasTypes;
+import cz.cvut.kbss.termit.util.Constants;
 import cz.cvut.kbss.termit.util.Vocabulary;
 import jakarta.validation.constraints.NotNull;
 
@@ -37,7 +40,7 @@
 
 @JsonTypeInfo(use = JsonTypeInfo.Id.CLASS, property = "javaClass")
 @OWLClass(iri = Vocabulary.s_c_vyskyt_termu)
-public abstract class TermOccurrence extends AbstractEntity implements HasTypes {
+public abstract class TermOccurrence extends AbstractEntity implements Copyable<TermOccurrence>, HasTypes {
 
     /**
      * Suffix used to identify term occurrence contexts (named graphs) in the repository.
@@ -60,7 +63,15 @@ public abstract class TermOccurrence extends AbstractEntity implements HasTypes
     @Types
     private Set<String> types;
 
-    private transient Double score;
+    @Transient
+    private Double score;
+
+    /**
+     * Value of the {@literal about} attribute of the HTML element representing the occurrence.
+     */
+    @Transient
+    @OWLDataProperty(iri = DC.Terms.IDENTIFIER)
+    private String elementAbout;
 
     public TermOccurrence() {
     }
@@ -119,6 +130,29 @@ public void setScore(Double score) {
         this.score = score;
     }
 
+    public String getElementAbout() {
+        return elementAbout;
+    }
+
+    public void setElementAbout(String elementAbout) {
+        this.elementAbout = elementAbout;
+    }
+
+    /**
+     * Determines the value of the {@literal about} attribute of the HTML element representing this occurrence.
+     * <p>
+     * The value is derived from the URI of the occurrence and a blank node prefix is prepended to it.
+     * <p>
+     * The value is also assigned to the field {@link #elementAbout} for further use.
+     *
+     * @return Value of the {@literal about} attribute
+     */
+    public String resolveElementAbout() {
+        final String strIri = getUri().toString();
+        this.elementAbout = Constants.BNODE_PREFIX + strIri.substring(strIri.lastIndexOf('/') + 1);
+        return elementAbout;
+    }
+
     /**
      * Marks this term occurrence as suggested by automation.
      * <p>

src/main/java/cz/cvut/kbss/termit/model/selector/CssSelector.java

@@ -49,6 +49,11 @@ public void setValue(String value) {
         this.value = value;
     }
 
+    @Override
+    public CssSelector copy() {
+        return new CssSelector(value);
+    }
+
     @Override
     public boolean equals(Object o) {
         if (this == o) {

src/main/java/cz/cvut/kbss/termit/model/selector/FragmentSelector.java

@@ -49,6 +49,11 @@ public void setValue(String value) {
         this.value = value;
     }
 
+    @Override
+    public FragmentSelector copy() {
+        return new FragmentSelector(value);
+    }
+
     @Override
     public boolean equals(Object o) {
         if (this == o) {

src/main/java/cz/cvut/kbss/termit/model/selector/Selector.java

@@ -20,9 +20,10 @@
 import com.fasterxml.jackson.annotation.JsonTypeInfo;
 import cz.cvut.kbss.jopa.model.annotations.OWLClass;
 import cz.cvut.kbss.termit.model.AbstractEntity;
+import cz.cvut.kbss.termit.model.util.Copyable;
 import cz.cvut.kbss.termit.util.Vocabulary;
 
 @JsonTypeInfo(use = JsonTypeInfo.Id.CLASS, property = "javaClass")
 @OWLClass(iri = Vocabulary.s_c_selektor)
-public abstract class Selector extends AbstractEntity {
+public abstract class Selector extends AbstractEntity implements Copyable<Selector> {
 }

src/main/java/cz/cvut/kbss/termit/model/selector/TextPositionSelector.java

@@ -70,6 +70,11 @@ public void setEnd(Integer end) {
         this.end = end;
     }
 
+    @Override
+    public TextPositionSelector copy() {
+        return new TextPositionSelector(start, end);
+    }
+
     @Override
     public boolean equals(Object o) {
         if (this == o) {

src/main/java/cz/cvut/kbss/termit/model/selector/TextQuoteSelector.java

@@ -83,6 +83,11 @@ public void setSuffix(String suffix) {
         this.suffix = suffix;
     }
 
+    @Override
+    public TextQuoteSelector copy() {
+        return new TextQuoteSelector(exactMatch, prefix, suffix);
+    }
+
     @Override
     public boolean equals(Object o) {
         if (this == o) {

src/main/java/cz/cvut/kbss/termit/model/selector/XPathSelector.java

@@ -49,6 +49,11 @@ public void setValue(String value) {
         this.value = value;
     }
 
+    @Override
+    public XPathSelector copy() {
+        return new XPathSelector(value);
+    }
+
     @Override
     public boolean equals(Object o) {
         if (this == o) {

src/main/java/cz/cvut/kbss/termit/model/util/Copyable.java

@@ -0,0 +1,18 @@
+package cz.cvut.kbss.termit.model.util;
+
+/**
+ * This interface is used to mark classes that can create a copy of themselves.
+ *
+ * @param <T> Type of the object
+ */
+public interface Copyable<T extends HasIdentifier> {
+
+    /**
+     * Creates a copy of this instance, copying the attribute values.
+     * <p>
+     * Note that the identifier should not be copied into the new instance, so that it can be persisted.
+     *
+     * @return New instance with values copied from this one
+     */
+    T copy();
+}

src/main/java/cz/cvut/kbss/termit/persistence/dao/TermOccurrenceDao.java

@@ -255,7 +255,7 @@ public void removeAll(Asset<?> target) {
         Objects.requireNonNull(target);
 
         final URI sourceContext = TermOccurrence.resolveContext(target.getUri());
-        LOG.debug("Removing all occurrences from {}", sourceContext);
+        LOG.debug("Removing all occurrences from {}", Utils.uriToString(sourceContext));
         em.createNativeQuery("DROP GRAPH ?context")
                 .setParameter("context", sourceContext)
                 .executeUpdate();

src/main/java/cz/cvut/kbss/termit/service/document/TermOccurrenceResolver.java

@@ -29,6 +29,7 @@
 
 import java.io.InputStream;
 import java.net.URI;
+import java.util.ArrayList;
 import java.util.Collections;
 import java.util.List;
 
@@ -39,7 +40,7 @@ public abstract class TermOccurrenceResolver {
 
     protected final TermRepositoryService termService;
 
-    protected List<TermOccurrence> existingOccurrences = Collections.emptyList();
+    protected List<TermOccurrence> existingApprovedOccurrences = Collections.emptyList();
 
     protected TermOccurrenceResolver(TermRepositoryService termService) {
         this.termService = termService;
@@ -58,11 +59,14 @@ protected TermOccurrenceResolver(TermRepositoryService termService) {
 
     /**
      * Sets occurrences that already existed on previous analyses.
+     * <p>
+     * The resolver uses only those that are approved.
      *
      * @param existingOccurrences Term occurrences from the previous analysis run
      */
     public void setExistingOccurrences(List<TermOccurrence> existingOccurrences) {
-        this.existingOccurrences = existingOccurrences;
+        this.existingApprovedOccurrences = new ArrayList<>(
+                existingOccurrences.stream().filter(to -> !to.isSuggested()).toList());
     }
 
     /**

src/main/java/cz/cvut/kbss/termit/service/document/TermOccurrenceSelectorCreator.java

@@ -0,0 +1,75 @@
+package cz.cvut.kbss.termit.service.document;
+
+import cz.cvut.kbss.termit.exception.SelectorGenerationException;
+import cz.cvut.kbss.termit.exception.UnsupportedOperationException;
+import cz.cvut.kbss.termit.model.assignment.DefinitionalOccurrenceTarget;
+import cz.cvut.kbss.termit.model.assignment.FileOccurrenceTarget;
+import cz.cvut.kbss.termit.model.assignment.OccurrenceTarget;
+import cz.cvut.kbss.termit.model.resource.File;
+import cz.cvut.kbss.termit.model.resource.Resource;
+import cz.cvut.kbss.termit.model.selector.Selector;
+import cz.cvut.kbss.termit.service.business.ResourceService;
+import cz.cvut.kbss.termit.service.document.html.HtmlSelectorGenerators;
+import cz.cvut.kbss.termit.util.Constants;
+import jakarta.annotation.Nonnull;
+import org.jsoup.Jsoup;
+import org.jsoup.nodes.Document;
+import org.jsoup.nodes.Element;
+import org.jsoup.select.Elements;
+import org.springframework.stereotype.Component;
+
+import java.util.Objects;
+import java.util.Set;
+
+/**
+ * Creates selectors for a new term occurrence.
+ */
+@Component
+public class TermOccurrenceSelectorCreator {
+
+    private final HtmlSelectorGenerators selectorGenerators;
+
+    private final DocumentManager documentManager;
+
+    private final ResourceService resourceService;
+
+    public TermOccurrenceSelectorCreator(HtmlSelectorGenerators selectorGenerators, DocumentManager documentManager,
+                                         ResourceService resourceService) {
+        this.selectorGenerators = selectorGenerators;
+        this.documentManager = documentManager;
+        this.resourceService = resourceService;
+    }
+
+    /**
+     * Creates selectors for a term occurrence in the specified target represented by an HTML element with the specified
+     * id.
+     *
+     * @param target       Asset in which the occurrence is to be found
+     * @param elementAbout Value of the {@literal about} attribute of the occurrence element
+     * @return Set of generated selectors for the occurrence element
+     * @throws UnsupportedOperationException If the specified target is not supported
+     * @throws SelectorGenerationException   If unable to generate selectors
+     */
+    public Set<Selector> createSelectors(@Nonnull OccurrenceTarget target, @Nonnull String elementAbout) {
+        Objects.requireNonNull(target);
+        Objects.requireNonNull(elementAbout);
+        if (target instanceof DefinitionalOccurrenceTarget) {
+            throw new UnsupportedOperationException("Cannot create selectors for a term definitional occurrence.");
+        }
+        final FileOccurrenceTarget ft = (FileOccurrenceTarget) target;
+        final Document targetContent = loadTargetContent(ft);
+        final Elements elements = targetContent.select(
+                "[" + Constants.RDFa.ABOUT + "=" + Constants.BNODE_PREFIX + elementAbout + "]");
+        if (elements.isEmpty()) {
+            throw new SelectorGenerationException("No element with id " + elementAbout + " found in " + ft.getSource());
+        }
+        return selectorGenerators.generateSelectors(elements.toArray(new Element[]{}));
+    }
+
+    private Document loadTargetContent(FileOccurrenceTarget ft) {
+        final Resource targetFile = resourceService.findRequired(ft.getSource());
+        assert targetFile instanceof File;
+        final String content = documentManager.loadFileContent((File) targetFile);
+        return Jsoup.parse(content);
+    }
+}