Predefined xpath functions

cli/src/main/java-templates/AbstractNormalize.java

@@ -12,6 +12,10 @@
 import java.util.concurrent.Callable;
 
 import net.sf.saxon.s9api.Processor;
+import net.sf.saxon.s9api.XPathCompiler;
+import net.sf.saxon.s9api.XsltCompiler;
+import net.sf.saxon.s9api.XsltPackage;
+import net.sf.saxon.s9api.SaxonApiException;
 
 import de.wwu.scdh.annotation.selection.resource.DOMResource;
 import de.wwu.scdh.annotation.selection.resource.ResourceBuilder;
@@ -26,7 +30,22 @@ abstract class AbstractNormalize {
 
     protected static final Processor PROC = new Processor();
 
-    enum Normalizer {
+	XPathCompiler compiler = PROC.newXPathCompiler();
+
+	protected void compileXPathLibrary() {
+		try {
+			File library = new File(DOMResource.class.getResource("/xslt/xpath.xsl").getPath());
+			XsltCompiler xsltCompiler = PROC.newXsltCompiler();
+			XsltPackage functionLibrary = xsltCompiler.compilePackage(library);
+			compiler.addXsltFunctionLibrary(functionLibrary);
+			compiler.declareNamespace("sel", "http://wwu.de/scdh/selection-engine/xpaths");
+		} catch (SaxonApiException e) {
+			System.err.println("failed to load xpath library: " + e.getMessage());
+			System.exit(2);
+		}
+	}
+
+	enum Normalizer {
 	FROM_ROOT_CLARK,
 	FROM_DEEPEST_ID_CLARK
     }
@@ -148,12 +167,10 @@ protected String getNormalizerXPath() throws CliException {
     }
 
     protected XPathNormalizerWithXPath getXPathNormalizer() throws CliException {
-	return new XPathNormalizerWithXPath(getNormalizerXPath());
+	return new XPathNormalizerWithXPath(PROC.newXPathCompiler(), getNormalizerXPath());
     }
 
-    protected NormalizerFactory getNormalizerFactory() {
-	return new NormalizerFactory();
-    }
+    protected NormalizerFactory getNormalizerFactory(){ return new NormalizerFactory(PROC.newXPathCompiler()); }
 
     protected RewriterConfig getRewriterConfig() throws CliException {
 	return new RewriterConfig(mode, false, getNormalizerXPath(), true, true);

cli/src/main/java-templates/Normalize.java

@@ -86,14 +86,14 @@ public Integer call() throws Exception {
 	XPathNormalizerWithXPath xpathNormalizer;
 	if (normalizer.equals(Normalizer.FROM_DEEPEST_ID_CLARK)) {
 	    try {
-		xpathNormalizer = new XPathNormalizerWithXPath(XPathNormalizerWithXPath.FROM_DEEPEST_ID_CLARK_XPATH);
+		xpathNormalizer = new XPathNormalizerWithXPath(compiler, XPathNormalizerWithXPath.FROM_DEEPEST_ID_CLARK_XPATH);
 	    } catch (Exception e) {
 		System.err.println(e.getMessage());
 		return 2;
 	    }
 	} else if (normalizer.equals(Normalizer.FROM_ROOT_CLARK)) {
 	    try {
-		xpathNormalizer = new XPathNormalizerWithXPath(XPathNormalizerWithXPath.FROM_ROOT_CLARK_XPATH);
+		xpathNormalizer = new XPathNormalizerWithXPath(compiler, XPathNormalizerWithXPath.FROM_ROOT_CLARK_XPATH);
 	    } catch (Exception e) {
 		System.err.println(e.getMessage());
 		return 2;

cli/src/main/java-templates/TransformSimple.java

@@ -128,13 +128,13 @@ public Integer call() throws Exception {
 	RewriterFactory factory;
 	Rewriter rewriter;
 	if (backward) {
-	    factory = new BackwardMappingFactory();
+	    factory = new BackwardMappingFactory(compiler);
 	    rewriter = factory.getRewriter
 		(preimage.getImage().getPointClass(),
 		 XPathRefinedByRFC5147CharScheme.class,
 		 getRewriterConfig());
 	} else {
-	    factory = new ForwardMappingFactory();
+	    factory = new ForwardMappingFactory(compiler);
 	    rewriter = factory.getRewriter
 		(XPathRefinedByRFC5147CharScheme.class,
 		 preimage.getImage().getPointClass(),

core/pom.xml

@@ -75,6 +75,36 @@
         </executions>
       </plugin>
 
+      <plugin>
+        <groupId>com.nkutsche</groupId>
+        <artifactId>xspec-maven-plugin</artifactId>
+        <version>2.1.1</version>
+        <dependencies>
+          <dependency>
+            <groupId>io.xspec</groupId>
+            <artifactId>xspec</artifactId>
+            <version>2.2.4</version>
+            <classifier>enduser-files</classifier>
+            <type>zip</type>
+          </dependency>
+          <dependency>
+            <groupId>net.sf.saxon</groupId>
+            <artifactId>Saxon-HE</artifactId>
+            <version>${saxon.version}</version>
+          </dependency>
+        </dependencies>
+        <executions>
+          <execution>
+            <goals>
+              <goal>run-xspec</goal>
+            </goals>
+            <configuration>
+              <testDir>${project.basedir}/src/main/resources/xslt</testDir>
+            </configuration>
+          </execution>
+        </executions>
+      </plugin>
+
     </plugins>
   </build>
 

core/src/main/java/de/wwu/scdh/annotation/selection/rewriter/BackwardMappingFactory.java

@@ -8,13 +8,24 @@
 import de.wwu.scdh.annotation.selection.RewriterFactory;
 import de.wwu.scdh.annotation.selection.point.RFC5147CharScheme;
 import de.wwu.scdh.annotation.selection.point.XPathRefinedByRFC5147CharScheme;
+import net.sf.saxon.s9api.XPathCompiler;
 import org.slf4j.Logger;
 import org.slf4j.LoggerFactory;
 
 public class BackwardMappingFactory implements RewriterFactory {
 
 	private static Logger Log = LoggerFactory.getLogger(ForwardMappingFactory.class);
 
+	private final XPathCompiler compiler;
+
+	/**
+	 * Creates a new {@link BackwardMappingFactory} with a {@link XPathCompiler}.
+	 * @param compiler - A compiler used for evaluating XPath expressions.
+	 */
+	public BackwardMappingFactory(XPathCompiler compiler) {
+		this.compiler = compiler;
+	}
+
 	// @SuppressWarnings("unchecked")
 	@Override
 	public <R extends Resource<?>, P1 extends Point, P2 extends Point, RW extends Rewriter<R, P1, P2>> RW getRewriter(
@@ -24,10 +35,10 @@ public <R extends Resource<?>, P1 extends Point, P2 extends Point, RW extends Re
 		RW rc;
 		if (XPathRefinedByRFC5147CharScheme.class.isAssignableFrom(point1)
 				&& XPathRefinedByRFC5147CharScheme.class.isAssignableFrom(point2)) {
-			rc = (RW) new XPathRefinedByRFC5147CharSchemeBackwardMapper(config.getXPath());
+			rc = (RW) new XPathRefinedByRFC5147CharSchemeBackwardMapper(compiler, config.getXPath());
 		} else if (RFC5147CharScheme.class.isAssignableFrom(point1)
 				&& XPathRefinedByRFC5147CharScheme.class.isAssignableFrom(point2)) {
-			rc = (RW) new XPathRefinedByRFC5147CharSchemeToTextBackwardMapper(config.getXPath());
+			rc = (RW) new XPathRefinedByRFC5147CharSchemeToTextBackwardMapper(compiler, config.getXPath());
 		} else {
 			Log.error("no backward mapping for {}, {}", point1.getCanonicalName(), point2.getCanonicalName());
 			throw new ConfigurationException(

core/src/main/java/de/wwu/scdh/annotation/selection/rewriter/ForwardMappingFactory.java

@@ -8,13 +8,24 @@
 import de.wwu.scdh.annotation.selection.RewriterFactory;
 import de.wwu.scdh.annotation.selection.point.RFC5147CharScheme;
 import de.wwu.scdh.annotation.selection.point.XPathRefinedByRFC5147CharScheme;
+import net.sf.saxon.s9api.XPathCompiler;
 import org.slf4j.Logger;
 import org.slf4j.LoggerFactory;
 
 public class ForwardMappingFactory implements RewriterFactory {
 
 	private static Logger Log = LoggerFactory.getLogger(ForwardMappingFactory.class);
 
+	private final XPathCompiler compiler;
+
+	/**
+	 * Creates a new {@link BackwardMappingFactory} with a {@link XPathCompiler}.
+	 * @param compiler - A compiler used for evaluating XPath expressions.
+	 */
+	public ForwardMappingFactory(XPathCompiler compiler) {
+		this.compiler = compiler;
+	}
+
 	// @SuppressWarnings("unchecked")
 	@Override
 	public <R extends Resource<?>, P1 extends Point, P2 extends Point, RW extends Rewriter<R, P1, P2>> RW getRewriter(
@@ -26,10 +37,10 @@ public <R extends Resource<?>, P1 extends Point, P2 extends Point, RW extends Re
 		RW rc;
 		if (XPathRefinedByRFC5147CharScheme.class.isAssignableFrom(point1)
 				&& XPathRefinedByRFC5147CharScheme.class.isAssignableFrom(point2)) {
-			rc = (RW) new XPathRefinedByRFC5147CharSchemeForwardMapper(config.getXPath());
+			rc = (RW) new XPathRefinedByRFC5147CharSchemeForwardMapper(compiler, config.getXPath());
 		} else if (XPathRefinedByRFC5147CharScheme.class.isAssignableFrom(point1)
 				&& RFC5147CharScheme.class.isAssignableFrom(point2)) {
-			rc = (RW) new XPathRefinedByRFC5147CharSchemeToTextForwardMapper(config.getXPath());
+			rc = (RW) new XPathRefinedByRFC5147CharSchemeToTextForwardMapper(compiler, config.getXPath());
 		} else {
 			Log.error("no forward mapping for {}, {}", point1.getCanonicalName(), point2.getCanonicalName());
 			throw new ConfigurationException(

core/src/main/java/de/wwu/scdh/annotation/selection/rewriter/NormalizerFactory.java

@@ -7,9 +7,20 @@
 import de.wwu.scdh.annotation.selection.RewriterConfig;
 import de.wwu.scdh.annotation.selection.RewriterFactory;
 import de.wwu.scdh.annotation.selection.point.XPathRefinedByRFC5147CharScheme;
+import net.sf.saxon.s9api.XPathCompiler;
 
 public class NormalizerFactory implements RewriterFactory {
 
+	private final XPathCompiler compiler;
+
+	/**
+	 * Creates a new {@link BackwardMappingFactory} with a {@link XPathCompiler}.
+	 * @param compiler - A compiler used for evaluating XPath expressions.
+	 */
+	public NormalizerFactory(XPathCompiler compiler) {
+		this.compiler = compiler;
+	}
+
 	// @SuppressWarnings("unchecked")
 	@Override
 	public <R extends Resource<?>, P1 extends Point, P2 extends Point, RW extends Rewriter<R, P1, P2>> RW getRewriter(
@@ -19,7 +30,7 @@ public <R extends Resource<?>, P1 extends Point, P2 extends Point, RW extends Re
 		RW rc;
 		if (XPathRefinedByRFC5147CharScheme.class.isAssignableFrom(point2)
 				&& XPathRefinedByRFC5147CharScheme.class.isAssignableFrom(point2)) {
-			rc = (RW) new XPathNormalizerWithXPath(config.getXPath());
+			rc = (RW) new XPathNormalizerWithXPath(compiler, config.getXPath());
 		} else {
 			throw new ConfigurationException(
 					"no rewriter for " + point1.getClass().getCanonicalName() + " ; "

core/src/main/java/de/wwu/scdh/annotation/selection/rewriter/XPathNormalizer.java

@@ -3,11 +3,11 @@
 import de.wwu.scdh.annotation.selection.*;
 import de.wwu.scdh.annotation.selection.point.XPathRefinedByRFC5147CharScheme;
 import de.wwu.scdh.annotation.selection.resource.DOMResource;
+import de.wwu.scdh.annotation.selection.resource.S9ApiResource;
 import java.util.List;
+import net.sf.saxon.s9api.XPathCompiler;
 import net.sf.saxon.s9api.XdmNode;
 import org.apache.commons.lang3.tuple.Pair;
-import org.slf4j.Logger;
-import org.slf4j.LoggerFactory;
 
 /**
  * The same position (or range) in a {@link DOMResource} can be
@@ -19,7 +19,7 @@
  *
  * Normalization of selectors is a 2-stage process: 1) In the first
  * stage, the text position (or range) which is referenced by the
- * selector, has to be found. A pair containing a text node and a RFC
+ * selector, has to be found. A pair containing a text node and an RFC
  * 5147 character scheme position is returned. 2) In the second stage,
  * this position is expressed as a selector again, i.e., the node is
  * referenced with an XPath, where the XPath may be written as a path
@@ -46,14 +46,12 @@
 public abstract class XPathNormalizer extends XPathRewriterBase
 		implements Rewriter<DOMResource, XPathRefinedByRFC5147CharScheme, XPathRefinedByRFC5147CharScheme> {
 
-	private static final Logger LOG = LoggerFactory.getLogger(XPathNormalizer.class);
-
-	// protected final DOMResource resource;
-
 	/**
-	 * Make a new {@link XPathNormalizer} for a {@link DOMResource}.
+	 * {@inheritDoc}
 	 */
-	public XPathNormalizer() {}
+	public XPathNormalizer(XPathCompiler xPathCompiler, String xpath) {
+		super(xPathCompiler, xpath);
+	}
 
 	/**
 	 * Normalize an XPath Selector refined by an RFC5147 character
@@ -78,18 +76,19 @@ public List<XPathRefinedByRFC5147CharScheme> rewrite(
 				getTextNodeAtPosition(resource, point.getXPath(), point.getChar(), config.getMode());
 		String normalizedXPath = getNormalizedXPath(resource, textNode.getLeft(), config.getEscaped());
 		XdmNode normalizedNode = getNode(resource, unespace(normalizedXPath));
-		Integer normalizedPos = posInNormalizedNode(textNode.getLeft(), textNode.getRight(), normalizedNode);
+		int normalizedPos = posInNormalizedNode(textNode.getLeft(), textNode.getRight(), normalizedNode);
 		return List.of(new XPathRefinedByRFC5147CharScheme(normalizedXPath, normalizedPos));
 	}
 
 	/**
 	 * This abstract method must be implemented by normalizers. It
 	 * returns the normalized path to the node given as argument.
 	 *
-	 * @param node  an {@link XdmNode} to which the normlized path must be generated to
-	 * @param escaped  whether or not the normalized XPath is to be escaped for X-processing, e.g., <code>'</code> escaped to <code>&amp;apos;</code>.
+	 * @param node  an {@link XdmNode} to which the normalized path must be generated to
+	 * @param escaped  whether or not the normalized XPath is to be escaped for X-processing, e.g.,
+	 *                    <code>'</code> escaped to <code>&amp;apos;</code>.
 	 * @return the normalized XPath expression as a String
-	 * @see XPathNormalizer#getTextNodeAtPosition(String, int, boolean)
+	 * @see XPathRewriterBase#getTextNodeAtPosition(S9ApiResource, String, int, Mode)
 	 */
 	protected abstract String getNormalizedXPath(DOMResource resource, XdmNode node, boolean escaped)
 			throws SelectorException;

core/src/main/java/de/wwu/scdh/annotation/selection/rewriter/XPathNormalizerWithXPath.java

@@ -2,6 +2,8 @@
 
 import de.wwu.scdh.annotation.selection.*;
 import de.wwu.scdh.annotation.selection.resource.DOMResource;
+import de.wwu.scdh.annotation.selection.resource.S9ApiResource;
+import net.sf.saxon.s9api.XPathCompiler;
 import net.sf.saxon.s9api.XdmNode;
 
 /**
@@ -10,14 +12,14 @@
  * context node in question.<P>
  *
  * The context item on which the XPath will be evaluated, is
- * determined by {@link XPathNormalizer#getTextNodeAtPosition(String, int,boolean)}.
+ * determined by {@link XPathRewriterBase#getTextNodeAtPosition(S9ApiResource, String, int, Mode)}.
  * This is generally a text node.
  */
 public class XPathNormalizerWithXPath extends XPathNormalizer {
 
 	/**
 	 * An XPath expression suitable for normalizing path expressions
-	 * to nodes.<P>
+	 * to nodes.<P/>
 	 *
 	 * Generates a path expression using the <code>path(.)</code>
 	 * XPath function. This result in a path expression stepping from
@@ -26,6 +28,13 @@ public class XPathNormalizerWithXPath extends XPathNormalizer {
 	 */
 	public static final String PATH_FUNCTION_XPATH = "path(.)";
 
+	/**
+	 * {@inheritDoc}
+	 */
+	public XPathNormalizerWithXPath(XPathCompiler xPathCompiler, String xpath) {
+		super(xPathCompiler, xpath);
+	}
+
 	/**
 	 * An XPath expression suitable for normalizing path expressions
 	 * to nodes.<P>
@@ -35,7 +44,7 @@ public class XPathNormalizerWithXPath extends XPathNormalizer {
 	 * notation.
 	 */
 	public static final String FROM_ROOT_CLARK_XPATH =
-			"let $ctx:=., $elsteps:=$ctx/ancestor-or-self::element(), $txtstep:=$ctx[self::text()] return ($elsteps ! (let $step:=., $ns:=namespace-uri($step), $name:=name($step), $pos:=count($step/preceding-sibling::node()[name() eq $name])+1 return concat('/Q{', $ns, '}', $name, '[', $pos, ']')), $txtstep ! (concat('/text()[', count($txtstep/preceding-sibling::text())+1, ']'))) => string-join('')";
+			"let $ctx:=., $elSteps:=$ctx/ancestor-or-self::element(), $txtStep:=$ctx[self::text()] return ($elSteps ! (let $step:=., $ns:=namespace-uri($step), $name:=name($step), $pos:=count($step/preceding-sibling::node()[name() eq $name])+1 return concat('/Q{', $ns, '}', $name, '[', $pos, ']')), $txtStep ! (concat('/text()[', count($txtStep/preceding-sibling::text())+1, ']'))) => string-join('')";
 
 	/**
 	 * An XPath expression suitable for normalizing path expressions
@@ -46,31 +55,14 @@ public class XPathNormalizerWithXPath extends XPathNormalizer {
 	 * node. Element names are QNames in Clark notation.
 	 */
 	public static final String FROM_DEEPEST_ID_CLARK_XPATH =
-			"let $ctx:=., $elsteps:=$ctx/ancestor-or-self::element(), $ids:=for $step in 1 to count($elsteps) return if ($elsteps[$step]/@xml:id) then $step else -1, $idat:=$ids[. ne -1][last()], $idstep:=if ($idat) then concat('id(&apos;', $elsteps[$idat]/@xml:id, '&apos;)') else '', $txtstep:=$ctx[self::text()] return concat($idstep, ($elsteps[position() gt $idat or empty($idat)] ! (let $step:=., $ns:=namespace-uri($step), $name:=name($step), $pos:=count($step/preceding-sibling::node()[name() eq $name])+1 return concat('/Q{', $ns, '}', $name, '[', $pos, ']')), $txtstep ! (concat('/text()[', count($txtstep/preceding-sibling::text())+1, ']'))) => string-join(''))";
-
-	protected final String xpath;
-
-	/**
-	 * Create a new {@link XPathNormalizerWithXPath} normalizer for a
-	 * {@link DOMResource}.
-	 *
-	 * @param xpath  an XPath expression which will be evaluated on a context node for normalization
-	 */
-	public XPathNormalizerWithXPath(String xpath) {
-		super();
-		this.xpath = xpath;
-	}
+			"let $ctx:=., $elSteps:=$ctx/ancestor-or-self::element(), $ids:=for $step in 1 to count($elSteps) return if ($elSteps[$step]/@xml:id) then $step else -1, $idAt:=$ids[. ne -1][last()], $idStep:=if ($idAt) then concat('id(&apos;', $elSteps[$idAt]/@xml:id, '&apos;)') else '', $txtStep:=$ctx[self::text()] return concat($idStep, ($elSteps[position() gt $idAt or empty($idAt)] ! (let $step:=., $ns:=namespace-uri($step), $name:=name($step), $pos:=count($step/preceding-sibling::node()[name() eq $name])+1 return concat('/Q{', $ns, '}', $name, '[', $pos, ']')), $txtStep ! (concat('/text()[', count($txtStep/preceding-sibling::text())+1, ']'))) => string-join(''))";
 
 	/**
 	 * Get the path to the {@link XdmNode} given as parameter by using
 	 * the XPath expressing the normalizer was initialized with.
 	 */
 	@Override
 	protected String getNormalizedXPath(DOMResource resource, XdmNode node, boolean escaped) throws SelectorException {
-		return pathExpressionWithXPath(getXPath(), node, escaped, resource.getProcessor());
-	}
-
-	public String getXPath() {
-		return xpath;
+		return pathExpressionWithXPath(getXPath(), node, escaped);
 	}
 }