rewrite(
getNode(preimage.getImage().getContents(), unespace(normalizedXPath), preimage.getProcessor());
Integer normalizedPos = posInNormalizedNode(imageNode, preimagePair.getRight(), normalizedNode);
transformed.add(
- new XPathRefinedByRFC5147CharScheme(replaceTraceTextLeaf(normalizedXPath, config), normalizedPos));
+ new XPathRefinedByRFC5147CharScheme(postProcForwardPaths(normalizedXPath, config), normalizedPos));
}
return transformed;
}
diff --git a/core/src/main/java/de/wwu/scdh/annotation/selection/rewriter/XPathRewriterBase.java b/core/src/main/java/de/wwu/scdh/annotation/selection/rewriter/XPathRewriterBase.java
index 03fa567..5a39333 100644
--- a/core/src/main/java/de/wwu/scdh/annotation/selection/rewriter/XPathRewriterBase.java
+++ b/core/src/main/java/de/wwu/scdh/annotation/selection/rewriter/XPathRewriterBase.java
@@ -41,12 +41,22 @@ public abstract class XPathRewriterBase {
public static Pattern TEXT_LEAF = Pattern.compile("text\\(\\)\\[(\\d+)]$");
+ public static final Pattern EMPTY_NAMESPACE = Pattern.compile("Q\\{}");
+
public static String replaceTraceTextLeaf(String xpath) {
Matcher matcher = TRACE_TEXT_LEAF.matcher(xpath);
return matcher.replaceAll("text()[$1]");
}
- public static String replaceTraceTextLeaf(String xpath, RewriterConfig config) {
+ public static String replaceEmptyNamespace(String xpath) {
+ Matcher matcher = EMPTY_NAMESPACE.matcher(xpath);
+ return matcher.replaceAll("");
+ }
+
+ public static String postProcForwardPaths(String xpath, RewriterConfig config) {
+ if (config.removeEmptyNamespaces()) {
+ xpath = replaceEmptyNamespace(xpath);
+ }
if (config.rewritesTraceLeaf()) {
return replaceTraceTextLeaf(xpath);
} else {
@@ -77,7 +87,7 @@ public XPathRewriterBase() {}
/**
* This method run the first stage of the normalization
* process. It gets the node where the pair of XPath and position
- * of an character-scheme-refined XPath selector points to. Either
+ * of a character-scheme-refined XPath selector points to. Either
* this node is a text node, or the selector is invalid in the
* context of the current document. In general, the resulting
* position is not the same as the input position.
@@ -95,8 +105,8 @@ public XPathRewriterBase() {}
* @param resource the {@link DOMResource} on the base of which the normalization is done
* @param xpath the XPath part of the XPath selector
* @param position the position following the character scheme of RFC5147
- * @param mode an normalization algorithm selected from {@link Mode}
- * @throws {@link SelectorException}
+ * @param mode a normalization algorithm selected from {@link Mode}
+ * @throws SelectorException if the XPath does not select exactly one node
* @return a pair of node and position
*/
protected Pair getTextNodeAtPosition(
@@ -117,7 +127,7 @@ protected Pair getTextNodeAtPosition(
}
/**
- * The implementation of step 1 of the normalization algorithm in
+ * The implementation of step 1 of the normalization algorithm
* in mode {@link Mode#FIRST}.
*/
protected final Pair getFirstNodeAtPosition(
@@ -132,7 +142,7 @@ protected final Pair getFirstNodeAtPosition(
}
/**
- * The implementation of step 1 of the normalization algorithm in
+ * The implementation of step 1 of the normalization algorithm
* in mode {@link Mode#SECOND}.
*/
protected final Pair getSecondNodeAtPosition(
@@ -149,7 +159,7 @@ protected final Pair getSecondNodeAtPosition(
}
/**
- * The implementation of step 1 of the normalization algorithm in
+ * The implementation of step 1 of the normalization algorithm
* in mode {@link Mode#FIRST_OF_DEEPEST_NODES}.
*/
protected final Pair getFirstOfDeepestNodesAtPosition(
@@ -162,7 +172,8 @@ protected final Pair getFirstOfDeepestNodesAtPosition(
return nodesAtPosition.get(0);
} else {
// we still have to get the text node with the deepest path
- LOG.debug("found {} nodes, getting deepest", nodesAtPosition.size());
+ LOG.debug(
+ "found {} nodes, getting deepest (get first of deepest nodes at position)", nodesAtPosition.size());
// note, that Stream.max() returns the first of the items with the maximum value
Optional> deepest =
nodesAtPosition.stream().max(Comparator.comparing(XPathNormalizer::getDepth));
@@ -171,7 +182,7 @@ protected final Pair getFirstOfDeepestNodesAtPosition(
}
/**
- * The implementation of step 1 of the normalization algorithm in
+ * The implementation of step 1 of the normalization algorithm
* in mode {@link Mode#LAST_OF_DEEPEST_NODES}.
*/
protected final Pair getLastOfDeepestNodesAtPosition(
@@ -200,7 +211,7 @@ protected final Pair getLastOfDeepestNodesAtPosition(
* @param resource the {@link DOMResource} on the base of which the normalization is done
* @param xpath the XPath part of the XPath selector
* @param position the position following the character scheme of RFC5147
- * @throws {@link SelectorException}
+ * @throws SelectorException if xpath does not evaluate to one node
* @return a pair of node and position
*/
protected final Pair getDeepTextNodeAtPositionStopAtEnd(
@@ -221,7 +232,7 @@ protected final Pair getDeepTextNodeAtPositionStopAtEnd(
* @param resource the {@link DOMResource} on the base of which the normalization is done
* @param xpath the XPath part of the XPath selector
* @param position the position following the character scheme of RFC5147
- * @throws {@link SelectorException}
+ * @throws SelectorException if xpath does not get one node
* @return a pair of node and position
*/
protected final Pair getDeepTextNodeAtPositionStepOverEnd(
@@ -238,7 +249,7 @@ protected final Pair getDeepTextNodeAtPositionStepOverEnd(
}
/**
- * Get the node from the DOM resource given by the the XPath
+ * Get the node from the DOM resource given by the XPath
* passed as argument. If the XPath does not evaluate to a single
* node, this method raises an {@link SelectorException}.
*
@@ -257,10 +268,8 @@ protected final XdmNode getNode(DOMResource resource, String xpath) throws Selec
// assert that the XPath selects exactly 1 node
if (nodes.size() != 1) {
LOG.error("XPath '{}' does not select exactly one node: selects {} nodes", xpath, nodes.size());
- throw new SelectorException("XPath '" + xpath
- + "' does not select exactly one node: selects "
- + String.valueOf(nodes.size())
- + " nodes");
+ throw new SelectorException(
+ "XPath '" + xpath + "' does not select exactly one node: selects " + nodes.size() + " nodes");
} else if (!nodes.itemAt(0).isNode()) {
LOG.error("Node selected by XPath '{}' does not select a node", xpath);
throw new SelectorException("XPath '" + xpath + "' does not select a node");
@@ -274,7 +283,7 @@ protected final XdmNode getNode(DOMResource resource, String xpath) throws Selec
}
/**
- * Get the node from the XDM value resource given by the the XPath
+ * Get the node from the XDM value resource given by the XPath
* passed as argument. If the XPath does not evaluate to a single
* node, this method raises an {@link SelectorException}.
*
@@ -287,7 +296,7 @@ protected final XdmNode getNode(XdmValueResource resource, String xpath) throws
}
/**
- * Get the node from the DOM resource given by the the XPath
+ * Get the node from the DOM resource given by the XPath
* passed as argument. If the XPath does not evaluate to a single
* node, this method raises an {@link SelectorException}.
*/
@@ -303,13 +312,13 @@ protected final XdmNode getNode(XdmValue value, String xpath, Processor processo
}
if (result.size() != 1) {
LOG.error(
- "XPath '{}' does not select exaclty one node in XdmValueResource: selects {} nodes",
+ "XPath '{}' does not select exactly one node in XdmValueResource: selects {} nodes",
xpath,
result.size());
throw new SelectorException(
"XPath '" + xpath + "' does not select exactly one node in XdmValueResource");
} else if (!result.itemAt(0).isNode()) {
- LOG.error("XPath '{}' does not select a node", xpath, result.size());
+ LOG.error("XPath '{}' does not select a node", xpath);
throw new SelectorException("XPath '" + xpath + "' does not select a node");
} else {
return (XdmNode) result.itemAt(0);
@@ -326,7 +335,7 @@ protected final XdmNode getNode(XdmValue value, String xpath, Processor processo
* a node and a position, that contain the character scheme
* position inside a given fragment.
*
- * This method collects all canditates in case of referential
+ * This method collects all candidates in case of referential
* ambiguity. See {@link Mode}.
*
* @param fragment as {@link XdmNode} inside a {@link DOMResource}
@@ -336,7 +345,7 @@ protected final List> getDescendantTextNodesWithPosition(
Iterator descendants = fragment.axisIterator(Axis.DESCENDANT_OR_SELF);
int charsEaten = 0;
XdmNode node = fragment;
- List> nodesAtPosition = new ArrayList>();
+ List> nodesAtPosition = new ArrayList<>();
while (descendants.hasNext()) {
node = descendants.next();
LOG.debug("investigating '{}' node", node.getUnderlyingNode().getLocalPart());
@@ -347,7 +356,7 @@ protected final List> getDescendantTextNodesWithPosition(
charsEaten += length;
} else {
// position is inside this text node or at its end
- nodesAtPosition.add(new ImmutablePair(node, position - charsEaten));
+ nodesAtPosition.add(new ImmutablePair<>(node, position - charsEaten));
if (position < charsEaten + length) {
// we can stop, since positions in all further text nodes will have greater positions
break;
@@ -367,14 +376,14 @@ protected final List> getDescendantTextNodesWithPosition(
* position inside a given fragment, or in the first text node
* before or after it.
*
- * This method collects all canditates in case of referential
+ * This method collects all candidates in case of referential
* ambiguity. See {@link Mode}.
*
* @param fragment as {@link XdmNode} inside a {@link DOMResource}
* @param position the RFC 5147 character scheme position inside the fragment
*/
protected final List> getTextNodesWithPosition(final XdmNode fragment, final int position) {
- List> nodesAtPosition = new ArrayList>();
+ List> nodesAtPosition = new ArrayList<>();
XdmNode node;
// 1. before the fragment, if position is zero
if (position == 0) {
@@ -386,7 +395,7 @@ protected final List> getTextNodesWithPosition(final XdmN
textNodeSeen = true;
int length =
node.getUnderlyingValue().getUnicodeStringValue().length32();
- nodesAtPosition.add(new ImmutablePair(node, length));
+ nodesAtPosition.add(new ImmutablePair<>(node, length));
}
}
}
@@ -404,7 +413,7 @@ protected final List> getTextNodesWithPosition(final XdmN
charsEaten += length;
} else {
// position is inside this text node or at its end
- nodesAtPosition.add(new ImmutablePair(node, position - charsEaten));
+ nodesAtPosition.add(new ImmutablePair<>(node, position - charsEaten));
if (position < charsEaten + length) {
// we can stop, since positions in all further text nodes will have greater positions
break;
@@ -422,7 +431,7 @@ protected final List> getTextNodesWithPosition(final XdmN
node = following.next();
if (node.getNodeKind().equals(XdmNodeKind.TEXT)) {
textNodeSeen = true;
- nodesAtPosition.add(new ImmutablePair(node, 0));
+ nodesAtPosition.add(new ImmutablePair<>(node, 0));
}
}
}
@@ -433,7 +442,6 @@ protected final List> getTextNodesWithPosition(final XdmN
* Get the new position value based on the old one and the
* fragment selected by the normalized XPath.
*
- * @param resource the {@link DOMResource} operating on
* @param textNode the text node as {@link XdmNode} gotten from step 1
* @param pos the position gotten form step 1
* @param fragment the {@link XdmNode} selected by the XPath resulting from step 2
@@ -446,9 +454,9 @@ protected int posInNormalizedNode(XdmNode textNode, int pos, XdmNode fragment) t
// iter over all descendant text nodes until we found textNode
boolean found = false;
int posAcc = pos;
- Iterator descenant = fragment.axisIterator(Axis.DESCENDANT);
- while (descenant.hasNext() && !found) {
- XdmNode node = descenant.next();
+ Iterator descendant = fragment.axisIterator(Axis.DESCENDANT);
+ while (descendant.hasNext() && !found) {
+ XdmNode node = descendant.next();
if (!node.getNodeKind().equals(XdmNodeKind.TEXT)) continue;
if (node.equals(textNode)) {
found = true;
@@ -479,17 +487,17 @@ protected Pair reportNotFound(String xpath, int position) thro
* its ancestors.
*/
protected static int getDepth(XdmNode node) {
- Iterator ascendents = node.axisIterator(Axis.ANCESTOR_OR_SELF);
+ Iterator ancestors = node.axisIterator(Axis.ANCESTOR_OR_SELF);
int depth = 0;
- while (ascendents.hasNext()) {
+ while (ancestors.hasNext()) {
depth += 1;
- ascendents.next();
+ ancestors.next();
}
return depth;
}
/**
- * A utiltity method that gets the depth of a node of a pair of
+ * A utility method that gets the depth of a node of a pair of
* node and position like used in this module.
*/
protected static int getDepth(Pair pair) {
@@ -497,7 +505,7 @@ protected static int getDepth(Pair pair) {
}
/**
- * A utility function for unescaping XPaths.
+ * A utility function for un-escaping XPaths.
*/
protected String unespace(String in) {
return in.replace("'", "'");
@@ -511,7 +519,7 @@ protected String unespace(String in) {
*
* @param xpath the XPath for generating a path expression, e.g., fn:xpath()
* @param node the {@link XdmNode} for which to generate the path expression
- * @param escaped whether or not the generated path expression is to be escaped
+ * @param escaped whether the generated path expression is to be escaped
* @param processor a Saxon {@link Processor}
*/
protected String pathExpressionWithXPath(String xpath, XdmNode node, boolean escaped, Processor processor)
@@ -524,17 +532,17 @@ protected String pathExpressionWithXPath(String xpath, XdmNode node, boolean esc
selector.setContextItem(node);
nodes = selector.evaluate();
} catch (SaxonApiException e) {
- LOG.error("failed to normalize XPath using '{}': ", xpath, e.getMessage());
+ LOG.error("failed to normalize XPath using '{}': ", xpath);
throw new SelectorException(e);
}
if (nodes.size() != 1) {
LOG.error("normalizing XPath '{}' did not return exactly one item: returned {} items", xpath, nodes.size());
throw new SelectorException("normalizing XPath '" + xpath
+ "' did not return exactly one item: returned "
- + String.valueOf(nodes.size())
+ + nodes.size()
+ " item");
} else if (!nodes.itemAt(0).isAtomicValue()) {
- LOG.error("normalizing XPath '{}' did not return an atomic value", xpath, nodes.size());
+ LOG.error("normalizing XPath '{}' did not return an atomic value", xpath);
throw new SelectorException("normalizing XPath '" + xpath + "' did not return an atomic value");
} else {
if (escaped) {
diff --git a/core/src/test/java/de/wwu/scdh/annotation/selection/rewriter/TestXPathNormalizerWithXPath.java b/core/src/test/java/de/wwu/scdh/annotation/selection/rewriter/TestXPathNormalizerWithXPath.java
index 507b036..2d437a7 100644
--- a/core/src/test/java/de/wwu/scdh/annotation/selection/rewriter/TestXPathNormalizerWithXPath.java
+++ b/core/src/test/java/de/wwu/scdh/annotation/selection/rewriter/TestXPathNormalizerWithXPath.java
@@ -23,7 +23,7 @@ public class TestXPathNormalizerWithXPath {
public static final URI GESANG_HTML = new File(TEST_DIR, "Gesang.tei.html").toURI();
public static final URI GESANG_XML = new File(TEST_DIR, "Gesang.tei.xml").toURI();
- RewriterConfig config = new RewriterConfig(null, false, null, false);
+ RewriterConfig config = new RewriterConfig(null, false, null, false, false);
@Test
public void testWithPathFunction() throws SelectorException, SaxonApiException, IOException {
diff --git a/core/src/test/java/de/wwu/scdh/annotation/selection/rewriter/TestXPathRefinedByRFC5147CharSchemeBackwardMapper.java b/core/src/test/java/de/wwu/scdh/annotation/selection/rewriter/TestXPathRefinedByRFC5147CharSchemeBackwardMapper.java
index 603ad71..0865e03 100644
--- a/core/src/test/java/de/wwu/scdh/annotation/selection/rewriter/TestXPathRefinedByRFC5147CharSchemeBackwardMapper.java
+++ b/core/src/test/java/de/wwu/scdh/annotation/selection/rewriter/TestXPathRefinedByRFC5147CharSchemeBackwardMapper.java
@@ -1,6 +1,5 @@
package de.wwu.scdh.annotation.selection.rewriter;
-import static org.hamcrest.CoreMatchers.*;
import static org.junit.jupiter.api.Assertions.*;
import de.wwu.scdh.annotation.selection.*;
@@ -33,10 +32,15 @@ public class TestXPathRefinedByRFC5147CharSchemeBackwardMapper {
public static final File ID_XSL =
Paths.get("src", "test", "resources", "xsl", "id.xsl").toFile();
- public static final File TEXT_WITH_TOC_XSL =
- Paths.get("src", "test", "resources", "xsl", "text-with-toc.xsl").toFile();
+ public static final File TEXT_WITH_TOC_XHTML_XSL = Paths.get(
+ "src", "test", "resources", "xsl", "text-with-toc-xhtml.xsl")
+ .toFile();
- RewriterConfig config = new RewriterConfig(Mode.FIRST, false, null, false);
+ public static final File TEXT_WITH_TOC_HTML_XSL = Paths.get(
+ "src", "test", "resources", "xsl", "text-with-toc-html.xsl")
+ .toFile();
+
+ RewriterConfig config = new RewriterConfig(Mode.FIRST, false, null, false, false);
public static XdmValue transform(DOMResource resource, File stylesheet, File pkg) throws SaxonApiException {
XsltCompiler compiler = PROC.newXsltCompiler();
@@ -127,7 +131,7 @@ public void testBackwardToCNode() throws ResourceException, SaxonApiException, S
DOMResource source = DOMResource.fromXMLwithXerces(GESANG_XML, PROC);
MappedDOMResource preimage = new MappedDOMResource(source);
XdmValueResource image =
- new XdmValueResource(GESANG_XML, transform(source, TEXT_WITH_TOC_XSL, LIBTRACE_XML), PROC);
+ new XdmValueResource(GESANG_XML, transform(source, TEXT_WITH_TOC_XHTML_XSL, LIBTRACE_XML), PROC);
preimage.setImage(image);
XPathRefinedByRFC5147CharSchemeBackwardMapper mapper =
new XPathRefinedByRFC5147CharSchemeBackwardMapper("path(.)");
@@ -149,7 +153,49 @@ public void testBackwardHeadNode() throws ResourceException, SaxonApiException,
DOMResource source = DOMResource.fromXMLwithXerces(GESANG_XML, PROC);
MappedDOMResource preimage = new MappedDOMResource(source);
XdmValueResource image =
- new XdmValueResource(GESANG_XML, transform(source, TEXT_WITH_TOC_XSL, LIBTRACE_XML), PROC);
+ new XdmValueResource(GESANG_XML, transform(source, TEXT_WITH_TOC_XHTML_XSL, LIBTRACE_XML), PROC);
+ preimage.setImage(image);
+ XPathRefinedByRFC5147CharSchemeBackwardMapper mapper =
+ new XPathRefinedByRFC5147CharSchemeBackwardMapper("path(.)");
+ List mapped = mapper.rewrite(preimage, imagePoint, config);
+ assertEquals(1, mapped.size());
+ assertEquals(
+ "/Q{http://www.tei-c.org/ns/1.0}TEI[1]/Q{http://www.tei-c.org/ns/1.0}text[1]/Q{http://www.tei-c.org/ns/1.0}body[1]/Q{http://www.tei-c.org/ns/1.0}lg[1]/Q{http://www.tei-c.org/ns/1.0}head[1]/text()[1]",
+ mapped.get(0).getXPath());
+ assertEquals(5, mapped.get(0).getChar());
+ }
+
+ @Test
+ public void testHtmlBackwardHeadNode() throws ResourceException, SaxonApiException, SelectorException {
+ // we transform a point from the H2, originating from a preimage node that occurs twice in the image
+ XPathRefinedByRFC5147CharScheme imagePoint =
+ new XPathRefinedByRFC5147CharScheme("/Q{}html[1]/Q{}body[1]/Q{}div[2]/Q{}div[1]/Q{}h2[1]", 5);
+ // ("/*:TEI[1]/*:text[1]/*:body[1]/*:lg[1]/*:head[1]", 5);
+ DOMResource source = DOMResource.fromXMLwithXerces(GESANG_XML, PROC);
+ MappedDOMResource preimage = new MappedDOMResource(source);
+ XdmValueResource image =
+ new XdmValueResource(GESANG_XML, transform(source, TEXT_WITH_TOC_HTML_XSL, LIBTRACE_XML), PROC);
+ preimage.setImage(image);
+ XPathRefinedByRFC5147CharSchemeBackwardMapper mapper =
+ new XPathRefinedByRFC5147CharSchemeBackwardMapper("path(.)");
+ List mapped = mapper.rewrite(preimage, imagePoint, config);
+ assertEquals(1, mapped.size());
+ assertEquals(
+ "/Q{http://www.tei-c.org/ns/1.0}TEI[1]/Q{http://www.tei-c.org/ns/1.0}text[1]/Q{http://www.tei-c.org/ns/1.0}body[1]/Q{http://www.tei-c.org/ns/1.0}lg[1]/Q{http://www.tei-c.org/ns/1.0}head[1]/text()[1]",
+ mapped.get(0).getXPath());
+ assertEquals(5, mapped.get(0).getChar());
+ }
+
+ @Test
+ public void testHtmlNoNsBackwardHeadNode() throws ResourceException, SaxonApiException, SelectorException {
+ // we transform a point from the H2, originating from a preimage node that occurs twice in the image
+ XPathRefinedByRFC5147CharScheme imagePoint =
+ new XPathRefinedByRFC5147CharScheme("/html[1]/body[1]/div[2]/div[1]/h2[1]", 5);
+ // ("/*:TEI[1]/*:text[1]/*:body[1]/*:lg[1]/*:head[1]", 5);
+ DOMResource source = DOMResource.fromXMLwithXerces(GESANG_XML, PROC);
+ MappedDOMResource preimage = new MappedDOMResource(source);
+ XdmValueResource image =
+ new XdmValueResource(GESANG_XML, transform(source, TEXT_WITH_TOC_HTML_XSL, LIBTRACE_XML), PROC);
preimage.setImage(image);
XPathRefinedByRFC5147CharSchemeBackwardMapper mapper =
new XPathRefinedByRFC5147CharSchemeBackwardMapper("path(.)");
diff --git a/core/src/test/java/de/wwu/scdh/annotation/selection/rewriter/TestXPathRefinedByRFC5147CharSchemeForwardMapper.java b/core/src/test/java/de/wwu/scdh/annotation/selection/rewriter/TestXPathRefinedByRFC5147CharSchemeForwardMapper.java
index 83567ae..c3a76d4 100644
--- a/core/src/test/java/de/wwu/scdh/annotation/selection/rewriter/TestXPathRefinedByRFC5147CharSchemeForwardMapper.java
+++ b/core/src/test/java/de/wwu/scdh/annotation/selection/rewriter/TestXPathRefinedByRFC5147CharSchemeForwardMapper.java
@@ -32,13 +32,20 @@ public class TestXPathRefinedByRFC5147CharSchemeForwardMapper {
public static final File ID_XSL =
Paths.get("src", "test", "resources", "xsl", "id.xsl").toFile();
- public static final File TEXT_WITH_TOC_XSL =
- Paths.get("src", "test", "resources", "xsl", "text-with-toc.xsl").toFile();
+ public static final File TEXT_WITH_TOC_XHTML_XSL = Paths.get(
+ "src", "test", "resources", "xsl", "text-with-toc-xhtml.xsl")
+ .toFile();
+
+ public static final File TEXT_WITH_TOC_HTML_XSL = Paths.get(
+ "src", "test", "resources", "xsl", "text-with-toc-html.xsl")
+ .toFile();
public static final File TEXT_XSL =
Paths.get("src", "test", "resources", "xsl", "text.xsl").toFile();
- RewriterConfig config = new RewriterConfig(Mode.FIRST, false, null, false);
+ public static final RewriterConfig CONFIG = new RewriterConfig(Mode.FIRST, false, null, false, false);
+
+ public static final RewriterConfig CONFIG_NO_NS = new RewriterConfig(Mode.FIRST, false, null, true, true);
public static XdmValue transform(DOMResource resource, File stylesheet, File pkg) throws SaxonApiException {
XsltCompiler compiler = PROC.newXsltCompiler();
@@ -59,7 +66,7 @@ public void testGesangMappedWithIdentity() throws ResourceException, SaxonApiExc
new XPathRefinedByRFC5147CharSchemeForwardMapper("path(.)");
XPathRefinedByRFC5147CharScheme preimagePoint =
new XPathRefinedByRFC5147CharScheme("/*:TEI[1]/*:text[1]/*:body[1]/*:lg[1]/*:head[1]", 5);
- List mapped = mapper.rewrite(preimage, preimagePoint, config);
+ List mapped = mapper.rewrite(preimage, preimagePoint, CONFIG);
assertEquals(1, mapped.size());
assertEquals(
mapped.get(0).getXPath(),
@@ -74,11 +81,12 @@ public void testForwardToMultipleNodes() throws ResourceException, SaxonApiExcep
new XPathRefinedByRFC5147CharScheme("/*:TEI[1]/*:text[1]/*:body[1]/*:lg[1]/*:head[1]", 5);
DOMResource source = DOMResource.fromXMLwithXerces(GESANG_XML, PROC);
MappedDOMResource preimage = new MappedDOMResource(source);
- XdmValueResource image = new XdmValueResource(GESANG_XML, transform(source, TEXT_WITH_TOC_XSL, LIBTRACE_XML));
+ XdmValueResource image =
+ new XdmValueResource(GESANG_XML, transform(source, TEXT_WITH_TOC_XHTML_XSL, LIBTRACE_XML));
preimage.setImage(image);
XPathRefinedByRFC5147CharSchemeForwardMapper mapper =
new XPathRefinedByRFC5147CharSchemeForwardMapper("path(.)");
- List mapped = mapper.rewrite(preimage, preimagePoint, config);
+ List mapped = mapper.rewrite(preimage, preimagePoint, CONFIG);
assertEquals(2, mapped.size());
assertEquals(
@@ -109,14 +117,55 @@ public void testForwardToZeroNodes() throws ResourceException, SaxonApiException
"/*:TEI[1]/*:teiHeader[1]/*:fileDesc[1]/*:titleStmt[1]/*:author[1]", 5);
DOMResource source = DOMResource.fromXMLwithXerces(GESANG_XML, PROC);
MappedDOMResource preimage = new MappedDOMResource(source);
- XdmValueResource image = new XdmValueResource(GESANG_XML, transform(source, TEXT_WITH_TOC_XSL, LIBTRACE_XML));
+ XdmValueResource image =
+ new XdmValueResource(GESANG_XML, transform(source, TEXT_WITH_TOC_XHTML_XSL, LIBTRACE_XML));
preimage.setImage(image);
XPathRefinedByRFC5147CharSchemeForwardMapper mapper =
new XPathRefinedByRFC5147CharSchemeForwardMapper("path(.)");
- List mapped = mapper.rewrite(preimage, preimagePoint, config);
+ List mapped = mapper.rewrite(preimage, preimagePoint, CONFIG);
assertEquals(0, mapped.size());
}
+ @Test
+ public void testHtmlForwardToOneNode() throws ResourceException, SaxonApiException, SelectorException {
+ // we transform a point that occurs twice in the output
+ XPathRefinedByRFC5147CharScheme preimagePoint =
+ new XPathRefinedByRFC5147CharScheme("/*:TEI[1]/*:text[1]/*:body[1]/*:lg[1]/*:l[2]/text()[2]", 7);
+ DOMResource source = DOMResource.fromXMLwithXerces(GESANG_XML, PROC);
+ MappedDOMResource preimage = new MappedDOMResource(source);
+ XdmValueResource image =
+ new XdmValueResource(GESANG_XML, transform(source, TEXT_WITH_TOC_HTML_XSL, LIBTRACE_XML));
+ preimage.setImage(image);
+ XPathRefinedByRFC5147CharSchemeForwardMapper mapper =
+ new XPathRefinedByRFC5147CharSchemeForwardMapper("path(.)");
+ List mapped = mapper.rewrite(preimage, preimagePoint, CONFIG);
+ assertEquals(1, mapped.size());
+ assertEquals(
+ "/Q{}html[1]/Q{}body[1]/Q{}div[2]/Q{}div[1]/Q{}p[2]/Q{http://wwu.de/scdh/selection-engine/node-tracing}text[2]",
+ mapped.get(0).getXPath(),
+ "fn:path returns QNames for elements on the '' namespace as Q{}*.");
+ }
+
+ @Test
+ public void testHtmlNoNsForwardToOneNode() throws ResourceException, SaxonApiException, SelectorException {
+ // we transform a point that occurs twice in the output
+ XPathRefinedByRFC5147CharScheme preimagePoint =
+ new XPathRefinedByRFC5147CharScheme("/*:TEI[1]/*:text[1]/*:body[1]/*:lg[1]/*:l[2]/text()[2]", 7);
+ DOMResource source = DOMResource.fromXMLwithXerces(GESANG_XML, PROC);
+ MappedDOMResource preimage = new MappedDOMResource(source);
+ XdmValueResource image =
+ new XdmValueResource(GESANG_XML, transform(source, TEXT_WITH_TOC_HTML_XSL, LIBTRACE_XML));
+ preimage.setImage(image);
+ XPathRefinedByRFC5147CharSchemeForwardMapper mapper =
+ new XPathRefinedByRFC5147CharSchemeForwardMapper("path(.)");
+ List mapped = mapper.rewrite(preimage, preimagePoint, CONFIG_NO_NS);
+ assertEquals(1, mapped.size());
+ assertEquals(
+ "/html[1]/body[1]/div[2]/div[1]/p[2]/text()[2]",
+ mapped.get(0).getXPath(),
+ "configured to return nice path expressions into HTML");
+ }
+
@Test
public void testTextToOneNode() throws ResourceException, SaxonApiException, SelectorException {
DOMResource source = DOMResource.fromXMLwithXerces(GESANG_XML, PROC);
@@ -127,7 +176,7 @@ public void testTextToOneNode() throws ResourceException, SaxonApiException, Sel
new XPathRefinedByRFC5147CharSchemeForwardMapper("path(.)");
XPathRefinedByRFC5147CharScheme preimagePoint =
new XPathRefinedByRFC5147CharScheme("/*:TEI[1]/*:text[1]/*:body[1]/*:lg[1]/*:head[1]", 5);
- List mapped = mapper.rewrite(preimage, preimagePoint, config);
+ List mapped = mapper.rewrite(preimage, preimagePoint, CONFIG);
assertEquals(1, mapped.size());
// the XdmValue output looks a bit weird: all text containers are directly in document node
assertTrue(mapped.get(0).getXPath().endsWith("/Q{http://wwu.de/scdh/selection-engine/node-tracing}text[22]"));
@@ -145,7 +194,7 @@ public void testTextToOneNodeWithRootOffset() throws ResourceException, SaxonApi
"path(ancestor-or-self::node()[ parent::node()[not(parent::node())] ]) ");
XPathRefinedByRFC5147CharScheme preimagePoint =
new XPathRefinedByRFC5147CharScheme("/*:TEI[1]/*:text[1]/*:body[1]/*:lg[1]/*:head[1]", 5);
- List mapped = mapper.rewrite(preimage, preimagePoint, config);
+ List mapped = mapper.rewrite(preimage, preimagePoint, CONFIG);
assertEquals(1, mapped.size());
assertTrue(mapped.get(0).getXPath().endsWith("/Q{http://wwu.de/scdh/selection-engine/node-tracing}text[22]"));
assertEquals(5, mapped.get(0).getChar());
diff --git a/core/src/test/java/de/wwu/scdh/annotation/selection/rewriter/TestXPathRefinedByRFC5147CharSchemeToTextBackwardMapper.java b/core/src/test/java/de/wwu/scdh/annotation/selection/rewriter/TestXPathRefinedByRFC5147CharSchemeToTextBackwardMapper.java
index 4c591b7..5e0cc60 100644
--- a/core/src/test/java/de/wwu/scdh/annotation/selection/rewriter/TestXPathRefinedByRFC5147CharSchemeToTextBackwardMapper.java
+++ b/core/src/test/java/de/wwu/scdh/annotation/selection/rewriter/TestXPathRefinedByRFC5147CharSchemeToTextBackwardMapper.java
@@ -39,7 +39,7 @@ public class TestXPathRefinedByRFC5147CharSchemeToTextBackwardMapper {
public static final File NO_HEADER_XSL =
Paths.get("src", "test", "resources", "xsl", "text-no-header.xsl").toFile();
- RewriterConfig config = new RewriterConfig(Mode.FIRST, false, null, false);
+ RewriterConfig config = new RewriterConfig(Mode.FIRST, false, null, false, false);
public static XdmValue transform(DOMResource resource, File stylesheet, File pkg) throws SaxonApiException {
XsltCompiler compiler = PROC.newXsltCompiler();
diff --git a/core/src/test/java/de/wwu/scdh/annotation/selection/rewriter/TestXPathRefinedByRFC5147CharSchemeToTextForwardMapper.java b/core/src/test/java/de/wwu/scdh/annotation/selection/rewriter/TestXPathRefinedByRFC5147CharSchemeToTextForwardMapper.java
index b41b7b4..f011835 100644
--- a/core/src/test/java/de/wwu/scdh/annotation/selection/rewriter/TestXPathRefinedByRFC5147CharSchemeToTextForwardMapper.java
+++ b/core/src/test/java/de/wwu/scdh/annotation/selection/rewriter/TestXPathRefinedByRFC5147CharSchemeToTextForwardMapper.java
@@ -39,7 +39,7 @@ public class TestXPathRefinedByRFC5147CharSchemeToTextForwardMapper {
public static final File NO_HEADER_XSL =
Paths.get("src", "test", "resources", "xsl", "text-no-header.xsl").toFile();
- RewriterConfig config = new RewriterConfig(Mode.FIRST, false, null, false);
+ RewriterConfig config = new RewriterConfig(Mode.FIRST, false, null, false, false);
public static XdmValue transform(DOMResource resource, File stylesheet, File pkg) throws SaxonApiException {
XsltCompiler compiler = PROC.newXsltCompiler();
diff --git a/core/src/test/java/de/wwu/scdh/annotation/selection/wadm/TestNormalizeAnnotation.java b/core/src/test/java/de/wwu/scdh/annotation/selection/wadm/TestNormalizeAnnotation.java
index 1aec7b0..027fda1 100644
--- a/core/src/test/java/de/wwu/scdh/annotation/selection/wadm/TestNormalizeAnnotation.java
+++ b/core/src/test/java/de/wwu/scdh/annotation/selection/wadm/TestNormalizeAnnotation.java
@@ -60,7 +60,7 @@ public void setupResource() throws URISyntaxException, ResourceException, Malfor
}
private static RewriterConfig makeConfig(String xpath) {
- return new RewriterConfig(null, false, xpath, false);
+ return new RewriterConfig(null, false, xpath, false, false);
}
// normalizing XPathSelector refinedBy FragmentSelector conforming to RFC5147 character scheme
diff --git a/core/src/test/resources/xsl/text-with-toc-html.xsl b/core/src/test/resources/xsl/text-with-toc-html.xsl
new file mode 100644
index 0000000..12ea13a
--- /dev/null
+++ b/core/src/test/resources/xsl/text-with-toc-html.xsl
@@ -0,0 +1,89 @@
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+ <xsl:apply-templates select="//teiHeader//title"/>
+
+
+
+
+
Table of contents:
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
diff --git a/core/src/test/resources/xsl/text-with-toc.xsl b/core/src/test/resources/xsl/text-with-toc-xhtml.xsl
similarity index 82%
rename from core/src/test/resources/xsl/text-with-toc.xsl
rename to core/src/test/resources/xsl/text-with-toc-xhtml.xsl
index 5d893c5..131c0dc 100644
--- a/core/src/test/resources/xsl/text-with-toc.xsl
+++ b/core/src/test/resources/xsl/text-with-toc-xhtml.xsl
@@ -50,14 +50,21 @@
-
+
-
+
-
+
+
+
+
+
+
+
+
@@ -71,15 +78,12 @@
-
-
+
+
-
-
-
-
-
+
+
diff --git a/test/Gesang.html b/test/Gesang.html
new file mode 100644
index 0000000..63c5bd0
--- /dev/null
+++ b/test/Gesang.html
@@ -0,0 +1,11 @@
+Trawr-Gesang von der noth Christi am Oelberg in dem GartenTable of contents:
Trawr-Gesang von der noth Christi am Oelberg in dem Garten
+
+
+
Trawr-Gesang von der noth Christi am Oelberg in dem Garten
+
Bey stiller nacht zur ersten wacht
+
Ein stimm sich gund zu klagen.
+
Jch nam in acht waß die doch sagt;
+
That hin mit Augen schlagen.
+
+
+
Table of contents:
Trawr-Gesang von der noth Christi am Oelberg in dem Garten
+
+
+
Trawr-Gesang von der noth Christi am Oelberg in dem Garten
+
Bey stiller nacht zur ersten wacht
+
Ein stimm sich gund zu klagen.
+
Jch nam in acht waß die doch sagt;
+
That hin mit Augen schlagen.
+
+
+