Revision: 14572
http://gate.svn.sourceforge.net/gate/?rev=14572&view=rev
Author: adamfunk
Date: 2011-11-18 15:20:14 +0000 (Fri, 18 Nov 2011)
Log Message:
-----------
Last (I think) round of changes to FlexibleGazetteer.
Loop for mapping back to original document is a little more efficient;
NodePosition (individual mapping) is now immutable.
Modified Paths:
--------------
gate/trunk/src/gate/creole/gazetteer/FlexibleGazetteer.java
gate/trunk/src/gate/creole/gazetteer/NodePosition.java
Modified: gate/trunk/src/gate/creole/gazetteer/FlexibleGazetteer.java
===================================================================
--- gate/trunk/src/gate/creole/gazetteer/FlexibleGazetteer.java 2011-11-18
12:15:35 UTC (rev 14571)
+++ gate/trunk/src/gate/creole/gazetteer/FlexibleGazetteer.java 2011-11-18
15:20:14 UTC (rev 14572)
@@ -77,11 +77,9 @@
* be called before the gazetteer can be used
*/
public Resource init() throws ResourceInstantiationException {
- // check for parameters
- if(gazetteerInst == null)
+ if(gazetteerInst == null) {
throw new ResourceInstantiationException("No Gazetteer Provided!");
-
- this.changedNodes = new ArrayList<NodePosition>();
+ }
return this;
}
@@ -90,7 +88,7 @@
* parameters are set. If they are not, an exception will be fired.
*/
public void execute() throws ExecutionException {
- changedNodes = new ArrayList<NodePosition>();
+ annotationMappings = new ArrayList<NodePosition>();
fireProgressChanged(0);
fireStatusChanged("Checking Document...");
if(document == null) {
@@ -108,7 +106,7 @@
long totalDeductedSpaces = 0;
fireStatusChanged("Replacing contents with the feature value...");
outer: for (Annotation currentToken :
Utils.inDocumentOrder(document.getAnnotations(inputAnnotationSetName))) {
- // check if it is a chinesesplit; if it is, replace no space character
with a single space
+ // Where ChineseSplits occur, insert a single space
if(currentToken.getType().equals(ANNIEConstants.SPACE_TOKEN_ANNOTATION_TYPE)
&&
((String)(currentToken.getFeatures().get(ANNIEConstants.TOKEN_KIND_FEATURE_NAME))).equals("ChineseSplit"))
{
@@ -119,12 +117,14 @@
// the endoffset will become newStartOffset + 1
long newStartOffset = startOffset - totalDeductedSpaces;
long newEndOffset = newStartOffset + 1;
- NodePosition newNode = new NodePosition(startOffset, startOffset,
- newStartOffset, newEndOffset, totalDeductedSpaces);
+ NodePosition mapping = new NodePosition(startOffset, startOffset,
+ newStartOffset, newEndOffset);
// here is the addition of space in the document
totalDeductedSpaces--;
- changedNodes.add(newNode);
+ // Should these Splits actually be mappable from the temp document
+ // back to the original one? (AF)
+ annotationMappings.add(mapping);
newdocString = newdocString.insert((int)newStartOffset, ' ');
continue outer;
} // chineseSplit if
@@ -146,23 +146,20 @@
long startOffset =
currentToken.getStartNode().getOffset().longValue();
long endOffset = currentToken.getEndNode().getOffset().longValue();
- // replacement code start
- long actualLength = endOffset - startOffset;
// let us find the difference between the lengths of the
// actual string and the newTokenValue
+ long actualLength = endOffset - startOffset;
long lengthDifference = actualLength - newTokenValue.length();
- // replacement code end
-
// so lets find out the new startOffset and endOffset
long newStartOffset = startOffset - totalDeductedSpaces;
long newEndOffset = newStartOffset + newTokenValue.length();
totalDeductedSpaces += lengthDifference;
// and make the entry for this
- NodePosition newNode = new NodePosition(startOffset, endOffset,
- newStartOffset, newEndOffset, totalDeductedSpaces);
- changedNodes.add(newNode);
+ NodePosition mapping = new NodePosition(startOffset, endOffset,
+ newStartOffset, newEndOffset);
+ annotationMappings.add(mapping);
// and finally replace the actual string in the document
// with the new document
@@ -174,6 +171,9 @@
}
} // END OF "inner" LOOP
} // END OF "outer" LOOP
+
+ // make sure the conversion table is in the right order
+ Collections.sort(annotationMappings, new NodePositionComparator());
fireStatusChanged("New Document to be processed with Gazetteer...");
try {
@@ -185,7 +185,6 @@
}
FeatureMap features = Factory.newFeatureMap();
- // Gate.setHiddenAttribute(features, true);
tempDoc = (Document)Factory.createResource("gate.corpora.DocumentImpl",
params, features);
}
@@ -194,7 +193,6 @@
}
// lets create the gazetteer based on the provided gazetteer name
- //FeatureMap params = Factory.newFeatureMap();
gazetteerInst.setDocument(tempDoc);
gazetteerInst.setAnnotationSetName(this.outputAnnotationSetName);
@@ -217,40 +215,35 @@
long originalStart = 0;
long originalEnd = tempDoc.getContent().size() - 1L;
-
- int i = 0;
- for(; i < changedNodes.size(); i++) {
- NodePosition np = changedNodes.get(i);
-
- // Find the last node whose temp start node is less than or equal
+ boolean foundStart = false;
+
+ for (NodePosition mapping : annotationMappings) {
+ // Find the last mapping whose temp start offset is less than or equal
// to the temp lookup's start
- if(np.getNewStartNode() <= startOffset) {
- originalStart = np.getOldStartNode();
- }
- else {
- break;
+ if (! foundStart) {
+ if (mapping.getNewStartOffset() <= startOffset) {
+ originalStart = mapping.getOriginalStartOffset();
+ }
+ else {
+ foundStart = true;
+ }
}
- }
-
- for(; i < changedNodes.size(); i++) {
- NodePosition np = changedNodes.get(i);
-
- // Find the first node whose temp end node is greater than or equal
- // to the temp lookup's end
- if(np.getNewEndNode() >= endOffset) {
- originalEnd = np.getOldEndNode();
+
+ // Find the first mapping whose temp end offset is greater than or
equal
+ // to the temp lookup's end; typically this will be the same as the
mapping found
+ // for the start offset
+ if (mapping.getNewEndOffset() >= endOffset) {
+ originalEnd = mapping.getOriginalEndOffset();
break;
}
}
-
+
try {
- original.add(originalStart, originalEnd,
- currentLookup.getType(), currentLookup.getFeatures());
+ original.add(originalStart, originalEnd, currentLookup.getType(),
currentLookup.getFeatures());
} // This should no longer happen
catch(InvalidOffsetException ioe) {
throw new ExecutionException(ioe);
}
-
}
// now remove the newDoc
@@ -357,5 +350,5 @@
private java.util.List<String> inputFeatureNames;
// parameters required within the program
- private ArrayList<NodePosition> changedNodes;
+ private ArrayList<NodePosition> annotationMappings;
}
Modified: gate/trunk/src/gate/creole/gazetteer/NodePosition.java
===================================================================
--- gate/trunk/src/gate/creole/gazetteer/NodePosition.java 2011-11-18
12:15:35 UTC (rev 14571)
+++ gate/trunk/src/gate/creole/gazetteer/NodePosition.java 2011-11-18
15:20:14 UTC (rev 14572)
@@ -1,7 +1,7 @@
/*
* NodePosition.java
*
- * Copyright (c) 2004, The University of Sheffield.
+ * Copyright (c) 2004--2011, The University of Sheffield.
*
* This file is part of GATE (see http://gate.ac.uk/), and is free
* software, licenced under the GNU Library General Public License,
@@ -11,43 +11,37 @@
* licence.html, and is also available at http://gate.ac.uk/gate/licence.html.
*
* Niraj Aswani 02/2002
+ * $Id$
*
+ * 2011-11-18: AF made this immutable.
*/
package gate.creole.gazetteer;
+import java.util.Comparator;
+
/**
* <p>Title: NodePosition.java </p>
* <p>Description: This class is used to store the information about the
* changes in the text and the addition or the subtraction of the spaces.
* It is used by FlexibleGazetteer. </p>
* @author Niraj Aswani
- * @version 1.0
*/
public class NodePosition {
/** The original start offset before changes */
- private long oldStartNode;
+ private long originalStartOffset;
/** The original end offset before changes */
- private long oldEndNode;
+ private long originalEndOffset;
/** The new start offset after the changes */
- private long newStartNode;
+ private long newStartOffset;
/** The new end offset after the changes */
- private long newEndNode;
+ private long newEndOffset;
- /** total deducted spaces due to change in the text before the start
- * offset in the document
- */
- private long deductedSpaces;
-
- /** Constructor */
- public NodePosition() {
- }
-
/**
* constructor
* @param osn - old start offset
@@ -57,91 +51,70 @@
* @param space - total deducted spaces due to change in the text before
* the start offset in the document
*/
- public NodePosition(long osn, long oen, long nsn, long nen, long space) {
- oldStartNode = osn;
- oldEndNode = oen;
- newStartNode = nsn;
- newEndNode = nen;
- deductedSpaces = space;
+ public NodePosition(long osn, long oen, long nsn, long nen) {
+ originalStartOffset = osn;
+ originalEndOffset = oen;
+ newStartOffset = nsn;
+ newEndOffset = nen;
}
/**
* Returns the old start offset
* @return a <tt>long</tt> value.
*/
- public long getOldStartNode() {
- return oldStartNode;
+ public long getOriginalStartOffset() {
+ return originalStartOffset;
}
/**
* Returns the old end offset
* @return a <tt>long</tt> value.
*/
- public long getOldEndNode() {
- return oldEndNode;
+ public long getOriginalEndOffset() {
+ return originalEndOffset;
}
/**
* Returns new start offset
* @return a <tt>long</tt> value.
*/
- public long getNewStartNode() {
- return newStartNode;
+ public long getNewStartOffset() {
+ return newStartOffset;
}
/**
* Returns the new end offset
* @return a <tt>long</tt> value.
*/
- public long getNewEndNode() {
- return newEndNode;
+ public long getNewEndOffset() {
+ return newEndOffset;
}
- /**
- * Sets the old start offset
- * @param node
- */
- public void setOldStartNode(long node) {
- oldStartNode = node;
- }
+}
- /**
- * Sets the old end offset
- * @param node
- */
- public void setOldEndNode(long node) {
- oldEndNode = node;
- }
- /**
- * sets the new start offset
- * @param node
- */
- public void setNewStartNode(long node) {
- newStartNode = node;
- }
+class NodePositionComparator implements Comparator<NodePosition> {
- /**
- * Sets the new end offset
- * @param node
- */
- public void setNewEndNode(long node) {
- newEndNode = node;
+ public int compare(NodePosition arg0, NodePosition arg1) {
+ long diff = arg0.getNewStartOffset() - arg1.getNewStartOffset();
+ if (diff != 0L) {
+ return (int) Long.signum(diff);
+ }
+ // implied else
+ diff = arg0.getNewEndOffset() - arg1.getNewEndOffset();
+ if (diff != 0L) {
+ return (int) Long.signum(diff);
+ }
+ // implied else
+ diff = arg0.getOriginalStartOffset() - arg1.getOriginalStartOffset();
+ if (diff != 0L) {
+ return (int) Long.signum(diff);
+ }
+ // implied else
+ diff = arg0.getOriginalEndOffset() - arg1.getOriginalEndOffset();
+ return (int) Long.signum(diff);
}
-
- /**
- * Sets the deducted spaces
- * @param space
- */
- public void setDeductedSpaces(long space) {
- deductedSpaces = space;
- }
-
- /**
- * Returns the total deducted spaces
- * @return a <tt>long</tt> value.
- */
- public long getDeductedSpaces() {
- return deductedSpaces;
- }
-}
\ No newline at end of file
+
+
+
+}
This was sent by the SourceForge.net collaborative development platform, the
world's largest Open Source development site.
------------------------------------------------------------------------------
All the data continuously generated in your IT infrastructure
contains a definitive record of customers, application performance,
security threats, fraudulent activity, and more. Splunk takes this
data and makes sense of it. IT sense. And common sense.
http://p.sf.net/sfu/splunk-novd2d
_______________________________________________
GATE-cvs mailing list
[email protected]
https://lists.sourceforge.net/lists/listinfo/gate-cvs
