Revision: 16607
http://sourceforge.net/p/gate/code/16607
Author: johann_p
Date: 2013-03-20 11:45:00 +0000 (Wed, 20 Mar 2013)
Log Message:
-----------
If using LST files, allow features inst and/or class to be set
from the LST file feares, if a gazetteerFeatureSeparator is
specified in the config.ttl file. Default is no separator and
behavior as before.
Modified Paths:
--------------
gate/trunk/plugins/Gazetteer_LKB/samples/dictionary_from_gazetteer/config.ttl
gate/trunk/plugins/Gazetteer_LKB/src/com/ontotext/kim/model/DataFeedFactory.java
gate/trunk/plugins/Gazetteer_LKB/src/com/ontotext/kim/model/Options.java
gate/trunk/plugins/Gazetteer_LKB/src/com/ontotext/kim/util/datastore/GazetteerListFeed.java
Modified:
gate/trunk/plugins/Gazetteer_LKB/samples/dictionary_from_gazetteer/config.ttl
===================================================================
---
gate/trunk/plugins/Gazetteer_LKB/samples/dictionary_from_gazetteer/config.ttl
2013-03-20 02:20:30 UTC (rev 16606)
+++
gate/trunk/plugins/Gazetteer_LKB/samples/dictionary_from_gazetteer/config.ttl
2013-03-20 11:45:00 UTC (rev 16607)
@@ -1,22 +1,31 @@
-# Gazetteer_LKB dictionary configuration file.
-@prefix lkbg: <http://www.ontotext.com/lkb_gazetteer#>.
-
-# The gazetteer-specific options follow.
-lkbg:DictionaryConfiguration
-
- # Whether the gazeetteer will be sensitive to case.
- # Possible values: caseinsensitive, casesensitive, casesensitiveAllUpper
- lkbg:caseSensitivity "caseinsensitive" ;
-
- # Whether the gazetteer will cache the dictionary after loading it from the
data source.
- # Put "enabled" to enable the cache, any other values will disable it.
- lkbg:caching "enabled" ;
-
- # The cache will be automatically reloaded on initialization if the
configuration have been
- # modified since the last initialization. However, changes in the
underlying datastore
- # can't be detected and the cache will not be automatically reloaded in
that case.
- # Thus, if you expect changes in your datastore, disable the cache, or make
use
- # of the modifications API.
-
- #A list of labels to be ignored, one per line.
- lkbg:ignoreList "ignoreList.txt" .
+# Gazetteer_LKB dictionary configuration file.
+@prefix lkbg: <http://www.ontotext.com/lkb_gazetteer#>.
+
+# The gazetteer-specific options follow.
+lkbg:DictionaryConfiguration
+
+ # Whether the gazeetteer will be sensitive to case.
+ # Possible values: caseinsensitive, casesensitive, casesensitiveAllUpper
+ lkbg:caseSensitivity "caseinsensitive" ;
+
+ # Whether the gazetteer will cache the dictionary after loading it from the
data source.
+ # Put "enabled" to enable the cache, any other values will disable it.
+ lkbg:caching "enabled" ;
+
+ # The cache will be automatically reloaded on initialization if the
configuration have been
+ # modified since the last initialization. However, changes in the
underlying datastore
+ # can't be detected and the cache will not be automatically reloaded in
that case.
+ # Thus, if you expect changes in your datastore, disable the cache, or make
use
+ # of the modifications API.
+
+ #A list of labels to be ignored, one per line.
+ lkbg:ignoreList "ignoreList.txt" ;
+
+ # if the cache is loaded from a gazetteer list file, the separator
character to use
+ # to split each line into entry, inst and class feature. If this is set to
+ # the empty string or not set, no splitting is done. Otherwise the string
is split]
+ # and the first part is taken to be the gazetteer entry. The remaining
parts are
+ # checked to see if they have the form "inst=URI" or "class=URI" and if
yes, are
+ # the URI is used to set the inst or class feature for that entry.
+ lkbg:gazetteerFeatureSeparator "" .
+
Modified:
gate/trunk/plugins/Gazetteer_LKB/src/com/ontotext/kim/model/DataFeedFactory.java
===================================================================
---
gate/trunk/plugins/Gazetteer_LKB/src/com/ontotext/kim/model/DataFeedFactory.java
2013-03-20 02:20:30 UTC (rev 16606)
+++
gate/trunk/plugins/Gazetteer_LKB/src/com/ontotext/kim/model/DataFeedFactory.java
2013-03-20 11:45:00 UTC (rev 16607)
@@ -62,7 +62,7 @@
}
if (result == null && !FileUtils.listFiles(dictionaryPath, new
String[]{"def"}, false).isEmpty()) {
- result = new GazetteerListFeed(dictionaryPath);
+ result = new GazetteerListFeed(dictionaryPath,opt);
}
if (result == null) {
result = new DummyFeed(dictionaryPath);
Modified:
gate/trunk/plugins/Gazetteer_LKB/src/com/ontotext/kim/model/Options.java
===================================================================
--- gate/trunk/plugins/Gazetteer_LKB/src/com/ontotext/kim/model/Options.java
2013-03-20 02:20:30 UTC (rev 16606)
+++ gate/trunk/plugins/Gazetteer_LKB/src/com/ontotext/kim/model/Options.java
2013-03-20 11:45:00 UTC (rev 16607)
@@ -1,5 +1,7 @@
package com.ontotext.kim.model;
+import gate.util.Strings;
+
import java.io.File;
import java.io.FileReader;
import java.io.IOException;
@@ -91,6 +93,12 @@
return res != null ? res : INSENSITIVE;
}
+ public String getSeparator() {
+ String ret = opts.get("gazetteerfeatureseparator");
+ if(ret == null) { ret = ""; }
+ return Strings.unescape(ret);
+ }
+
public Map<String, String> getMap() {
return opts; // unmodifiable
}
Modified:
gate/trunk/plugins/Gazetteer_LKB/src/com/ontotext/kim/util/datastore/GazetteerListFeed.java
===================================================================
---
gate/trunk/plugins/Gazetteer_LKB/src/com/ontotext/kim/util/datastore/GazetteerListFeed.java
2013-03-20 02:20:30 UTC (rev 16606)
+++
gate/trunk/plugins/Gazetteer_LKB/src/com/ontotext/kim/util/datastore/GazetteerListFeed.java
2013-03-20 11:45:00 UTC (rev 16607)
@@ -21,6 +21,7 @@
import com.ontotext.kim.client.query.KIMQueryException;
import com.ontotext.kim.client.semanticrepository.QueryResultListener;
import com.ontotext.kim.client.semanticrepository.QueryResultListener.Feed;
+import com.ontotext.kim.model.Options;
/**
* @author mnozchev
@@ -33,9 +34,11 @@
private static Logger log = Logger.getLogger(GazetteerListFeed.class);
private final File dictionaryPath;
+ private Options options;
- public GazetteerListFeed(File dictionaryPath) {
+ public GazetteerListFeed(File dictionaryPath, Options opt) {
this.dictionaryPath = dictionaryPath;
+ this.options = opt;
}
/**
@@ -85,6 +88,7 @@
LinearDefinition definition) throws ResourceInstantiationException,
IOException {
+ String separator = options.getSeparator();
Map listsByNode = definition.loadLists();
Iterator inodes = definition.iterator();
LinearNode node;
@@ -107,19 +111,43 @@
Iterator iline = gazList.iterator();
while(iline.hasNext()) {
GazetteerNode gazNode = (GazetteerNode)iline.next();
- addEntity(listener, node.getMinorType(), node.getMajorType(),
- gazNode.getEntry());
+ String clazz = "urn:"+node.getMajorType();
+ String inst = "urn:"+node.getMinorType();
+ String entry = gazNode.getEntry();
+ // if a separator string is set, attempt to split the gazetteer
+ // line by that separator and try to get the inst and/or class feature
+ // values.
+ if(!separator.equals("")) {
+ // If the entry contains the separator, split it up and try to find
+ // features class and inst.
+ if(entry.contains(separator)) {
+ String fields[] = entry.split(separator);
+ entry = fields[0];
+ for(int i = 1; i<fields.length; i++) {
+ if(fields[i].startsWith("inst=")) {
+ inst = fields[i].substring(5);
+ } else if(fields[i].startsWith("class=")) {
+ clazz = fields[i].substring(6);
+ }
+ }
+ }
+ }
+
+ addEntity(listener, inst, clazz,entry);
}
+
+
+
}
}
- private void addEntity(QueryResultListener listener, String minorType,
- String majorType, String label) throws IOException {
+ private void addEntity(QueryResultListener listener, String inst,
+ String clazz, String label) throws IOException {
listener.startTuple();
listener.tupleValue(new LiteralImpl(label));
- listener.tupleValue(new URIImpl("urn:" + minorType));
- listener.tupleValue(new URIImpl("urn:" + majorType));
+ listener.tupleValue(new URIImpl(inst));
+ listener.tupleValue(new URIImpl(clazz));
listener.endTuple();
}
This was sent by the SourceForge.net collaborative development platform, the
world's largest Open Source development site.
------------------------------------------------------------------------------
Everyone hates slow websites. So do we.
Make your web apps faster with AppDynamics
Download AppDynamics Lite for free today:
http://p.sf.net/sfu/appdyn_d2d_mar
_______________________________________________
GATE-cvs mailing list
[email protected]
https://lists.sourceforge.net/lists/listinfo/gate-cvs
