Changeset 6689

Timestamp:

10/20/15 13:15:51 (9 years ago)

Author:

Twan Goosen

Message:

improved query sanitising for highlighting

File:

• vlo/branches/vlo-3.3/vlo-web-app/src/main/java/eu/clarin/cmdi/vlo/wicket/HighlightSearchTermScriptFactory.java (modified) (3 diffs)

vlo/branches/vlo-3.3/vlo-web-app/src/main/java/eu/clarin/cmdi/vlo/wicket/HighlightSearchTermScriptFactory.java

@@ -19 +19 @@
 import com.google.common.collect.ImmutableSet;
 import java.util.Collection;
-import org.apache.wicket.util.string.StringValue;
 
 /**
@@ -33 +32 @@
     public String createScript(String componentSelector, final String words) {
         return String.format(HIGHLIGHT_FUNCTION,
-                componentSelector, 
+                componentSelector,
                 makeWordListArray(words),
                 getSearchWordClass()
@@ -63 +62 @@
 
     private String sanitise(String word) {
-        return word.replaceAll("^[\\s'\"]+|[\\s'\"]+$", "");
+        //remove everything up to first colon and strip off quotation marks and white space
+        return word.replaceAll(
+                //match beginning 
+                "^("
+                //case with colon (also strip quotes + optional whitespace after quotes)
+                + "[^:\"']+:(['\"])?"
+                //or case without colon (strip quotes and white space)
+                + "|['\"]+"
+                + ")"
+                //match end
+                + "|("
+                //quotes
+                + "['\"]+"
+                //or boosting values
+                + "|['\"]?\\^.*"
+                + ")$",
+                //replace with empty string
+                "");
     }