Improve brat writer newline

This handle the case when multiple newlines are within a annotation
dkpro · Jul 1, 2018 · 2db002e · 2db002e
1 parent 742f4ae
commit 2db002e
Show file tree

Hide file tree

Showing 3 changed files with 25 additions and 29 deletions.
diff --git a/dkpro-core-io-brat-asl/src/main/java/de/tudarmstadt/ukp/dkpro/core/io/brat/BratWriter.java b/dkpro-core-io-brat-asl/src/main/java/de/tudarmstadt/ukp/dkpro/core/io/brat/BratWriter.java
@@ -615,27 +615,32 @@ private void writeRelationAnnotation(BratAnnotationDocument aDoc, FeatureStructu
         }
     }
 
+
+
     private BratTextAnnotation splitNewline(AnnotationFS aFS)
     {
-        String[] textSplit = explodeNewlines(aFS.getCoveredText());
-        int[] begins = new int[textSplit.length];
-        int[] ends = new int[textSplit.length];
-        int pos = aFS.getBegin();
-        int end = aFS.getBegin();
-        for (int i = 0; i < textSplit.length; i++) {
-            end += textSplit[i].length();
-            begins[i] = pos;
-            ends[i] = end;
-            end++;
-            pos = end;
+
+        Pattern p = Pattern.compile("(.+?)(?:\\R|$)+", Pattern.DOTALL);
+        Matcher m = p.matcher(aFS.getCoveredText());
+        // counts the number of matches to initialize arrays sized
+        int i = 0;
+        while (m.find()) {
+            i++;
+        }
+        // initialize arrays
+        int[] begins = new int[i];
+        int[] ends = new int[i];
+        m = p.matcher(aFS.getCoveredText());
+        i = 0;
+        while (m.find()) {
+            System.out.println(m.group(1));
+            begins[i] = m.start(1) + aFS.getBegin();
+            ends[i] = m.end(1) + aFS.getBegin();
+            i++;
         }
 
         return new BratTextAnnotation(nextTextAnnotationId, getBratType(aFS.getType()), begins,
-                ends, new String[] { aFS.getCoveredText().replaceAll("\\R", " ") });
-    }
-
-    private String[] explodeNewlines(String str) {
-        return str.split("\\R");
+                ends, new String[] { aFS.getCoveredText().replaceAll("\\R+", " ") });
     }
 
     private void writeTextAnnotation(BratAnnotationDocument aDoc, AnnotationFS aFS)

diff --git a/...io-brat-asl/src/test/java/de/tudarmstadt/ukp/dkpro/core/io/brat/BratReaderWriterTest.java b/...io-brat-asl/src/test/java/de/tudarmstadt/ukp/dkpro/core/io/brat/BratReaderWriterTest.java
@@ -132,7 +132,7 @@ public void testWithLongNames() throws Exception
     }
 
     @Test
-    public void testBratWithNewlines() throws Exception
+    public void testBratWithDiscontinuousFragmentNear() throws Exception
     {
         testRoundTrip(createReaderDescription(BratReader.class,
                 BratReader.PARAM_TEXT_ANNOTATION_TYPE_MAPPINGS,
@@ -145,7 +145,7 @@ public void testBratWithNewlines() throws Exception
     }
 
     @Test
-    public void testBratWithDiscontinousTwo() throws Exception
+    public void testBratWithDiscontinuousFragmentFar() throws Exception
     {
         testOneWay(createReaderDescription(BratReader.class,
                 BratReader.PARAM_TEXT_ANNOTATION_TYPE_MAPPINGS,

diff --git a/...est/java/de/tudarmstadt/ukp/dkpro/core/io/brat/internal/model/BratTextAnnotationTest.java b/...est/java/de/tudarmstadt/ukp/dkpro/core/io/brat/internal/model/BratTextAnnotationTest.java
@@ -40,21 +40,12 @@ public void parseTestZeroLength()
     }
 
     @Test
-    public void parseTestDiscontinous1()
+    public void parseTestDiscontinousMergeFragments()
     {
         final String in = "T1\tOrganization 0 13;14 43\tInternational Business Machines Corporation";
         final String out = "T1\tOrganization 0 43\tInternational Business Machines Corporation";
         BratTextAnnotation v = BratTextAnnotation.parse(in);
         assertEquals(out, v.toString());
     }
-
-    @Test
-    public void parseTestDiscontinous2()
-    {
-        final String in = "T1\tOrganization 0 13;15 43\tInternational Business Machines Corporation";
-        final String out = "T1\tOrganization 0 13;15 43\tInternational";
-        BratTextAnnotation v = BratTextAnnotation.parse(in);
-        System.out.println(v);
-        assertEquals(out, v.toString());
-    }
+
 }