Merge pull request #689 from schen149/master

Fixing inefficient string concat in core-utilities

Author: CogComp Dev

core-utilities/src/main/java/edu/illinois/cs/cogcomp/annotation/BasicTextAnnotationBuilder.java

@@ -79,11 +79,12 @@ public static TextAnnotation createTextAnnotationFromListofListofTokens(List<Lis
     public static TextAnnotation createTextAnnotationFromTokens(String corpusId, String textId,
             List<String[]> tokenizedSentences) {
         Tokenization tokenization = tokenizeTextSpan(tokenizedSentences);
-        String text = "";
+        StringBuilder text = new StringBuilder();
         for (String[] sentenceTokens : tokenizedSentences)
-            text += StringUtils.join(sentenceTokens, ' ') + System.lineSeparator();
+            text.append(StringUtils.join(sentenceTokens, ' '))
+                    .append(System.lineSeparator());
 
-        return new TextAnnotation(corpusId, textId, text, tokenization.getCharacterOffsets(),
+        return new TextAnnotation(corpusId, textId, text.toString(), tokenization.getCharacterOffsets(),
                 tokenization.getTokens(), tokenization.getSentenceEndTokenIndexes());
     }
 

core-utilities/src/main/java/edu/illinois/cs/cogcomp/core/datastructures/trees/Tree.java

@@ -297,27 +297,29 @@ public boolean hasEdgeLabels() {
      */
 
     private String toString(boolean firstChild, int spaces) {
-        String treeString = "";
+        StringBuilder treeString = new StringBuilder();
         if (!firstChild) {
             for (int i = 0; i < spaces; i++) {
-                treeString += " ";
+                treeString.append(" ");
             }
         }
 
         if (!this.isLeaf())
-            treeString += "(";
+            treeString.append("(");
 
         int position = this.getPositionAmongParentsChildren();
 
         if (this.parent != null && this.parent.childrenEdgeLabels != null
                 && this.parent.childrenEdgeLabels.get(position) != null) {
-            treeString += (":LABEL:" + this.parent.childrenEdgeLabels.get(position) + " ");
+            treeString.append(":LABEL:")
+                    .append(this.parent.childrenEdgeLabels.get(position))
+                    .append(" ");
         }
 
-        treeString += label.toString();
+        treeString.append(label.toString());
 
         if (children.size() > 0)
-            treeString += " ";
+            treeString.append(" ");
 
         int len = treeString.length();
 
@@ -326,19 +328,19 @@ private String toString(boolean firstChild, int spaces) {
         for (Iterator<Tree<T>> iterator = children.iterator(); iterator.hasNext();) {
             Tree<T> child = iterator.next();
 
-            treeString += child.toString(first, len);
+            treeString.append(child.toString(first, len));
 
             first = false;
 
             if (iterator.hasNext())
-                treeString += "\n";
+                treeString.append("\n");
             index++;
         }
 
         if (!this.isLeaf())
-            treeString += ")";
+            treeString.append(")");
 
-        return treeString;
+        return treeString.toString();
 
     }
 

core-utilities/src/main/java/edu/illinois/cs/cogcomp/core/utilities/Table.java

@@ -82,9 +82,9 @@ public String toTextTable() {
             }
         }
 
-        String formatterString = "";
+        StringBuilder formatterString = new StringBuilder();
         for (int i = 0; i < formatterList.size(); i ++) {
-            formatterString += formatterList.get(i);
+            formatterString.append(formatterList.get(i));
         }
 
         int lenColSep = 0;
@@ -99,7 +99,7 @@ public String toTextTable() {
                 buffer.append(colSep);
                 buffer.append("|\n");
             }
-            String appendStr = String.format(formatterString, (allTableRows.get(i)).toArray());
+            String appendStr = String.format(formatterString.toString(), (allTableRows.get(i)).toArray());
             buffer.append(appendStr);
             if (i == 0) {
                 lenColSep = appendStr.length();

core-utilities/src/test/java/edu/illinois/cs/cogcomp/nlp/utilities/TextCleanerTest.java

@@ -159,12 +159,12 @@ public void testCleanLongText() {
             fail(e.getMessage());
         }
 
-        String badText = "";
+        StringBuilder badText = new StringBuilder();
 
         for (String line : badTextLines)
-            badText += line + "\n";
+            badText.append(line).append("\n");
 
-        String cleanText = TextCleaner.replaceControlSequence(badText);
+        String cleanText = TextCleaner.replaceControlSequence(badText.toString());
 
         boolean isGood = !(cleanText.contains("^@"));