Issue #10924: Update SingleSpaceSeparatorCheck to use code points

pom.xml

@@ -3292,6 +3292,9 @@
                 <param>
                   com.puppycrawl.tools.checkstyle.checks.AvoidEscapedUnicodeCharactersCheckTest
                 </param>
+                <param>
+                  com.puppycrawl.tools.checkstyle.checks.whitespace.SingleSpaceSeparatorCheckTest
+                </param>
                 <!-- 2% mutation in ScopeUtil -->
                 <param>
                   com.puppycrawl.tools.checkstyle.checks.javadoc.JavadocStyleCheckTest

src/main/java/com/puppycrawl/tools/checkstyle/checks/whitespace/SingleSpaceSeparatorCheck.java

@@ -19,9 +19,12 @@
 
 package com.puppycrawl.tools.checkstyle.checks.whitespace;
 
+import java.util.Arrays;
+
 import com.puppycrawl.tools.checkstyle.StatelessCheck;
 import com.puppycrawl.tools.checkstyle.api.AbstractCheck;
 import com.puppycrawl.tools.checkstyle.api.DetailAST;
+import com.puppycrawl.tools.checkstyle.utils.CodePointUtil;
 import com.puppycrawl.tools.checkstyle.utils.CommonUtil;
 
 /**
@@ -188,7 +191,7 @@ private void visitEachToken(DetailAST node) {
 
             if (columnNo >= minSecondWhitespaceColumnNo
                     && !isTextSeparatedCorrectlyFromPrevious(
-                            getLine(currentNode.getLineNo() - 1),
+                            getLineCodePoints(currentNode.getLineNo() - 1),
                             columnNo)) {
                 log(currentNode, MSG_KEY);
             }
@@ -218,14 +221,14 @@ private void visitEachToken(DetailAST node) {
      * end of a block comment. </li>
      * </ul>
      *
-     * @param line The line in the file to examine.
+     * @param line Unicode code point array of line in the file to examine.
      * @param columnNo The column position in the {@code line} to examine.
      * @return {@code true} if the text at {@code columnNo} is separated
      *         correctly from the previous token.
      */
-    private boolean isTextSeparatedCorrectlyFromPrevious(String line, int columnNo) {
+    private boolean isTextSeparatedCorrectlyFromPrevious(int[] line, int columnNo) {
         return isSingleSpace(line, columnNo)
-                || !isWhitespace(line, columnNo)
+                || !CommonUtil.isCodePointWhitespace(line, columnNo)
                 || isFirstInLine(line, columnNo)
                 || !validateComments && isBlockCommentEnd(line, columnNo);
     }
@@ -234,61 +237,51 @@ private boolean isTextSeparatedCorrectlyFromPrevious(String line, int columnNo)
      * Checks if the {@code line} at {@code columnNo} is a single space, and not
      * preceded by another space.
      *
-     * @param line The line in the file to examine.
+     * @param line Unicode code point array of line in the file to examine.
      * @param columnNo The column position in the {@code line} to examine.
      * @return {@code true} if the character at {@code columnNo} is a space, and
      *         not preceded by another space.
      */
-    private static boolean isSingleSpace(String line, int columnNo) {
-        return isSpace(line, columnNo) && !Character.isWhitespace(line.charAt(columnNo - 1));
+    private static boolean isSingleSpace(int[] line, int columnNo) {
+        return isSpace(line, columnNo) && !CommonUtil.isCodePointWhitespace(line, columnNo - 1);
     }
 
     /**
      * Checks if the {@code line} at {@code columnNo} is a space.
      *
-     * @param line The line in the file to examine.
+     * @param line Unicode code point array of line in the file to examine.
      * @param columnNo The column position in the {@code line} to examine.
      * @return {@code true} if the character at {@code columnNo} is a space.
      */
-    private static boolean isSpace(String line, int columnNo) {
-        return line.charAt(columnNo) == ' ';
-    }
-
-    /**
-     * Checks if the {@code line} at {@code columnNo} is a whitespace character.
-     *
-     * @param line The line in the file to examine.
-     * @param columnNo The column position in the {@code line} to examine.
-     * @return {@code true} if the character at {@code columnNo} is a
-     *         whitespace.
-     */
-    private static boolean isWhitespace(String line, int columnNo) {
-        return Character.isWhitespace(line.charAt(columnNo));
+    private static boolean isSpace(int[] line, int columnNo) {
+        return line[columnNo] == ' ';
     }
 
     /**
      * Checks if the {@code line} up to and including {@code columnNo} is all
      * non-whitespace text encountered.
      *
-     * @param line The line in the file to examine.
+     * @param line Unicode code point array of line in the file to examine.
      * @param columnNo The column position in the {@code line} to examine.
      * @return {@code true} if the column position is the first non-whitespace
      *         text on the {@code line}.
      */
-    private static boolean isFirstInLine(String line, int columnNo) {
-        return CommonUtil.isBlank(line.substring(0, columnNo));
+    private static boolean isFirstInLine(int[] line, int columnNo) {
+        return CodePointUtil.isBlank(Arrays.copyOfRange(line, 0, columnNo));
     }
 
     /**
      * Checks if the {@code line} at {@code columnNo} is the end of a comment,
      * '*&#47;'.
      *
-     * @param line The line in the file to examine.
+     * @param line Unicode code point array of line in the file to examine.
      * @param columnNo The column position in the {@code line} to examine.
      * @return {@code true} if the previous text is a end comment block.
      */
-    private static boolean isBlockCommentEnd(String line, int columnNo) {
-        return line.substring(0, columnNo).trim().endsWith("*/");
+    private static boolean isBlockCommentEnd(int[] line, int columnNo) {
+        final int[] strippedLine = CodePointUtil
+                .stripTrailing(Arrays.copyOfRange(line, 0, columnNo));
+        return CodePointUtil.endsWith(strippedLine, "*/");
     }
 
 }

src/main/java/com/puppycrawl/tools/checkstyle/utils/CodePointUtil.java

@@ -55,4 +55,32 @@ public static boolean hasWhitespaceBefore(int index, int... codePoints) {
                 .allMatch(Character::isWhitespace);
     }
 
+    /**
+     * Removes trailing whitespaces.
+     *
+     * @param codePoints array of unicode code points
+     * @return unicode code points array with trailing whitespaces removed
+     */
+    public static int[] stripTrailing(int... codePoints) {
+        int lastIndex = codePoints.length;
+        while (CommonUtil.isCodePointWhitespace(codePoints, lastIndex - 1)) {
+            lastIndex--;
+        }
+        return Arrays.copyOfRange(codePoints, 0, lastIndex);
+    }
+
+    /**
+     * Tests if the unicode code points array
+     * ends with the specified suffix.
+     *
+     * @param suffix the suffix
+     * @param codePoints the array of unicode code points to check
+     * @return {@code true}, if the unicode code points array ends with provided suffix
+     */
+    public static boolean endsWith(int[] codePoints, String suffix) {
+        final int startIndex = codePoints.length - suffix.length();
+        return startIndex > -1 && Arrays.equals(Arrays
+                    .copyOfRange(codePoints, startIndex, codePoints.length),
+                    suffix.codePoints().toArray());
+    }
 }

src/test/java/com/puppycrawl/tools/checkstyle/checks/whitespace/SingleSpaceSeparatorCheckTest.java

@@ -159,4 +159,41 @@ public void testEmpty() throws Exception {
         verifyWithInlineConfigParser(
                 getPath("InputSingleSpaceSeparatorEmpty.java"), expected);
     }
+
+    @Test
+    public void testSpaceErrorsWithEmoji() throws Exception {
+        final String[] expected = {
+            "14:18: " + getCheckMessage(MSG_KEY),
+            "16:17: " + getCheckMessage(MSG_KEY),
+            "18:27: " + getCheckMessage(MSG_KEY),
+            "24:46: " + getCheckMessage(MSG_KEY),
+            "27:9: " + getCheckMessage(MSG_KEY),
+            "33:17: " + getCheckMessage(MSG_KEY),
+            "36:14: " + getCheckMessage(MSG_KEY),
+            "36:25: " + getCheckMessage(MSG_KEY),
+            "36:37: " + getCheckMessage(MSG_KEY),
+            "37:43: " + getCheckMessage(MSG_KEY),
+            "37:46: " + getCheckMessage(MSG_KEY),
+            "38:15: " + getCheckMessage(MSG_KEY),
+            "40:16: " + getCheckMessage(MSG_KEY),
+        };
+        verifyWithInlineConfigParser(
+                getPath("InputSingleSpaceSeparatorWithEmoji.java"), expected);
+    }
+
+    @Test
+    public void testSpaceErrorsAroundCommentsWithEmoji() throws Exception {
+        final String[] expected = {
+            "25:22: " + getCheckMessage(MSG_KEY),
+            "25:26: " + getCheckMessage(MSG_KEY),
+            "26:26: " + getCheckMessage(MSG_KEY),
+            "27:13: " + getCheckMessage(MSG_KEY),
+            "34:8: " + getCheckMessage(MSG_KEY),
+            "36:37: " + getCheckMessage(MSG_KEY),
+            "38:46: " + getCheckMessage(MSG_KEY),
+            "40:24: " + getCheckMessage(MSG_KEY),
+        };
+        verifyWithInlineConfigParser(
+            getPath("InputSingleSpaceSeparatorCommentsWithEmoji.java"), expected);
+    }
 }

src/test/resources/com/puppycrawl/tools/checkstyle/checks/whitespace/singlespaceseparator/InputSingleSpaceSeparatorCommentsWithEmoji.java

@@ -0,0 +1,46 @@
+/*
+SingleSpaceSeparator
+validateComments = true
+
+
+*/
+
+package com.puppycrawl.tools.checkstyle.checks.whitespace.singlespaceseparator;
+
+public class InputSingleSpaceSeparatorCommentsWithEmoji {
+
+
+    String s1 = "🧐😉           assa  "; // ok
+    String s2 = "🧐😉" + "          " + "assa  "; // ok
+    String s3 = "🧐" + "🎄 "; // ok
+    String s4
+        = "da🎄"; // ok
+    private void foo(String s) {
+        if (s.equals("🤩🎄") ){ // ok
+            foo(s);
+        }
+    }
+
+    /* ok */ /*🧐*/ String s = "🎄😉";
+    String /* 🧐  ok  */  st = "🎄  assa"; // 2 violations
+    String str = "🤩🎄";   // violation 'Use a single space to separate non-whitespace characters'
+    ;       String j = ""; // violation 'Use a single space to separate non-whitespace characters'
+    /**
+     * ok
+     */
+    void foo2() {
+        /*🧐 🧐 🧐 🧐*/ /* ok */
+        String s = "🧐  🧐";
+    }  // violation 'Use a single space to separate non-whitespace characters'
+
+    private void foo3 (String s) {  // ok
+        // violation above 'Use a single space to separate non-whitespace characters'
+        if (s.substring(0).equals("da🎄")) {  // ok
+            // violation above 'Use a single space to separate non-whitespace characters'
+            /*🧐 🧐 🧐*/  /* comment */ if ("🧐".isEmpty()){ // ok
+            //  violation above 'Use a single space to separate non-whitespace characters'
+            }
+        }
+
+    }
+}

src/test/resources/com/puppycrawl/tools/checkstyle/checks/whitespace/singlespaceseparator/InputSingleSpaceSeparatorWithEmoji.java

@@ -0,0 +1,45 @@
+/*
+SingleSpaceSeparator
+validateComments = (default)false
+
+
+*/
+
+package com.puppycrawl.tools.checkstyle.checks.whitespace.singlespaceseparator;
+
+public class InputSingleSpaceSeparatorWithEmoji {
+
+    String s1 = "🧐😉           assa  "; // ok
+    String s2 = "🧐😉" + "          " + "assa  "; // ok
+    String s3 =  "🧐" + "🎄 "; // violation 'Use a single space to separate non-whitespace characters'
+    String s4
+            =   "da🎄"; // violation 'Use a single space to separate non-whitespace characters'
+    private void foo(String s) {
+      if (s.equals("🤩🎄")  ){ // violation 'Use a single space to separate non-whitespace characters'
+                foo(s);
+      }
+    }
+
+    /* ok */ /*🧐*/ String s = "🎄😉";
+    String /* 🧐  block comment ok */   st =  "🎄  assa";
+    // violation above 'Use a single space to separate non-whitespace characters'
+    String str = "🤩🎄";   // Multiple whitespaces before comment
+;       String j = ""; // violation 'Use a single space to separate non-whitespace characters'
+    /**
+     * ok
+     */
+    void foo2() {
+        /* ok */
+        String  s = "🧐  🧐"; // violation 'Use a single space to separate non-whitespace characters'
+    }  // ok
+
+    private  void foo3  (String     s) { // 3 violations
+        if (s.substring(0).equals("da🎄")  )  { // 2 violations
+          if  ("🧐". // violation 'Use a single space to separate non-whitespace characters'
+                  isEmpty()){
+/*🧐*/       }  else { // violation 'Use a single space to separate non-whitespace characters'
+          }
+        }
+
+    }
+}