Merge pull request #621 from rootvector2/quote-unicode-whitespace

Quote value with leading or trailing Unicode whitespace

Author: garydgregory

src/main/java/org/apache/commons/csv/CSVFormat.java

@@ -2514,11 +2514,12 @@ private void printWithQuotes(final Object object, final CharSequence charSeq, fi
                 }
             } else {
                 char c = charSeq.charAt(pos);
-                if (c <= Constants.COMMENT || isCommentMarkerSet() && c == commentMarker.charValue()) {
+                if (c <= Constants.COMMENT || Character.isWhitespace(c) || isCommentMarkerSet() && c == commentMarker.charValue()) {
                     // Some other chars at the start of a value caused the parser to fail, so for now
                     // encapsulate if we start in anything less than '#'. We are being conservative
                     // by including the default comment char and any configured comment marker too,
-                    // which the parser would otherwise read back as a comment line.
+                    // which the parser would otherwise read back as a comment line. A leading Unicode
+                    // whitespace above ' ' is stripped by ignoreSurroundingSpaces on read, so quote it too.
                     quote = true;
                 } else {
                     while (pos < len) {
@@ -2534,8 +2535,9 @@ private void printWithQuotes(final Object object, final CharSequence charSeq, fi
                         pos = len - 1;
                         c = charSeq.charAt(pos);
                         // Some other chars at the end caused the parser to fail, so for now
-                        // encapsulate if we end in anything less than ' '
-                        if (isTrimChar(c)) {
+                        // encapsulate if we end in anything less than ' '. A trailing Unicode whitespace
+                        // above ' ' is stripped by ignoreSurroundingSpaces on read, so quote it too.
+                        if (isTrimChar(c) || Character.isWhitespace(c)) {
                             quote = true;
                         } else if (endsWithDelimiterPrefix(charSeq, delim, delimLength)) {
                             // A trailing partial multi-character delimiter would merge with the following delimiter on read.

src/test/java/org/apache/commons/csv/CSVPrinterTest.java

@@ -1963,6 +1963,30 @@ void testQuoteValueEndingWithSupplementaryDelimiterPrefix() throws IOException {
         }
     }
 
+    @Test
+    void testQuoteValueWithLeadingOrTrailingUnicodeWhitespace() throws IOException {
+        // U+3000 IDEOGRAPHIC SPACE is Character.isWhitespace but greater than ' ', so ignoreSurroundingSpaces
+        // strips it on read. It must be quoted or the surrounding whitespace is lost on a round trip.
+        final String idSpace = "　";
+        final CSVFormat format = CSVFormat.DEFAULT.builder().setIgnoreSurroundingSpaces(true).get();
+        final StringWriter sw = new StringWriter();
+        try (CSVPrinter printer = new CSVPrinter(sw, format)) {
+            printer.printRecord(idSpace + "a" + idSpace, "b");
+            // A value without surrounding whitespace stays unquoted.
+            printer.printRecord("ab", "c");
+        }
+        final String string = sw.toString();
+        assertEquals("\"" + idSpace + "a" + idSpace + "\",b" + RECORD_SEPARATOR + "ab,c" + RECORD_SEPARATOR, string);
+        try (CSVParser parser = CSVParser.parse(string, format)) {
+            final List<CSVRecord> records = parser.getRecords();
+            assertEquals(2, records.size());
+            assertEquals(idSpace + "a" + idSpace, records.get(0).get(0));
+            assertEquals("b", records.get(0).get(1));
+            assertEquals("ab", records.get(1).get(0));
+            assertEquals("c", records.get(1).get(1));
+        }
+    }
+
     @Test
     void testRandomDefault() throws Exception {
         doRandom(CSVFormat.DEFAULT, ITERATIONS_FOR_RANDOM_TEST);