Merge pull request #614 from rootvector2/escape-comment-marker-first-char

Escape leading comment marker in printWithEscapes

Author: garydgregory

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

@@ -2329,12 +2329,16 @@ private void printWithEscapes(final CharSequence charSeq, final Appendable appen
         final char escape = getEscapeChar();
         final boolean quoteSet = isQuoteCharacterSet();
         final char quote = quoteSet ? getQuoteCharacter().charValue() : 0;
+        final boolean commentMarkerSet = isCommentMarkerSet();
+        final char commentChar = commentMarkerSet ? commentMarker.charValue() : 0; // Explicit unboxing is intentional
         while (pos < end) {
             char c = charSeq.charAt(pos);
             final boolean isDelimiterStart = isDelimiter(c, charSeq, pos, delimArray, delimLength);
             final boolean isCr = c == Constants.CR;
             final boolean isLf = c == Constants.LF;
-            if (isCr || isLf || c == escape || quoteSet && c == quote || isDelimiterStart) {
+            // A leading comment marker would be read back as a comment, so escape it.
+            final boolean isComment = commentMarkerSet && pos == 0 && c == commentChar;
+            if (isCr || isLf || c == escape || quoteSet && c == quote || isDelimiterStart || isComment) {
                 // write out segment up until this char
                 if (pos > start) {
                     appendable.append(charSeq, start, pos);
@@ -2375,8 +2379,11 @@ private void printWithEscapes(final Reader reader, final Appendable appendable)
         final char escape = getEscapeChar();
         final boolean quoteSet = isQuoteCharacterSet();
         final char quote = quoteSet ? getQuoteCharacter().charValue() : 0;
+        final boolean commentMarkerSet = isCommentMarkerSet();
+        final char commentChar = commentMarkerSet ? commentMarker.charValue() : 0; // Explicit unboxing is intentional
         final StringBuilder builder = new StringBuilder(IOUtils.DEFAULT_BUFFER_SIZE);
         int c;
+        boolean firstChar = true;
         final char[] lookAheadBuffer = new char[delimLength - 1];
         while (EOF != (c = bufferedReader.read())) {
             builder.append((char) c);
@@ -2386,7 +2393,10 @@ private void printWithEscapes(final Reader reader, final Appendable appendable)
             final boolean isDelimiterStart = isDelimiter((char) c, test, pos, delimArray, delimLength);
             final boolean isCr = c == Constants.CR;
             final boolean isLf = c == Constants.LF;
-            if (isCr || isLf || c == escape || quoteSet && c == quote || isDelimiterStart) {
+            // A leading comment marker would be read back as a comment, so escape it.
+            final boolean isComment = commentMarkerSet && firstChar && c == commentChar;
+            firstChar = false;
+            if (isCr || isLf || c == escape || quoteSet && c == quote || isDelimiterStart || isComment) {
                 // write out segment up until this char
                 if (pos > start) {
                     append(builder.substring(start, pos), appendable);

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

@@ -569,6 +569,57 @@ void testEscapeBackslash5() throws IOException {
         assertEquals("\\\\", sw.toString());
     }
 
+    @Test
+    void testEscapeCommentMarkerFirstChar() throws IOException {
+        // No quoting available in escape mode, so a leading comment marker must be escaped or the
+        // record reads back as a comment and is dropped. Mirrors the quoting fix for QuoteMode.MINIMAL.
+        final CSVFormat format = CSVFormat.DEFAULT.builder().setQuote(null).setEscape('\\').setCommentMarker(';').get();
+        final StringWriter sw = new StringWriter();
+        final String col1 = ";comment-like";
+        try (CSVPrinter printer = new CSVPrinter(sw, format)) {
+            printer.printRecord(col1, "b");
+            printer.printRecord(new StringReader(col1), new StringReader("b"));
+            // The marker past the first character does not start a comment and is left alone.
+            printer.printRecord("a;b", ";c");
+        }
+        final String string = sw.toString();
+        assertEquals("\\;comment-like,b" + RECORD_SEPARATOR +
+                "\\;comment-like,b" + RECORD_SEPARATOR +
+                "a;b,\\;c" + RECORD_SEPARATOR, string);
+        // The emitted records must read back as the original values, none parsed as a comment.
+        try (CSVParser parser = CSVParser.parse(string, format)) {
+            final List<CSVRecord> records = parser.getRecords();
+            assertEquals(3, records.size());
+            assertEquals(col1, records.get(0).get(0));
+            assertEquals("b", records.get(0).get(1));
+            assertEquals(col1, records.get(1).get(0));
+            assertEquals("b", records.get(1).get(1));
+            assertEquals("a;b", records.get(2).get(0));
+            assertEquals(";c", records.get(2).get(1));
+        }
+    }
+
+    @Test
+    void testEscapeCommentMarkerFirstCharWithQuoteModeNone() throws IOException {
+        final CSVFormat format = CSVFormat.DEFAULT.builder().setEscape('\\').setQuoteMode(QuoteMode.NONE).setCommentMarker(';').get();
+        final StringWriter sw = new StringWriter();
+        final String col1 = ";bar";
+        try (CSVPrinter printer = new CSVPrinter(sw, format)) {
+            printer.printRecord(col1, "b");
+            printer.printRecord(new StringReader(col1), new StringReader("b"));
+        }
+        final String string = sw.toString();
+        assertEquals("\\;bar,b" + RECORD_SEPARATOR + "\\;bar,b" + RECORD_SEPARATOR, string);
+        try (CSVParser parser = CSVParser.parse(string, format)) {
+            final List<CSVRecord> records = parser.getRecords();
+            assertEquals(2, records.size());
+            for (final CSVRecord record : records) {
+                assertEquals(col1, record.get(0));
+                assertEquals("b", record.get(1));
+            }
+        }
+    }
+
     @Test
     void testEscapeNull1() throws IOException {
         final StringWriter sw = new StringWriter();