keep an escaped value that equals the null string

Author: saleemno1

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

@@ -806,8 +806,10 @@ private String handleNull(final String input) {
         final String nullString = format.getNullString();
         final boolean strictQuoteMode = isStrictQuoteMode();
         if (input.equals(nullString)) {
+            // An escaped value is the null string itself, not the null marker: printing null writes the null string
+            // verbatim, so "\\N" for nullString "\N" can only have come from a value that really is "\N".
             // nullString = NULL(String), distinguish between "NULL" and NULL in ALL_NON_NULL or NON_NUMERIC quote mode
-            return strictQuoteMode && isQuoted ? input : null;
+            return reusableToken.isEscaped || strictQuoteMode && isQuoted ? input : null;
         }
         // don't set nullString, distinguish between "" and ,, (absent values) in All_NON_NULL or NON_NUMERIC quote mode
         return strictQuoteMode && nullString == null && input.isEmpty() && !isQuoted ? null : input;

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

@@ -76,12 +76,15 @@ final class Lexer implements Closeable {
     private void appendNextEscapedCharacterToToken(final Token token) throws IOException {
         if (isEscapeDelimiter()) {
             token.content.append(delimiter);
+            token.isEscaped = true;
         } else {
             final int unescaped = readEscape();
             if (unescaped == EOF) { // unexpected char after escape
+                // The escape character is kept verbatim, so nothing was translated.
                 token.content.append((char) escape).append((char) reader.getLastChar());
             } else {
                 token.content.append((char) unescaped);
+                token.isEscaped = true;
             }
         }
     }

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

@@ -61,11 +61,15 @@ enum Type {
 
     boolean isQuoted;
 
+    /** True when an escape sequence in the input was translated while building {@link #content}. */
+    boolean isEscaped;
+
     void reset() {
         content.setLength(0);
         type = INVALID;
         isReady = false;
         isQuoted = false;
+        isEscaped = false;
     }
 
     /**

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

@@ -584,6 +584,22 @@ void testEndOfFileBehaviorExcel() throws Exception {
         }
     }
 
+    @Test
+    void testEscapedNullStringIsAValue() throws Exception {
+        // "\N" is the MySQL and PostgreSQL null marker, "\\N" is the value "\N", which is what the printer writes for it.
+        for (final CSVFormat format : new CSVFormat[] { CSVFormat.MYSQL, CSVFormat.POSTGRESQL_TEXT, CSVFormat.ORACLE }) {
+            final StringWriter writer = new StringWriter();
+            try (CSVPrinter printer = new CSVPrinter(writer, format)) {
+                printer.printRecord("\\N", null);
+            }
+            try (CSVParser parser = CSVParser.parse(writer.toString(), format)) {
+                final CSVRecord record = parser.nextRecord();
+                assertEquals("\\N", record.get(0), format.toString());
+                assertNull(record.get(1), format.toString());
+            }
+        }
+    }
+
     @Test
     void testExcelFormat1() throws IOException {
         final String code = "value1,value2,value3,value4\r\na,b,c,d\r\n  x,,,\r\n\r\n\"\"\"hello\"\"\",\"  \"\"world\"\"\",\"abc\ndef\",\r\n";

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

@@ -157,13 +157,17 @@ private void doRandom(final CSVFormat format, final int iter) throws Exception {
     }
 
     /**
-     * Converts an input CSV array into expected output values, including NULLs. NULL strings are converted to null values because the parser will convert
-     * these strings to null.
+     * Converts an input CSV array into expected output values, including NULLs. A value equal to the null string is converted to null only when the printer
+     * writes that value unchanged; when the printer escapes or quotes it, the parser reads it back as the value it is.
      */
-    private <T> T[] expectNulls(final T[] original, final CSVFormat csvFormat) {
+    private <T> T[] expectNulls(final T[] original, final CSVFormat csvFormat) throws IOException {
         final T[] fixed = original.clone();
+        final String nullString = csvFormat.getNullString();
+        if (nullString == null || !printsVerbatim(csvFormat, nullString)) {
+            return fixed;
+        }
         for (int i = 0; i < fixed.length; i++) {
-            if (Objects.equals(csvFormat.getNullString(), fixed[i])) {
+            if (Objects.equals(nullString, fixed[i])) {
                 fixed[i] = null;
             }
         }
@@ -187,6 +191,13 @@ private Connection getH2Connection() throws SQLException, ClassNotFoundException
         return DriverManager.getConnection("jdbc:h2:mem:my_test;", "sa", "");
     }
 
+    /** Tests whether the format prints the given value unchanged, in which case the parser cannot tell it from the null string. */
+    private boolean printsVerbatim(final CSVFormat csvFormat, final String value) throws IOException {
+        final StringBuilder sb = new StringBuilder();
+        csvFormat.print(value, sb, true);
+        return value.contentEquals(sb);
+    }
+
     private CSVPrinter printWithHeaderComments(final StringWriter sw, final Date now, final CSVFormat baseFormat) throws IOException {
         // Use withHeaderComments first to test CSV-145
         // @formatter:off