Merge pull request apache#608 from rootvector2/escape-delimiter-buffer-clear

Clear escape delimiter buffer before peek in isEscapeDelimiter

Author: garydgregory

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

@@ -191,6 +191,7 @@ boolean isEscape(final int ch) {
      * @throws IOException If an I/O error occurs.
      */
     boolean isEscapeDelimiter() throws IOException {
+        Arrays.fill(escapeDelimiterBuf, '\0');
         reader.peek(escapeDelimiterBuf);
         if (escapeDelimiterBuf[0] != delimiter[0]) {
             return false;

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

@@ -1682,6 +1682,20 @@ void testPartialMultiCharacterDelimiterAtEOF() throws IOException {
         }
     }
 
+    /**
+     * A truncated escaped multi-character delimiter at EOF must stay literal data and not be completed from a stale
+     * escape delimiter look-ahead.
+     */
+    @Test
+    void testPartialEscapedMultiCharacterDelimiterAtEOF() throws IOException {
+        final CSVFormat format = CSVFormat.DEFAULT.builder().setDelimiter("[|]").setEscape('!').get();
+        try (CSVParser parser = format.parse(new StringReader("x![!|!]y![!|"))) {
+            final CSVRecord record = parser.nextRecord();
+            assertEquals("x[|]y![!|", record.get(0));
+            assertEquals(1, record.size());
+        }
+    }
+
     @Test
     void testProvidedHeader() throws Exception {
         final Reader in = new StringReader("a,b,c\n1,2,3\nx,y,z");

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

@@ -421,6 +421,18 @@ void testPartialMultiCharacterDelimiterAtEOF() throws IOException {
         }
     }
 
+    /**
+     * A truncated escaped multi-character delimiter at EOF must not be accepted by reusing the previous escape delimiter
+     * look-ahead in {@link Lexer#isEscapeDelimiter()}.
+     */
+    @Test
+    void testPartialEscapedMultiCharacterDelimiterAtEOF() throws IOException {
+        final CSVFormat format = CSVFormat.DEFAULT.builder().setDelimiter("[|]").setEscape('!').get();
+        try (Lexer lexer = createLexer("x![!|!]y![!|", format)) {
+            assertNextToken(EOF, "x[|]y![!|", lexer);
+        }
+    }
+
     @Test
     void testReadEscapeBackspace() throws IOException {
         try (Lexer lexer = createLexer("b", CSVFormat.DEFAULT.withEscape('\b'))) {