Support reading from CSV tables with null escape and quote characters

Author: mcarmonagonzalez

lib/trino-hive-formats/src/main/java/io/trino/hive/formats/line/csv/CsvDeserializer.java

@@ -60,7 +60,11 @@ public CsvDeserializer(List<Column> columns, char separatorChar, char quoteChar,
         checkArgument(separatorChar != '\0', "Separator cannot be the null character (ASCII 0)");
         checkArgument(separatorChar != quoteChar, "Separator and quote character cannot be the same");
         checkArgument(separatorChar != escapeChar, "Separator and escape character cannot be the same");
-        checkArgument(quoteChar != escapeChar, "Quote and escape character cannot be the same");
+
+        // Quote and escape character can be the same when both are the null character (quoting and escaping are disabled)
+        if (quoteChar != '\0' || escapeChar != '\0') {
+            checkArgument(quoteChar != escapeChar, "Quote and escape character cannot be the same");
+        }
         this.separatorChar = separatorChar;
         this.quoteChar = quoteChar;
         this.escapeChar = escapeChar;

lib/trino-hive-formats/src/test/java/io/trino/hive/formats/line/csv/TestCsvFormat.java

@@ -126,6 +126,10 @@ public void testCsv()
         assertTrinoHiveByteForByte(true, Arrays.asList("f**", "b*r", "b*z"), Optional.of('\t'), Optional.of('*'), Optional.of('#'));
         assertTrinoHiveByteForByte(false, Arrays.asList("f**", "b*r", "b*z"), Optional.of('\t'), Optional.of('*'), Optional.of('\0'));
 
+        // If both the quote character and escape character are `\0` then quoting and escaping is simply disabled, even if this would cause output that does not round trip
+        assertTrinoHiveByteForByte(true, Arrays.asList("foo", "bar", "baz"), Optional.of('\t'), Optional.of('\0'), Optional.of('\0'));
+        assertTrinoHiveByteForByte(false, Arrays.asList("f\t\t", "\tbar\t", "baz"), Optional.of('\t'), Optional.of('\0'), Optional.of('\0'));
+
         // These cases don't round trip, because Hive uses different default escape characters for serialization and deserialization.
         // For serialization the pipe character is escaped with a quote char, but for deserialization escape character is the backslash character
         assertTrinoHiveByteForByte(false, Arrays.asList("|", "a", "b"), Optional.empty(), Optional.of('|'), Optional.empty());