Fix NaN deserialization for Quoted escaping rule

src/DataTypes/Serializations/SerializationNullable.cpp

@@ -394,12 +394,65 @@ template<typename ReturnType>
 ReturnType SerializationNullable::deserializeTextQuotedImpl(IColumn & column, ReadBuffer & istr, const FormatSettings & settings,
                                                    const SerializationPtr & nested)
 {
-    return safeDeserialize<ReturnType>(column, *nested,
-        [&istr]
+    if (istr.eof() || (*istr.position() != 'N' && *istr.position() != 'n'))
+    {
+        /// This is not null, surely.
+        return safeDeserialize<ReturnType>(column, *nested,
+            [] { return false; },
+            [&nested, &istr, &settings] (IColumn & nested_column) { nested->deserializeTextQuoted(nested_column, istr, settings); });
+    }
+
+    /// Check if we have enough data in buffer to check if it's a null.
+    if (istr.available() >= 4)
+    {
+        auto check_for_null = [&istr]()
         {
-            return checkStringByFirstCharacterAndAssertTheRestCaseInsensitive("NULL", istr);
-        },
-        [&nested, &istr, &settings] (IColumn & nested_column) { nested->deserializeTextQuoted(nested_column, istr, settings); });
+            auto * pos = istr.position();
+            if (checkStringCaseInsensitive("NULL", istr))
+                return true;
+            istr.position() = pos;
+            return false;
+        };
+        auto deserialize_nested = [&nested, &settings, &istr] (IColumn & nested_column)
+        {
+            nested->deserializeTextQuoted(nested_column, istr, settings);
+        };
+        return safeDeserialize<ReturnType>(column, *nested, check_for_null, deserialize_nested);
+    }
+
+    /// We don't have enough data in buffer to check if it's a NULL
+    /// and we cannot check it just by one symbol (otherwise we won't be able
+    /// to differentiate for example NULL and NaN for float)
+    /// Use PeekableReadBuffer to make a checkpoint before checking
+    /// null and rollback if the check was failed.
+    PeekableReadBuffer buf(istr, true);
+    auto check_for_null = [&buf]()
+    {
+        buf.setCheckpoint();
+        SCOPE_EXIT(buf.dropCheckpoint());
+        if (checkStringCaseInsensitive("NULL", buf))
+            return true;
+
+        buf.rollbackToCheckpoint();
+        return false;
+    };
+
+    auto deserialize_nested = [&nested, &settings, &buf] (IColumn & nested_column)
+    {
+        nested->deserializeTextQuoted(nested_column, buf, settings);
+        /// Check that we don't have any unread data in PeekableReadBuffer own memory.
+        if (likely(!buf.hasUnreadData()))
+            return;
+
+        /// We have some unread data in PeekableReadBuffer own memory.
+        /// It can happen only if there is an unquoted string instead of a number.
+        throw DB::ParsingException(
+            ErrorCodes::CANNOT_READ_ALL_DATA,
+            "Error while parsing Nullable: got an unquoted string {} instead of a number",
+            String(buf.position(), std::min(10ul, buf.available())));
+    };
+
+    return safeDeserialize<ReturnType>(column, *nested, check_for_null, deserialize_nested);
 }
 
 

src/Processors/Formats/Impl/CustomSeparatedRowInputFormat.cpp

@@ -205,7 +205,7 @@ void CustomSeparatedRowInputFormat::syncAfterError()
 
 bool CustomSeparatedRowInputFormat::parseRowStartWithDiagnosticInfo(WriteBuffer & out)
 {
-    return parseDelimiterWithDiagnosticInfo(out, buf, format_settings.custom.row_before_delimiter, "delimiter before first firld", ignore_spaces);
+    return parseDelimiterWithDiagnosticInfo(out, buf, format_settings.custom.row_before_delimiter, "delimiter before first field", ignore_spaces);
 }
 
 bool CustomSeparatedRowInputFormat::parseFieldDelimiterWithDiagnosticInfo(WriteBuffer & out)

tests/queries/0_stateless/02130_parse_quoted_null.reference

@@ -0,0 +1,12 @@
+\N	1
+nan	2
+42.42	3
+\N	4
+\N	5
+\N	6
+\N	7
+nan	8
+nan	9
+nan	10
+nan	11
+OK

tests/queries/0_stateless/02130_parse_quoted_null.sh

@@ -0,0 +1,56 @@
+#!/usr/bin/env bash
+# Tags: no-parallel
+
+CURDIR=$(cd "$(dirname "${BASH_SOURCE[0]}")" && pwd)
+# shellcheck source=../shell_config.sh
+. "$CURDIR"/../shell_config.sh
+
+
+USER_FILES_PATH=$(clickhouse-client --query "select _path,_file from file('nonexist.txt', 'CSV', 'val1 char')" 2>&1 | grep Exception | awk '{gsub("/nonexist.txt","",$9); print $9}')
+DATA_FILE=$USER_FILES_PATH/test_02130.data
+SELECT_QUERY="select * from file('test_02130.data', 'CustomSeparated', 'x Nullable(Float64), y Nullable(UInt64)') settings input_format_parallel_parsing=0, format_custom_escaping_rule='Quoted'"
+
+
+$CLICKHOUSE_CLIENT -q "drop table if exists test_02130"
+$CLICKHOUSE_CLIENT -q "create table test_02130 (x Nullable(Float64), y Nullable(UInt64)) engine=Memory()"
+
+echo -e "null\t1" > $DATA_FILE
+$CLICKHOUSE_CLIENT -q "$SELECT_QUERY"
+
+echo -e "nan\t2" > $DATA_FILE
+$CLICKHOUSE_CLIENT -q "$SELECT_QUERY"
+
+echo -e "42.42\t3" > $DATA_FILE
+$CLICKHOUSE_CLIENT -q "$SELECT_QUERY"
+
+echo -e "null\t4" > $DATA_FILE
+$CLICKHOUSE_CLIENT -q "$SELECT_QUERY" --max_read_buffer_size=1
+
+echo -e "null\t5" > $DATA_FILE
+$CLICKHOUSE_CLIENT -q "$SELECT_QUERY" --max_read_buffer_size=2
+
+echo -e "null\t6" > $DATA_FILE
+$CLICKHOUSE_CLIENT -q "$SELECT_QUERY" --max_read_buffer_size=3
+
+echo -e "null\t7" > $DATA_FILE
+$CLICKHOUSE_CLIENT -q "$SELECT_QUERY" --max_read_buffer_size=4
+
+echo -e "nan\t8" > $DATA_FILE
+$CLICKHOUSE_CLIENT -q "$SELECT_QUERY" --max_read_buffer_size=1
+
+echo -e "nan\t9" > $DATA_FILE
+$CLICKHOUSE_CLIENT -q "$SELECT_QUERY" --max_read_buffer_size=2
+
+echo -e "nan\t10" > $DATA_FILE
+$CLICKHOUSE_CLIENT -q "$SELECT_QUERY" --max_read_buffer_size=3
+
+echo -e "nan\t11" > $DATA_FILE
+$CLICKHOUSE_CLIENT -q "$SELECT_QUERY" --max_read_buffer_size=4
+
+echo -e "42\tnan" > $DATA_FILE
+$CLICKHOUSE_CLIENT -q "$SELECT_QUERY" --max_read_buffer_size=4 2>&1 | grep -F -q "CANNOT_READ_ALL_DATA" && echo 'OK' || echo 'FAIL'
+
+$CLICKHOUSE_CLIENT -q "select * from test_02130 order by y"
+$CLICKHOUSE_CLIENT -q "drop table test_02130"
+
+rm $DATA_FILE