Merge pull request ClickHouse#78919 from Avogar/csv-date-time-better-exception

Fix parsing of bad DateTime values in CSV format

Author: Avogar

src/DataTypes/Serializations/SerializationDateTime.cpp

@@ -13,6 +13,11 @@
 namespace DB
 {
 
+namespace ErrorCodes
+{
+    extern const int UNEXPECTED_DATA_AFTER_PARSED_VALUE;
+}
+
 namespace
 {
 
@@ -253,7 +258,11 @@ void SerializationDateTime::deserializeTextCSV(IColumn & column, ReadBuffer & is
             ReadBufferFromString buf(datetime_str);
             readText(x, buf, settings, time_zone, utc_time_zone);
             if (!buf.eof())
-                throwUnexpectedDataAfterParsedValue(column, istr, settings, "DateTime");
+                throw Exception(
+                    ErrorCodes::UNEXPECTED_DATA_AFTER_PARSED_VALUE,
+                    "Unexpected data '{}' after parsed DateTime value '{}'",
+                    String(buf.position(), buf.buffer().end()),
+                    String(buf.buffer().begin(), buf.position()));
         }
     }
 

src/DataTypes/Serializations/SerializationDateTime64.cpp

@@ -13,6 +13,11 @@
 namespace DB
 {
 
+namespace ErrorCodes
+{
+    extern const int UNEXPECTED_DATA_AFTER_PARSED_VALUE;
+}
+
 SerializationDateTime64::SerializationDateTime64(
     UInt32 scale_, const TimezoneMixin & time_zone_)
     : SerializationDecimalBase<DateTime64>(DecimalUtils::max_precision<DateTime64>, scale_)
@@ -243,6 +248,12 @@ void SerializationDateTime64::deserializeTextCSV(IColumn & column, ReadBuffer &
             readCSVString(datetime_str, istr, settings.csv);
             ReadBufferFromString buf(datetime_str);
             readText(x, scale, buf, settings, time_zone, utc_time_zone);
+            if (!buf.eof())
+                throw Exception(
+                    ErrorCodes::UNEXPECTED_DATA_AFTER_PARSED_VALUE,
+                    "Unexpected data '{}' after parsed DateTime64 value '{}'",
+                    String(buf.position(), buf.buffer().end()),
+                    String(buf.buffer().begin(), buf.position()));
         }
     }
 

tests/queries/0_stateless/03407_csv_bad_date_time_parsing.reference

@@ -0,0 +1,3 @@
+select * from format(CSV, 'd DateTime64(3)', '1744042005 797') settings date_time_input_format='best_effort'; -- {serverError UNEXPECTED_DATA_AFTER_PARSED_VALUE}
+select * from format(CSV, 'd DateTime', '1744042005 797') settings date_time_input_format='best_effort'; -- {serverError UNEXPECTED_DATA_AFTER_PARSED_VALUE}
+