Basics of implicit time/timestamp cast

awitten1 · Apr 19, 2024 · 239a1d1 · 239a1d1
1 parent 1601d94
commit 239a1d1
Show file tree

Hide file tree

Showing 3 changed files with 43 additions and 27 deletions.
diff --git a/src/execution/operator/csv_scanner/scanner/string_value_scanner.cpp b/src/execution/operator/csv_scanner/scanner/string_value_scanner.cpp
@@ -215,16 +215,32 @@ void StringValueResult::AddValueToVector(const char *value_ptr, const idx_t size
 		                               false, state_machine.options.decimal_separator[0]);
 		break;
 	case LogicalTypeId::DATE: {
-		idx_t pos;
-		bool special;
-		success = Date::TryConvertDate(value_ptr, size, pos,
-		                               static_cast<date_t *>(vector_ptr[chunk_col_id])[number_of_rows], special, false);
+		if (!state_machine.dialect_options.date_format.find(LogicalTypeId::DATE)->second.GetValue().Empty()) {
+			string error_message;
+			success = state_machine.dialect_options.date_format.find(LogicalTypeId::DATE)
+			              ->second.GetValue()
+			              .TryParseDate(value_ptr, size,
+			                            static_cast<date_t *>(vector_ptr[chunk_col_id])[number_of_rows], error_message);
+		} else {
+			idx_t pos;
+			bool special;
+			success = Date::TryConvertDate(
+			    value_ptr, size, pos, static_cast<date_t *>(vector_ptr[chunk_col_id])[number_of_rows], special, false);
+		}
 		break;
 	}
 	case LogicalTypeId::TIMESTAMP: {
-		success = Timestamp::TryConvertTimestamp(
-		              value_ptr, size, static_cast<timestamp_t *>(vector_ptr[chunk_col_id])[number_of_rows]) ==
-		          TimestampCastResult::SUCCESS;
+		if (!state_machine.dialect_options.date_format.find(LogicalTypeId::TIMESTAMP)->second.GetValue().Empty()) {
+			timestamp_t result;
+			string error_message;
+			//			success =  state_machine.dialect_options.date_format.find(LogicalTypeId::TIMESTAMP)
+			//			    ->second.GetValue()
+			//			    .TryParseTimestamp(value, result, error_message);
+		} else {
+			success = Timestamp::TryConvertTimestamp(
+			              value_ptr, size, static_cast<timestamp_t *>(vector_ptr[chunk_col_id])[number_of_rows]) ==
+			          TimestampCastResult::SUCCESS;
+		}
 		break;
 	}
 	default: {
@@ -1185,7 +1201,6 @@ bool StringValueScanner::CanDirectlyCast(const LogicalType &type,
                                          const map<LogicalTypeId, CSVOption<StrpTimeFormat>> &format_options) {
 
 	switch (type.id()) {
-		// All Integers (Except HugeInt)
 	case LogicalTypeId::TINYINT:
 	case LogicalTypeId::SMALLINT:
 	case LogicalTypeId::INTEGER:
@@ -1196,20 +1211,8 @@ bool StringValueScanner::CanDirectlyCast(const LogicalType &type,
 	case LogicalTypeId::UBIGINT:
 	case LogicalTypeId::DOUBLE:
 	case LogicalTypeId::FLOAT:
-		return true;
 	case LogicalTypeId::DATE:
-		// We can only internally cast YYYY-MM-DD
-		if (format_options.at(LogicalTypeId::DATE).GetValue().format_specifier == "%Y-%m-%d") {
-			return true;
-		} else {
-			return false;
-		}
 	case LogicalTypeId::TIMESTAMP:
-		if (format_options.at(LogicalTypeId::TIMESTAMP).GetValue().format_specifier == "%Y-%m-%d %H:%M:%S") {
-			return true;
-		} else {
-			return false;
-		}
 	case LogicalType::VARCHAR:
 		return true;
 	default:

diff --git a/src/function/scalar/strftime_format.cpp b/src/function/scalar/strftime_format.cpp
@@ -740,8 +740,7 @@ int32_t StrpTimeFormat::TryParseCollection(const char *data, idx_t &pos, idx_t s
 	return -1;
 }
 
-//! Parses a timestamp using the given specifier
-bool StrpTimeFormat::Parse(string_t str, ParseResult &result) const {
+bool StrpTimeFormat::Parse(const char *data, size_t size, ParseResult &result) const {
 	auto &result_data = result.data;
 	auto &error_message = result.error_message;
 	auto &error_position = result.error_position;
@@ -755,15 +754,11 @@ bool StrpTimeFormat::Parse(string_t str, ParseResult &result) const {
 	result_data[5] = 0;
 	result_data[6] = 0;
 	result_data[7] = 0;
-
-	auto data = str.GetData();
-	idx_t size = str.GetSize();
 	// skip leading spaces
 	while (StringUtil::CharacterIsSpace(*data)) {
 		data++;
 		size--;
 	}
-
 	//	Check for specials
 	//	Precheck for alphas for performance.
 	idx_t pos = 0;
@@ -1067,7 +1062,6 @@ bool StrpTimeFormat::Parse(string_t str, ParseResult &result) const {
 				case StrTimeSpecifier::YEAR_DECIMAL:
 					// Just validate, don't use
 					break;
-					break;
 				case StrTimeSpecifier::WEEKDAY_DECIMAL:
 					// First offset specifier
 					offset_specifier = specifiers[i];
@@ -1324,6 +1318,13 @@ bool StrpTimeFormat::Parse(string_t str, ParseResult &result) const {
 	return true;
 }
 
+//! Parses a timestamp using the given specifier
+bool StrpTimeFormat::Parse(string_t str, ParseResult &result) const {
+	auto data = str.GetData();
+	idx_t size = str.GetSize();
+	return Parse(data, size, result);
+}
+
 StrpTimeFormat::ParseResult StrpTimeFormat::Parse(const string &format_string, const string &text) {
 	StrpTimeFormat format;
 	format.format_specifier = format_string;
@@ -1413,6 +1414,14 @@ bool StrpTimeFormat::TryParseDate(string_t input, date_t &result, string &error_
 	return parse_result.TryToDate(result);
 }
 
+bool StrpTimeFormat::TryParseDate(const char *data, size_t size, date_t &result, string &error_message) const {
+	ParseResult parse_result;
+	if (!Parse(data, size, parse_result)) {
+		return false;
+	}
+	return parse_result.TryToDate(result);
+}
+
 bool StrpTimeFormat::TryParseTime(string_t input, dtime_t &result, string &error_message) const {
 	ParseResult parse_result;
 	if (!Parse(input, parse_result)) {

diff --git a/src/include/duckdb/function/scalar/strftime_format.hpp b/src/include/duckdb/function/scalar/strftime_format.hpp
@@ -161,6 +161,10 @@ struct StrpTimeFormat : public StrTimeFormat { // NOLINT: work-around bug in cla
 
 	DUCKDB_API bool Parse(string_t str, ParseResult &result) const;
 
+	DUCKDB_API bool Parse(const char *data, size_t size, ParseResult &result) const;
+
+	DUCKDB_API bool TryParseDate(const char *data, size_t size, date_t &result, string &error_message) const;
+
 	DUCKDB_API bool TryParseDate(string_t str, date_t &result, string &error_message) const;
 	DUCKDB_API bool TryParseTime(string_t str, dtime_t &result, string &error_message) const;
 	DUCKDB_API bool TryParseTimestamp(string_t str, timestamp_t &result, string &error_message) const;