dbms: development.

dbms/include/DB/Columns/ColumnFixedArray.h

@@ -23,7 +23,7 @@ public:
 	ColumnFixedArray(SharedPtr<IColumn> nested_column, size_t n_)
 		: data(nested_column), n(n_)
 	{
-		data.clear();
+		clear();
 	}
 
 	SharedPtr<IColumn> cloneEmpty() const
@@ -68,7 +68,7 @@ public:
 
 	void clear()
 	{
-		data.clear();
+		data->clear();
 	}
 
 	/** Более эффективные методы манипуляции */
@@ -82,7 +82,7 @@ public:
 		return *data;
 	}
 
-private:
+protected:
 	SharedPtr<IColumn> data;
 	const size_t n;
 };

dbms/include/DB/Columns/ColumnFixedString.h

@@ -0,0 +1,54 @@
+#pragma once
+
+#include <string.h> // memcpy
+
+#include <DB/Columns/ColumnFixedArray.h>
+#include <DB/Columns/ColumnsNumber.h>
+
+
+namespace DB
+{
+
+/** Cтолбeц значений типа "строка фиксированной длины".
+  * Отличается от массива UInt8 фиксированной длины только получением элемента (в виде String, а не Array)
+  * Если вставить строку меньшей длины, то она будет дополнена нулевыми байтами.
+  */
+class ColumnFixedString : public ColumnFixedArray
+{
+private:
+	ColumnUInt8::Container_t & char_data;
+
+public:
+	/** Создать пустой столбец строк фиксированной длины n */
+	ColumnFixedString(size_t n)
+		: ColumnFixedArray(new ColumnUInt8(), n),
+		char_data(dynamic_cast<ColumnUInt8 &>(*data).getData())
+	{
+	}
+
+	SharedPtr<IColumn> cloneEmpty() const
+	{
+		return new ColumnFixedString(n);
+	}
+	
+	Field operator[](size_t index) const
+	{
+		return String(reinterpret_cast<const char *>(&char_data[n * index]), n);
+	}
+
+	void insert(const Field & x)
+	{
+		const String & s = boost::get<const String &>(x);
+		size_t old_size = char_data.size();
+		char_data.resize(old_size + n);
+		memcpy(&char_data[old_size], s.data(), s.size());
+	}
+
+	void insertDefault()
+	{
+		char_data.resize(char_data.size() + n);
+	}
+};
+
+
+}

dbms/include/DB/Core/ErrorCodes.h

@@ -48,6 +48,7 @@ namespace ErrorCodes
 		CANNOT_PARSE_DATE,
 		TOO_LARGE_SIZE_COMPRESSED,
 		CHECKSUM_DOESNT_MATCH,
+		CANNOT_PARSE_DATETIME,
 	};
 }
 

dbms/include/DB/DataTypes/DataTypeDate.h

@@ -1,10 +1,4 @@
-#ifndef DBMS_DATA_TYPES_NUMBER_FIXED_H
-#define DBMS_DATA_TYPES_NUMBER_FIXED_H
-
-#include <Poco/DateTimeParser.h>
-
-#include <DB/Core/Exception.h>
-#include <DB/Core/ErrorCodes.h>
+#pragma once
 
 #include <DB/IO/ReadHelpers.h>
 #include <DB/IO/WriteHelpers.h>
@@ -16,36 +10,24 @@
 namespace DB
 {
 
-class DataTypeDate : public IDataTypeNumberFixed<UInt16, ColumnUInt64>
+class DataTypeDate : public IDataTypeNumberFixed<UInt16, ColumnUInt16>
 {
-private:
-	DateLUTSingleton & date_lut;
-
 public:
-	DataTypeDate() : date_lut(DateLUTSingleton::instance()) {}
+	DataTypeDate() {}
 
 	std::string getName() const { return "Date"; }
 	SharedPtr<IDataType> clone() const { return new DataTypeDate; }
 
 	void serializeText(const Field & field, WriteBuffer & ostr) const
 	{
-		DateLUT::Values & values = date_lut.getValues(boost::get<UInt16>(field));
-		writeIntText(values.year, ostr);
-		writeChar('-', ostr);
-		writeIntText(values.month, ostr);
-		writeChar('-', ostr);
-		writeIntText(values.day_of_month, ostr);
+		writeDateText(boost::get<UInt64>(field), ostr);
 	}
 	
 	void deserializeText(Field & field, ReadBuffer & istr) const
 	{
-		std::string s;
-		readString(s, istr);
-
-		// TODO: тормоза
-		int time_zone_diff = 0;
-		field = date_lut.toDayNum(Poco::DateTimeParser::parse(
-			s, time_zone_diff).timestamp().epochTime());
+		Yandex::DayNum_t x;
+		readDateText(x, istr);
+		field = x;
 	}
 
 	void serializeTextEscaped(const Field & field, WriteBuffer & ostr) const
@@ -74,5 +56,3 @@ public:
 };
 
 }
-
-#endif

dbms/include/DB/DataTypes/DataTypeDateTime.h

@@ -0,0 +1,58 @@
+#pragma once
+
+#include <DB/IO/ReadHelpers.h>
+#include <DB/IO/WriteHelpers.h>
+
+#include <DB/Columns/ColumnsNumber.h>
+#include <DB/DataTypes/IDataTypeNumberFixed.h>
+
+
+namespace DB
+{
+
+class DataTypeDateTime : public IDataTypeNumberFixed<UInt32, ColumnUInt32>
+{
+public:
+	DataTypeDateTime() {}
+
+	std::string getName() const { return "DateTime"; }
+	SharedPtr<IDataType> clone() const { return new DataTypeDateTime; }
+
+	void serializeText(const Field & field, WriteBuffer & ostr) const
+	{
+		writeDateTimeText(boost::get<UInt64>(field), ostr);
+	}
+	
+	void deserializeText(Field & field, ReadBuffer & istr) const
+	{
+		time_t x;
+		readDateTimeText(x, istr);
+		field = x;
+	}
+
+	void serializeTextEscaped(const Field & field, WriteBuffer & ostr) const
+	{
+		serializeText(field, ostr);
+	}
+	
+	void deserializeTextEscaped(Field & field, ReadBuffer & istr) const
+	{
+		deserializeText(field, istr);
+	}
+	
+	void serializeTextQuoted(const Field & field, WriteBuffer & ostr, bool compatible = false) const
+	{
+		writeChar('\'', ostr);
+		serializeText(field, ostr);
+		writeChar('\'', ostr);
+	}
+	
+	void deserializeTextQuoted(Field & field, ReadBuffer & istr, bool compatible = false) const
+	{
+		assertString("'", istr);
+		deserializeText(field, istr);
+		assertString("'", istr);
+	}
+};
+
+}

dbms/include/DB/DataTypes/DataTypeFixedString.h

@@ -0,0 +1,52 @@
+#pragma once
+
+#include <ostream>
+
+#include <Poco/SharedPtr.h>
+#include <Poco/NumberFormatter.h>
+
+#include <DB/DataTypes/IDataType.h>
+
+
+namespace DB
+{
+
+using Poco::SharedPtr;
+
+
+class DataTypeFixedString : public IDataType
+{
+private:
+	size_t n;
+	
+public:
+	DataTypeFixedString(size_t n_) : n(n_) {}
+	
+	std::string getName() const
+	{
+		return "FixedString(" + Poco::NumberFormatter::format(n) + ")";
+	}
+
+	SharedPtr<IDataType> clone() const
+	{
+		return new DataTypeFixedString(n);
+	}
+
+	void serializeBinary(const Field & field, WriteBuffer & ostr) const;
+	void deserializeBinary(Field & field, ReadBuffer & istr) const;
+	void serializeBinary(const IColumn & column, WriteBuffer & ostr) const;
+	void deserializeBinary(IColumn & column, ReadBuffer & istr, size_t limit) const;
+
+	void serializeText(const Field & field, WriteBuffer & ostr) const;
+	void deserializeText(Field & field, ReadBuffer & istr) const;
+
+	void serializeTextEscaped(const Field & field, WriteBuffer & ostr) const;
+	void deserializeTextEscaped(Field & field, ReadBuffer & istr) const;
+
+	void serializeTextQuoted(const Field & field, WriteBuffer & ostr, bool compatible = false) const;
+	void deserializeTextQuoted(Field & field, ReadBuffer & istr, bool compatible = false) const;
+
+	SharedPtr<IColumn> createColumn() const;
+};
+
+}

dbms/include/DB/IO/ReadHelpers.h

@@ -246,6 +246,30 @@ inline void readDateText(Yandex::DayNum_t & date, ReadBuffer & buf)
 	date = Yandex::DateLUTSingleton::instance().makeDayNum(year, month, day);
 }
 
+
+/// в формате YYYY-MM-DD HH:MM:SS, согласно текущему часовому поясу
+inline void readDateTimeText(time_t & datetime, ReadBuffer & buf)
+{
+	char s[19];
+	size_t size = buf.read(s, 19);
+	if (19 != size)
+	{
+		s[size] = 0;
+		throw Exception(std::string("Cannot parse datetime ") + s, ErrorCodes::CANNOT_PARSE_DATETIME);
+	}
+
+	UInt16 year = (s[0] - '0') * 1000 + (s[1] - '0') * 100 + (s[2] - '0') * 10 + (s[3] - '0');
+	UInt8 month = (s[5] - '0') * 10 + (s[6] - '0');
+	UInt8 day = (s[8] - '0') * 10 + (s[9] - '0');
+
+	UInt8 hour = (s[11] - '0') * 10 + (s[12] - '0');
+	UInt8 minute = (s[14] - '0') * 10 + (s[15] - '0');
+	UInt8 second = (s[17] - '0') * 10 + (s[18] - '0');
+
+	datetime = Yandex::DateLUTSingleton::instance().makeDateTime(year, month, day, hour, minute, second);
+}
+
+
 }
 
 #endif

dbms/include/DB/IO/WriteHelpers.h

@@ -127,6 +127,43 @@ inline void writeDateText(Yandex::DayNum_t date, WriteBuffer & buf)
 }
 
 
+/// в формате YYYY-MM-DD HH:MM:SS, согласно текущему часовому поясу
+inline void writeDateTimeText(time_t datetime, WriteBuffer & buf)
+{
+	char s[19];
+
+	Yandex::DateLUTSingleton & date_lut = Yandex::DateLUTSingleton::instance();
+	const Yandex::DateLUT::Values & values = date_lut.getValues(datetime);
+
+	s[0] = '0' + values.year / 1000;
+	s[1] = '0' + (values.year / 100) % 10;
+	s[2] = '0' + (values.year / 10) % 10;
+	s[3] = '0' + values.year % 10;
+	s[4] = '-';
+	s[5] = '0' + values.month / 10;
+	s[6] = '0' + values.month % 10;
+	s[7] = '-';
+	s[8] = '0' + values.day_of_month / 10;
+	s[9] = '0' + values.day_of_month % 10;
+
+	UInt8 hour = date_lut.toHourInaccurate(datetime);
+	UInt8 minute = date_lut.toMinute(datetime);
+	UInt8 second = date_lut.toSecond(datetime);
+
+	s[10] = ' ';
+	s[11] = '0' + hour / 10;
+	s[12] = '0' + hour % 10;
+	s[13] = ':';
+	s[14] = '0' + minute / 10;
+	s[15] = '0' + minute % 10;
+	s[16] = ':';
+	s[17] = '0' + second / 10;
+	s[18] = '0' + second % 10;
+
+	buf.write(s, 19);
+}
+
+
 }
 
 #endif

dbms/src/DataTypes/DataTypeFixedString.cpp

@@ -0,0 +1,113 @@
+#include <Poco/SharedPtr.h>
+
+#include <DB/Columns/ColumnFixedArray.h>
+#include <DB/Columns/ColumnFixedString.h>
+#include <DB/Columns/ColumnsNumber.h>
+
+#include <DB/DataTypes/DataTypeFixedString.h>
+
+#include <DB/IO/ReadHelpers.h>
+#include <DB/IO/WriteHelpers.h>
+#include <DB/IO/VarInt.h>
+
+
+namespace DB
+{
+
+using Poco::SharedPtr;
+
+
+void DataTypeFixedString::serializeBinary(const Field & field, WriteBuffer & ostr) const
+{
+	const String & s = boost::get<String>(field);
+	ostr.write(s.data(), std::min(s.size(), n));
+	if (s.size() < n)
+		for (size_t i = s.size(); i < n; ++i)
+			ostr.write(0);
+}
+
+
+void DataTypeFixedString::deserializeBinary(Field & field, ReadBuffer & istr) const
+{
+	field = String();
+	String & s = boost::get<String>(field);
+	s.resize(n);
+	/// непереносимо, но (действительно) быстрее
+	istr.readStrict(const_cast<char*>(s.data()), n);
+}
+
+
+void DataTypeFixedString::serializeBinary(const IColumn & column, WriteBuffer & ostr) const
+{
+	const ColumnFixedArray & column_array = dynamic_cast<const ColumnFixedArray &>(column);
+	const ColumnUInt8::Container_t & data = dynamic_cast<const ColumnUInt8 &>(column_array.getData()).getData();
+
+	ostr.write(reinterpret_cast<const char *>(&data[0]), data.size());
+}
+
+
+void DataTypeFixedString::deserializeBinary(IColumn & column, ReadBuffer & istr, size_t limit) const
+{
+	ColumnFixedArray & column_array = dynamic_cast<ColumnFixedArray &>(column);
+	ColumnUInt8::Container_t & data = dynamic_cast<ColumnUInt8 &>(column_array.getData()).getData();
+
+	size_t max_bytes = limit * n;
+	data.resize(max_bytes);
+	size_t read_bytes = istr.read(reinterpret_cast<char *>(&data[0]), max_bytes);
+
+	if (read_bytes % n != 0)
+		throw Exception("Cannot read all data of type FixedString",
+			ErrorCodes::CANNOT_READ_ALL_DATA);
+		
+	data.resize(read_bytes);
+}
+
+
+void DataTypeFixedString::serializeText(const Field & field, WriteBuffer & ostr) const
+{
+	writeString(boost::get<const String &>(field), ostr);
+}
+
+
+void DataTypeFixedString::deserializeText(Field & field, ReadBuffer & istr) const
+{
+	String s;
+	readString(s, istr);
+	field = s;
+}
+
+
+void DataTypeFixedString::serializeTextEscaped(const Field & field, WriteBuffer & ostr) const
+{
+	writeEscapedString(boost::get<const String &>(field), ostr);
+}
+
+
+void DataTypeFixedString::deserializeTextEscaped(Field & field, ReadBuffer & istr) const
+{
+	String s;
+	readEscapedString(s, istr);
+	field = s;
+}
+
+
+void DataTypeFixedString::serializeTextQuoted(const Field & field, WriteBuffer & ostr, bool compatible) const
+{
+	writeQuotedString(boost::get<const String &>(field), ostr);
+}
+
+
+void DataTypeFixedString::deserializeTextQuoted(Field & field, ReadBuffer & istr, bool compatible) const
+{
+	String s;
+	readQuotedString(s, istr);
+	field = s;
+}
+
+
+SharedPtr<IColumn> DataTypeFixedString::createColumn() const
+{
+	return new ColumnFixedString(n);
+}
+
+}

dbms/src/DataTypes/DataTypeString.cpp

@@ -29,7 +29,7 @@ void DataTypeString::deserializeBinary(Field & field, ReadBuffer & istr) const
 {
 	UInt64 size;
 	readVarUInt(size, istr);
-	field = String("");
+	field = String();
 	String & s = boost::get<String>(field);
 	s.resize(size);
 	/// непереносимо, но (действительно) быстрее

dbms/src/Storages/tests/hit_log.cpp

@@ -11,6 +11,8 @@
 
 #include <DB/DataTypes/DataTypesNumberFixed.h>
 #include <DB/DataTypes/DataTypeString.h>
+#include <DB/DataTypes/DataTypeFixedString.h>
+#include <DB/DataTypes/DataTypeDateTime.h>
 
 #include <DB/DataStreams/TabSeparatedRowInputStream.h>
 #include <DB/DataStreams/TabSeparatedRowOutputStream.h>
@@ -33,18 +35,14 @@ int main(int argc, char ** argv)
 
 		boost::assign::push_back(names_and_types_list)
 			("WatchID",				new DB::DataTypeUInt64)
-			("ChunkID",				new DB::DataTypeUInt64)
-			("Random",				new DB::DataTypeUInt32)
 			("JavaEnable",			new DB::DataTypeUInt8)
-			("FrameEnable",			new DB::DataTypeUInt8)
 			("Title",				new DB::DataTypeString)
 			("GoodEvent",			new DB::DataTypeUInt32)
-//			("EventTime",			new DB::DataTypeDateTime)
+			("EventTime",			new DB::DataTypeDateTime)
 			("CounterID",			new DB::DataTypeUInt32)
 			("ClientIP",			new DB::DataTypeUInt32)
 			("RegionID",			new DB::DataTypeUInt32)
 			("UniqID",				new DB::DataTypeUInt64)
-			("SessID",				new DB::DataTypeUInt32)
 			("CounterClass",		new DB::DataTypeUInt8)
 			("OS",					new DB::DataTypeUInt8)
 			("UserAgent",			new DB::DataTypeUInt8)
@@ -60,7 +58,7 @@ int main(int argc, char ** argv)
 			("NetMajor",			new DB::DataTypeUInt8)
 			("NetMinor",			new DB::DataTypeUInt8)
 			("UserAgentMajor",		new DB::DataTypeUInt16)
-//			("UserAgentMinor",		new DB::DataTypeFixedString(2))
+			("UserAgentMinor",		new DB::DataTypeFixedString(2))
 			("CookieEnable",		new DB::DataTypeUInt8)
 			("JavascriptEnable",	new DB::DataTypeUInt8)
 			("IsMobile",			new DB::DataTypeUInt8)
@@ -76,12 +74,24 @@ int main(int argc, char ** argv)
 			("WindowClientWidth",	new DB::DataTypeUInt16)
 			("WindowClientHeight",	new DB::DataTypeUInt16)
 			("ClientTimeZone",		new DB::DataTypeInt16)
-//			("ClientEventTime",		new DB::DataTypeDateTime)
+			("ClientEventTime",		new DB::DataTypeDateTime)
 			("SilverlightVersion1",	new DB::DataTypeUInt8)
 			("SilverlightVersion2",	new DB::DataTypeUInt8)
 			("SilverlightVersion3",	new DB::DataTypeUInt32)
 			("SilverlightVersion4",	new DB::DataTypeUInt16)
 			("PageCharset",			new DB::DataTypeString)
+			("CodeVersion",			new DB::DataTypeUInt32) 
+			("IsLink",				new DB::DataTypeUInt8)
+			("IsDownload",			new DB::DataTypeUInt8)
+			("IsNotBounce",			new DB::DataTypeUInt8)
+			("FUniqID",				new DB::DataTypeUInt64)
+			("OriginalURL",			new DB::DataTypeString)
+			("HID",					new DB::DataTypeUInt32)
+			("IsOldCounter",		new DB::DataTypeUInt8)
+			("IsEvent",				new DB::DataTypeUInt8)
+			("IsParameter",			new DB::DataTypeUInt8)
+			("DontCountHits",		new DB::DataTypeUInt8)
+			("WithHash",			new DB::DataTypeUInt8)
 		;
 
 		SharedPtr<DB::NamesAndTypes> names_and_types_map = new DB::NamesAndTypes;