ClickHouse/dbms/include/DB/IO/ReadHelpers.h

930 lines
22 KiB
C++
Raw Normal View History

2011-10-30 05:19:41 +00:00
#pragma once
2010-05-28 19:13:55 +00:00
#include <cmath>
2010-06-01 13:35:09 +00:00
#include <cstring>
2010-06-01 13:41:51 +00:00
#include <limits>
2010-05-28 19:13:55 +00:00
#include <algorithm>
2014-01-08 16:33:28 +00:00
#include <type_traits>
2015-09-29 19:19:54 +00:00
#include <common/Common.h>
#include <common/DateLUT.h>
2011-05-05 19:10:17 +00:00
#include <common/LocalDate.h>
#include <common/LocalDateTime.h>
#include <common/exp10.h>
2010-05-28 19:13:55 +00:00
#include <DB/Core/Types.h>
#include <DB/Core/StringRef.h>
2015-10-05 01:35:28 +00:00
#include <DB/Common/Exception.h>
#include <DB/Common/StringUtils.h>
2016-09-22 23:26:08 +00:00
#include <DB/Common/Arena.h>
2010-05-28 19:13:55 +00:00
2010-06-04 18:25:25 +00:00
#include <DB/IO/ReadBuffer.h>
2017-01-03 01:42:17 +00:00
#include <DB/IO/ReadBufferFromMemory.h>
2011-08-19 18:31:14 +00:00
#include <DB/IO/VarInt.h>
2014-04-02 13:45:39 +00:00
#include <city.h>
2011-08-19 18:31:14 +00:00
#define DEFAULT_MAX_STRING_SIZE 0x00FFFFFFULL
2010-05-28 19:13:55 +00:00
namespace DB
{
namespace ErrorCodes
{
extern const int CANNOT_PARSE_DATE;
extern const int CANNOT_PARSE_DATETIME;
extern const int CANNOT_READ_ARRAY_FROM_TEXT;
}
2016-10-21 00:19:49 +00:00
/// Helper functions for formatted input.
2010-06-01 13:35:09 +00:00
inline char parseEscapeSequence(char c)
2010-06-01 13:35:09 +00:00
{
switch(c)
2010-05-28 19:13:55 +00:00
{
case 'a':
return '\a';
2010-06-01 13:35:09 +00:00
case 'b':
return '\b';
case 'f':
return '\f';
case 'n':
return '\n';
case 'r':
return '\r';
case 't':
return '\t';
case 'v':
return '\v';
2010-06-01 13:35:09 +00:00
case '0':
return '\0';
default:
return c;
2010-05-28 19:13:55 +00:00
}
2010-06-01 13:35:09 +00:00
}
2010-05-28 19:13:55 +00:00
inline char unhex(char c)
{
switch (c)
{
case '0' ... '9':
return c - '0';
case 'a' ... 'f':
return c - 'a' + 10;
case 'A' ... 'F':
return c - 'A' + 10;
default:
return 0;
}
}
2016-10-21 00:19:49 +00:00
/// These functions are located in VarInt.h
/// inline void throwReadAfterEOF()
inline void readChar(char & x, ReadBuffer & buf)
{
if (!buf.eof())
{
x = *buf.position();
++buf.position();
}
else
throwReadAfterEOF();
}
2010-06-04 18:25:25 +00:00
2016-10-21 00:19:49 +00:00
/// Read POD-type in native format
2011-07-04 18:22:37 +00:00
template <typename T>
inline void readPODBinary(T & x, ReadBuffer & buf)
2011-07-04 18:22:37 +00:00
{
buf.readStrict(reinterpret_cast<char *>(&x), sizeof(x));
}
2011-08-09 17:24:17 +00:00
template <typename T>
inline void readIntBinary(T & x, ReadBuffer & buf)
{
readPODBinary(x, buf);
2011-08-09 17:24:17 +00:00
}
template <typename T>
inline void readFloatBinary(T & x, ReadBuffer & buf)
{
readPODBinary(x, buf);
2011-08-09 17:24:17 +00:00
}
2011-07-04 18:22:37 +00:00
2011-08-19 18:31:14 +00:00
2013-10-10 23:06:51 +00:00
inline void readStringBinary(std::string & s, ReadBuffer & buf, size_t MAX_STRING_SIZE = DEFAULT_MAX_STRING_SIZE)
2011-08-19 18:31:14 +00:00
{
size_t size = 0;
2013-10-10 23:06:51 +00:00
readVarUInt(size, buf);
2011-08-19 18:31:14 +00:00
if (size > MAX_STRING_SIZE)
throw Poco::Exception("Too large string size.");
s.resize(size);
buf.readStrict(&s[0], size);
2011-08-19 18:31:14 +00:00
}
2016-09-22 23:26:08 +00:00
inline StringRef readStringBinaryInto(Arena & arena, ReadBuffer & buf)
{
size_t size = 0;
readVarUInt(size, buf);
char * data = arena.alloc(size);
buf.readStrict(data, size);
return StringRef(data, size);
}
template <typename T>
void readVectorBinary(std::vector<T> & v, ReadBuffer & buf, size_t MAX_VECTOR_SIZE = DEFAULT_MAX_STRING_SIZE)
{
size_t size = 0;
2013-10-10 23:06:51 +00:00
readVarUInt(size, buf);
if (size > MAX_VECTOR_SIZE)
throw Poco::Exception("Too large vector size.");
v.resize(size);
for (size_t i = 0; i < size; ++i)
readBinary(v[i], buf);
}
2010-06-01 14:12:28 +00:00
void assertString(const char * s, ReadBuffer & buf);
void assertEOF(ReadBuffer & buf);
void assertChar(char symbol, ReadBuffer & buf);
2010-06-01 13:35:09 +00:00
2014-03-24 12:10:47 +00:00
inline void assertString(const String & s, ReadBuffer & buf)
2014-03-22 14:44:44 +00:00
{
assertString(s.c_str(), buf);
}
bool checkString(const char * s, ReadBuffer & buf);
inline bool checkString(const String & s, ReadBuffer & buf)
{
return checkString(s.c_str(), buf);
}
inline bool checkChar(char c, ReadBuffer & buf)
{
if (buf.eof() || *buf.position() != c)
return false;
++buf.position();
return true;
}
bool checkStringCaseInsensitive(const char * s, ReadBuffer & buf);
inline bool checkStringCaseInsensitive(const String & s, ReadBuffer & buf)
{
return checkStringCaseInsensitive(s.c_str(), buf);
}
void assertStringCaseInsensitive(const char * s, ReadBuffer & buf);
inline void assertStringCaseInsensitive(const String & s, ReadBuffer & buf)
{
return assertStringCaseInsensitive(s.c_str(), buf);
}
/** Check that next character in buf matches first character of s.
* If true, then check all characters in s and throw exception if it doesn't match.
* If false, then return false, and leave position in buffer unchanged.
*/
bool checkStringByFirstCharacterAndAssertTheRest(const char * s, ReadBuffer & buf);
bool checkStringByFirstCharacterAndAssertTheRestCaseInsensitive(const char * s, ReadBuffer & buf);
inline bool checkStringByFirstCharacterAndAssertTheRest(const String & s, ReadBuffer & buf)
{
return checkStringByFirstCharacterAndAssertTheRest(s.c_str(), buf);
}
inline bool checkStringByFirstCharacterAndAssertTheRestCaseInsensitive(const String & s, ReadBuffer & buf)
{
return checkStringByFirstCharacterAndAssertTheRestCaseInsensitive(s.c_str(), buf);
}
inline void readBoolText(bool & x, ReadBuffer & buf)
{
char tmp = '0';
readChar(tmp, buf);
x = tmp != '0';
}
template <typename T, typename ReturnType = void>
ReturnType readIntTextImpl(T & x, ReadBuffer & buf)
2010-06-01 13:35:09 +00:00
{
static constexpr bool throw_exception = std::is_same<ReturnType, void>::value;
2010-06-01 13:35:09 +00:00
bool negative = false;
x = 0;
2010-06-04 18:25:25 +00:00
if (buf.eof())
{
if (throw_exception)
throwReadAfterEOF();
else
return ReturnType(false);
}
2010-06-01 13:35:09 +00:00
while (!buf.eof())
{
switch (*buf.position())
2010-05-28 19:13:55 +00:00
{
2010-06-01 13:35:09 +00:00
case '+':
break;
case '-':
2017-03-13 18:01:46 +00:00
if (std::is_signed<T>::value)
negative = true;
else
return ReturnType(false);
2010-06-01 13:35:09 +00:00
break;
case '0':
case '1':
case '2':
case '3':
case '4':
case '5':
case '6':
case '7':
case '8':
case '9':
x *= 10;
x += *buf.position() - '0';
break;
default:
if (negative)
2010-05-28 19:13:55 +00:00
x = -x;
return ReturnType(true);
2010-05-28 19:13:55 +00:00
}
2010-06-01 13:35:09 +00:00
++buf.position();
2010-05-28 19:13:55 +00:00
}
2010-06-01 13:35:09 +00:00
if (negative)
x = -x;
return ReturnType(true);
2010-06-01 13:35:09 +00:00
}
template <typename T>
void readIntText(T & x, ReadBuffer & buf)
{
readIntTextImpl<T, void>(x, buf);
2010-06-01 13:35:09 +00:00
}
template <typename T>
bool tryReadIntText(T & x, ReadBuffer & buf)
{
return readIntTextImpl<T, bool>(x, buf);
}
2016-10-21 00:19:49 +00:00
/** More efficient variant (about 1.5 times on real dataset).
* Differs in following:
* - for numbers starting with zero, parsed only zero;
* - symbol '+' before number is not supported;
* - symbols :;<=>? are parsed as some numbers.
*/
template <typename T, bool throw_on_error = true>
void readIntTextUnsafe(T & x, ReadBuffer & buf)
{
bool negative = false;
x = 0;
auto on_error = []
{
if (throw_on_error)
throwReadAfterEOF();
};
if (unlikely(buf.eof()))
return on_error();
2014-01-08 16:33:28 +00:00
if (std::is_signed<T>::value && *buf.position() == '-')
{
++buf.position();
negative = true;
if (unlikely(buf.eof()))
return on_error();
}
2016-10-21 00:19:49 +00:00
if (*buf.position() == '0') /// There are many zeros in real datasets.
{
++buf.position();
return;
}
while (!buf.eof())
{
2016-10-21 00:19:49 +00:00
if ((*buf.position() & 0xF0) == 0x30) /// It makes sense to have this condition inside loop.
{
x *= 10;
x += *buf.position() & 0x0F;
++buf.position();
}
else
break;
}
2014-01-08 16:33:28 +00:00
if (std::is_signed<T>::value && negative)
x = -x;
}
template<typename T>
void tryReadIntTextUnsafe(T & x, ReadBuffer & buf)
{
return readIntTextUnsafe<T, false>(x, buf);
}
template <bool throw_exception, class ExcepFun, class NoExcepFun, class... Args>
bool exceptionPolicySelector(ExcepFun && excep_f, NoExcepFun && no_excep_f, Args &&... args)
{
if (throw_exception)
{
excep_f(std::forward<Args>(args)...);
return true;
}
else
return no_excep_f(std::forward<Args>(args)...);
};
/// Returns true, iff parsed.
bool parseInfinity(ReadBuffer & buf);
bool parseNaN(ReadBuffer & buf);
void assertInfinity(ReadBuffer & buf);
void assertNaN(ReadBuffer & buf);
2016-08-16 20:21:28 +00:00
/// Rough: not exactly nearest machine representable number is returned.
2016-08-16 21:39:49 +00:00
/// Some garbage may be successfully parsed, examples: '.' parsed as 0; 123Inf parsed as inf.
template <typename T, typename ReturnType, char point_symbol = '.'>
ReturnType readFloatTextImpl(T & x, ReadBuffer & buf)
2010-06-01 13:35:09 +00:00
{
static constexpr bool throw_exception = std::is_same<ReturnType, void>::value;
2010-06-01 13:35:09 +00:00
bool negative = false;
x = 0;
bool after_point = false;
double power_of_ten = 1;
2010-05-28 19:13:55 +00:00
2010-06-04 18:25:25 +00:00
if (buf.eof())
2015-10-06 12:04:50 +00:00
{
if (throw_exception)
throwReadAfterEOF();
else
return ReturnType(false);
2015-10-06 12:04:50 +00:00
}
2010-06-04 18:25:25 +00:00
2010-06-01 13:35:09 +00:00
while (!buf.eof())
2010-05-28 19:13:55 +00:00
{
2010-06-01 13:35:09 +00:00
switch (*buf.position())
2010-05-28 19:13:55 +00:00
{
2010-06-01 13:35:09 +00:00
case '+':
break;
case '-':
negative = true;
break;
case point_symbol:
after_point = true;
break;
2010-06-01 13:35:09 +00:00
case '0':
case '1':
case '2':
case '3':
case '4':
case '5':
case '6':
case '7':
case '8':
case '9':
if (after_point)
{
power_of_ten /= 10;
x += (*buf.position() - '0') * power_of_ten;
}
else
{
x *= 10;
x += *buf.position() - '0';
}
break;
case 'e':
case 'E':
2010-05-28 19:13:55 +00:00
{
2010-06-01 13:35:09 +00:00
++buf.position();
Int32 exponent = 0;
bool res = exceptionPolicySelector<throw_exception>(readIntText<Int32>, tryReadIntText<Int32>, exponent, buf);
if (res)
{
x *= exp10(exponent);
if (negative)
x = -x;
}
return ReturnType(res);
2010-05-28 19:13:55 +00:00
}
case 'i':
2010-06-01 13:35:09 +00:00
case 'I':
{
bool res = exceptionPolicySelector<throw_exception>(assertInfinity, parseInfinity, buf);
if (res)
{
x = std::numeric_limits<T>::infinity();
if (negative)
x = -x;
}
return ReturnType(res);
}
2010-06-01 13:35:09 +00:00
case 'n':
case 'N':
{
bool res = exceptionPolicySelector<throw_exception>(assertNaN, parseNaN, buf);
if (res)
{
x = std::numeric_limits<T>::quiet_NaN();
if (negative)
x = -x;
}
return ReturnType(res);
}
2010-06-01 13:35:09 +00:00
default:
{
2010-06-01 13:35:09 +00:00
if (negative)
x = -x;
return ReturnType(true);
}
2010-05-28 19:13:55 +00:00
}
2010-06-01 13:35:09 +00:00
++buf.position();
2010-05-28 19:13:55 +00:00
}
2010-06-01 13:35:09 +00:00
if (negative)
x = -x;
return ReturnType(true);
2010-06-01 13:35:09 +00:00
}
2010-05-28 19:13:55 +00:00
template <class T>
inline bool tryReadFloatText(T & x, ReadBuffer & buf)
{
return readFloatTextImpl<T, bool>(x, buf);
2010-06-01 13:35:09 +00:00
}
2010-05-28 19:13:55 +00:00
template <class T>
inline void readFloatText(T & x, ReadBuffer & buf)
{
readFloatTextImpl<T, void>(x, buf);
}
2011-08-09 17:24:17 +00:00
2016-10-21 00:19:49 +00:00
/// rough; all until '\n' or '\t'
2010-06-01 14:12:28 +00:00
void readString(String & s, ReadBuffer & buf);
2010-05-28 19:13:55 +00:00
2010-06-01 14:12:28 +00:00
void readEscapedString(String & s, ReadBuffer & buf);
2010-05-28 19:13:55 +00:00
2010-06-01 14:12:28 +00:00
void readQuotedString(String & s, ReadBuffer & buf);
2010-05-28 19:13:55 +00:00
2011-06-15 18:54:18 +00:00
void readDoubleQuotedString(String & s, ReadBuffer & buf);
void readJSONString(String & s, ReadBuffer & buf);
2011-11-01 17:57:37 +00:00
void readBackQuotedString(String & s, ReadBuffer & buf);
void readStringUntilEOF(String & s, ReadBuffer & buf);
2010-05-28 19:13:55 +00:00
2016-10-21 00:19:49 +00:00
/** Read string in CSV format.
* Parsing rules:
* - string could be placed in quotes; quotes could be single: ' or double: ";
* - or string could be unquoted - this is determined by first character;
* - if string is unquoted, then it is read until next delimiter,
* either until end of line (CR or LF),
* or until end of stream;
* but spaces and tabs at begin and end of unquoted string are consumed but ignored (note that this behaviour differs from RFC).
* - if string is in quotes, then it will be read until closing quote,
* but sequences of two consecutive quotes are parsed as single quote inside string;
*/
void readCSVString(String & s, ReadBuffer & buf, const char delimiter = ',');
2016-10-21 00:19:49 +00:00
/// Read and append result to array of characters.
template <typename Vector>
void readStringInto(Vector & s, ReadBuffer & buf);
template <typename Vector>
void readEscapedStringInto(Vector & s, ReadBuffer & buf);
template <typename Vector>
void readQuotedStringInto(Vector & s, ReadBuffer & buf);
template <typename Vector>
void readDoubleQuotedStringInto(Vector & s, ReadBuffer & buf);
template <typename Vector>
void readBackQuotedStringInto(Vector & s, ReadBuffer & buf);
template <typename Vector>
void readStringUntilEOFInto(Vector & s, ReadBuffer & buf);
template <typename Vector>
void readCSVStringInto(Vector & s, ReadBuffer & buf, const char delimiter = ',');
template <typename Vector>
void readJSONStringInto(Vector & s, ReadBuffer & buf);
2016-10-21 00:19:49 +00:00
/// This could be used as template parameter for functions above, if you want to just skip data.
struct NullSink
{
void append(const char *, size_t) {};
void push_back(char) {};
};
2016-10-21 00:19:49 +00:00
/// In YYYY-MM-DD format
inline void readDateText(DayNum_t & date, ReadBuffer & buf)
2011-05-05 19:10:17 +00:00
{
char s[10];
2011-05-27 18:46:29 +00:00
size_t size = buf.read(s, 10);
if (10 != size)
{
s[size] = 0;
throw Exception(std::string("Cannot parse date ") + s, ErrorCodes::CANNOT_PARSE_DATE);
}
2011-05-05 19:10:17 +00:00
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');
2014-07-08 23:52:53 +00:00
date = DateLUT::instance().makeDayNum(year, month, day);
2011-05-05 19:10:17 +00:00
}
inline void readDateText(LocalDate & date, ReadBuffer & buf)
{
char s[10];
size_t size = buf.read(s, 10);
if (10 != size)
{
s[size] = 0;
throw Exception(std::string("Cannot parse date ") + s, ErrorCodes::CANNOT_PARSE_DATE);
}
date.year((s[0] - '0') * 1000 + (s[1] - '0') * 100 + (s[2] - '0') * 10 + (s[3] - '0'));
date.month((s[5] - '0') * 10 + (s[6] - '0'));
date.day((s[8] - '0') * 10 + (s[9] - '0'));
}
2011-08-07 02:08:22 +00:00
template <typename T>
inline T parse(const char * data, size_t size);
void readDateTimeTextFallback(time_t & datetime, ReadBuffer & buf, const DateLUTImpl & date_lut);
2016-10-21 00:19:49 +00:00
/** In YYYY-MM-DD hh:mm:ss format, according to current time zone.
* As an exception, also supported parsing of unix timestamp in form of decimal number.
*/
inline void readDateTimeText(time_t & datetime, ReadBuffer & buf, const DateLUTImpl & date_lut = DateLUT::instance())
2011-08-07 02:08:22 +00:00
{
2016-10-21 00:19:49 +00:00
/** Read 10 characters, that could represent unix timestamp.
* Only unix timestamp of 5-10 characters is supported.
* Then look at 5th charater. If it is a number - treat whole as unix timestamp.
* If it is not a number - then parse datetime in YYYY-MM-DD hh:mm:ss format.
*/
2016-10-21 00:19:49 +00:00
/// Optimistic path, when whole value are in buffer.
const char * s = buf.position();
if (s + 19 < buf.buffer().end())
{
if (s[4] < '0' || s[4] > '9')
{
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');
2011-08-07 02:08:22 +00:00
if (unlikely(year == 0))
datetime = 0;
else
datetime = date_lut.makeDateTime(year, month, day, hour, minute, second);
2011-08-07 02:08:22 +00:00
buf.position() += 19;
}
else
2016-10-21 00:19:49 +00:00
/// Why not readIntTextUnsafe? Because for needs of AdFox, parsing of unix timestamp with leading zeros is supported: 000...NNNN.
readIntText(datetime, buf);
}
2012-06-07 20:02:41 +00:00
else
readDateTimeTextFallback(datetime, buf, date_lut);
2011-08-07 02:08:22 +00:00
}
inline void readDateTimeText(LocalDateTime & 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);
}
datetime.year((s[0] - '0') * 1000 + (s[1] - '0') * 100 + (s[2] - '0') * 10 + (s[3] - '0'));
datetime.month((s[5] - '0') * 10 + (s[6] - '0'));
datetime.day((s[8] - '0') * 10 + (s[9] - '0'));
datetime.hour((s[11] - '0') * 10 + (s[12] - '0'));
datetime.minute((s[14] - '0') * 10 + (s[15] - '0'));
datetime.second((s[17] - '0') * 10 + (s[18] - '0'));
}
2016-10-21 00:19:49 +00:00
/// Generic methods to read value in native binary format.
2017-03-13 19:23:56 +00:00
template <typename T>
inline typename std::enable_if<std::is_arithmetic<T>::value, void>::type
readBinary(T & x, ReadBuffer & buf) { readPODBinary(x, buf); }
inline void readBinary(String & x, ReadBuffer & buf) { readStringBinary(x, buf); }
2014-04-02 13:45:39 +00:00
inline void readBinary(uint128 & x, ReadBuffer & buf) { readPODBinary(x, buf); }
inline void readBinary(LocalDate & x, ReadBuffer & buf) { readPODBinary(x, buf); }
inline void readBinary(LocalDateTime & x, ReadBuffer & buf) { readPODBinary(x, buf); }
2016-10-21 00:19:49 +00:00
/// Generic methods to read value in text tab-separated format.
2017-03-13 19:23:56 +00:00
template <typename T>
inline typename std::enable_if<std::is_integral<T>::value, void>::type
readText(T & x, ReadBuffer & buf) { readIntText(x, buf); }
template <typename T>
inline typename std::enable_if<std::is_floating_point<T>::value, void>::type
readText(T & x, ReadBuffer & buf) { readFloatText(x, buf); }
2017-03-13 19:23:56 +00:00
inline void readText(bool & x, ReadBuffer & buf) { readBoolText(x, buf); }
inline void readText(String & x, ReadBuffer & buf) { readEscapedString(x, buf); }
inline void readText(LocalDate & x, ReadBuffer & buf) { readDateText(x, buf); }
inline void readText(LocalDateTime & x, ReadBuffer & buf) { readDateTimeText(x, buf); }
2011-08-07 02:08:22 +00:00
2016-10-21 00:19:49 +00:00
/// Generic methods to read value in text format,
/// possibly in single quotes (only for data types that use quotes in VALUES format of INSERT statement in SQL).
2017-03-13 19:23:56 +00:00
template <typename T>
inline typename std::enable_if<std::is_arithmetic<T>::value, void>::type
readQuoted(T & x, ReadBuffer & buf) { readText(x, buf); }
2017-03-13 19:23:56 +00:00
inline void readQuoted(String & x, ReadBuffer & buf) { readQuotedString(x, buf); }
inline void readQuoted(LocalDate & x, ReadBuffer & buf)
{
assertChar('\'', buf);
readDateText(x, buf);
assertChar('\'', buf);
}
inline void readQuoted(LocalDateTime & x, ReadBuffer & buf)
{
assertChar('\'', buf);
readDateTimeText(x, buf);
assertChar('\'', buf);
}
2016-10-21 00:19:49 +00:00
/// Same as above, but in double quotes.
2017-03-13 19:23:56 +00:00
template <typename T>
inline typename std::enable_if<std::is_arithmetic<T>::value, void>::type
readDoubleQuoted(T & x, ReadBuffer & buf) { readText(x, buf); }
2017-03-13 19:23:56 +00:00
inline void readDoubleQuoted(String & x, ReadBuffer & buf) { readDoubleQuotedString(x, buf); }
inline void readDoubleQuoted(LocalDate & x, ReadBuffer & buf)
{
assertChar('"', buf);
readDateText(x, buf);
assertChar('"', buf);
}
inline void readDoubleQuoted(LocalDateTime & x, ReadBuffer & buf)
{
assertChar('"', buf);
readDateTimeText(x, buf);
assertChar('"', buf);
}
2016-10-21 00:19:49 +00:00
/// CSV, for numbers, dates, datetimes: quotes are optional, no special escaping rules.
template <typename T>
inline void readCSVSimple(T & x, ReadBuffer & buf)
{
if (buf.eof())
throwReadAfterEOF();
char maybe_quote = *buf.position();
if (maybe_quote == '\'' || maybe_quote == '\"')
++buf.position();
readText(x, buf);
if (maybe_quote == '\'' || maybe_quote == '\"')
assertChar(maybe_quote, buf);
}
2017-03-13 19:23:56 +00:00
template <typename T>
inline typename std::enable_if<std::is_arithmetic<T>::value, void>::type
readCSV(T & x, ReadBuffer & buf) { readCSVSimple(x, buf); }
inline void readCSV(String & x, ReadBuffer & buf, const char delimiter = ',') { readCSVString(x, buf, delimiter); }
inline void readCSV(LocalDate & x, ReadBuffer & buf) { readCSVSimple(x, buf); }
inline void readCSV(LocalDateTime & x, ReadBuffer & buf) { readCSVSimple(x, buf); }
template <typename T>
void readBinary(std::vector<T> & x, ReadBuffer & buf)
{
size_t size = 0;
readVarUInt(size, buf);
if (size > DEFAULT_MAX_STRING_SIZE)
throw Poco::Exception("Too large vector size.");
x.resize(size);
for (size_t i = 0; i < size; ++i)
readBinary(x[i], buf);
}
template <typename T>
void readQuoted(std::vector<T> & x, ReadBuffer & buf)
{
bool first = true;
assertChar('[', buf);
while (!buf.eof() && *buf.position() != ']')
{
if (!first)
{
if (*buf.position() == ',')
++buf.position();
else
throw Exception("Cannot read array from text", ErrorCodes::CANNOT_READ_ARRAY_FROM_TEXT);
}
first = false;
x.push_back(T());
readQuoted(x.back(), buf);
}
assertChar(']', buf);
}
template <typename T>
void readDoubleQuoted(std::vector<T> & x, ReadBuffer & buf)
{
bool first = true;
assertChar('[', buf);
while (!buf.eof() && *buf.position() != ']')
{
if (!first)
{
if (*buf.position() == ',')
++buf.position();
else
throw Exception("Cannot read array from text", ErrorCodes::CANNOT_READ_ARRAY_FROM_TEXT);
}
first = false;
x.push_back(T());
readDoubleQuoted(x.back(), buf);
}
assertChar(']', buf);
}
template <typename T>
void readText(std::vector<T> & x, ReadBuffer & buf)
{
readQuoted(x, buf);
}
2016-10-21 00:19:49 +00:00
/// Skip whitespace characters.
2011-10-30 05:19:41 +00:00
inline void skipWhitespaceIfAny(ReadBuffer & buf)
{
while (!buf.eof() && isWhitespaceASCII(*buf.position()))
2011-10-30 05:19:41 +00:00
++buf.position();
2010-05-28 19:13:55 +00:00
}
/// Skips json value. If the value contains objects (i.e. {...} sequence), an exception will be thrown.
void skipJSONFieldPlain(ReadBuffer & buf, const StringRef & name_of_filed);
2012-05-08 05:42:05 +00:00
2016-10-21 00:19:49 +00:00
/** Read serialized exception.
* During serialization/deserialization some information is lost
* (type is cut to base class, 'message' replaced by 'displayText', and stack trace is appended to 'message')
* Some additional message could be appended to exception (example: you could add information about from where it was received).
2012-05-08 05:42:05 +00:00
*/
void readException(Exception & e, ReadBuffer & buf, const String & additional_message = "");
void readAndThrowException(ReadBuffer & buf, const String & additional_message = "");
2016-10-21 00:19:49 +00:00
/** Helper function for implementation.
*/
template <typename T>
static inline const char * tryReadIntText(T & x, const char * pos, const char * end)
{
bool negative = false;
x = 0;
if (pos >= end)
return pos;
while (pos < end)
{
switch (*pos)
{
case '+':
break;
case '-':
2014-01-08 16:33:28 +00:00
if (std::is_signed<T>::value)
negative = true;
else
return pos;
break;
case '0':
case '1':
case '2':
case '3':
case '4':
case '5':
case '6':
case '7':
case '8':
case '9':
x *= 10;
x += *pos - '0';
break;
default:
if (negative)
x = -x;
return pos;
}
++pos;
}
if (negative)
x = -x;
return pos;
}
2013-06-21 20:34:19 +00:00
2016-10-21 00:19:49 +00:00
/// Convenient methods for reading something from string in text format.
2013-06-21 20:34:19 +00:00
template <typename T>
inline T parse(const char * data, size_t size)
{
T res;
2017-01-03 01:42:17 +00:00
ReadBufferFromMemory buf(data, size);
2013-06-21 20:34:19 +00:00
readText(res, buf);
return res;
}
2013-06-21 21:05:16 +00:00
template <typename T>
inline T parse(const char * data)
{
return parse<T>(data, strlen(data));
}
2013-06-21 20:34:19 +00:00
template <typename T>
inline T parse(const String & s)
{
return parse<T>(s.data(), s.size());
}
/** Skip UTF-8 BOM if it is under cursor.
* As BOM is usually located at start of stream, and buffer size is usually larger than three bytes,
* the function expects, that all three bytes of BOM is fully in buffer (otherwise it don't skip anything).
*/
inline void skipBOMIfExists(ReadBuffer & buf)
{
if (!buf.eof()
&& buf.position() + 3 < buf.buffer().end()
&& buf.position()[0] == '\xEF'
&& buf.position()[1] == '\xBB'
&& buf.position()[2] == '\xBF')
{
buf.position() += 3;
}
}
Text formats allow to skip errors (#407) * Allow to skip errors in text formats: added settings 'input_format_allow_errors_num' and 'input_format_allow_errors_ratio' [#CLICKHOUSE-2778]. https://github.com/yandex/ClickHouse/issues/134 * Allow to skip errors in text formats: added settings 'input_format_allow_errors_num' and 'input_format_allow_errors_ratio' [#CLICKHOUSE-2778]. https://github.com/yandex/ClickHouse/issues/134 * Allow to skip errors in text formats: added settings 'input_format_allow_errors_num' and 'input_format_allow_errors_ratio' [#CLICKHOUSE-2778]. https://github.com/yandex/ClickHouse/issues/134 * Allow to skip errors in text formats: added settings 'input_format_allow_errors_num' and 'input_format_allow_errors_ratio' [#CLICKHOUSE-2778]. https://github.com/yandex/ClickHouse/issues/134 * Allow to skip errors in text formats: added settings 'input_format_allow_errors_num' and 'input_format_allow_errors_ratio' [#CLICKHOUSE-2778]. https://github.com/yandex/ClickHouse/issues/134 * Allow to skip errors in text formats: added settings 'input_format_allow_errors_num' and 'input_format_allow_errors_ratio' [#CLICKHOUSE-2778]. https://github.com/yandex/ClickHouse/issues/134 * Allow to skip errors in text formats: added settings 'input_format_allow_errors_num' and 'input_format_allow_errors_ratio' [#CLICKHOUSE-2778]. https://github.com/yandex/ClickHouse/issues/134 * Allow to skip errors in text formats: added settings 'input_format_allow_errors_num' and 'input_format_allow_errors_ratio' [#CLICKHOUSE-2778]. https://github.com/yandex/ClickHouse/issues/134 * Allow to skip errors in text formats: added settings 'input_format_allow_errors_num' and 'input_format_allow_errors_ratio' [#CLICKHOUSE-2778]. https://github.com/yandex/ClickHouse/issues/134 * Allow to skip errors in text formats: added settings 'input_format_allow_errors_num' and 'input_format_allow_errors_ratio' [#CLICKHOUSE-2778]. https://github.com/yandex/ClickHouse/issues/134 * Added test [#CLICKHOUSE-2778].
2017-01-27 04:29:47 +00:00
/// Skip to next character after next \n. If no \n in stream, skip to end.
void skipToNextLineOrEOF(ReadBuffer & buf);
/// Skip to next character after next unescaped \n. If no \n in stream, skip to end. Does not throw on invalid escape sequences.
void skipToUnescapedNextLineOrEOF(ReadBuffer & buf);
2011-10-30 05:19:41 +00:00
}