1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
|
#pragma once
#include <limits>
#include <IO/ReadHelpers.h>
#include <Common/intExp.h>
#include <base/wide_integer_to_string.h>
namespace DB
{
namespace ErrorCodes
{
extern const int CANNOT_PARSE_NUMBER;
extern const int ARGUMENT_OUT_OF_BOUND;
}
/// Try to read Decimal into underlying type T from ReadBuffer. Throws if 'digits_only' is set and there's unexpected symbol in input.
/// Returns integer 'exponent' factor that x should be multiplied by to get correct Decimal value: result = x * 10^exponent.
/// Use 'digits' input as max allowed meaning decimal digits in result. Place actual number of meaning digits in 'digits' output.
/// Does not care about decimal scale, only about meaningful digits in decimal text representation.
template <bool _throw_on_error, typename T>
inline bool readDigits(ReadBuffer & buf, T & x, uint32_t & digits, int32_t & exponent, bool digits_only = false)
{
x = T(0);
exponent = 0;
uint32_t max_digits = digits;
digits = 0;
uint32_t places = 0;
typename T::NativeType sign = 1;
bool leading_zeroes = true;
bool after_point = false;
if (buf.eof())
{
if constexpr (_throw_on_error)
throwReadAfterEOF();
return false;
}
switch (*buf.position())
{
case '-':
sign = -1;
[[fallthrough]];
case '+':
++buf.position();
break;
}
bool stop = false;
while (!buf.eof() && !stop)
{
const char & byte = *buf.position();
switch (byte)
{
case '.':
after_point = true;
leading_zeroes = false;
break;
case '0':
{
if (leading_zeroes)
break;
if (after_point)
{
++places; /// Count trailing zeroes. They would be used only if there's some other digit after them.
break;
}
[[fallthrough]];
}
case '1': [[fallthrough]];
case '2': [[fallthrough]];
case '3': [[fallthrough]];
case '4': [[fallthrough]];
case '5': [[fallthrough]];
case '6': [[fallthrough]];
case '7': [[fallthrough]];
case '8': [[fallthrough]];
case '9':
{
leading_zeroes = false;
++places; // num zeroes before + current digit
if (digits + places > max_digits)
{
if (after_point)
{
/// Simply cut excessive digits.
break;
}
else
{
if constexpr (_throw_on_error)
throw Exception(ErrorCodes::ARGUMENT_OUT_OF_BOUND, "Too many digits ({} > {}) in decimal value",
std::to_string(digits + places), std::to_string(max_digits));
return false;
}
}
else
{
digits += places;
if (after_point)
exponent -= places;
// TODO: accurate shift10 for big integers
x *= intExp10OfSize<typename T::NativeType>(places);
places = 0;
x += (byte - '0');
break;
}
}
case 'e': [[fallthrough]];
case 'E':
{
++buf.position();
Int32 addition_exp = 0;
if (!tryReadIntText(addition_exp, buf))
{
if constexpr (_throw_on_error)
throw ParsingException(ErrorCodes::CANNOT_PARSE_NUMBER, "Cannot parse exponent while reading decimal");
else
return false;
}
exponent += addition_exp;
stop = true;
continue;
}
default:
if (digits_only)
{
if constexpr (_throw_on_error)
throw ParsingException(ErrorCodes::CANNOT_PARSE_NUMBER, "Unexpected symbol while reading decimal");
return false;
}
stop = true;
continue;
}
++buf.position();
}
x *= sign;
return true;
}
template <typename T, typename ReturnType=void>
inline ReturnType readDecimalText(ReadBuffer & buf, T & x, uint32_t precision, uint32_t & scale, bool digits_only = false)
{
static constexpr bool throw_exception = std::is_same_v<ReturnType, void>;
uint32_t digits = precision;
int32_t exponent;
auto ok = readDigits<throw_exception>(buf, x, digits, exponent, digits_only);
if (!throw_exception && !ok)
return ReturnType(false);
if (static_cast<int32_t>(digits) + exponent > static_cast<int32_t>(precision - scale))
{
if constexpr (throw_exception)
{
static constexpr auto pattern = "Decimal value is too big: {} digits were read: {}e{}."
" Expected to read decimal with scale {} and precision {}";
if constexpr (is_big_int_v<typename T::NativeType>)
throw Exception(ErrorCodes::ARGUMENT_OUT_OF_BOUND, pattern, digits, x.value, exponent, scale, precision);
else
throw Exception(ErrorCodes::ARGUMENT_OUT_OF_BOUND, pattern, digits, x, exponent, scale, precision);
}
else
return ReturnType(false);
}
if (static_cast<int32_t>(scale) + exponent < 0)
{
auto divisor_exp = -exponent - static_cast<int32_t>(scale);
if (divisor_exp >= std::numeric_limits<typename T::NativeType>::digits10)
{
/// Too big negative exponent
x.value = 0;
scale = 0;
return ReturnType(true);
}
else
{
/// Too many digits after point. Just cut off excessive digits.
auto divisor = intExp10OfSize<typename T::NativeType>(divisor_exp);
assert(divisor > 0); /// This is for Clang Static Analyzer. It is not smart enough to infer it automatically.
x.value /= divisor;
scale = 0;
return ReturnType(true);
}
}
scale += exponent;
return ReturnType(true);
}
template <typename T>
inline bool tryReadDecimalText(ReadBuffer & buf, T & x, uint32_t precision, uint32_t & scale)
{
return readDecimalText<T, bool>(buf, x, precision, scale, true);
}
template <typename T>
inline void readCSVDecimalText(ReadBuffer & buf, T & x, uint32_t precision, uint32_t & scale)
{
if (buf.eof())
throwReadAfterEOF();
char maybe_quote = *buf.position();
if (maybe_quote == '\'' || maybe_quote == '\"')
++buf.position();
readDecimalText(buf, x, precision, scale, false);
if (maybe_quote == '\'' || maybe_quote == '\"')
assertChar(maybe_quote, buf);
}
}
|
