/
description.rs
380 lines (368 loc) · 16.6 KB
/
description.rs
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
228
229
230
231
232
233
234
235
236
237
238
239
240
241
242
243
244
245
246
247
248
249
250
251
252
253
254
255
256
257
258
259
260
261
262
263
264
265
266
267
268
269
270
271
272
273
274
275
276
277
278
279
280
281
282
283
284
285
286
287
288
289
290
291
292
293
294
295
296
297
298
299
300
301
302
303
304
305
306
307
308
309
310
311
312
313
314
315
316
317
318
319
320
321
322
323
324
325
326
327
328
329
330
331
332
333
334
335
336
337
338
339
340
341
342
343
344
345
346
347
348
349
350
351
352
353
354
355
356
357
358
359
360
361
362
363
364
365
366
367
368
369
370
371
372
373
374
375
376
377
378
379
380
use std::mem::size_of;
use odbc_sys::{Date, Time, Timestamp};
use crate::{Bit, DataType};
/// Describes a column of a [`crate::buffers::ColumnarBuffer`].
///
/// While related to to the [`crate::DataType`] of the column this is bound to, the Buffer type is
/// different as it does not describe the type of the data source but the format the data is going
/// to be represented in memory. While the data source is often considered to choose the buffer type
/// the kind of processing which is supposed to be applied to the data may be even more important
/// if choosing the a buffer for the cursor type. E.g. if you intend to print a date to standard out
/// it may be more reasonable to bind it as `Text` rather than `Date`.
#[derive(Clone, Copy, Debug, PartialEq, Eq)]
pub enum BufferDesc {
/// Variable sized binary buffer, holding up to `length` bytes per value.
Binary {
/// Maximum number of bytes per value.
length: usize,
},
/// Text buffer holding strings with binary length of up to `max_str_len`.
///
/// Consider an upper bound choosing this based on the information in a [`DataType::Varchar`]
/// column. E.g. PostgreSQL may return a field size of several GiB for individual values if a
/// column is specified as `TEXT`, or Microsoft SQL Server may return `0` for a column of type
/// `VARCHAR(max)`. In such situations, if values are truly that large, bulk fetching data is
/// not recommended, but streaming individual fields one by one. Usually though, the actual
/// cells of the table in the database contain much shorter values. The best thing todo is to
/// adapt the database schema to better reflect the actual size of the values. Lacking control
/// over the database schema, you can always choose a smaller buffer size than initializing the
/// buffer in disagreement with the database schema.
Text {
/// Maximum string length. Terminating zero is excluded, i.e. memory for it will be
/// implicitly allocated if required.
max_str_len: usize,
},
/// UTF-16 encoded text buffer holding strings with length of up to `max_str_len`. Length is in
/// terms of 2-Byte characters.
WText {
/// Maximum string length. Terminating zero is excluded, i.e. memory for it will be
/// implicitly allocated if required.
max_str_len: usize,
},
/// 64 bit floating point
F64 {
/// This indicates whether or not the buffer will be able to represent NULL values. This will
/// cause an indicator buffer to be bound.
nullable: bool,
},
/// 32 bit floating point
F32 {
/// This indicates whether or not the buffer will be able to represent NULL values. This will
/// cause an indicator buffer to be bound.
nullable: bool,
},
/// Describes a buffer holding [`crate::sys::Date`] values.
Date {
/// This indicates whether or not the buffer will be able to represent NULL values. This will
/// cause an indicator buffer to be bound.
nullable: bool,
},
/// Describes a buffer holding [`crate::sys::Time`] values.
Time {
/// This indicates whether or not the buffer will be able to represent NULL values. This will
/// cause an indicator buffer to be bound.
nullable: bool,
},
/// Describes a buffer holding [`crate::sys::Timestamp`] values.
Timestamp {
/// This indicates whether or not the buffer will be able to represent NULL values. This will
/// cause an indicator buffer to be bound.
nullable: bool,
},
/// Signed 8 Bit integer
I8 {
/// This indicates whether or not the buffer will be able to represent NULL values. This will
/// cause an indicator buffer to be bound.
nullable: bool,
},
/// Signed 16 Bit integer
I16 {
/// This indicates whether or not the buffer will be able to represent NULL values. This will
/// cause an indicator buffer to be bound.
nullable: bool,
},
/// Signed 32 Bit integer
I32 {
/// This indicates whether or not the buffer will be able to represent NULL values. This will
/// cause an indicator buffer to be bound.
nullable: bool,
},
/// Signed 64 Bit integer
I64 {
/// This indicates whether or not the buffer will be able to represent NULL values. This will
/// cause an indicator buffer to be bound.
nullable: bool,
},
/// Unsigned 8 Bit integer
U8 {
/// This indicates whether or not the buffer will be able to represent NULL values. This will
/// cause an indicator buffer to be bound.
nullable: bool,
},
/// Can either be zero or one
Bit {
/// This indicates whether or not the buffer will be able to represent NULL values. This will
/// cause an indicator buffer to be bound.
nullable: bool,
},
}
/// Describes a column of a [`crate::buffers::ColumnarBuffer`].
///
/// While related to to the [`crate::DataType`] of the column this is bound to, the Buffer type is
/// different as it does not describe the type of the data source but the format the data is going
/// to be represented in memory. While the data source is often considered to choose the buffer type
/// the kind of processing which is supposed to be applied to the data may be even more important
/// if choosing the a buffer for the cursor type. E.g. if you intend to print a date to standard out
/// it may be more reasonable to bind it as `Text` rather than `Date`.
#[derive(Clone, Copy, Debug)]
pub struct BufferDescription {
/// This indicates whether or not the buffer will be able to represent NULL values. This will
/// cause an indicator buffer to be bound.
pub nullable: bool,
/// The type of CData the buffer will be holding.
pub kind: BufferKind,
}
impl BufferDescription {
/// Returns the element size of such a buffer if bound as a columnar row. Can be used to
/// estimate memory for columnar bindings.
pub fn bytes_per_row(&self) -> usize {
let indicator = size_of::<isize>();
let opt_indicator = if self.nullable { indicator } else { 0 };
match self.kind {
BufferKind::Binary { length } => length + indicator,
BufferKind::Text { max_str_len } => max_str_len + 1 + indicator,
BufferKind::WText { max_str_len } => (max_str_len + 1) * 2 + indicator,
BufferKind::F64 => size_of::<f64>() + opt_indicator,
BufferKind::F32 => size_of::<f32>() + opt_indicator,
BufferKind::Date => size_of::<Date>() + opt_indicator,
BufferKind::Time => size_of::<Time>() + opt_indicator,
BufferKind::Timestamp => size_of::<Timestamp>() + opt_indicator,
BufferKind::I8 => size_of::<i8>() + opt_indicator,
BufferKind::I16 => size_of::<i16>() + opt_indicator,
BufferKind::I32 => size_of::<i32>() + opt_indicator,
BufferKind::I64 => size_of::<i64>() + opt_indicator,
BufferKind::U8 => size_of::<u8>() + opt_indicator,
BufferKind::Bit => size_of::<Bit>() + opt_indicator,
}
}
}
impl From<BufferDescription> for BufferDesc {
fn from(source: BufferDescription) -> Self {
let nullable = source.nullable;
match source.kind {
BufferKind::Binary { length } => BufferDesc::Binary { length },
BufferKind::Text { max_str_len } => BufferDesc::Text { max_str_len },
BufferKind::WText { max_str_len } => BufferDesc::WText { max_str_len },
BufferKind::F64 => BufferDesc::F64 { nullable },
BufferKind::F32 => BufferDesc::F32 { nullable },
BufferKind::Date => BufferDesc::Date { nullable },
BufferKind::Time => BufferDesc::Time { nullable },
BufferKind::Timestamp => BufferDesc::Timestamp { nullable },
BufferKind::I8 => BufferDesc::I8 { nullable },
BufferKind::I16 => BufferDesc::I16 { nullable },
BufferKind::I32 => BufferDesc::I32 { nullable },
BufferKind::I64 => BufferDesc::I64 { nullable },
BufferKind::U8 => BufferDesc::U8 { nullable },
BufferKind::Bit => BufferDesc::Bit { nullable },
}
}
}
/// This class is used together with [`BufferDescription`] to specify the layout of buffers bound to
/// ODBC cursors and statements.
#[derive(Clone, Copy, Debug, PartialEq, Eq)]
pub enum BufferKind {
/// Variable sized binary buffer, holding up to `length` bytes per value.
Binary {
/// Maximum number of bytes per value.
length: usize,
},
/// Text buffer holding strings with binary length of up to `max_str_len`.
///
/// Consider an upper bound choosing this based on the information in a [`DataType::Varchar`]
/// column. E.g. PostgreSQL may return a field size of several GiB for individual values if a
/// column is specified as `TEXT`, or Microsoft SQL Server may return `0` for a column of type
/// `VARCHAR(max)`. In such situations, if values are truly that large, bulk fetching data is
/// not recommended, but streaming individual fields one by one. Usually though, the actual
/// cells of the table in the database contain much shorter values. The best thing todo is to
/// adapt the database schema to better reflect the actual size of the values. Lacking control
/// over the database schema, you can always choose a smaller buffer size than initializing the
/// buffer in disagreement with the database schema.
Text {
/// Maximum string length. Terminating zero is excluded, i.e. memory for it will be
/// implicitly allocated if required.
max_str_len: usize,
},
/// UTF-16 encoded text buffer holding strings with length of up to `max_str_len`. Length is in
/// terms of 2-Byte characters.
WText {
/// Maximum string length. Terminating zero is excluded, i.e. memory for it will be
/// implicitly allocated if required.
max_str_len: usize,
},
/// 64 bit floating point
F64,
/// 32 bit floating point
F32,
/// Describes a buffer holding [`crate::sys::Date`] values.
Date,
/// Describes a buffer holding [`crate::sys::Time`] values.
Time,
/// Describes a buffer holding [`crate::sys::Timestamp`] values.
Timestamp,
/// Signed 8 Bit integer
I8,
/// Signed 16 Bit integer
I16,
/// Signed 32 Bit integer
I32,
/// Signed 64 Bit integer
I64,
/// Unsigned 8 Bit integer
U8,
/// Can either be zero or one
Bit,
}
impl BufferKind {
/// Describe a buffer which fits best the SQL Data Type.
///
/// ```
/// use odbc_api::{DataType, buffers::BufferKind};
///
/// assert_eq!(BufferKind::from_data_type(DataType::Unknown), None);
/// assert_eq!(
/// BufferKind::from_data_type(DataType::Numeric { precision: 2, scale: 0 }),
/// Some(BufferKind::I8)
/// );
/// assert_eq!(
/// BufferKind::from_data_type(DataType::Numeric { precision: 9, scale: 0 }),
/// Some(BufferKind::I32)
/// );
/// assert_eq!(
/// BufferKind::from_data_type(DataType::Numeric { precision: 18, scale: 0 }),
/// Some(BufferKind::I64)
/// );
/// assert_eq!(
/// BufferKind::from_data_type(DataType::Numeric { precision: 20, scale: 5 }),
/// Some(BufferKind::Text { max_str_len: 20 + 2 })
/// );
/// assert_eq!(
/// BufferKind::from_data_type(DataType::Varchar { length: 42 }),
/// Some(BufferKind::Text { max_str_len: 42 })
/// );
/// // We do not care about the encoding in the datasource. WVarchar is mapped to `Text`, too
/// // (instead of `WText`).
/// assert_eq!(
/// BufferKind::from_data_type(DataType::WVarchar { length: 42 }),
/// Some(BufferKind::Text { max_str_len: 42 })
/// );
/// assert_eq!(
/// BufferKind::from_data_type(DataType::BigInt),
/// Some(BufferKind::I64)
/// );
/// assert_eq!(
/// BufferKind::from_data_type(DataType::Integer),
/// Some(BufferKind::I32)
/// );
/// assert_eq!(
/// BufferKind::from_data_type(DataType::SmallInt),
/// Some(BufferKind::I16)
/// );
/// assert_eq!(
/// BufferKind::from_data_type(DataType::TinyInt),
/// Some(BufferKind::I8)
/// );
/// assert_eq!(
/// BufferKind::from_data_type(DataType::Float { precision : 24 }),
/// Some(BufferKind::F32)
/// );
/// assert_eq!(
/// BufferKind::from_data_type(DataType::Float { precision : 53 }),
/// Some(BufferKind::F64)
/// );
/// assert_eq!(
/// BufferKind::from_data_type(DataType::Double),
/// Some(BufferKind::F64)
/// );
/// assert_eq!(
/// BufferKind::from_data_type(DataType::Date),
/// Some(BufferKind::Date)
/// );
/// assert_eq!(
/// BufferKind::from_data_type(DataType::Time { precision: 0 }),
/// Some(BufferKind::Time)
/// );
/// assert_eq!(
/// BufferKind::from_data_type(DataType::Bit),
/// Some(BufferKind::Bit)
/// );
/// assert_eq!(
/// BufferKind::from_data_type(DataType::Time { precision: 3 }),
/// Some(BufferKind::Text { max_str_len: 12 })
/// );
/// assert_eq!(
/// BufferKind::from_data_type(DataType::Timestamp { precision: 3 }),
/// Some(BufferKind::Timestamp)
/// );
/// ```
pub fn from_data_type(data_type: DataType) -> Option<Self> {
let buffer_kind = match data_type {
DataType::Numeric { precision, scale }
| DataType::Decimal { precision, scale } if scale == 0 && precision < 3 => BufferKind::I8,
DataType::Numeric { precision, scale }
| DataType::Decimal { precision, scale } if scale == 0 && precision < 10 => BufferKind::I32,
DataType::Numeric { precision, scale }
| DataType::Decimal { precision, scale } if scale == 0 && precision < 19 => BufferKind::I64,
DataType::Integer => BufferKind::I32,
DataType::SmallInt => BufferKind::I16,
DataType::Float { precision: 0..=24 } | DataType::Real => BufferKind::F32,
DataType::Float { precision: 25..=53 } |DataType::Double => BufferKind::F64,
DataType::Date => BufferKind::Date,
DataType::Time { precision: 0 } => BufferKind::Time,
DataType::Timestamp { precision: _ } => BufferKind::Timestamp,
DataType::BigInt => BufferKind::I64,
DataType::TinyInt => BufferKind::I8,
DataType::Bit => BufferKind::Bit,
DataType::Varbinary { length }
| DataType::Binary { length }
| DataType::LongVarbinary { length } => BufferKind::Binary { length },
DataType::Varchar { length }
| DataType::WVarchar { length }
// Currently no special buffers for fixed lengths text implemented.
| DataType::WChar {length }
| DataType::Char { length }
| DataType::LongVarchar { length } => BufferKind::Text { max_str_len : length },
// Specialized buffers for Numeric and decimal are not yet supported.
| DataType::Numeric { precision: _, scale: _ }
| DataType::Decimal { precision: _, scale: _ }
| DataType::Time { precision: _ } => BufferKind::Text { max_str_len: data_type.display_size().unwrap() },
DataType::Unknown
| DataType::Float { precision: _ }
| DataType::Other { data_type: _, column_size: _, decimal_digits: _ } => return None,
};
Some(buffer_kind)
}
}
#[cfg(test)]
mod tests {
use super::*;
#[test]
#[cfg(target_pointer_width = "64")] // Indicator size is platform dependent.
fn bytes_per_row() {
let bpr = |kind, nullable| BufferDescription { nullable, kind }.bytes_per_row();
assert_eq!(5 + 8, bpr(BufferKind::Binary { length: 5 }, false));
assert_eq!(5 + 1 + 8, bpr(BufferKind::Text { max_str_len: 5 }, false));
assert_eq!(10 + 2 + 8, bpr(BufferKind::WText { max_str_len: 5 }, false));
assert_eq!(6, bpr(BufferKind::Date, false));
assert_eq!(6, bpr(BufferKind::Time, false));
assert_eq!(16, bpr(BufferKind::Timestamp, false));
assert_eq!(1, bpr(BufferKind::Bit, false));
assert_eq!(1 + 8, bpr(BufferKind::Bit, true));
assert_eq!(4, bpr(BufferKind::F32, false));
assert_eq!(8, bpr(BufferKind::F64, false));
assert_eq!(1, bpr(BufferKind::I8, false));
assert_eq!(2, bpr(BufferKind::I16, false));
assert_eq!(4, bpr(BufferKind::I32, false));
assert_eq!(8, bpr(BufferKind::I64, false));
assert_eq!(1, bpr(BufferKind::U8, false));
}
}