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 99 100 101 102 103 104 105 106 107 108 109 110 111 112 113 114 115 116 117 118 119 120 121 122 123 124 125 126 127 128 129 130 131 132 133 134 135 136 137 138 139 140 141 142 143 144 145 146 147 148 149 150 151 152 153 154 155 156 157 158 159 160 161 162 163 164 165 166 167 168 169 170 171 172 173 174 175 176 177 178 179 180 181 182 183 184 185 186 187 188 189 190 191 192 193 194 195 196 197 198 199 200 201 202 203 204 205 206 207 208 209 210 211 212 213 214 215 216 217 218 219 220 221 222 223 224 225 226 227 228 229 230 231 232 233 234 235 236 237 238 239 240 241 242 243 244 245 246 247 248 249 250 251 252 253 254 255 256 257 258 259 260 261 262 263 264 265 266 267 268 269 270 271 272 273 274 275 276 277 278 279 280 281 282 283 284 285 286 287 288 289 290 291 292 293 294 295 296 297 298 299 300 301 302 303 304 305 306 307 308 309 310 311 312 313 314 315 316 317 318 319 320 321 322 323 324 325 326 327 328 329 330 331 332 333 334 335 336 337 338 339 340 341 342 343 344 345 346 347 348 349 350 351 352 353 354 355 356 357 358 359 360 361 362 363 364 365 366 367 368 369 370 371 372 373 374 375 376 377 378 379 380 381 382 383 384 385 386 387 388 389 390 391 392 393 394 395 396 397 398 399 400 401 402 403 404 405 406 407 408 409 410 411 412 413 414 415 416 417 418 419 420 421 422 423 424 425 426 427 428 429 430 431 432 433 434 435 436 437 438 439
use crate::{
buffers::Indicator,
handles::{CData, CDataMut, HasDataType},
DataType,
};
use log::debug;
use odbc_sys::{CDataType, NULL_DATA};
use std::{cmp::min, ffi::c_void};
/// A buffer intended to be bound to a column of a cursor. Elements of the buffer will contain a
/// variable amount of bytes up to a maximum length. Since elements of this type have variable
/// length an additional indicator buffer is also maintained, whether the column is nullable or not.
/// Therefore this buffer type is used for variable sized binary data whether it is nullable or not.
#[derive(Debug)]
pub struct BinColumn {
/// Maximum element length.
max_len: usize,
values: Vec<u8>,
/// Elements in this buffer are either `NULL_DATA` or hold the length of the element in value
/// with the same index. Please note that this value may be larger than `max_len` if the value
/// has been truncated.
indicators: Vec<isize>,
}
impl BinColumn {
/// This will allocate a value and indicator buffer for `batch_size` elements. Each value may
/// have a maximum length of `max_len`.
pub fn new(batch_size: usize, max_len: usize) -> Self {
BinColumn {
max_len,
values: vec![0; max_len * batch_size],
indicators: vec![0; batch_size],
}
}
/// Return the value for the given row index.
///
/// The column buffer does not know how many elements were in the last row group, and therefore
/// can not guarantee the accessed element to be valid and in a defined state. It also can not
/// panic on accessing an undefined element. It will panic however if `row_index` is larger or
/// equal to the maximum number of elements in the buffer.
pub fn value_at(&self, row_index: usize) -> Option<&[u8]> {
self.content_length_at(row_index).map(|length| {
let offset = row_index * self.max_len;
&self.values[offset..offset + length]
})
}
/// Indicator value at the specified position. Useful to detect truncation of data.
///
/// The column buffer does not know how many elements were in the last row group, and therefore
/// can not guarantee the accessed element to be valid and in a defined state. It also can not
/// panic on accessing an undefined element. It will panic however if `row_index` is larger or
/// equal to the maximum number of elements in the buffer.
pub fn indicator_at(&self, row_index: usize) -> Indicator {
Indicator::from_isize(self.indicators[row_index])
}
/// Length of value at the specified position. This is different from an indicator as it refers
/// to the length of the value in the buffer, not to the length of the value in the datasource.
/// The two things are different for truncated values.
pub fn content_length_at(&self, row_index: usize) -> Option<usize> {
match self.indicator_at(row_index) {
Indicator::Null => None,
// Seen no total in the wild then binding shorter buffer to fixed sized CHAR in MSSQL.
Indicator::NoTotal => Some(self.max_len),
Indicator::Length(length) => {
let length = min(self.max_len, length);
Some(length)
}
}
}
/// Changes the maximum element length the buffer can hold. This operation is useful if you find
/// an unexpected large input during insertion. All values in the buffer will be set to NULL.
///
/// # Parameters
///
/// * `new_max_len`: New maximum string length without terminating zero.
pub fn set_max_len(&mut self, new_max_len: usize) {
let batch_size = self.indicators.len();
// Allocate a new buffer large enough to hold a batch of strings with maximum length.
let new_values = vec![0u8; new_max_len * batch_size];
// Set all indicators to NULL
self.fill_null(0, batch_size);
self.values = new_values;
self.max_len = new_max_len;
}
/// Maximum length of elements in bytes.
pub fn max_len(&self) -> usize {
self.max_len
}
/// View of the first `num_rows` values of a binary column.
///
/// Num rows may not exceed the actually amount of valid num_rows filled be the ODBC API. The
/// column buffer does not know how many elements were in the last row group, and therefore can
/// not guarantee the accessed element to be valid and in a defined state. It also can not panic
/// on accessing an undefined element. It will panic however if `row_index` is larger or equal
/// to the maximum number of elements in the buffer.
pub fn view(&self, num_rows: usize) -> BinColumnView<'_> {
BinColumnView {
num_rows,
col: self,
}
}
/// Sets the value of the buffer at index to NULL or the specified bytes. This method will panic
/// on out of bounds index, or if input holds a value which is longer than the maximum allowed
/// element length.
pub fn set_value(&mut self, index: usize, input: Option<&[u8]>) {
if let Some(input) = input {
self.indicators[index] = input.len().try_into().unwrap();
if input.len() > self.max_len {
panic!(
"Tried to insert a value into a binary buffer which is larger than the maximum \
allowed element length for the buffer."
);
}
let start = self.max_len * index;
let end = start + input.len();
let buf = &mut self.values[start..end];
buf.copy_from_slice(input);
} else {
self.indicators[index] = NULL_DATA;
}
}
/// Fills the column with NULL, between From and To
pub fn fill_null(&mut self, from: usize, to: usize) {
for index in from..to {
self.indicators[index] = NULL_DATA;
}
}
/// A writer able to fill the first `n` elements of the buffer, from an iterator.
pub fn writer_n(&mut self, n: usize) -> BinColumnWriter<'_> {
BinColumnWriter {
column: self,
to: n,
}
}
/// Changes the maximum number of bytes per row the buffer can hold. This operation is useful if
/// you find an unexpected large input during insertion.
///
/// This is however costly, as not only does the new buffer have to be allocated, but all values
/// have to copied from the old to the new buffer.
///
/// This method could also be used to reduce the maximum length, which would truncate values in
/// the process.
///
/// This method does not adjust indicator buffers as these might hold values larger than the
/// maximum length.
///
/// # Parameters
///
/// * `new_max_len`: New maximum element length in bytes.
/// * `num_rows`: Number of valid rows currently stored in this buffer.
pub fn resize_max_element_length(&mut self, new_max_len: usize, num_rows: usize) {
debug!(
"Rebinding binary column buffer with {} elements. Maximum length {} => {}",
num_rows, self.max_len, new_max_len
);
let batch_size = self.indicators.len();
// Allocate a new buffer large enough to hold a batch of elements with maximum length.
let mut new_values = vec![0; new_max_len * batch_size];
// Copy values from old to new buffer.
let max_copy_length = min(self.max_len, new_max_len);
for ((&indicator, old_value), new_value) in self
.indicators
.iter()
.zip(self.values.chunks_exact_mut(self.max_len))
.zip(new_values.chunks_exact_mut(new_max_len))
.take(num_rows)
{
match Indicator::from_isize(indicator) {
Indicator::Null => (),
Indicator::NoTotal => {
// There is no good choice here in case we are expanding the buffer. Since
// NO_TOTAL indicates that we use the entire buffer, but in truth it would now
// be padded with 0. I currently cannot think of any use case there it would
// matter.
new_value[..max_copy_length].clone_from_slice(&old_value[..max_copy_length]);
}
Indicator::Length(num_bytes_len) => {
let num_bytes_to_copy = min(num_bytes_len, max_copy_length);
new_value[..num_bytes_to_copy].copy_from_slice(&old_value[..num_bytes_to_copy]);
}
}
}
self.values = new_values;
self.max_len = new_max_len;
}
/// Appends a new element to the column buffer. Rebinds the buffer to increase maximum element
/// length should the input be to large.
///
/// # Parameters
///
/// * `index`: Zero based index of the new row position. Must be equal to the number of rows
/// currently in the buffer.
/// * `bytes`: Value to store.
pub fn append(&mut self, index: usize, bytes: Option<&[u8]>) {
if let Some(bytes) = bytes {
if bytes.len() > self.max_len {
let new_max_len = (bytes.len() as f64 * 1.2) as usize;
self.resize_max_element_length(new_max_len, index)
}
let offset = index * self.max_len;
self.values[offset..offset + bytes.len()].copy_from_slice(bytes);
// And of course set the indicator correctly.
self.indicators[index] = bytes.len().try_into().unwrap();
} else {
self.indicators[index] = NULL_DATA;
}
}
/// Maximum number of elements this buffer can hold.
pub fn capacity(&self) -> usize {
self.indicators.len()
}
}
#[derive(Debug, Clone, Copy)]
pub struct BinColumnView<'c> {
num_rows: usize,
col: &'c BinColumn,
}
impl<'c> BinColumnView<'c> {
/// The number of valid elements in the text column.
pub fn len(&self) -> usize {
self.num_rows
}
/// True if, and only if there are no valid rows in the column buffer.
pub fn is_empty(&self) -> bool {
self.num_rows == 0
}
/// Slice of text at the specified row index without terminating zero.
pub fn get(&self, index: usize) -> Option<&'c [u8]> {
self.col.value_at(index)
}
/// Iterator over the valid elements of the text buffer
pub fn iter(&self) -> BinColumnIt<'c> {
BinColumnIt {
pos: 0,
num_rows: self.num_rows,
col: self.col,
}
}
}
/// Iterator over a binary column. See [`crate::buffers::AnyColumnView`]
#[derive(Debug)]
pub struct BinColumnIt<'c> {
pos: usize,
num_rows: usize,
col: &'c BinColumn,
}
impl<'c> Iterator for BinColumnIt<'c> {
type Item = Option<&'c [u8]>;
fn next(&mut self) -> Option<Self::Item> {
if self.pos == self.num_rows {
None
} else {
let ret = Some(self.col.value_at(self.pos));
self.pos += 1;
ret
}
}
fn size_hint(&self) -> (usize, Option<usize>) {
let len = self.num_rows - self.pos;
(len, Some(len))
}
}
impl<'c> ExactSizeIterator for BinColumnIt<'c> {}
/// Fills a binary column buffer with elements from an Iterator. See
/// [`crate::buffers::AnyColumnViewMut`]
#[derive(Debug)]
pub struct BinColumnWriter<'a> {
column: &'a mut BinColumn,
/// Upper limit, the binary column writer will not write beyond this index.
to: usize,
}
impl<'a> BinColumnWriter<'a> {
/// Fill the binary column with values by consuming the iterator and copying its items into the
/// buffer. It will not extract more items from the iterator than the buffer may hold. This
/// method panics if elements of the iterator are larger than the maximum element length of the
/// buffer.
pub fn write<'b>(&mut self, it: impl Iterator<Item = Option<&'b [u8]>>) {
for (index, item) in it.enumerate().take(self.to) {
self.column.set_value(index, item)
}
}
/// Changes the maximum element length the buffer can hold. This operation is useful if you find
/// an unexpected large input during insertion. All values in the buffer will be set to NULL.
///
/// # Parameters
///
/// * `new_max_len`: New maximum element length
pub fn set_max_len(&mut self, new_max_len: usize) {
self.column.set_max_len(new_max_len)
}
/// Maximum length
pub fn max_len(&self) -> usize {
self.column.max_len()
}
/// Changes the maximum element length the buffer can hold. This operation is useful if you find
/// an unexpected large input during insertion.
///
/// This is however costly, as not only does the new buffer have to be allocated, but all values
/// have to copied from the old to the new buffer.
///
/// This method could also be used to reduce the maximum element length, which would truncate
/// values in the process.
///
/// This method does not adjust indicator buffers as these might hold values larger than the
/// maximum element length.
///
/// # Parameters
///
/// * `new_max_len`: New maximum element length.
/// * `num_rows`: Number of valid rows currently stored in this buffer.
pub fn resize_max_element_length(&mut self, new_max_len: usize, num_rows: usize) {
self.column.resize_max_element_length(new_max_len, num_rows)
}
/// Inserts a new element to the column buffer. Rebinds the buffer to increase maximum element
/// length should the value be larger than the maximum allowed element length. The number of
/// rows the column buffer can hold stays constant, but during rebind only values before `index`
/// would be copied to the new memory location. Therefore this method is intended to be used to
/// fill the buffer element-wise and in order. Hence the name `append`.
///
/// # Parameters
///
/// * `index`: Zero based index of the new row position. Must be equal to the number of rows
/// currently in the buffer.
/// * `bytes`: Value to store
///
/// # Example
///
/// ```
/// # use odbc_api::buffers::{
/// BufferDescription, BufferKind, AnyColumnViewMut, AnyColumnView, buffer_from_description
/// };
/// # use std::iter;
/// #
/// let desc = BufferDescription {
/// // Buffer size purposefully chosen too small, so we need to increase the buffer size if we
/// // encounter larger inputs.
/// kind: BufferKind::Binary { length: 1 },
/// nullable: true,
/// };
///
/// // Input values to insert.
/// let input : [Option<&[u8]>; 5]= [
/// Some(&[1]),
/// Some(&[2,3]),
/// Some(&[4,5,6]),
/// None,
/// Some(&[7,8,9,10,11,12]),
/// ];
///
/// let mut buffer = buffer_from_description(input.len(), iter::once(desc));
///
/// buffer.set_num_rows(input.len());
/// if let AnyColumnViewMut::Binary(mut writer) = buffer.column_mut(0) {
/// for (index, &bytes) in input.iter().enumerate() {
/// writer.append(index, bytes)
/// }
/// } else {
/// panic!("Expected binary column writer");
/// }
///
/// let col_view = buffer.column(0).as_bin_view().expect("Expected binary column slice");
/// assert!(
/// col_view
/// .iter()
/// .zip(input.iter().copied())
/// .all(|(expected, actual)| expected == actual)
/// )
/// ```
pub fn append(&mut self, index: usize, bytes: Option<&[u8]>) {
self.column.append(index, bytes)
}
}
unsafe impl CData for BinColumn {
fn cdata_type(&self) -> CDataType {
CDataType::Binary
}
fn indicator_ptr(&self) -> *const isize {
self.indicators.as_ptr()
}
fn value_ptr(&self) -> *const c_void {
self.values.as_ptr() as *const c_void
}
fn buffer_length(&self) -> isize {
self.max_len.try_into().unwrap()
}
}
impl HasDataType for BinColumn {
fn data_type(&self) -> DataType {
DataType::Varbinary {
length: self.max_len,
}
}
}
unsafe impl CDataMut for BinColumn {
fn mut_indicator_ptr(&mut self) -> *mut isize {
self.indicators.as_mut_ptr()
}
fn mut_value_ptr(&mut self) -> *mut c_void {
self.values.as_mut_ptr() as *mut c_void
}
}