Merge remote-tracking branch 'origin/main' into optimize-operation

This commit is contained in:
Yuriy Dupyn 2024-01-28 00:26:56 +01:00
commit 900608b2f4
17 changed files with 814 additions and 198 deletions

View file

@ -7,3 +7,4 @@ edition = "2021"
[dependencies]
bimap = "0.6.3"
thiserror = "1.0.50"

View file

@ -1,11 +1,33 @@
use std::num::{ParseFloatError, ParseIntError};
use std::str::Utf8Error;
use thiserror::Error;
use crate::internals::row::ColumnPosition;
use crate::schema::{ColumnName, TableName};
use crate::type_system::{DbType, Uuid, Value};
#[derive(Debug)]
#[derive(Debug, Error)]
pub enum Error {
#[error("column position {1} of table {0} does not exist")]
ColumnPositionDoesNotExist(TableName, ColumnPosition),
#[error("column {1} of table {0} has unexpected type {2:?} and value {3:?}")]
ValueDoesNotMatchExpectedType(TableName, ColumnName, DbType, Value),
#[error("table {0} already contains row with id {1}")]
AttemptingToInsertAlreadyPresentId(TableName, Uuid),
#[error("table {0} cannot be indexed on column {1}")]
AttemptToIndexNonIndexableColumn(TableName, ColumnName),
}
#[derive(Debug, Error)]
pub enum TypeConversionError {
#[error("failed to decode bytes to string")]
TextDecodeFailed(#[from] Utf8Error),
#[error("failed to parse float from text")]
NumberDecodeFailed(#[from] ParseFloatError),
#[error("failed to parse int from text")]
IntDecodeFailed(#[from] ParseIntError),
#[error("unknown type with oid {oid} and size {size}")]
UnknownType {
oid: i32,
size: i16
}
}

View file

@ -2,6 +2,7 @@ use crate::type_system::Value;
use crate::operation::InsertionValues;
use std::ops::{Index, IndexMut};
use std::slice::SliceIndex;
use crate::restricted_row::RestrictedRow;
pub type ColumnPosition = usize;
@ -63,14 +64,15 @@ impl Row {
self.0.get(column_position)
}
pub fn restrict_columns(&self, columns: &Vec<ColumnPosition>) -> Row {
pub fn restrict_columns(&self, columns: &Vec<ColumnPosition>) -> RestrictedRow {
// If the index from `columns` is non-existant in `row`, it will just ignore it.
let mut subrow: Row = Row::new();
let mut subrow: Vec<(ColumnPosition, Value)> = vec![];
for column_position in columns {
if let Some(value) = self.get(*column_position) {
subrow.0.push(value.clone())
subrow.push((*column_position, value.clone()));
}
}
subrow
subrow.into()
}
}

View file

@ -3,6 +3,7 @@ use std::collections::{BTreeMap, HashMap, HashSet};
use crate::error::Error;
use crate::internals::column_index::ColumnIndex;
use crate::internals::row::{ColumnPosition, Row};
use crate::restricted_row::RestrictedRow;
use crate::schema::{ColumnName, TableSchema, TableName};
use crate::result::DbResult;
use crate::type_system::{IndexableValue, Uuid, Value};
@ -67,7 +68,7 @@ impl Table {
.collect()
}
pub fn select_all_rows<'a>(&'a self, selected_column_positions: Vec<ColumnPosition>) -> impl Iterator<Item=Row> + 'a {
pub fn select_all_rows<'a>(&'a self, selected_column_positions: Vec<ColumnPosition>) -> impl Iterator<Item=RestrictedRow> + 'a {
self.rows
.values()
.map(move |row| row.restrict_columns(&selected_column_positions))
@ -78,7 +79,7 @@ impl Table {
selected_column_positions: Vec<ColumnPosition>,
column_position: ColumnPosition,
value: Value,
) -> DbResult<impl Iterator<Item=Row> + 'a> {
) -> DbResult<impl Iterator<Item=RestrictedRow> + 'a> {
let restrict_columns_of_row = move |row: Row| row.restrict_columns(&selected_column_positions);
match value {
Value::Indexable(value) => match self.fetch_ids_from_index(column_position, &value)? {

View file

@ -1,9 +1,10 @@
use crate::internals::row::Row;
use crate::internals::row::ColumnPosition;
use crate::schema::{TableName, TableSchema};
use crate::internals::table::Table;
use crate::operation::{Operation, Condition};
use crate::result::DbResult;
use bimap::BiMap;
use crate::restricted_row::RestrictedRow;
// Use `TablePosition` as index
pub type Tables = Vec<Table>;
@ -18,7 +19,7 @@ pub struct State {
// #[derive(Debug)]
pub enum Response<'a> {
Selected(&'a TableSchema, Box<dyn Iterator<Item=Row> + 'a + Send>),
Selected(&'a TableSchema, Box<dyn Iterator<Item=RestrictedRow> + 'a + Send>),
Inserted,
Deleted(usize), // how many were deleted
TableCreated,
@ -89,8 +90,8 @@ impl State {
let selected_rows = match maybe_condition {
None => {
let x = table.select_all_rows(column_selection);
Box::new(x) as Box<dyn Iterator<Item=Row> + 'a + Send>
let rows = table.select_all_rows(column_selection);
Box::new(rows) as Box<dyn Iterator<Item=RestrictedRow> + 'a + Send>
},
Some(Condition::Eq(eq_column, value)) => {
@ -100,7 +101,7 @@ impl State {
eq_column,
value,
)?;
Box::new(x) as Box<dyn Iterator<Item=Row> + 'a + Send>
Box::new(x) as Box<dyn Iterator<Item=RestrictedRow> + 'a + Send>
}
};
@ -246,9 +247,9 @@ mod tests {
let row = &rows[0];
assert!(row.len() == 3);
assert!(row[0] == id);
assert!(row[1] == name);
assert!(row[2] == age);
assert!(row[0].1 == id);
assert!(row[1].1 == name);
assert!(row[2].1 == age);
}
#[test]
@ -305,23 +306,24 @@ mod tests {
let response: Response = state.interpret(Select(users_position, users_schema.all_selection(), None)).unwrap();
assert!(matches!(response, Response::Selected(_, _)));
let Response::Selected(_schema, rows) = response else {
let Response::Selected(_, rows) = response else {
panic!()
};
let rows: Vec<_> = rows.collect();
let rows: Vec<_> = rows.collect();
assert!(rows.len() == 2);
let row0 = &rows[0];
let row1 = &rows[1];
assert!(row0.len() == 3);
assert!(row0[0] == id0);
assert!(row0[1] == name0);
assert!(row0[2] == age0);
assert!(row0[0].1 == id0);
assert!(row0[1].1 == name0);
assert!(row0[2].1 == age0);
assert!(row1.len() == 3);
assert!(row1[0] == id1);
assert!(row1[1] == name1);
assert!(row1[2] == age1);
assert!(row1[0].1 == id1);
assert!(row1[1].1 == name1);
assert!(row1[2].1 == age1);
}
{
@ -333,7 +335,7 @@ mod tests {
))
.unwrap();
assert!(matches!(response, Response::Selected(_, _)));
let Response::Selected(_schema, rows) = response else {
let Response::Selected(_, rows) = response else {
panic!()
};
let rows: Vec<_> = rows.collect();
@ -341,9 +343,9 @@ mod tests {
let row0 = &rows[0];
assert!(row0.len() == 3);
assert!(row0[0] == id0);
assert!(row0[1] == name0);
assert!(row0[2] == age0);
assert!(row0[0].1 == id0);
assert!(row0[1].1 == name0);
assert!(row0[2].1 == age0);
}
{
@ -355,7 +357,7 @@ mod tests {
))
.unwrap();
assert!(matches!(response, Response::Selected(_, _)));
let Response::Selected(_schema, rows) = response else {
let Response::Selected(_, rows) = response else {
panic!()
};
let rows: Vec<_> = rows.collect();
@ -363,8 +365,8 @@ mod tests {
let row0 = &rows[0];
assert!(row0.len() == 2);
assert!(row0[0] == name0);
assert!(row0[1] == id0);
assert!(row0[0].1 == name0);
assert!(row0[1].1 == id0);
}
}
@ -430,7 +432,7 @@ mod tests {
let response: Response = state.interpret(Select(users_position, users_schema.all_selection(), None)).unwrap();
assert!(matches!(response, Response::Selected(_, _)));
let Response::Selected(_schema, rows) = response else {
let Response::Selected(_, rows) = response else {
panic!()
};
let rows: Vec<_> = rows.collect();
@ -438,9 +440,9 @@ mod tests {
let row = &rows[0];
assert!(row.len() == 3);
assert!(row[0] == id1);
assert!(row[1] == name1);
assert!(row[2] == age1);
assert!(row[0].1 == id1);
assert!(row[1].1 == name1);
assert!(row[2].1 == age1);
}
#[test]
@ -516,7 +518,6 @@ mod tests {
pub fn example() {
use crate::type_system::{IndexableValue, Value, DbType};
use crate::internals::row::ColumnPosition;
use Condition::*;
use IndexableValue::*;
use Operation::*;

View file

@ -5,3 +5,4 @@ pub mod type_system;
mod error;
mod internals;
mod result;
pub mod restricted_row;

View file

@ -4,6 +4,7 @@ use crate::internals::row::ColumnPosition;
use crate::interpreter::TablePosition;
// Validated operation. Constructed by validation crate.
#[derive(Debug)]
pub enum Operation {
Select(TablePosition, ColumnSelection, Option<Condition>),
Insert(TablePosition, InsertionValues),
@ -16,6 +17,7 @@ pub type InsertionValues = Vec<Value>;
pub type ColumnSelection = Vec<ColumnPosition>;
#[derive(Debug)]
pub enum Condition {
Eq(ColumnPosition, Value),
}

View file

@ -0,0 +1,35 @@
use std::ops::Index;
use std::slice::SliceIndex;
use crate::internals::row::ColumnPosition;
use crate::type_system::Value;
#[derive(Debug, Clone)]
pub struct RestrictedRow(Vec<(ColumnPosition, Value)>);
impl<Idx> Index<Idx> for RestrictedRow
where
Idx: SliceIndex<[(ColumnPosition, Value)]>,
{
type Output = Idx::Output;
fn index(&self, index: Idx) -> &Self::Output {
&self.0[index]
}
}
impl From<Vec<(ColumnPosition, Value)>> for RestrictedRow {
fn from(v: Vec<(ColumnPosition, Value)>) -> Self {
RestrictedRow(v)
}
}
impl RestrictedRow {
pub fn len(&self) -> usize {
self.0.len()
}
pub fn iter(&self) -> impl Iterator<Item=&(ColumnPosition, Value)> {
self.0.iter()
}
}

View file

@ -1,3 +1,5 @@
use crate::error::TypeConversionError;
// ==============Types================
#[derive(Debug, Clone, Copy, PartialEq, Eq)]
pub enum DbType {
@ -39,4 +41,156 @@ impl Value {
},
}
}
pub fn type_oid(&self) -> i32 {
match self {
Self::Number(_) => 701,
Self::Indexable(val) => match val {
IndexableValue::String(_) => 25,
IndexableValue::Int(_) => 23,
IndexableValue::Uuid(_) => 2950,
},
}
}
pub fn type_size(&self) -> i16 {
match self {
Self::Number(_) => 8,
Self::Indexable(val) => match val {
IndexableValue::String(_) => -2, // null terminated string
IndexableValue::Int(_) => 8,
IndexableValue::Uuid(_) => 16,
},
}
}
pub fn as_text_bytes(&self) -> Vec<u8> {
match self {
Self::Number(n) => format!("{n}").into_bytes(),
Self::Indexable(i) => match i {
IndexableValue::String(s) => format!("{s}\0").into_bytes(),
IndexableValue::Int(i) => format!("{i}").into_bytes(),
IndexableValue::Uuid(u) => format!("{u}").into_bytes(),
},
}
}
pub fn from_text_bytes(bytes: &[u8], type_oid: i32, type_size: i16) -> Result<Value, TypeConversionError> {
match (type_oid, type_size) {
(701, 8) => {
let s = std::str::from_utf8(bytes)?;
let n = s.parse::<f64>()?;
Ok(Value::Number(n))
}
(25, -2) => {
let s = std::str::from_utf8(bytes)?;
let s = &s[..s.len() - 1]; // remove null terminator
Ok(Value::Indexable(IndexableValue::String(s.to_string())))
}
(23, 8) => {
let s = std::str::from_utf8(bytes)?;
let n = s.parse::<u64>()?;
Ok(Value::Indexable(IndexableValue::Int(n)))
}
(2950, 16) => {
let s = std::str::from_utf8(bytes)?;
let n = s.parse::<Uuid>()?;
Ok(Value::Indexable(IndexableValue::Uuid(n)))
}
(oid, size) => Err(TypeConversionError::UnknownType { oid, size }),
}
}
}
#[cfg(test)]
mod tests {
use crate::error::TypeConversionError::UnknownType;
use super::{Value, IndexableValue};
#[test]
fn test_encode_number() {
let value = Value::Number(123.456);
let oid = value.type_oid();
let size = value.type_size();
let bytes = value.as_text_bytes();
let from_bytes = Value::from_text_bytes(&bytes, oid, size).unwrap();
assert_eq!(value, from_bytes);
assert_eq!(oid, 701);
assert_eq!(size, 8);
}
#[test]
fn test_encode_string() {
let value = Value::Indexable(IndexableValue::String("hello".to_string()));
let oid = value.type_oid();
let size = value.type_size();
let bytes = value.as_text_bytes();
let from_bytes = Value::from_text_bytes(&bytes, oid, size).unwrap();
assert_eq!(value, from_bytes);
assert_eq!(oid, 25);
assert_eq!(size, -2);
}
#[test]
fn test_encode_string_utf8() {
let value = Value::Indexable(IndexableValue::String("#速度与激情9 早上好中国 现在我有冰激淋 我很喜欢冰激淋 但是《速度与激情9》比冰激淋 🍧🍦🍨".to_string()));
let oid = value.type_oid();
let size = value.type_size();
let bytes = value.as_text_bytes();
let from_bytes = Value::from_text_bytes(&bytes, oid, size).unwrap();
assert_eq!(value, from_bytes);
assert_eq!(oid, 25);
assert_eq!(size, -2);
}
#[test]
fn test_encode_int() {
let value = Value::Indexable(IndexableValue::Int(123));
let oid = value.type_oid();
let size = value.type_size();
let bytes = value.as_text_bytes();
let from_bytes = Value::from_text_bytes(&bytes, oid, size).unwrap();
assert_eq!(value, from_bytes);
assert_eq!(oid, 23);
assert_eq!(size, 8);
}
#[test]
fn test_encode_uuid() {
let value = Value::Indexable(IndexableValue::Uuid(123));
let oid = value.type_oid();
let size = value.type_size();
let bytes = value.as_text_bytes();
let from_bytes = Value::from_text_bytes(&bytes, oid, size).unwrap();
assert_eq!(value, from_bytes);
assert_eq!(oid, 2950);
assert_eq!(size, 16);
}
#[test]
fn test_mismatched_size() {
let value = Value::Indexable(IndexableValue::Uuid(123));
let oid = value.type_oid();
let size = 8;
let bytes = value.as_text_bytes();
let from_bytes = Value::from_text_bytes(&bytes, oid, size);
assert!(matches!(from_bytes, Err(UnknownType { oid: 2950, size: 8 })))
}
}