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
//! User-facing query representation.
use crate::document::Document;
use crate::index::IndexSchema;
use crate::value::Value;
use std::collections::HashMap;
use std::ops::Bound;
pub type FieldPath = Vec<String>;
/// The order that the results should be sorted in.
pub enum ResultOrder {
/// Sorted ascending according to the given field.
Asc(FieldPath),
/// Sorted descending according to the given field.
Desc(FieldPath),
}
/// Document type modeling query constraints.
pub type ConstraintDocument = HashMap<FieldPath, Constraint>;
/// Exists so that we can implement functions on ConstraintDocument.
pub trait ConstraintDocumentTrait {
/// Constructs a reduced ConstraintDocument by removing the fields
/// that are part of the IndexSchema.
fn remove_index_fields(&self, index_schema: &IndexSchema) -> Self;
/// Returns whether the current constraint document matches a document.
fn matches_document(&self, doc: &Document) -> bool;
}
impl ConstraintDocumentTrait for ConstraintDocument {
fn remove_index_fields(&self, _index_schema: &IndexSchema) -> ConstraintDocument {
// For the time being, don't remove index fields, due to the issue with
// inclusive/exclusive bounds
self.clone()
// let mut reduced_constraints = HashMap::new();
// let constraints_set: HashSet<&FieldPath> = self.keys().collect();
// let index_constraints_set: HashSet<&FieldPath> =
// HashSet::from_iter(index_schema.get_fields().iter().clone());
// constraints_set
// .difference(&index_constraints_set)
// .for_each(|field_path| {
// match self.get(field_path.clone()) {
// None => panic!("missing field path"),
// Some(constraint) => {
// reduced_constraints.insert(field_path.clone().clone(), constraint.clone())
// }
// };
// });
// reduced_constraints
}
/// Check if a Document matches the constraints. If a Document is missing a Value,
/// it automatically doesn't match the Constraint.
fn matches_document(&self, doc: &Document) -> bool {
for (path, constraint) in self.iter() {
match doc.get(path) {
Some(value) => {
if !constraint.matches(&value) {
return false;
}
}
None => return false,
}
}
true
}
}
/// A single query constraint on a field.
///
/// Note that Constraints applied to an array
/// value will map the constraint over the array.
#[derive(Clone, Debug)]
pub enum Constraint {
/// Constraints on subdocuments (hashtables).
MatchesDocument(ConstraintDocument),
/// Equality constraint on a value.
Equals(Value),
/// Less-than constraint on a value.
LessThan(Value),
/// Greater-than constraint on a value.
GreaterThan(Value),
/// Constraint if value is in specified list of values.
In(Vec<Value>),
/// Disjunction of constraints on a single field.
Or(Box<Constraint>, Box<Constraint>),
/// Conjunction of constraints on a single field.
And(Box<Constraint>, Box<Constraint>),
}
impl Constraint {
/// Determine the type of a constraint.
/// Return None if the Constraint is invalid due to mismatched subconstraints.
pub fn get_value_type(&self) -> Option<Value> {
match self {
Self::Equals(value) | Self::LessThan(value) | Self::GreaterThan(value) => {
Some(value.clone())
}
Self::Or(constraint1, constraint2) | Self::And(constraint1, constraint2) => {
match (constraint1.get_value_type(), constraint2.get_value_type()) {
(None, _) | (_, None) => None,
(Some(value1), Some(value2)) => {
if value1.is_variant_equal(&value2) {
Some(value1)
} else {
None
}
}
}
}
Self::In(values) => {
for i in 1..values.len() {
if !values[i - 1].is_variant_equal(&values[i]) {
return None;
}
}
return Some(values[0].clone());
}
_ => panic!("currently unsupported constraint type"),
}
}
/// Whether a Value matches a constraint.
pub fn matches(&self, value: &Value) -> bool {
match self {
Constraint::MatchesDocument(constraint_doc) => match value {
Value::Dict(doc) => constraint_doc.matches_document(doc),
_ => false,
},
Constraint::Equals(value2) => value == value2,
Constraint::LessThan(value2) => value < value2,
Constraint::GreaterThan(value2) => value > value2,
Constraint::And(constraint1, constraint2) => {
constraint1.matches(value) && constraint2.matches(value)
}
Constraint::Or(constraint1, constraint2) => {
constraint1.matches(value) || constraint2.matches(value)
}
Constraint::In(values) => values.iter().any(|value2| value == value2),
}
}
/// Generate the value range(s) for this constraint.
///
/// Note: for the time being, this generates inclusive ranges only, because
/// we cannot mix inclusive and exclusive ranges in a multi-field index.
/// Note that since we also cannot mix unbounded ranges, one-sided ranges
/// (such as inequalities) use a predefined min/max value. For unbounded numbers,
/// this limits the range of valid numbers that can be returned with a range.
/// This can be considered a "feature" of our implementation due to the limitations
/// of using the builtin BTreeMap implementation. An alternative implementation
/// is to create a special value type that represents the extrema of a type
/// (which would probably be a cleaner and more robust solution if only I had
/// thought of it earlier).
///
/// Note: disjunction (OR) operator assumes ranges are non-overlapping.
pub fn generate_value_ranges(&self) -> Vec<(Bound<Value>, Bound<Value>)> {
match self {
Constraint::Equals(value) => vec![(
Bound::Included(value.clone()),
Bound::Included(value.clone()),
)],
Constraint::LessThan(value) => vec![(
Bound::Included(value.get_min_value()),
Bound::Included(value.clone()),
)],
Constraint::GreaterThan(value) => vec![(
Bound::Included(value.clone()),
Bound::Included(value.get_max_value()),
)],
// Conjunction: Combines each pair of ranges
Constraint::And(constraint1, constraint2) => {
let value_ranges1 = constraint1.generate_value_ranges();
let value_ranges2 = constraint2.generate_value_ranges();
let mut value_ranges = Vec::new();
for (value_range1_min, value_range1_max) in &value_ranges1 {
for (value_range2_min, value_range2_max) in &value_ranges2 {
// Assert that range bounds are inclusive.
let (min1, max1, min2, max2) = match (
value_range1_min,
value_range1_max,
value_range2_min,
value_range2_max,
) {
(
Bound::Included(min1),
Bound::Included(max1),
Bound::Included(min2),
Bound::Included(max2),
) => (min1, max1, min2, max2),
_ => panic!("non-inclusive bounds"),
};
// Combine range if overlapping
let (range_min, range_max) =
(std::cmp::max(min1, min2), std::cmp::min(max1, max2));
if range_max > range_min {
value_ranges.push((
Bound::Included(range_min.clone()),
Bound::Included(range_max.clone()),
));
}
}
}
value_ranges
}
// Disjunction: Returns disjoint ranges
Constraint::Or(constraint1, constraint2) => {
let mut value_ranges = constraint1.generate_value_ranges();
value_ranges.append(&mut constraint2.generate_value_ranges());
// Remove overlapping ranges. Sort ranges by higher element,
// iterate through this and determine overlapping ranges.
value_ranges.sort_by(|vr1, vr2| match (vr1, vr2) {
((_, Bound::Included(higher1)), (_, Bound::Included(higher2))) => {
higher1.cmp(higher2)
}
_ => panic!("non-inclusive bounds"),
});
// If any of the ranges from the left overlap any of the
// ranges from the right, then join them into a single range.
let mut value_ranges_nonoverlap = Vec::new();
let mut current_interval = value_ranges[0].clone();
for i in 1..value_ranges.len() {
match (current_interval.clone(), value_ranges[i].clone()) {
(
(_, Bound::Included(higher1)),
(Bound::Included(lower2), Bound::Included(higher2)),
) => {
if lower2 <= higher1 {
// Join intervals
current_interval.1 = Bound::Included(higher2);
} else {
// Intervals split
value_ranges_nonoverlap.push(current_interval);
current_interval =
(Bound::Included(lower2), Bound::Included(higher2));
}
}
_ => panic!("non-inclusive bounds"),
}
}
value_ranges_nonoverlap.push(current_interval);
value_ranges_nonoverlap
}
_ => panic!("unsupported range type"),
}
}
}
/// Projection of fields during a query (analogous to SQL `SELECT`).
pub type ProjectionDocument = HashMap<FieldPath, Projection>;
/// Projection of a single field of the projection document.
pub enum Projection {
// Recursive projections on subdocuments
ProjectDocument(ProjectionDocument),
// Exclude projecting a value
Exclude,
// Project a value
Include,
}
/// Complete query operation.
pub struct Query {
// Constraint document (`WHERE`)
pub constraints: ConstraintDocument,
// Projection document (`SELECT`)
pub projection: ProjectionDocument,
// Ordering document (`ORDER BY`)
pub order: Option<Vec<ResultOrder>>,
}