use crate::sparql::model::Variable; use crate::sparql::GraphPattern; use crate::store::numeric_encoder::{ EncodedQuad, EncodedTerm, Encoder, MemoryStrStore, StrContainer, StrLookup, ENCODED_DEFAULT_GRAPH, }; use crate::store::StoreConnection; use crate::Result; use std::cell::{RefCell, RefMut}; use std::collections::BTreeSet; #[derive(Eq, PartialEq, Debug, Clone, Hash)] pub enum PlanNode { Init, StaticBindings { tuples: Vec, }, Service { service_name: PatternValue, variables: Vec, child: Box, graph_pattern: Box, silent: bool, }, QuadPatternJoin { child: Box, subject: PatternValue, predicate: PatternValue, object: PatternValue, graph_name: PatternValue, }, PathPatternJoin { child: Box, subject: PatternValue, path: PlanPropertyPath, object: PatternValue, graph_name: PatternValue, }, Join { left: Box, right: Box, }, AntiJoin { left: Box, right: Box, }, Filter { child: Box, expression: PlanExpression, }, Union { children: Vec, }, LeftJoin { left: Box, right: Box, possible_problem_vars: Vec, //Variables that should not be part of the entry of the left join }, Extend { child: Box, position: usize, expression: PlanExpression, }, Sort { child: Box, by: Vec, }, HashDeduplicate { child: Box, }, Skip { child: Box, count: usize, }, Limit { child: Box, count: usize, }, Project { child: Box, mapping: Vec<(usize, usize)>, // pairs of (variable key in child, variable key in output) }, Aggregate { // By definition the group by key are the range 0..key_mapping.len() child: Box, key_mapping: Vec<(usize, usize)>, // aggregate key pairs of (variable key in child, variable key in output) aggregates: Vec<(PlanAggregation, usize)>, }, } impl PlanNode { /// Returns variables that might be bound in the result set pub fn maybe_bound_variables(&self) -> BTreeSet { let mut set = BTreeSet::default(); self.add_maybe_bound_variables(&mut set); set } fn add_maybe_bound_variables(&self, set: &mut BTreeSet) { match self { PlanNode::Init => (), PlanNode::StaticBindings { tuples } => { for tuple in tuples { for (key, value) in tuple.iter().enumerate() { if value.is_some() { set.insert(key); } } } } PlanNode::QuadPatternJoin { child, subject, predicate, object, graph_name, } => { if let PatternValue::Variable(var) = subject { set.insert(*var); } if let PatternValue::Variable(var) = predicate { set.insert(*var); } if let PatternValue::Variable(var) = object { set.insert(*var); } if let PatternValue::Variable(var) = graph_name { set.insert(*var); } child.add_maybe_bound_variables(set); } PlanNode::PathPatternJoin { child, subject, object, graph_name, .. } => { if let PatternValue::Variable(var) = subject { set.insert(*var); } if let PatternValue::Variable(var) = object { set.insert(*var); } if let PatternValue::Variable(var) = graph_name { set.insert(*var); } child.add_maybe_bound_variables(set); } PlanNode::Filter { child, .. } => child.add_maybe_bound_variables(set), PlanNode::Union { children } => { for child in children { child.add_maybe_bound_variables(set); } } PlanNode::Join { left, right, .. } | PlanNode::AntiJoin { left, right, .. } | PlanNode::LeftJoin { left, right, .. } => { left.add_maybe_bound_variables(set); right.add_maybe_bound_variables(set); } PlanNode::Extend { child, position, .. } => { set.insert(*position); child.add_maybe_bound_variables(set); } PlanNode::Service { child, .. } => child.add_maybe_bound_variables(set), PlanNode::Sort { child, .. } => child.add_maybe_bound_variables(set), PlanNode::HashDeduplicate { child } => child.add_maybe_bound_variables(set), PlanNode::Skip { child, .. } => child.add_maybe_bound_variables(set), PlanNode::Limit { child, .. } => child.add_maybe_bound_variables(set), PlanNode::Project { mapping, child } => { let child_bound = child.maybe_bound_variables(); for (child_i, output_i) in mapping.iter() { if child_bound.contains(child_i) { set.insert(*output_i); } } } PlanNode::Aggregate { key_mapping, aggregates, .. } => { set.extend(key_mapping.iter().map(|(_, o)| o)); for (_, var) in aggregates { set.insert(*var); } } } } } #[derive(Eq, PartialEq, Debug, Clone, Copy, Hash)] pub enum PatternValue { Constant(EncodedTerm), Variable(usize), } impl PatternValue { pub fn is_var(&self) -> bool { match self { PatternValue::Constant(_) => false, PatternValue::Variable(_) => true, } } } #[derive(Eq, PartialEq, Debug, Clone, Hash)] pub enum PlanExpression { Constant(EncodedTerm), Variable(usize), Exists(Box), Or(Box, Box), And(Box, Box), Equal(Box, Box), NotEqual(Box, Box), Greater(Box, Box), GreaterOrEq(Box, Box), Lower(Box, Box), LowerOrEq(Box, Box), In(Box, Vec), Add(Box, Box), Sub(Box, Box), Mul(Box, Box), Div(Box, Box), UnaryPlus(Box), UnaryMinus(Box), UnaryNot(Box), Str(Box), Lang(Box), LangMatches(Box, Box), Datatype(Box), Bound(usize), IRI(Box), BNode(Option>), Rand, Abs(Box), Ceil(Box), Floor(Box), Round(Box), Concat(Vec), SubStr( Box, Box, Option>, ), StrLen(Box), Replace( Box, Box, Box, Option>, ), UCase(Box), LCase(Box), EncodeForURI(Box), Contains(Box, Box), StrStarts(Box, Box), StrEnds(Box, Box), StrBefore(Box, Box), StrAfter(Box, Box), Year(Box), Month(Box), Day(Box), Hours(Box), Minutes(Box), Seconds(Box), Timezone(Box), Tz(Box), Now, UUID, StrUUID, MD5(Box), SHA1(Box), SHA256(Box), SHA384(Box), SHA512(Box), Coalesce(Vec), If( Box, Box, Box, ), StrLang(Box, Box), StrDT(Box, Box), SameTerm(Box, Box), IsIRI(Box), IsBlank(Box), IsLiteral(Box), IsNumeric(Box), Regex( Box, Box, Option>, ), BooleanCast(Box), DoubleCast(Box), FloatCast(Box), DecimalCast(Box), IntegerCast(Box), DateCast(Box), TimeCast(Box), DateTimeCast(Box), DurationCast(Box), StringCast(Box), } #[derive(Eq, PartialEq, Debug, Clone, Hash)] pub struct PlanAggregation { pub function: PlanAggregationFunction, pub parameter: Option, pub distinct: bool, } #[derive(Eq, PartialEq, Debug, Clone, Hash)] pub enum PlanAggregationFunction { Count, Sum, Min, Max, Avg, Sample, GroupConcat { separator: String }, } #[derive(Eq, PartialEq, Debug, Clone, Hash)] pub enum PlanPropertyPath { PredicatePath(EncodedTerm), InversePath(Box), SequencePath(Box, Box), AlternativePath(Box, Box), ZeroOrMorePath(Box), OneOrMorePath(Box), ZeroOrOnePath(Box), NegatedPropertySet(Vec), } #[derive(Eq, PartialEq, Debug, Clone, Hash)] pub enum Comparator { Asc(PlanExpression), Desc(PlanExpression), } #[derive(Eq, PartialEq, Debug, Clone, Copy, Hash)] pub struct TripleTemplate { pub subject: TripleTemplateValue, pub predicate: TripleTemplateValue, pub object: TripleTemplateValue, } #[derive(Eq, PartialEq, Debug, Clone, Copy, Hash)] pub enum TripleTemplateValue { Constant(EncodedTerm), BlankNode(usize), Variable(usize), } #[derive(Eq, PartialEq, Debug, Clone, Hash)] pub struct EncodedTuple { inner: Vec>, } impl EncodedTuple { pub fn with_capacity(capacity: usize) -> Self { Self { inner: Vec::with_capacity(capacity), } } pub fn capacity(&self) -> usize { self.inner.capacity() } pub fn contains(&self, index: usize) -> bool { self.inner.get(index).map_or(false, |v| v.is_some()) } pub fn get(&self, index: usize) -> Option { self.inner.get(index).cloned().unwrap_or(None) } pub fn iter<'a>(&'a self) -> impl Iterator> + 'a { self.inner.iter().cloned() } pub fn set(&mut self, index: usize, value: EncodedTerm) { if self.inner.len() <= index { self.inner.resize(index + 1, None); } self.inner[index] = Some(value); } pub fn unset(&mut self, index: usize) { if let Some(v) = self.inner.get_mut(index) { *v = None; } } pub fn combine_with(&self, other: &EncodedTuple) -> Option { if self.inner.len() < other.inner.len() { let mut result = other.inner.to_owned(); for (key, self_value) in self.inner.iter().enumerate() { if let Some(self_value) = self_value { match other.inner[key] { Some(ref other_value) => { if self_value != other_value { return None; } } None => result[key] = Some(*self_value), } } } Some(EncodedTuple { inner: result }) } else { let mut result = self.inner.to_owned(); for (key, other_value) in other.inner.iter().enumerate() { if let Some(other_value) = other_value { match self.inner[key] { Some(ref self_value) => { if self_value != other_value { return None; } } None => result[key] = Some(*other_value), } } } Some(EncodedTuple { inner: result }) } } } pub struct DatasetView { store: S, extra: RefCell, default_graph_as_union: bool, } impl DatasetView { pub fn new(store: S, default_graph_as_union: bool) -> Self { Self { store, extra: RefCell::new(MemoryStrStore::default()), default_graph_as_union, } } pub fn quads_for_pattern<'a>( &'a self, subject: Option, predicate: Option, object: Option, graph_name: Option, ) -> Box> + 'a> { if graph_name == None { Box::new( self.store .quads_for_pattern(subject, predicate, object, None) .filter(|quad| match quad { Err(_) => true, Ok(quad) => quad.graph_name != ENCODED_DEFAULT_GRAPH, }), ) } else if graph_name == Some(ENCODED_DEFAULT_GRAPH) && self.default_graph_as_union { Box::new( self.store .quads_for_pattern(subject, predicate, object, None) .map(|quad| { let quad = quad?; Ok(EncodedQuad::new( quad.subject, quad.predicate, quad.object, ENCODED_DEFAULT_GRAPH, )) }), ) } else { self.store .quads_for_pattern(subject, predicate, object, graph_name) } } pub fn encoder<'a>(&'a self) -> impl Encoder + StrContainer + 'a { DatasetViewStrContainer { store: &self.store, extra: self.extra.borrow_mut(), } } } impl StrLookup for DatasetView { fn get_str(&self, id: u128) -> Result> { if let Some(value) = self.extra.borrow().get_str(id)? { Ok(Some(value)) } else { self.store.get_str(id) } } } struct DatasetViewStrContainer<'a, S: StoreConnection> { store: &'a S, extra: RefMut<'a, MemoryStrStore>, } impl<'a, S: StoreConnection> StrContainer for DatasetViewStrContainer<'a, S> { fn insert_str(&mut self, key: u128, value: &str) -> Result<()> { if self.store.get_str(key)?.is_none() { self.extra.insert_str(key, value) } else { Ok(()) } } }