use crate::sparql::model::Variable; use crate::storage::numeric_encoder::EncodedTerm; use spargebra::algebra::GraphPattern; use std::collections::BTreeSet; use std::rc::Rc; #[derive(Eq, PartialEq, Debug, Clone, Hash)] pub enum PlanNode { Init, StaticBindings { tuples: Vec, }, Service { service_name: PatternValue, variables: Rc>, child: Rc, graph_pattern: Rc, silent: bool, }, QuadPatternJoin { child: Rc, subject: PatternValue, predicate: PatternValue, object: PatternValue, graph_name: PatternValue, }, PathPatternJoin { child: Rc, subject: PatternValue, path: Rc, object: PatternValue, graph_name: PatternValue, }, Join { left: Rc, right: Rc, }, AntiJoin { left: Rc, right: Rc, }, Filter { child: Rc, expression: Rc, }, Union { children: Vec>, }, LeftJoin { left: Rc, right: Rc, possible_problem_vars: Rc>, //Variables that should not be part of the entry of the left join }, Extend { child: Rc, position: usize, expression: Rc, }, Sort { child: Rc, by: Vec, }, HashDeduplicate { child: Rc, }, Skip { child: Rc, count: usize, }, Limit { child: Rc, count: usize, }, Project { child: Rc, mapping: Rc>, // 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: Rc, key_mapping: Rc>, // aggregate key pairs of (variable key in child, variable key in output) aggregates: Rc>, }, } 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 } pub 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, expression } => { expression.add_maybe_bound_variables(set); child.add_maybe_bound_variables(set); } PlanNode::Union { children } => { for child in children.iter() { 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, expression, } => { set.insert(*position); expression.add_maybe_bound_variables(set); child.add_maybe_bound_variables(set); } PlanNode::Service { child, .. } | PlanNode::Sort { child, .. } | PlanNode::HashDeduplicate { child } | PlanNode::Skip { child, .. } | 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.iter() { set.insert(*var); } } } } } #[derive(Eq, PartialEq, Debug, Clone, Hash)] pub enum PatternValue { Constant(EncodedTerm), Variable(usize), Triple(Box), } #[derive(Eq, PartialEq, Debug, Clone, Hash)] pub struct TriplePatternValue { pub subject: PatternValue, pub predicate: PatternValue, pub object: PatternValue, } #[derive(Eq, PartialEq, Debug, Clone, Hash)] pub enum PlanExpression { Constant(EncodedTerm), Variable(usize), Exists(Rc), Or(Box, Box), And(Box, Box), Equal(Box, Box), Greater(Box, Box), GreaterOrEqual(Box, Box), Less(Box, Box), LessOrEqual(Box, Box), In(Box, Vec), Add(Box, Box), Subtract(Box, Box), Multiply(Box, Box), Divide(Box, Box), UnaryPlus(Box), UnaryMinus(Box), Not(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>, ), Triple( Box, Box, Box, ), Subject(Box), Predicate(Box), Object(Box), IsTriple(Box), BooleanCast(Box), DoubleCast(Box), FloatCast(Box), DecimalCast(Box), IntegerCast(Box), DateCast(Box), TimeCast(Box), DateTimeCast(Box), DurationCast(Box), YearMonthDurationCast(Box), DayTimeDurationCast(Box), StringCast(Box), } impl PlanExpression { pub fn add_maybe_bound_variables(&self, set: &mut BTreeSet) { match self { PlanExpression::Variable(v) | PlanExpression::Bound(v) => { set.insert(*v); } PlanExpression::Constant(_) | PlanExpression::Rand | PlanExpression::Now | PlanExpression::Uuid | PlanExpression::StrUuid | PlanExpression::BNode(None) => (), PlanExpression::UnaryPlus(e) | PlanExpression::UnaryMinus(e) | PlanExpression::Not(e) | PlanExpression::BNode(Some(e)) | PlanExpression::Str(e) | PlanExpression::Lang(e) | PlanExpression::Datatype(e) | PlanExpression::Iri(e) | PlanExpression::Abs(e) | PlanExpression::Ceil(e) | PlanExpression::Floor(e) | PlanExpression::Round(e) | PlanExpression::UCase(e) | PlanExpression::LCase(e) | PlanExpression::StrLen(e) | PlanExpression::EncodeForUri(e) | PlanExpression::Year(e) | PlanExpression::Month(e) | PlanExpression::Day(e) | PlanExpression::Hours(e) | PlanExpression::Minutes(e) | PlanExpression::Seconds(e) | PlanExpression::Timezone(e) | PlanExpression::Tz(e) | PlanExpression::Md5(e) | PlanExpression::Sha1(e) | PlanExpression::Sha256(e) | PlanExpression::Sha384(e) | PlanExpression::Sha512(e) | PlanExpression::IsIri(e) | PlanExpression::IsBlank(e) | PlanExpression::IsLiteral(e) | PlanExpression::IsNumeric(e) | PlanExpression::IsTriple(e) | PlanExpression::Subject(e) | PlanExpression::Predicate(e) | PlanExpression::Object(e) | PlanExpression::BooleanCast(e) | PlanExpression::DoubleCast(e) | PlanExpression::FloatCast(e) | PlanExpression::DecimalCast(e) | PlanExpression::IntegerCast(e) | PlanExpression::DateCast(e) | PlanExpression::TimeCast(e) | PlanExpression::DateTimeCast(e) | PlanExpression::DurationCast(e) | PlanExpression::YearMonthDurationCast(e) | PlanExpression::DayTimeDurationCast(e) | PlanExpression::StringCast(e) => e.add_maybe_bound_variables(set), PlanExpression::Or(a, b) | PlanExpression::And(a, b) | PlanExpression::Equal(a, b) | PlanExpression::Greater(a, b) | PlanExpression::GreaterOrEqual(a, b) | PlanExpression::Less(a, b) | PlanExpression::LessOrEqual(a, b) | PlanExpression::Add(a, b) | PlanExpression::Subtract(a, b) | PlanExpression::Multiply(a, b) | PlanExpression::Divide(a, b) | PlanExpression::LangMatches(a, b) | PlanExpression::Contains(a, b) | PlanExpression::StrStarts(a, b) | PlanExpression::StrEnds(a, b) | PlanExpression::StrBefore(a, b) | PlanExpression::StrAfter(a, b) | PlanExpression::StrLang(a, b) | PlanExpression::StrDt(a, b) | PlanExpression::SameTerm(a, b) | PlanExpression::SubStr(a, b, None) | PlanExpression::Regex(a, b, None) => { a.add_maybe_bound_variables(set); b.add_maybe_bound_variables(set); } PlanExpression::If(a, b, c) | PlanExpression::SubStr(a, b, Some(c)) | PlanExpression::Regex(a, b, Some(c)) | PlanExpression::Replace(a, b, c, None) | PlanExpression::Triple(a, b, c) => { a.add_maybe_bound_variables(set); b.add_maybe_bound_variables(set); c.add_maybe_bound_variables(set); } PlanExpression::Replace(a, b, c, Some(d)) => { a.add_maybe_bound_variables(set); b.add_maybe_bound_variables(set); c.add_maybe_bound_variables(set); d.add_maybe_bound_variables(set); } PlanExpression::Concat(es) | PlanExpression::Coalesce(es) => { for e in es { e.add_maybe_bound_variables(set); } } PlanExpression::In(a, bs) => { a.add_maybe_bound_variables(set); for b in bs { b.add_maybe_bound_variables(set); } } PlanExpression::Exists(e) => { e.add_maybe_bound_variables(set); } } } } #[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: Rc }, } #[derive(Eq, PartialEq, Debug, Clone, Hash)] pub enum PlanPropertyPath { Path(EncodedTerm), Reverse(Rc), Sequence(Rc, Rc), Alternative(Rc, Rc), ZeroOrMore(Rc), OneOrMore(Rc), ZeroOrOne(Rc), NegatedPropertySet(Rc>), } #[derive(Eq, PartialEq, Debug, Clone, Hash)] pub enum Comparator { Asc(PlanExpression), Desc(PlanExpression), } #[derive(Eq, PartialEq, Debug, Clone, Hash)] pub struct TripleTemplate { pub subject: TripleTemplateValue, pub predicate: TripleTemplateValue, pub object: TripleTemplateValue, } #[derive(Eq, PartialEq, Debug, Clone, Hash)] pub enum TripleTemplateValue { Constant(EncodedTerm), BlankNode(usize), Variable(usize), Triple(Box), } #[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, Option::is_some) } pub fn get(&self, index: usize) -> Option<&EncodedTerm> { self.inner.get(index).unwrap_or(&None).as_ref() } pub fn iter(&self) -> impl Iterator> + '_ { 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.clone()), } } } 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.clone()), } } } Some(EncodedTuple { inner: result }) } } } impl IntoIterator for EncodedTuple { type Item = Option; type IntoIter = std::vec::IntoIter>; fn into_iter(self) -> Self::IntoIter { self.inner.into_iter() } }