Evaluates more SPARQL built-in functions

lib/Cargo.toml

@@ -27,6 +27,7 @@ ordered-float = "1"
 num-traits = "0.2"
 rust_decimal = "0.10"
 chrono = "0.4"
+language-tags = "0.2"
 
 [build-dependencies]
 peg = "0.5"

lib/src/lib.rs

@@ -37,6 +37,7 @@ extern crate error_chain;
 #[macro_use]
 extern crate lazy_static;
 extern crate chrono;
+extern crate language_tags;
 extern crate num_traits;
 extern crate ordered_float;
 extern crate quick_xml;

lib/src/sparql/eval.rs

@@ -1,5 +1,7 @@
 use chrono::DateTime;
 use chrono::NaiveDateTime;
+use language_tags::LanguageTag;
+use model::BlankNode;
 use num_traits::identities::Zero;
 use num_traits::FromPrimitive;
 use num_traits::One;
@@ -8,10 +10,12 @@ use ordered_float::OrderedFloat;
 use rust_decimal::Decimal;
 use sparql::algebra::*;
 use sparql::plan::*;
+use std::cmp::Ordering;
 use std::collections::HashSet;
 use std::iter::once;
 use std::iter::Iterator;
 use std::str;
+use std::str::FromStr;
 use std::sync::Arc;
 use store::encoded::EncodedQuadsStore;
 use store::numeric_encoder::*;
@@ -277,23 +281,47 @@ impl<S: EncodedQuadsStore> SimpleEvaluator<S> {
                 },
             },
             PlanExpression::Equal(a, b) => {
-                Some((self.eval_expression(a, tuple)? == self.eval_expression(b, tuple)?).into())
+                let a = self.eval_expression(a, tuple)?;
+                let b = self.eval_expression(b, tuple)?;
+                Some((a == b || self.partial_cmp_terms(a, b) == Some(Ordering::Equal)).into())
             }
             PlanExpression::NotEqual(a, b) => {
-                Some((self.eval_expression(a, tuple)? != self.eval_expression(b, tuple)?).into())
+                let a = self.eval_expression(a, tuple)?;
-            }
+                let b = self.eval_expression(b, tuple)?;
-            PlanExpression::Greater(a, b) => {
+                Some((a != b && self.partial_cmp_terms(a, b) != Some(Ordering::Equal)).into())
-                Some((self.eval_expression(a, tuple)? > self.eval_expression(b, tuple)?).into())
+            }
-            }
+            PlanExpression::Greater(a, b) => Some(
-            PlanExpression::GreaterOrEq(a, b) => {
+                (self.partial_cmp_terms(
-                Some((self.eval_expression(a, tuple)? >= self.eval_expression(b, tuple)?).into())
+                    self.eval_expression(a, tuple)?,
-            }
+                    self.eval_expression(b, tuple)?,
-            PlanExpression::Lower(a, b) => {
+                )? == Ordering::Greater)
-                Some((self.eval_expression(a, tuple)? < self.eval_expression(b, tuple)?).into())
+                    .into(),
-            }
+            ),
-            PlanExpression::LowerOrEq(a, b) => {
+            PlanExpression::GreaterOrEq(a, b) => Some(
-                Some((self.eval_expression(a, tuple)? <= self.eval_expression(b, tuple)?).into())
+                match self.partial_cmp_terms(
-            }
+                    self.eval_expression(a, tuple)?,
+                    self.eval_expression(b, tuple)?,
+                )? {
+                    Ordering::Greater | Ordering::Equal => true,
+                    _ => false,
+                }.into(),
+            ),
+            PlanExpression::Lower(a, b) => Some(
+                (self.partial_cmp_terms(
+                    self.eval_expression(a, tuple)?,
+                    self.eval_expression(b, tuple)?,
+                )? == Ordering::Less)
+                    .into(),
+            ),
+            PlanExpression::LowerOrEq(a, b) => Some(
+                match self.partial_cmp_terms(
+                    self.eval_expression(a, tuple)?,
+                    self.eval_expression(b, tuple)?,
+                )? {
+                    Ordering::Less | Ordering::Equal => true,
+                    _ => false,
+                }.into(),
+            ),
             PlanExpression::Add(a, b) => Some(match self.parse_numeric_operands(a, b, tuple)? {
                 NumericBinaryOperands::Float(v1, v2) => (v1 + v2).into(),
                 NumericBinaryOperands::Double(v1, v2) => (v1 + v2).into(),
@@ -344,6 +372,7 @@ impl<S: EncodedQuadsStore> SimpleEvaluator<S> {
                         value_id: language_id,
                     })
                 }
+                e if e.is_literal() => Some(ENCODED_EMPTY_SIMPLE_LITERAL),
                 _ => None,
             },
             PlanExpression::Datatype(e) => self.eval_expression(e, tuple)?.datatype(),
@@ -356,6 +385,10 @@ impl<S: EncodedQuadsStore> SimpleEvaluator<S> {
                 }
                 _ => None,
             },
+            PlanExpression::BNode(id) => match id {
+                Some(id) => unimplemented!(),
+                None => Some(BlankNode::default().into()),
+            },
             PlanExpression::SameTerm(a, b) => {
                 Some((self.eval_expression(a, tuple)? == self.eval_expression(b, tuple)?).into())
             }
@@ -368,6 +401,31 @@ impl<S: EncodedQuadsStore> SimpleEvaluator<S> {
             PlanExpression::IsLiteral(e) => {
                 Some(self.eval_expression(e, tuple)?.is_literal().into())
             }
+            PlanExpression::IsNumeric(e) => Some(
+                match self.eval_expression(e, tuple)? {
+                    EncodedTerm::FloatLiteral(_)
+                    | EncodedTerm::DoubleLiteral(_)
+                    | EncodedTerm::IntegerLiteral(_)
+                    | EncodedTerm::DecimalLiteral(_) => true,
+                    _ => false,
+                }.into(),
+            ),
+            PlanExpression::LangMatches(language_tag, language_range) => {
+                let language_tag =
+                    self.to_simple_string(self.eval_expression(language_tag, tuple)?)?;
+                let language_range =
+                    self.to_simple_string(self.eval_expression(language_range, tuple)?)?;
+                Some(
+                    if language_range == "*" {
+                        !language_tag.is_empty()
+                    } else {
+                        LanguageTag::from_str(&language_range)
+                            .ok()?
+                            .matches(&LanguageTag::from_str(&language_tag).ok()?)
+                    }.into(),
+                )
+            }
+            PlanExpression::Regex(text, pattern, flags) => unimplemented!(),
             PlanExpression::BooleanCast(e) => match self.eval_expression(e, tuple)? {
                 EncodedTerm::BooleanLiteral(value) => Some(value.into()),
                 EncodedTerm::SimpleLiteral { value_id }
@@ -473,7 +531,6 @@ impl<S: EncodedQuadsStore> SimpleEvaluator<S> {
             PlanExpression::StringCast(e) => Some(EncodedTerm::StringLiteral {
                 value_id: self.to_string_id(self.eval_expression(e, tuple)?)?,
             }),
-            e => unimplemented!(),
         }
     }
 
@@ -517,6 +574,18 @@ impl<S: EncodedQuadsStore> SimpleEvaluator<S> {
         }
     }
 
+    fn to_simple_string(&self, term: EncodedTerm) -> Option<String> {
+        if let EncodedTerm::SimpleLiteral { value_id } = term {
+            Some(
+                str::from_utf8(&self.store.get_bytes(value_id).ok()??)
+                    .ok()?
+                    .to_owned(),
+            )
+        } else {
+            None
+        }
+    }
+
     fn parse_numeric_operands(
         &self,
         e1: &PlanExpression,
@@ -600,6 +669,46 @@ impl<S: EncodedQuadsStore> SimpleEvaluator<S> {
             })),
         )
     }
+
+    fn partial_cmp_terms(&self, a: EncodedTerm, b: EncodedTerm) -> Option<Ordering> {
+        match a {
+            EncodedTerm::SimpleLiteral { value_id: a }
+            | EncodedTerm::StringLiteral { value_id: a } => match b {
+                EncodedTerm::SimpleLiteral { value_id: b }
+                | EncodedTerm::StringLiteral { value_id: b } => a.partial_cmp(&b),
+                _ => None,
+            },
+            EncodedTerm::FloatLiteral(a) => match b {
+                EncodedTerm::FloatLiteral(b) => (*a).partial_cmp(&*b),
+                EncodedTerm::DoubleLiteral(b) => a.to_f64()?.partial_cmp(&*b),
+                EncodedTerm::IntegerLiteral(b) => (*a).partial_cmp(&b.to_f32()?),
+                EncodedTerm::DecimalLiteral(b) => (*a).partial_cmp(&b.to_f32()?),
+                _ => None,
+            },
+            EncodedTerm::DoubleLiteral(a) => match b {
+                EncodedTerm::FloatLiteral(b) => (*a).partial_cmp(&b.to_f64()?),
+                EncodedTerm::DoubleLiteral(b) => (*a).partial_cmp(&*b),
+                EncodedTerm::IntegerLiteral(b) => (*a).partial_cmp(&b.to_f64()?),
+                EncodedTerm::DecimalLiteral(b) => (*a).partial_cmp(&b.to_f64()?),
+                _ => None,
+            },
+            EncodedTerm::IntegerLiteral(a) => match b {
+                EncodedTerm::FloatLiteral(b) => a.to_f32()?.partial_cmp(&*b),
+                EncodedTerm::DoubleLiteral(b) => a.to_f64()?.partial_cmp(&*b),
+                EncodedTerm::IntegerLiteral(b) => a.partial_cmp(&b),
+                EncodedTerm::DecimalLiteral(b) => Decimal::from_i128(a)?.partial_cmp(&b),
+                _ => None,
+            },
+            EncodedTerm::DecimalLiteral(a) => match b {
+                EncodedTerm::FloatLiteral(b) => a.to_f32()?.partial_cmp(&*b),
+                EncodedTerm::DoubleLiteral(b) => a.to_f64()?.partial_cmp(&*b),
+                EncodedTerm::IntegerLiteral(b) => a.partial_cmp(&Decimal::from_i128(b)?),
+                EncodedTerm::DecimalLiteral(b) => a.partial_cmp(&b),
+                _ => None,
+            },
+            _ => None,
+        }
+    }
 }
 
 enum NumericBinaryOperands {

lib/src/store/numeric_encoder.rs

@@ -234,6 +234,12 @@ impl From<NaiveDateTime> for EncodedTerm {
     }
 }
 
+impl From<BlankNode> for EncodedTerm {
+    fn from(node: BlankNode) -> Self {
+        EncodedTerm::BlankNode(*node.as_uuid())
+    }
+}
+
 #[derive(Eq, PartialEq, Ord, PartialOrd, Debug, Clone, Hash)]
 pub struct EncodedQuad {
     pub subject: EncodedTerm,

lib/tests/sparql_test_cases.rs

@@ -96,6 +96,12 @@ fn sparql_w3c_query_evaluation_testsuite() {
         Url::parse("http://www.w3.org/2001/sw/DataAccess/tests/data-r2/cast/manifest.ttl").unwrap(),
         Url::parse("http://www.w3.org/2001/sw/DataAccess/tests/data-r2/distinct/manifest.ttl")
             .unwrap(),
+        Url::parse("http://www.w3.org/2001/sw/DataAccess/tests/data-r2/expr-builtin/manifest.ttl")
+            .unwrap(),
+        Url::parse("http://www.w3.org/2001/sw/DataAccess/tests/data-r2/expr-equals/manifest.ttl")
+            .unwrap(),
+        Url::parse("http://www.w3.org/2001/sw/DataAccess/tests/data-r2/expr-ops/manifest.ttl")
+            .unwrap(),
         Url::parse("http://www.w3.org/2001/sw/DataAccess/tests/data-r2/i18n/manifest.ttl").unwrap(),
         Url::parse(
             "http://www.w3.org/2001/sw/DataAccess/tests/data-r2/optional-filter/manifest.ttl",
@@ -113,6 +119,31 @@ fn sparql_w3c_query_evaluation_testsuite() {
         NamedNode::from_str(
             "http://www.w3.org/2001/sw/DataAccess/tests/data-r2/distinct/manifest#distinct-9",
         ).unwrap(),
+        NamedNode::from_str(
+            "http://www.w3.org/2001/sw/DataAccess/tests/data-r2/expr-builtin/manifest#dawg-str-1",
+        ).unwrap(),
+        NamedNode::from_str(
+            "http://www.w3.org/2001/sw/DataAccess/tests/data-r2/expr-builtin/manifest#dawg-str-2",
+        ).unwrap(),
+        NamedNode::from_str(
+            "http://www.w3.org/2001/sw/DataAccess/tests/data-r2/expr-equals/manifest#eq-graph-1",
+        ).unwrap(),
+        NamedNode::from_str(
+            "http://www.w3.org/2001/sw/DataAccess/tests/data-r2/expr-equals/manifest#eq-graph-2",
+        ).unwrap(),
+        //Multiple writing of the same xsd:double. Our system does strong normalization.
+        NamedNode::from_str(
+            "http://www.w3.org/2001/sw/DataAccess/tests/data-r2/expr-builtin/manifest#sameTerm",
+        ).unwrap(),
+        NamedNode::from_str(
+            "http://www.w3.org/2001/sw/DataAccess/tests/data-r2/expr-builtin/manifest#sameTerm-simple",
+        ).unwrap(),
+        NamedNode::from_str(
+            "http://www.w3.org/2001/sw/DataAccess/tests/data-r2/expr-builtin/manifest#sameTerm-eq",
+        ).unwrap(),
+        NamedNode::from_str(
+            "http://www.w3.org/2001/sw/DataAccess/tests/data-r2/expr-builtin/manifest#sameTerm-not-eq",
+        ).unwrap(),
         //URI normalization: we are normalizing more strongly
         NamedNode::from_str(
             "http://www.w3.org/2001/sw/DataAccess/tests/data-r2/i18n/manifest#normalization-3",
@@ -121,6 +152,27 @@ fn sparql_w3c_query_evaluation_testsuite() {
         NamedNode::from_str(
             "http://www.w3.org/2001/sw/DataAccess/tests/data-r2/optional-filter/manifest#dawg-optional-filter-005-not-simplified",
         ).unwrap(),
+        //Case insensitive language tag comparison
+        NamedNode::from_str(
+            "http://www.w3.org/2001/sw/DataAccess/tests/data-r2/expr-builtin/manifest#lang-case-insensitive-eq",
+        ).unwrap(),
+        NamedNode::from_str(
+            "http://www.w3.org/2001/sw/DataAccess/tests/data-r2/expr-builtin/manifest#lang-case-insensitive-ne",
+        ).unwrap(),
+        NamedNode::from_str(
+            "http://www.w3.org/2001/sw/DataAccess/tests/data-r2/expr-builtin/manifest#dawg-lang-3",
+        ).unwrap(),
+        //Difference in language matching
+        NamedNode::from_str(
+            "http://www.w3.org/2001/sw/DataAccess/tests/data-r2/expr-builtin/manifest#dawg-langMatches-basic",
+        ).unwrap(),
+        NamedNode::from_str(
+            "http://www.w3.org/2001/sw/DataAccess/tests/data-r2/expr-builtin/manifest#dawg-langMatches-basic",
+        ).unwrap(),
+        //DATATYPE("foo"@en) returns rdf:langString in SPARQL 1.1
+        NamedNode::from_str(
+            "http://www.w3.org/2001/sw/DataAccess/tests/data-r2/expr-builtin/manifest#dawg-datatype-2",
+        ).unwrap(),
     ];
     let client = RDFClient::default();
 
@@ -170,14 +222,14 @@ fn sparql_w3c_query_evaluation_testsuite() {
                         .load_sparql_query_result_graph(test.result.clone().unwrap())
                         .unwrap();
                     assert!(
-                            actual_graph.is_isomorphic(&expected_graph).unwrap(),
+                        actual_graph.is_isomorphic(&expected_graph).unwrap(),
-                            "Failure on {}.\nExpected file:\n{}\nOutput file:\n{}\nParsed query:\n{}\nData:\n{}\n",
+                        "Failure on {}.\nExpected file:\n{}\nOutput file:\n{}\nParsed query:\n{}\nData:\n{}\n",
-                            test,
+                        test,
-                            expected_graph,
+                        expected_graph,
-                            actual_graph,
+                        actual_graph,
-                            client.load_sparql_query(test.query.clone()).unwrap(),
+                        client.load_sparql_query(test.query.clone()).unwrap(),
-                            data
+                        data
-                        )
+                    )
                 }
             }
         } else {