NextGraph
/
oxigraph
1
0
Fork 0
Code Issues Pull Requests Packages Projects Releases Activity

Adds support of SPARQL DISTINCT and UNION evaluation

Browse Source 
Uses the Literal::new_simple_literal constructors where relevant
pull/10/head
Tpt 6 years ago
parent 5cff79c372
commit c1ea33b5aa
7 changed files with 154 additions and 33 deletions
Show all changes Ignore whitespace when comparing lines Ignore changes in amount of whitespace Ignore changes in whitespace at EOL
Show Stats Download Patch File Download Diff File
  1. 2
      src/rio/ntriples/ntriples_grammar.rustpeg
  2. 2
      src/rio/turtle/turtle_grammar.rustpeg
  3. 2
      src/sparql/sparql_grammar.rustpeg
  4. 38
      src/sparql/xml_results.rs
  5. 38
      src/store/numeric_encoder.rs
  6. 89
      src/store/sparql.rs
  7. 16
      tests/sparql_test_cases.rs

2
src/rio/ntriples/ntriples_grammar.rustpeg
Unescape View File

@ -33,7 +33,7 @@ object -> Term =
literal -> Literal =
v: STRING_LITERAL_QUOTE _ "^^" _ t:IRIREF { Literal::new_typed_literal(v, t) } /
v: STRING_LITERAL_QUOTE _ l:LANGTAG { Literal::new_language_tagged_literal(v, l) } /
v: STRING_LITERAL_QUOTE { v.into() }
v: STRING_LITERAL_QUOTE { Literal::new_simple_literal(v) }
//[144s]

2
src/rio/turtle/turtle_grammar.rustpeg
Unescape View File

@ -128,7 +128,7 @@ NumericLiteral -> Literal =
RDFLiteral -> Literal =
v:String _ "^^" _ t:iri { Literal::new_typed_literal(v, t) } /
v:String _ l:LANGTAG { Literal::new_language_tagged_literal(v, l) } /
v:String { v.into() }
v:String { Literal::new_simple_literal(v) }
//[133s]
BooleanLiteral -> Literal =

2
src/sparql/sparql_grammar.rustpeg
Unescape View File

@ -864,7 +864,7 @@ iriOrFunction -> Expression = i: iri _ a: ArgList? {
RDFLiteral -> Literal =
v:String _ "^^" _ t:iri { Literal::new_typed_literal(v, t) } /
v:String _ l:LANGTAG { Literal::new_language_tagged_literal(v, l) } /
v:String { v.into() }
v:String { Literal::new_simple_literal(v) }
//[130]
NumericLiteral -> Literal = NumericLiteralUnsigned / NumericLiteralPositive / NumericLiteralNegative

38
src/sparql/xml_results.rs
Unescape View File

@ -279,20 +279,7 @@ impl<R: BufRead> Iterator for ResultsIterator<R> {
}
State::Literal => {
let value = self.reader.decode(&data).to_string();
term = Some(
match datatype {
Some(ref datatype) => {
Literal::new_typed_literal(value, datatype.clone())
}
None => match lang {
Some(ref lang) => Literal::new_language_tagged_literal(
value,
lang.clone(),
),
None => Literal::new_simple_literal(value),
},
}.into(),
)
term = Some(build_literal(value, &lang, &datatype).into());
}
_ => {
return Some(Err(format!(
@ -322,7 +309,14 @@ impl<R: BufRead> Iterator for ResultsIterator<R> {
term = None;
state = State::Result;
}
State::Uri | State::BNode | State::Literal => state = State::Binding,
State::Uri | State::BNode => state = State::Binding,
State::Literal => {
if term.is_none() {
//We default to the empty literal
term = Some(build_literal("", &lang, &datatype).into())
}
state = State::Binding;
}
_ => (),
},
Event::Eof => return None,
@ -331,3 +325,17 @@ impl<R: BufRead> Iterator for ResultsIterator<R> {
}
}
}
fn build_literal(
value: impl Into<String>,
lang: &Option<String>,
datatype: &Option<NamedNode>,
) -> Literal {
match datatype {
Some(datatype) => Literal::new_typed_literal(value, datatype.clone()),
None => match lang {
Some(lang) => Literal::new_language_tagged_literal(value, lang.clone()),
None => Literal::new_simple_literal(value),
},
}
}

38
src/store/numeric_encoder.rs
Unescape View File

@ -19,8 +19,9 @@ pub trait BytesStore {
const TYPE_DEFAULT_GRAPH_ID: u8 = 0;
const TYPE_NAMED_NODE_ID: u8 = 1;
const TYPE_BLANK_NODE_ID: u8 = 2;
const TYPE_LANG_STRING_LITERAL_ID: u8 = 3;
const TYPE_TYPED_LITERAL_ID: u8 = 4;
const TYPE_SIMPLE_LITERAL_ID: u8 = 3;
const TYPE_LANG_STRING_LITERAL_ID: u8 = 4;
const TYPE_TYPED_LITERAL_ID: u8 = 5;
pub static ENCODED_DEFAULT_GRAPH: EncodedTerm = EncodedTerm::DefaultGraph {};
@ -29,6 +30,7 @@ pub enum EncodedTerm {
DefaultGraph {},
NamedNode { iri_id: u64 },
BlankNode(Uuid),
SimpleLiteral { value_id: u64 },
LangStringLiteral { value_id: u64, language_id: u64 },
TypedLiteral { value_id: u64, datatype_id: u64 },
}
@ -39,6 +41,7 @@ impl EncodedTerm {
EncodedTerm::DefaultGraph { .. } => TYPE_DEFAULT_GRAPH_ID,
EncodedTerm::NamedNode { .. } => TYPE_NAMED_NODE_ID,
EncodedTerm::BlankNode(_) => TYPE_BLANK_NODE_ID,
EncodedTerm::SimpleLiteral { .. } => TYPE_SIMPLE_LITERAL_ID,
EncodedTerm::LangStringLiteral { .. } => TYPE_LANG_STRING_LITERAL_ID,
EncodedTerm::TypedLiteral { .. } => TYPE_TYPED_LITERAL_ID,
}
@ -88,6 +91,9 @@ impl<R: Read> TermReader for R {
self.read_exact(&mut uuid_buffer)?;
Ok(EncodedTerm::BlankNode(Uuid::from_bytes(uuid_buffer)))
}
TYPE_SIMPLE_LITERAL_ID => Ok(EncodedTerm::SimpleLiteral {
value_id: self.read_u64::<NetworkEndian>()?,
}),
TYPE_LANG_STRING_LITERAL_ID => Ok(EncodedTerm::LangStringLiteral {
language_id: self.read_u64::<NetworkEndian>()?,
value_id: self.read_u64::<NetworkEndian>()?,
@ -154,6 +160,9 @@ impl<R: Write> TermWriter for R {
EncodedTerm::DefaultGraph {} => {}
EncodedTerm::NamedNode { iri_id } => self.write_u64::<NetworkEndian>(iri_id)?,
EncodedTerm::BlankNode(id) => self.write_all(id.as_bytes())?,
EncodedTerm::SimpleLiteral { value_id } => {
self.write_u64::<NetworkEndian>(value_id)?;
}
EncodedTerm::LangStringLiteral {
value_id,
language_id,
@ -217,17 +226,23 @@ impl<S: BytesStore> Encoder<S> {
}
pub fn encode_literal(&self, literal: &Literal) -> Result<EncodedTerm> {
if let Some(language) = literal.language() {
Ok(EncodedTerm::LangStringLiteral {
value_id: self.encode_str_value(&literal.value())?,
language_id: self.encode_str_value(language)?,
})
Ok(if literal.is_plain() {
if let Some(language) = literal.language() {
EncodedTerm::LangStringLiteral {
value_id: self.encode_str_value(&literal.value())?,
language_id: self.encode_str_value(language)?,
}
} else {
EncodedTerm::SimpleLiteral {
value_id: self.encode_str_value(&literal.value())?,
}
}
} else {
Ok(EncodedTerm::TypedLiteral {
EncodedTerm::TypedLiteral {
value_id: self.encode_str_value(&literal.value())?,
datatype_id: self.encode_str_value(literal.datatype().as_str())?,
})
}
}
})
}
pub fn encode_named_or_blank_node(&self, term: &NamedOrBlankNode) -> Result<EncodedTerm> {
@ -277,6 +292,9 @@ impl<S: BytesStore> Encoder<S> {
Ok(NamedNode::from(self.decode_url_value(iri_id)?).into())
}
EncodedTerm::BlankNode(id) => Ok(BlankNode::from(id).into()),
EncodedTerm::SimpleLiteral { value_id } => {
Ok(Literal::new_simple_literal(self.decode_str_value(value_id)?).into())
}
EncodedTerm::LangStringLiteral {
value_id,
language_id,

89
src/store/sparql.rs
Unescape View File

@ -1,4 +1,5 @@
use sparql::algebra::*;
use std::collections::BTreeSet;
use std::iter::once;
use std::iter::Iterator;
use std::sync::Arc;
@ -49,6 +50,85 @@ impl EncodedBindingsIterator {
})),
}
}
fn unique(self) -> Self {
let EncodedBindingsIterator { variables, iter } = self;
let mut oks = BTreeSet::default();
let mut errors = Vec::default();
for element in iter {
match element {
Ok(ok) => {
oks.insert(ok);
}
Err(error) => errors.push(error),
}
}
EncodedBindingsIterator {
variables,
iter: Box::new(errors.into_iter().map(Err).chain(oks.into_iter().map(Ok))),
}
}
fn chain(self, other: Self) -> Self {
let EncodedBindingsIterator {
variables: variables1,
iter: iter1,
} = self;
let EncodedBindingsIterator {
variables: variables2,
iter: iter2,
} = other;
let mut variables = variables1;
let mut map_2_to_1 = Vec::with_capacity(variables2.len());
for var in variables2 {
map_2_to_1.push(match slice_key(&variables, &var) {
Some(key) => key,
None => {
variables.push(var);
variables.len() - 1
}
})
}
let variables_len = variables.len();
EncodedBindingsIterator {
variables,
iter: Box::new(iter1.chain(iter2.map(move |binding| {
let binding = binding?;
let mut new_binding = binding.clone();
new_binding.resize(variables_len, None);
for (old_key, new_key) in map_2_to_1.iter().enumerate() {
new_binding[*new_key] = binding[old_key];
}
Ok(new_binding)
}))),
}
}
fn duplicate(self) -> (Self, Self) {
let EncodedBindingsIterator { variables, iter } = self;
//TODO: optimize
let mut oks = Vec::default();
let mut errors = Vec::default();
for element in iter {
match element {
Ok(ok) => {
oks.push(ok);
}
Err(error) => errors.push(error),
}
}
(
EncodedBindingsIterator {
variables: variables.clone(),
iter: Box::new(oks.clone().into_iter().map(Ok)),
},
EncodedBindingsIterator {
variables: variables,
iter: Box::new(errors.into_iter().map(Err).chain(oks.into_iter().map(Ok))),
},
)
}
}
impl Default for EncodedBindingsIterator {
@ -101,7 +181,7 @@ impl<S: EncodedQuadsStore> SparqlEvaluator<S> {
ListPattern::Project(l, new_variables) => Ok(self
.eval_list_pattern(l, from)?
.project(new_variables.to_vec())),
ListPattern::Distinct(l) => self.eval_list_pattern(l, from), //TODO
ListPattern::Distinct(l) => Ok(self.eval_list_pattern(l, from)?.unique()),
ListPattern::Reduced(l) => self.eval_list_pattern(l, from),
ListPattern::Slice(l, start, length) => {
let mut iter = self.eval_list_pattern(l, from)?;
@ -139,7 +219,12 @@ impl<S: EncodedQuadsStore> SparqlEvaluator<S> {
}
MultiSetPattern::LeftJoin(a, b, e) => unimplemented!(),
MultiSetPattern::Filter(e, p) => unimplemented!(),
MultiSetPattern::Union(a, b) => unimplemented!(),
MultiSetPattern::Union(a, b) => {
let (from1, from2) = from.duplicate();
Ok(self
.eval_multi_set_pattern(a, from1)?
.chain(self.eval_multi_set_pattern(b, from2)?))
}
MultiSetPattern::Graph(g, p) => unimplemented!(),
MultiSetPattern::Extend(p, v, e) => unimplemented!(),
MultiSetPattern::Minus(a, b) => unimplemented!(),

16
tests/sparql_test_cases.rs
Unescape View File

@ -83,8 +83,18 @@ fn sparql_w3c_query_evaluation_testsuite() {
.unwrap(),
Url::parse("http://www.w3.org/2001/sw/DataAccess/tests/data-r2/triple-match/manifest.ttl")
.unwrap(),
Url::parse("http://www.w3.org/2001/sw/DataAccess/tests/data-r2/distinct/manifest.ttl")
.unwrap(),
];
let test_blacklist = vec![
//With LeftJoin
NamedNode::from_str(
"http://www.w3.org/2001/sw/DataAccess/tests/data-r2/distinct/manifest#distinct-4",
).unwrap(),
NamedNode::from_str(
"http://www.w3.org/2001/sw/DataAccess/tests/data-r2/distinct/manifest#no-distinct-4",
).unwrap(),
];
let test_blacklist = vec![];
let client = RDFClient::default();
for test_result in manifest_10_urls
@ -232,7 +242,7 @@ fn to_graph(result: QueryResult) -> Result<MemoryGraph> {
graph.insert(&Triple::new(
result_set.clone(),
rs::RESULT_VARIABLE.clone(),
Literal::from(variable.name()?),
Literal::new_simple_literal(variable.name()?),
))?;
}
for binding_values in iter {
@ -259,7 +269,7 @@ fn to_graph(result: QueryResult) -> Result<MemoryGraph> {
graph.insert(&Triple::new(
binding.clone(),
rs::VARIABLE.clone(),
Literal::from(variables[i].name()?),
Literal::new_simple_literal(variables[i].name()?),
))?;
}
}

Write Preview
Loading…
Cancel
Save