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

Adds Tokio async to SPARQL XML results parser

Browse Source
pull/776/merge
Tpt 10 months ago committed by Thomas Tanon
parent c13cb8db7c
commit 7d45ea43f5
3 changed files with 534 additions and 361 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. 6
      lib/sparesults/src/error.rs
  2. 27
      lib/sparesults/src/parser.rs
  3. 512
      lib/sparesults/src/xml.rs

6
lib/sparesults/src/error.rs
Unescape View File

@ -47,6 +47,12 @@ impl From<quick_xml::Error> for QueryResultsParseError {
}
}
impl From<quick_xml::escape::EscapeError> for QueryResultsParseError {
#[inline]
fn from(error: quick_xml::escape::EscapeError) -> Self {
quick_xml::Error::from(error).into()
}
}
/// An error in the syntax of the parsed file.
#[derive(Debug, thiserror::Error)]
#[error(transparent)]

27
lib/sparesults/src/parser.rs
Unescape View File

@ -7,7 +7,9 @@ use crate::json::{
FromTokioAsyncReadJsonQueryResultsReader, FromTokioAsyncReadJsonSolutionsReader,
};
use crate::solution::QuerySolution;
use crate::xml::{XmlQueryResultsReader, XmlSolutionsReader};
use crate::xml::{FromReadXmlQueryResultsReader, FromReadXmlSolutionsReader};
#[cfg(feature = "async-tokio")]
use crate::xml::{FromTokioAsyncReadXmlQueryResultsReader, FromTokioAsyncReadXmlSolutionsReader};
use oxrdf::Variable;
use std::io::Read;
use std::sync::Arc;
@ -81,9 +83,9 @@ impl QueryResultsParser {
reader: R,
) -> Result<FromReadQueryResultsReader<R>, QueryResultsParseError> {
Ok(match self.format {
QueryResultsFormat::Xml => match XmlQueryResultsReader::read(reader)? {
XmlQueryResultsReader::Boolean(r) => FromReadQueryResultsReader::Boolean(r),
XmlQueryResultsReader::Solutions {
QueryResultsFormat::Xml => match FromReadXmlQueryResultsReader::read(reader)? {
FromReadXmlQueryResultsReader::Boolean(r) => FromReadQueryResultsReader::Boolean(r),
FromReadXmlQueryResultsReader::Solutions {
solutions,
variables,
} => FromReadQueryResultsReader::Solutions(FromReadSolutionsReader {
@ -128,7 +130,7 @@ impl QueryResultsParser {
/// Reads are automatically buffered.
///
/// Example in XML (the API is the same for JSON and TSV):
/// ```no_run
/// ```
/// use sparesults::{QueryResultsFormat, QueryResultsParser, FromTokioAsyncReadQueryResultsReader};
/// use oxrdf::{Literal, Variable};
///
@ -157,7 +159,16 @@ impl QueryResultsParser {
reader: R,
) -> Result<FromTokioAsyncReadQueryResultsReader<R>, QueryResultsParseError> {
Ok(match self.format {
QueryResultsFormat::Xml => return Err(QueryResultsSyntaxError::msg("The XML query results parser does not support Tokio AsyncRead yet").into()),
QueryResultsFormat::Xml => match FromTokioAsyncReadXmlQueryResultsReader::read(reader).await? {
FromTokioAsyncReadXmlQueryResultsReader::Boolean(r) => FromTokioAsyncReadQueryResultsReader::Boolean(r),
FromTokioAsyncReadXmlQueryResultsReader::Solutions {
solutions,
variables,
} => FromTokioAsyncReadQueryResultsReader::Solutions(FromTokioAsyncReadSolutionsReader {
variables: variables.into(),
solutions: FromTokioAsyncReadSolutionsReaderKind::Xml(solutions),
}),
},
QueryResultsFormat::Json => match FromTokioAsyncReadJsonQueryResultsReader::read(reader).await? {
FromTokioAsyncReadJsonQueryResultsReader::Boolean(r) => FromTokioAsyncReadQueryResultsReader::Boolean(r),
FromTokioAsyncReadJsonQueryResultsReader::Solutions {
@ -248,7 +259,7 @@ pub struct FromReadSolutionsReader<R: Read> {
}
enum FromReadSolutionsReaderKind<R: Read> {
Xml(XmlSolutionsReader<R>),
Xml(FromReadXmlSolutionsReader<R>),
Json(FromReadJsonSolutionsReader<R>),
Tsv(TsvSolutionsReader<R>),
}
@ -381,6 +392,7 @@ pub struct FromTokioAsyncReadSolutionsReader<R: AsyncRead + Unpin> {
#[cfg(feature = "async-tokio")]
enum FromTokioAsyncReadSolutionsReaderKind<R: AsyncRead + Unpin> {
Json(FromTokioAsyncReadJsonSolutionsReader<R>),
Xml(FromTokioAsyncReadXmlSolutionsReader<R>),
}
#[cfg(feature = "async-tokio")]
@ -422,6 +434,7 @@ impl<R: AsyncRead + Unpin> FromTokioAsyncReadSolutionsReader<R> {
Some(
match &mut self.solutions {
FromTokioAsyncReadSolutionsReaderKind::Json(reader) => reader.read_next().await,
FromTokioAsyncReadSolutionsReaderKind::Xml(reader) => reader.read_next().await,
}
.transpose()?
.map(|values| (Arc::clone(&self.variables), values).into()),

512
lib/sparesults/src/xml.rs
Unescape View File

@ -3,15 +3,15 @@
use crate::error::{QueryResultsParseError, QueryResultsSyntaxError};
use oxrdf::vocab::rdf;
use oxrdf::*;
use quick_xml::escape::unescape;
use quick_xml::events::{BytesDecl, BytesEnd, BytesStart, BytesText, Event};
use quick_xml::{Reader, Writer};
use std::borrow::Cow;
use quick_xml::{Decoder, Reader, Writer};
use std::collections::BTreeMap;
use std::io::{self, BufReader, Read, Write};
use std::str;
use std::mem::take;
use std::sync::Arc;
#[cfg(feature = "async-tokio")]
use tokio::io::AsyncWrite;
use tokio::io::{AsyncRead, AsyncWrite, BufReader as AsyncBufReader};
pub fn write_boolean_xml_result<W: Write>(write: W, value: bool) -> io::Result<W> {
let mut writer = Writer::new(write);
@ -154,7 +154,6 @@ impl InnerXmlSolutionsWriter {
}
#[allow(clippy::unused_self)]
fn write<'a>(
&self,
output: &mut Vec<Event<'a>>,
@ -218,17 +217,149 @@ fn write_xml_term<'a>(output: &mut Vec<Event<'a>>, term: TermRef<'a>) {
}
}
pub enum XmlQueryResultsReader<R: Read> {
pub enum FromReadXmlQueryResultsReader<R: Read> {
Solutions {
variables: Vec<Variable>,
solutions: XmlSolutionsReader<R>,
solutions: FromReadXmlSolutionsReader<R>,
},
Boolean(bool),
}
impl<R: Read> XmlQueryResultsReader<R> {
pub fn read(source: R) -> Result<Self, QueryResultsParseError> {
enum State {
impl<R: Read> FromReadXmlQueryResultsReader<R> {
pub fn read(read: R) -> Result<Self, QueryResultsParseError> {
let mut reader = Reader::from_reader(BufReader::new(read));
reader.trim_text(true);
reader.expand_empty_elements(true);
let mut reader_buffer = Vec::new();
let mut inner = XmlInnerQueryResultsReader {
state: ResultsState::Start,
variables: Vec::new(),
decoder: reader.decoder(),
};
loop {
reader_buffer.clear();
let event = reader.read_event_into(&mut reader_buffer)?;
if let Some(result) = inner.read_event(event)? {
return Ok(match result {
XmlInnerQueryResults::Solutions {
variables,
solutions,
} => Self::Solutions {
variables,
solutions: FromReadXmlSolutionsReader {
reader,
inner: solutions,
reader_buffer,
},
},
XmlInnerQueryResults::Boolean(value) => Self::Boolean(value),
});
}
}
}
}
pub struct FromReadXmlSolutionsReader<R: Read> {
reader: Reader<BufReader<R>>,
inner: XmlInnerSolutionsReader,
reader_buffer: Vec<u8>,
}
impl<R: Read> FromReadXmlSolutionsReader<R> {
pub fn read_next(&mut self) -> Result<Option<Vec<Option<Term>>>, QueryResultsParseError> {
loop {
self.reader_buffer.clear();
let event = self.reader.read_event_into(&mut self.reader_buffer)?;
if event == Event::Eof {
return Ok(None);
}
if let Some(solution) = self.inner.read_event(event)? {
return Ok(Some(solution));
}
}
}
}
#[cfg(feature = "async-tokio")]
pub enum FromTokioAsyncReadXmlQueryResultsReader<R: AsyncRead + Unpin> {
Solutions {
variables: Vec<Variable>,
solutions: FromTokioAsyncReadXmlSolutionsReader<R>,
},
Boolean(bool),
}
#[cfg(feature = "async-tokio")]
impl<R: AsyncRead + Unpin> FromTokioAsyncReadXmlQueryResultsReader<R> {
pub async fn read(read: R) -> Result<Self, QueryResultsParseError> {
let mut reader = Reader::from_reader(AsyncBufReader::new(read));
reader.trim_text(true);
reader.expand_empty_elements(true);
let mut reader_buffer = Vec::new();
let mut inner = XmlInnerQueryResultsReader {
state: ResultsState::Start,
variables: Vec::new(),
decoder: reader.decoder(),
};
loop {
reader_buffer.clear();
let event = reader.read_event_into_async(&mut reader_buffer).await?;
if let Some(result) = inner.read_event(event)? {
return Ok(match result {
XmlInnerQueryResults::Solutions {
variables,
solutions,
} => Self::Solutions {
variables,
solutions: FromTokioAsyncReadXmlSolutionsReader {
reader,
inner: solutions,
reader_buffer,
},
},
XmlInnerQueryResults::Boolean(value) => Self::Boolean(value),
});
}
}
}
}
#[cfg(feature = "async-tokio")]
pub struct FromTokioAsyncReadXmlSolutionsReader<R: AsyncRead + Unpin> {
reader: Reader<AsyncBufReader<R>>,
inner: XmlInnerSolutionsReader,
reader_buffer: Vec<u8>,
}
#[cfg(feature = "async-tokio")]
impl<R: AsyncRead + Unpin> FromTokioAsyncReadXmlSolutionsReader<R> {
pub async fn read_next(&mut self) -> Result<Option<Vec<Option<Term>>>, QueryResultsParseError> {
loop {
self.reader_buffer.clear();
let event = self
.reader
.read_event_into_async(&mut self.reader_buffer)
.await?;
if event == Event::Eof {
return Ok(None);
}
if let Some(solution) = self.inner.read_event(event)? {
return Ok(Some(solution));
}
}
}
}
enum XmlInnerQueryResults {
Solutions {
variables: Vec<Variable>,
solutions: XmlInnerSolutionsReader,
},
Boolean(bool),
}
#[derive(Clone, Copy)]
enum ResultsState {
Start,
Sparql,
Head,
@ -236,105 +367,126 @@ impl<R: Read> XmlQueryResultsReader<R> {
Boolean,
}
let mut reader = Reader::from_reader(BufReader::new(source));
reader.trim_text(true);
reader.expand_empty_elements(true);
let mut buffer = Vec::default();
let mut variables = Vec::default();
let mut state = State::Start;
struct XmlInnerQueryResultsReader {
state: ResultsState,
variables: Vec<Variable>,
decoder: Decoder,
}
// Read header
loop {
buffer.clear();
let event = reader.read_event_into(&mut buffer)?;
impl XmlInnerQueryResultsReader {
pub fn read_event(
&mut self,
event: Event<'_>,
) -> Result<Option<XmlInnerQueryResults>, QueryResultsParseError> {
match event {
Event::Start(event) => match state {
State::Start => {
Event::Start(event) => match self.state {
ResultsState::Start => {
if event.local_name().as_ref() == b"sparql" {
state = State::Sparql;
self.state = ResultsState::Sparql;
Ok(None)
} else {
return Err(QueryResultsSyntaxError::msg(format!("Expecting <sparql> tag, found <{}>", decode(&reader, &event.name())?)).into());
Err(QueryResultsSyntaxError::msg(format!("Expecting <sparql> tag, found <{}>", self.decoder.decode(event.name().as_ref())?)).into())
}
}
State::Sparql => {
ResultsState::Sparql => {
if event.local_name().as_ref() == b"head" {
state = State::Head;
self.state = ResultsState::Head;
Ok(None)
} else {
return Err(QueryResultsSyntaxError::msg(format!("Expecting <head> tag, found <{}>", decode(&reader, &event.name())?)).into());
Err(QueryResultsSyntaxError::msg(format!("Expecting <head> tag, found <{}>", self.decoder.decode(event.name().as_ref())?)).into())
}
}
State::Head => {
ResultsState::Head => {
if event.local_name().as_ref() == b"variable" {
let name = event.attributes()
.filter_map(Result::ok)
.find(|attr| attr.key.local_name().as_ref() == b"name")
.ok_or_else(|| QueryResultsSyntaxError::msg("No name attribute found for the <variable> tag"))?
.decode_and_unescape_value(&reader)?;
.ok_or_else(|| QueryResultsSyntaxError::msg("No name attribute found for the <variable> tag"))?;
let name = unescape(&self.decoder.decode(&name.value)?)?.into_owned();
let variable = Variable::new(name).map_err(|e| QueryResultsSyntaxError::msg(format!("Invalid variable name: {e}")))?;
if variables.contains(&variable) {
if self.variables.contains(&variable) {
return Err(QueryResultsSyntaxError::msg(format!(
"The variable {variable} is declared twice"
))
.into());
}
variables.push(variable);
self.variables.push(variable);
Ok(None)
} else if event.local_name().as_ref() == b"link" {
// no op
Ok(None)
} else {
return Err(QueryResultsSyntaxError::msg(format!("Expecting <variable> or <link> tag, found <{}>", decode(&reader, &event.name())?)).into());
Err(QueryResultsSyntaxError::msg(format!("Expecting <variable> or <link> tag, found <{}>", self.decoder.decode(event.name().as_ref())?)).into())
}
}
State::AfterHead => {
ResultsState::AfterHead => {
if event.local_name().as_ref() == b"boolean" {
state = State::Boolean
self.state = ResultsState::Boolean;
Ok(None)
} else if event.local_name().as_ref() == b"results" {
let mut mapping = BTreeMap::default();
for (i, var) in variables.iter().enumerate() {
for (i, var) in self.variables.iter().enumerate() {
mapping.insert(var.clone().into_string(), i);
}
return Ok(Self::Solutions { variables,
solutions: XmlSolutionsReader {
reader,
buffer,
Ok(Some(XmlInnerQueryResults::Solutions {
variables: take(&mut self.variables),
solutions: XmlInnerSolutionsReader {
decoder: self.decoder,
mapping,
stack: Vec::new(),
state_stack: vec![State::Start, State::Start],
new_bindings: Vec::new(),
current_var: None,
term: None,
lang: None,
datatype: None,
subject_stack: Vec::new(),
predicate_stack: Vec::new(),
object_stack: Vec::new(),
}});
},
}))
} else if event.local_name().as_ref() != b"link" && event.local_name().as_ref() != b"results" && event.local_name().as_ref() != b"boolean" {
return Err(QueryResultsSyntaxError::msg(format!("Expecting sparql tag, found <{}>", decode(&reader, &event.name())?)).into());
Err(QueryResultsSyntaxError::msg(format!("Expecting sparql tag, found <{}>", self.decoder.decode(event.name().as_ref())?)).into())
} else {
Ok(None)
}
}
State::Boolean => return Err(QueryResultsSyntaxError::msg(format!("Unexpected tag inside of <boolean> tag: <{}>", decode(&reader, &event.name())?)).into())
ResultsState::Boolean => Err(QueryResultsSyntaxError::msg(format!("Unexpected tag inside of <boolean> tag: <{}>", self.decoder.decode(event.name().as_ref())?)).into())
},
Event::Text(event) => {
let value = event.unescape()?;
return match state {
State::Boolean => {
return if value == "true" {
Ok(Self::Boolean(true))
match self.state {
ResultsState::Boolean => {
if value == "true" {
Ok(Some(XmlInnerQueryResults::Boolean(true)))
} else if value == "false" {
Ok(Self::Boolean(false))
Ok(Some(XmlInnerQueryResults::Boolean(false)))
} else {
Err(QueryResultsSyntaxError::msg(format!("Unexpected boolean value. Found '{value}'")).into())
};
}
}
_ => Err(QueryResultsSyntaxError::msg(format!("Unexpected textual value found: '{value}'")).into())
};
},
}
}
Event::End(event) => {
if let State::Head = state {
if let ResultsState::Head = self.state {
if event.local_name().as_ref() == b"head" {
state = State::AfterHead
self.state = ResultsState::AfterHead
}
Ok(None)
} else {
return Err(QueryResultsSyntaxError::msg("Unexpected early file end. All results file should have a <head> and a <result> or <boolean> tag").into());
Err(QueryResultsSyntaxError::msg("Unexpected early file end. All results file should have a <head> and a <result> or <boolean> tag").into())
}
},
Event::Eof => return Err(QueryResultsSyntaxError::msg("Unexpected early file end. All results file should have a <head> and a <result> or <boolean> tag").into()),
_ => (),
}
Event::Eof => Err(QueryResultsSyntaxError::msg("Unexpected early file end. All results file should have a <head> and a <result> or <boolean> tag").into()),
Event::Comment(_) | Event::Decl(_) | Event::PI(_) | Event::DocType(_) => {
Ok(None)
}
Event::Empty(_) => unreachable!("Empty events are expended"),
Event::CData(_) => {
Err(QueryResultsSyntaxError::msg(
"<![CDATA[...]]> are not supported in SPARQL XML results",
)
.into())
}
}
}
@ -351,135 +503,150 @@ enum State {
Subject,
Predicate,
Object,
End,
}
pub struct XmlSolutionsReader<R: Read> {
reader: Reader<BufReader<R>>,
buffer: Vec<u8>,
struct XmlInnerSolutionsReader {
decoder: Decoder,
mapping: BTreeMap<String, usize>,
stack: Vec<State>,
state_stack: Vec<State>,
new_bindings: Vec<Option<Term>>,
current_var: Option<String>,
term: Option<Term>,
lang: Option<String>,
datatype: Option<NamedNode>,
subject_stack: Vec<Term>,
predicate_stack: Vec<Term>,
object_stack: Vec<Term>,
}
impl<R: Read> XmlSolutionsReader<R> {
pub fn read_next(&mut self) -> Result<Option<Vec<Option<Term>>>, QueryResultsParseError> {
let mut state = State::Start;
let mut new_bindings = vec![None; self.mapping.len()];
let mut current_var = None;
let mut term: Option<Term> = None;
let mut lang = None;
let mut datatype = None;
loop {
self.buffer.clear();
let event = self.reader.read_event_into(&mut self.buffer)?;
impl XmlInnerSolutionsReader {
#[allow(clippy::unwrap_in_result)]
pub fn read_event(
&mut self,
event: Event<'_>,
) -> Result<Option<Vec<Option<Term>>>, QueryResultsParseError> {
match event {
Event::Start(event) => match state {
Event::Start(event) => match self.state_stack.last().unwrap() {
State::Start => {
if event.local_name().as_ref() == b"result" {
state = State::Result;
self.new_bindings = vec![None; self.mapping.len()];
self.state_stack.push(State::Result);
Ok(None)
} else {
return Err(QueryResultsSyntaxError::msg(format!(
Err(QueryResultsSyntaxError::msg(format!(
"Expecting <result>, found <{}>",
decode(&self.reader, &event.name())?
self.decoder.decode(event.name().as_ref())?
))
.into());
.into())
}
}
State::Result => {
if event.local_name().as_ref() == b"binding" {
match event
let Some(attr) = event
.attributes()
.filter_map(Result::ok)
.find(|attr| attr.key.local_name().as_ref() == b"name")
{
Some(attr) => {
current_var = Some(
attr.decode_and_unescape_value(&self.reader)?.to_string(),
)
}
None => {
else {
return Err(QueryResultsSyntaxError::msg(
"No name attribute found for the <binding> tag",
)
.into());
}
}
state = State::Binding;
};
self.current_var =
Some(unescape(&self.decoder.decode(&attr.value)?)?.into_owned());
self.state_stack.push(State::Binding);
Ok(None)
} else {
return Err(QueryResultsSyntaxError::msg(format!(
Err(QueryResultsSyntaxError::msg(format!(
"Expecting <binding>, found <{}>",
decode(&self.reader, &event.name())?
self.decoder.decode(event.name().as_ref())?
))
.into());
.into())
}
}
State::Binding | State::Subject | State::Predicate | State::Object => {
if term.is_some() {
if self.term.is_some() {
return Err(QueryResultsSyntaxError::msg(
"There is already a value for the current binding",
)
.into());
}
self.stack.push(state);
if event.local_name().as_ref() == b"uri" {
state = State::Uri;
self.state_stack.push(State::Uri);
Ok(None)
} else if event.local_name().as_ref() == b"bnode" {
state = State::BNode;
self.state_stack.push(State::BNode);
Ok(None)
} else if event.local_name().as_ref() == b"literal" {
for attr in event.attributes() {
let attr = attr.map_err(quick_xml::Error::from)?;
if attr.key.as_ref() == b"xml:lang" {
lang = Some(
attr.decode_and_unescape_value(&self.reader)?.to_string(),
self.lang = Some(
unescape(&self.decoder.decode(&attr.value)?)?.into_owned(),
);
} else if attr.key.local_name().as_ref() == b"datatype" {
let iri = attr.decode_and_unescape_value(&self.reader)?;
datatype =
Some(NamedNode::new(iri.to_string()).map_err(|e| {
let iri = self.decoder.decode(&attr.value)?;
let iri = unescape(&iri)?;
self.datatype =
Some(NamedNode::new(iri.as_ref()).map_err(|e| {
QueryResultsSyntaxError::msg(format!(
"Invalid datatype IRI '{iri}': {e}"
))
})?);
}
}
state = State::Literal;
self.state_stack.push(State::Literal);
Ok(None)
} else if event.local_name().as_ref() == b"triple" {
state = State::Triple;
self.state_stack.push(State::Triple);
Ok(None)
} else {
return Err(QueryResultsSyntaxError::msg(format!(
Err(QueryResultsSyntaxError::msg(format!(
"Expecting <uri>, <bnode> or <literal> found <{}>",
decode(&self.reader, &event.name())?
self.decoder.decode(event.name().as_ref())?
))
.into());
.into())
}
}
State::Triple => {
if event.local_name().as_ref() == b"subject" {
state = State::Subject
self.state_stack.push(State::Subject);
Ok(None)
} else if event.local_name().as_ref() == b"predicate" {
state = State::Predicate
self.state_stack.push(State::Predicate);
Ok(None)
} else if event.local_name().as_ref() == b"object" {
state = State::Object
self.state_stack.push(State::Object);
Ok(None)
} else {
return Err(QueryResultsSyntaxError::msg(format!(
Err(QueryResultsSyntaxError::msg(format!(
"Expecting <subject>, <predicate> or <object> found <{}>",
decode(&self.reader, &event.name())?
self.decoder.decode(event.name().as_ref())?
))
.into());
.into())
}
}
_ => (),
State::Uri => Err(QueryResultsSyntaxError::msg(format!(
"<uri> must only contain a string, found <{}>",
self.decoder.decode(event.name().as_ref())?
))
.into()),
State::BNode => Err(QueryResultsSyntaxError::msg(format!(
"<uri> must only contain a string, found <{}>",
self.decoder.decode(event.name().as_ref())?
))
.into()),
State::Literal => Err(QueryResultsSyntaxError::msg(format!(
"<uri> must only contain a string, found <{}>",
self.decoder.decode(event.name().as_ref())?
))
.into()),
},
Event::Text(event) => {
let data = event.unescape()?;
match state {
match self.state_stack.last().unwrap() {
State::Uri => {
term = Some(
self.term = Some(
NamedNode::new(data.to_string())
.map_err(|e| {
QueryResultsSyntaxError::msg(format!(
@ -487,10 +654,11 @@ impl<R: Read> XmlSolutionsReader<R> {
))
})?
.into(),
)
);
Ok(None)
}
State::BNode => {
term = Some(
self.term = Some(
BlankNode::new(data.to_string())
.map_err(|e| {
QueryResultsSyntaxError::msg(format!(
@ -498,26 +666,28 @@ impl<R: Read> XmlSolutionsReader<R> {
))
})?
.into(),
)
);
Ok(None)
}
State::Literal => {
term = Some(build_literal(data, lang.take(), datatype.take())?.into());
self.term = Some(
build_literal(data, self.lang.take(), self.datatype.take())?.into(),
);
Ok(None)
}
_ => {
return Err(QueryResultsSyntaxError::msg(format!(
_ => Err(QueryResultsSyntaxError::msg(format!(
"Unexpected textual value found: {data}"
))
.into());
.into()),
}
}
}
Event::End(_) => match state {
State::Start => state = State::End,
State::Result => return Ok(Some(new_bindings)),
Event::End(_) => match self.state_stack.pop().unwrap() {
State::Start | State::Uri => Ok(None),
State::Result => Ok(Some(take(&mut self.new_bindings))),
State::Binding => {
if let Some(var) = &current_var {
if let Some(var) = &self.current_var {
if let Some(var) = self.mapping.get(var) {
new_bindings[*var] = term.take()
self.new_bindings[*var] = self.term.take()
} else {
return Err(
QueryResultsSyntaxError::msg(format!("The variable '{var}' is used in a binding but not declared in the variables list")).into()
@ -529,51 +699,40 @@ impl<R: Read> XmlSolutionsReader<R> {
)
.into());
}
state = State::Result;
Ok(None)
}
State::Subject => {
if let Some(subject) = term.take() {
if let Some(subject) = self.term.take() {
self.subject_stack.push(subject)
}
state = State::Triple;
Ok(None)
}
State::Predicate => {
if let Some(predicate) = term.take() {
if let Some(predicate) = self.term.take() {
self.predicate_stack.push(predicate)
}
state = State::Triple;
Ok(None)
}
State::Object => {
if let Some(object) = term.take() {
if let Some(object) = self.term.take() {
self.object_stack.push(object)
}
state = State::Triple;
}
State::Uri => {
state = self
.stack
.pop()
.ok_or_else(|| QueryResultsSyntaxError::msg("Empty stack"))?
Ok(None)
}
State::BNode => {
if term.is_none() {
if self.term.is_none() {
// We default to a random bnode
term = Some(BlankNode::default().into())
self.term = Some(BlankNode::default().into())
}
state = self
.stack
.pop()
.ok_or_else(|| QueryResultsSyntaxError::msg("Empty stack"))?
Ok(None)
}
State::Literal => {
if term.is_none() {
if self.term.is_none() {
// We default to the empty literal
term = Some(build_literal("", lang.take(), datatype.take())?.into())
self.term =
Some(build_literal("", self.lang.take(), self.datatype.take())?.into())
}
state = self
.stack
.pop()
.ok_or_else(|| QueryResultsSyntaxError::msg("Empty stack"))?;
Ok(None)
}
State::Triple => {
#[cfg(feature = "rdf-star")]
@ -582,7 +741,7 @@ impl<R: Read> XmlSolutionsReader<R> {
self.predicate_stack.pop(),
self.object_stack.pop(),
) {
term = Some(
self.term = Some(
Triple::new(
match subject {
Term::NamedNode(subject) => subject.into(),
@ -592,7 +751,7 @@ impl<R: Read> XmlSolutionsReader<R> {
return Err(QueryResultsSyntaxError::msg(
"The <subject> value should not be a <literal>",
)
.into())
.into());
}
},
match predicate {
@ -601,35 +760,37 @@ impl<R: Read> XmlSolutionsReader<R> {
return Err(QueryResultsSyntaxError::msg(
"The <predicate> value should be an <uri>",
)
.into())
.into());
}
},
object,
)
.into(),
);
state = self
.stack
.pop()
.ok_or_else(|| QueryResultsSyntaxError::msg("Empty stack"))?;
Ok(None)
} else {
return Err(
QueryResultsSyntaxError::msg("A <triple> should contain a <subject>, a <predicate> and an <object>").into()
);
Err(QueryResultsSyntaxError::msg(
"A <triple> should contain a <subject>, a <predicate> and an <object>",
)
.into())
}
#[cfg(not(feature = "rdf-star"))]
{
return Err(QueryResultsSyntaxError::msg(
Err(QueryResultsSyntaxError::msg(
"The <triple> tag is only supported with RDF-star",
)
.into());
.into())
}
}
State::End => (),
},
Event::Eof => return Ok(None),
_ => (),
Event::Eof | Event::Comment(_) | Event::Decl(_) | Event::PI(_) | Event::DocType(_) => {
Ok(None)
}
Event::Empty(_) => unreachable!("Empty events are expended"),
Event::CData(_) => Err(QueryResultsSyntaxError::msg(
"<![CDATA[...]]> are not supported in SPARQL XML results",
)
.into()),
}
}
}
@ -661,13 +822,6 @@ fn build_literal(
}
}
fn decode<'a, T>(
reader: &Reader<T>,
data: &'a impl AsRef<[u8]>,
) -> Result<Cow<'a, str>, QueryResultsParseError> {
Ok(reader.decoder().decode(data.as_ref())?)
}
fn map_xml_error(error: quick_xml::Error) -> io::Error {
match error {
quick_xml::Error::Io(error) => {

Write Preview
Loading…
Cancel
Save