NQuads, TriG: Support writing to the default graph (#615)

Author: mmarx

.gitignore

@@ -26,6 +26,9 @@ flycheck_*.el
 /nemo/test-files
 /nemo-benches/test-files
 
+# default output directory
+/results
+
 # nix build
 /result
 

CONTRIBUTING.md

@@ -72,12 +72,12 @@ Add `quickcheck` tests whenever it is applicable.
 
 ### Integration testing
 
-Integration testing is done in the related `tests` directory on the top-level of this crate.
+Integration testing is done in the related `tests` directory on the top-level of the `nemo-cli` crate.
 
 ## Coding conventions
 
 Start reading our code and you'll get the hang of it. Code format and essential coding guidelines are already ensured
-by our use of `rstufmt` and `clippy` (as mentioned above). Some further conventions are listed below.
+by our use of `rustfmt` and `clippy` (as mentioned above). Some further conventions are listed below.
 
   * We try to reduce redundancies in enumeration-variant names.
   * We try to use the `where` clause over embedded clauses for better readability

Cargo.toml

@@ -1,7 +1,7 @@
 [workspace]
 resolver = "2"
 default-members = [
-       "nemo",
+    "nemo",
     "nemo-cli",
     "nemo-physical",
     "nemo-python",

nemo-language-server/Cargo.toml

@@ -8,7 +8,6 @@ homepage.workspace = true
 license.workspace = true
 readme = "README.md"
 repository.workspace = true
-default-run = "nemo-language-server"
 
 [[bin]]
 name = "nemo-language-server"

nemo/src/io/formats/rdf.rs

@@ -44,6 +44,13 @@ use crate::{
 use super::FileFormatMeta;
 use super::{ExportHandler, FormatBuilder, ImportHandler, TableWriter};
 
+/// IRI to be used for the default graph used by Nemo when loading RDF data with
+/// named graphs (quads).
+///
+/// SPARQL 1.1 has failed to provide any standard identifier for this purpose.
+/// If future SPARQL or RDF versions are adding this, we could align accordingly.
+const DEFAULT_GRAPH_IRI: &str = "tag:nemo:defaultgraph";
+
 /// The different supported variants of the RDF format.
 #[derive(Assoc, Debug, Clone, Copy, PartialEq, Eq, VariantArray)]
 #[func(pub fn media_type(&self) -> &'static str)]

nemo/src/io/formats/rdf/reader.rs

@@ -27,16 +27,9 @@ use crate::io::formats::PROGRESS_NOTIFY_INCREMENT;
 use super::{
     error::RdfFormatError,
     value_format::{RdfValueFormat, RdfValueFormats},
-    RdfVariant,
+    RdfVariant, DEFAULT_GRAPH_IRI,
 };
 
-/// IRI to be used for the default graph used by Nemo when loading RDF data with
-/// named graphs (quads).
-///
-/// SPARQL 1.1 has failed to provide any standard identifier for this purpose.
-/// If future SPARQL or RDF versions are adding this, we could align accordingly.
-const DEFAULT_GRAPH: &str = "tag:nemo:defaultgraph";
-
 /// A [TableProvider] for RDF 1.1 files containing triples.
 pub(super) struct RdfReader {
     /// Buffer from which content is read
@@ -157,7 +150,7 @@ impl RdfReader {
         value: Option<GraphName<'_>>,
     ) -> Result<AnyDataValue, RdfFormatError> {
         match value {
-            None => Ok(AnyDataValue::new_iri(DEFAULT_GRAPH.to_string())),
+            None => Ok(AnyDataValue::new_iri(DEFAULT_GRAPH_IRI.to_string())),
             Some(GraphName::NamedNode(nn)) => Ok(Self::datavalue_from_named_node(nn)),
             Some(GraphName::BlankNode(bn)) => {
                 Ok(Self::datavalue_from_blank_node(bnode_map, tuple_writer, bn))
@@ -368,7 +361,7 @@ impl ByteSized for RdfReader {
 
 #[cfg(test)]
 mod test {
-    use super::{RdfReader, DEFAULT_GRAPH};
+    use super::{RdfReader, DEFAULT_GRAPH_IRI};
     use std::cell::RefCell;
 
     use nemo_physical::{
@@ -454,7 +447,7 @@ mod test {
         let dict = RefCell::new(Dict::default());
         let mut tuple_writer = TupleWriter::new(&dict, 3);
         let mut null_map = NullMap::default();
-        let graph_dv = AnyDataValue::new_iri(DEFAULT_GRAPH.to_string());
+        let graph_dv = AnyDataValue::new_iri(DEFAULT_GRAPH_IRI.to_string());
 
         // check that we use our own default graph IRI
         assert_eq!(
@@ -463,115 +456,10 @@ mod test {
         );
         // check that our default graph is a valid IRI in the first place
         assert_eq!(
-            Iri::parse(DEFAULT_GRAPH.to_string()).unwrap().to_string(),
-            DEFAULT_GRAPH.to_string()
+            Iri::parse(DEFAULT_GRAPH_IRI.to_string())
+                .unwrap()
+                .to_string(),
+            DEFAULT_GRAPH_IRI.to_string()
         );
     }
-
-    // #[test]
-    // fn example_1() {
-    //     macro_rules! parse_example_with_rdf_parser {
-    //         ($data:tt, $make_parser:expr) => {
-    //             let $data = r#"<http://one.example/subject1> <http://one.example/predicate1> <http://one.example/object1> . # comments here
-    //                   # or on a line by themselves
-    //                   _:subject1 <http://an.example/predicate1> "object1" .
-    //                   _:subject2 <http://an.example/predicate2> "object2" .
-    //                   "#.as_bytes();
-
-    //             let dict = RefCell::new(Dict::default());
-    //             let mut builders = vec![
-    //                 PhysicalBuilderProxyEnum::String(PhysicalStringColumnBuilderProxy::new(&dict)),
-    //                 PhysicalBuilderProxyEnum::String(PhysicalStringColumnBuilderProxy::new(&dict)),
-    //                 PhysicalBuilderProxyEnum::String(PhysicalStringColumnBuilderProxy::new(&dict)),
-    //             ];
-    //             let reader = RDFReader::new(ResourceProviders::empty(), String::new(), None, vec![PrimitiveType::Any, PrimitiveType::Any, PrimitiveType::Any]);
-
-    //             let result = reader.read_triples_with_parser(&mut builders, $make_parser);
-    //             assert!(result.is_ok());
-
-    //             let columns = builders
-    //                 .into_iter()
-    //                 .map(|builder| match builder {
-    //                     PhysicalBuilderProxyEnum::String(b) => b.finalize(),
-    //                     _ => unreachable!("only string columns here"),
-    //                 })
-    //                 .collect::<Vec<_>>();
-
-    //             log::debug!("columns: {columns:?}");
-    //             let triples = (0..=2)
-    //                 .map(|idx| {
-    //                     columns
-    //                         .iter()
-    //                         .map(|column| {
-    //                             column
-    //                                 .get(idx)
-    //                                 .and_then(|value| value.try_into().ok())
-    //                                 .and_then(|u64: u64| usize::try_from(u64).ok())
-    //                                 .and_then(|usize| dict.borrow_mut().get(usize))
-    //                                 .unwrap()
-    //                         })
-    //                         .map(PhysicalString::from)
-    //                         .collect::<Vec<_>>()
-    //                 })
-    //                 .collect::<Vec<_>>();
-    //             log::debug!("triple: {triples:?}");
-    //             for (value, expected) in PrimitiveType::Any.serialize_output(DataValueIteratorT::String(Box::new(triples[0].iter().cloned()))).zip(vec!["http://one.example/subject1", "http://one.example/predicate1", "http://one.example/object1"]) {
-    //                 assert_eq!(value, expected);
-    //             }
-    //             for (value, expected) in PrimitiveType::Any.serialize_output(DataValueIteratorT::String(Box::new(triples[1].iter().cloned()))).zip(vec!["_:subject1", "http://an.example/predicate1", r#""object1""#]) {
-    //                 assert_eq!(value, expected);
-    //             }
-    //             for (value, expected) in PrimitiveType::Any.serialize_output(DataValueIteratorT::String(Box::new(triples[2].iter().cloned()))).zip(vec!["_:subject2", "http://an.example/predicate2", r#""object2""#]) {
-    //                 assert_eq!(value, expected);
-    //             }
-    //         };
-    //     }
-
-    //     parse_example_with_rdf_parser!(reader, || NTriplesParser::new(reader));
-    //     parse_example_with_rdf_parser!(reader, || TurtleParser::new(reader, None));
-    // }
-
-    // #[test]
-    // fn rollback() {
-    //     let data = r#"<http://example.org/> <http://example.org/> <http://example.org/> .
-    //                       malformed <http://example.org/> <http://example.org/>
-    //                       <http://example.org/> malformed <http://example.org/> .
-    //                       <http://example.org/> <http://example.org/> malformed .
-    //                       <http://example.org/> <http://example.org/> "123"^^<http://www.w3.org/2001/XMLSchema#integer> .
-    //                       <http://example.org/> <http://example.org/> "123.45"^^<http://www.w3.org/2001/XMLSchema#integer> .
-    //                       <http://example.org/> <http://example.org/> "123.45"^^<http://www.w3.org/2001/XMLSchema#decimal> .
-    //                       <http://example.org/> <http://example.org/> "123.45a"^^<http://www.w3.org/2001/XMLSchema#decimal> .
-    //                       <https://example.org/> <https://example.org/> <https://example.org/> .
-    //                   "#
-    //     .as_bytes();
-
-    //     let dict = RefCell::new(Dict::default());
-    //     let mut builders = vec![
-    //         PhysicalBuilderProxyEnum::String(PhysicalStringColumnBuilderProxy::new(&dict)),
-    //         PhysicalBuilderProxyEnum::String(PhysicalStringColumnBuilderProxy::new(&dict)),
-    //         PhysicalBuilderProxyEnum::String(PhysicalStringColumnBuilderProxy::new(&dict)),
-    //     ];
-    //     let reader = RDFReader::new(
-    //         ResourceProviders::empty(),
-    //         String::new(),
-    //         None,
-    //         vec![PrimitiveType::Any, PrimitiveType::Any, PrimitiveType::Any],
-    //     );
-
-    //     let result = reader.read_triples_with_parser(&mut builders, || NTriplesParser::new(data));
-    //     assert!(result.is_ok());
-
-    //     let columns = builders
-    //         .into_iter()
-    //         .map(|builder| match builder {
-    //             PhysicalBuilderProxyEnum::String(b) => b.finalize(),
-    //             _ => unreachable!("only string columns here"),
-    //         })
-    //         .collect::<Vec<_>>();
-
-    //     assert_eq!(columns.len(), 3);
-    //     assert_eq!(columns[0].len(), 4);
-    //     assert_eq!(columns[1].len(), 4);
-    //     assert_eq!(columns[2].len(), 4);
-    // }
 }

nemo/src/io/formats/rdf/writer.rs

@@ -16,7 +16,7 @@ use crate::{
 
 use super::{
     value_format::{RdfValueFormat, RdfValueFormats},
-    RdfVariant,
+    RdfVariant, DEFAULT_GRAPH_IRI,
 };
 
 /// Private struct to record the type of an RDF term that
@@ -31,13 +31,43 @@ enum RdfTermType {
     SimpleStringLiteral,
 }
 
+#[derive(Debug, Default)]
+enum QuadGraphName {
+    #[default]
+    DefaultGraph,
+    NamedNode(String),
+    BlankNode(String),
+}
+
+#[derive(Debug)]
+struct InvalidGraphNameError;
+
+impl TryFrom<&AnyDataValue> for QuadGraphName {
+    type Error = InvalidGraphNameError;
+
+    fn try_from(value: &AnyDataValue) -> Result<Self, Self::Error> {
+        match value.value_domain() {
+            ValueDomain::Iri => {
+                let iri = value.to_iri_unchecked();
+
+                if iri == DEFAULT_GRAPH_IRI {
+                    Ok(Self::DefaultGraph)
+                } else {
+                    Ok(Self::NamedNode(iri))
+                }
+            }
+            ValueDomain::Null => Ok(Self::BlankNode(value.lexical_value())),
+            _ => Err(InvalidGraphNameError),
+        }
+    }
+}
+
 /// Struct to store information of one quad (or triple) for export.
 /// This is necessary since all RIO RDF term implementations use `&str`
 /// pointers internally, that must be owned elsewhere.
 #[derive(Debug, Default)]
 struct QuadBuffer {
-    graph_name_is_blank: bool,
-    graph_name: String,
+    graph_name: QuadGraphName,
     subject_is_blank: bool,
     subject: String,
     predicate: String,
@@ -88,15 +118,15 @@ impl<'a> QuadBuffer {
         }
     }
 
-    fn graph_name(&'a self) -> GraphName<'a> {
-        if self.graph_name_is_blank {
-            GraphName::BlankNode(BlankNode {
-                id: self.graph_name.as_str(),
-            })
-        } else {
-            GraphName::NamedNode(NamedNode {
-                iri: self.graph_name.as_str(),
-            })
+    fn graph_name(&'a self) -> Option<GraphName<'a>> {
+        match &self.graph_name {
+            QuadGraphName::DefaultGraph => None,
+            QuadGraphName::NamedNode(iri) => {
+                Some(GraphName::NamedNode(NamedNode { iri: iri.as_str() }))
+            }
+            QuadGraphName::BlankNode(id) => {
+                Some(GraphName::BlankNode(BlankNode { id: id.as_str() }))
+            }
         }
     }
 
@@ -170,20 +200,13 @@ impl<'a> QuadBuffer {
         true
     }
 
-    fn set_graph_name_from_datavalue(&mut self, datavalue: &AnyDataValue) -> bool {
-        match datavalue.value_domain() {
-            ValueDomain::Iri => {
-                self.graph_name = datavalue.to_iri_unchecked();
-                self.graph_name_is_blank = false;
-                true
-            }
-            ValueDomain::Null => {
-                self.graph_name = datavalue.lexical_value();
-                self.graph_name_is_blank = true;
-                true
-            }
-            _ => false,
-        }
+    fn set_graph_name_from_datavalue(
+        &mut self,
+        datavalue: &AnyDataValue,
+    ) -> Result<(), InvalidGraphNameError> {
+        self.graph_name = QuadGraphName::try_from(datavalue)?;
+
+        Ok(())
     }
 }
 
@@ -318,14 +341,17 @@ impl RdfWriter {
             if !buffer.set_object_from_datavalue(&record[o_pos]) {
                 continue;
             }
-            if !buffer.set_graph_name_from_datavalue(&record[g_pos]) {
+            if buffer
+                .set_graph_name_from_datavalue(&record[g_pos])
+                .is_err()
+            {
                 continue;
             }
             if let Err(e) = formatter.format(&Quad {
                 subject: buffer.subject(),
                 predicate: buffer.predicate(),
                 object: buffer.object(),
-                graph_name: Some(buffer.graph_name()),
+                graph_name: buffer.graph_name(),
             }) {
                 log::debug!("failed to write quad: {e}");
                 drop_count += 1;