Merge pull request #5 from gtfierro/gtf-python-fixes

gtfierro · web-flow · commit 5561be4e2bcb · 2025-01-23T12:04:57.000-07:00
Gtf python fixes
diff --git a/cli/src/main.rs b/cli/src/main.rs
@@ -207,7 +207,12 @@ fn main() -> Result<()> {
                 .get_ontology_by_name(iri.as_ref())
                 .ok_or(anyhow::anyhow!(format!("Ontology {} not found", iri)))?;
             let closure = env.get_dependency_closure(ont.id())?;
-            let graph = env.get_union_graph(&closure, rewrite_sh_prefixes, remove_owl_imports)?;
+            let (graph, _successful, failed_imports) = env.get_union_graph(&closure, rewrite_sh_prefixes, remove_owl_imports)?;
+            if let Some(failed_imports) = failed_imports {
+                for imp in failed_imports {
+                    eprintln!("{}", imp);
+                }
+            }
             // write the graph to a file
             if let Some(destination) = destination {
                 write_dataset_to_file(&graph, &destination)?;
diff --git a/lib/src/lib.rs b/lib/src/lib.rs
@@ -30,7 +30,7 @@ use std::fs;
 use std::io::{BufReader, Write};
 use std::path::Path;
 use walkdir::WalkDir;
-use std::fmt;
+use std::fmt::{self, Display};
 
 // custom derive for ontologies field as vec of Ontology
 fn ontologies_ser<S>(
@@ -56,6 +56,23 @@ where
     Ok(map)
 }
 
+pub struct FailedImport {
+    ontology: GraphIdentifier,
+    error: String,
+}
+
+impl FailedImport {
+    pub fn new(ontology: GraphIdentifier, error: String) -> Self {
+        Self { ontology, error }
+    }
+}
+
+impl Display for FailedImport {
+    fn fmt(&self, f: &mut fmt::Formatter) -> fmt::Result {
+        write!(f, "Failed to import ontology {}: {}", self.ontology, self.error)
+    }
+}
+
 pub struct EnvironmentStatus {
     // true if there is an environment that ontoenv can find
     exists: bool,
@@ -831,25 +848,31 @@ impl OntoEnv {
         Ok(closure)
     }
 
-    /// Returns a graph containing the union of all graphs_ids
+    /// Returns a graph containing the union of all graphs_ids, along with a list of
+    /// graphs that could and could not be imported.
     pub fn get_union_graph(
         &self,
         graph_ids: &[GraphIdentifier],
         rewrite_sh_prefixes: Option<bool>,
         remove_owl_imports: Option<bool>,
-        // TODO: remove_ontology_declarations
-    ) -> Result<Dataset> {
+    ) -> Result<(Dataset, Vec<GraphIdentifier>, Option<Vec<FailedImport>>)> {
         // compute union of all graphs
         let mut union: Dataset = Dataset::new();
         let store = self.store();
+        let mut failed_imports: Vec<FailedImport> = vec![];
+        let mut successful_imports: Vec<GraphIdentifier> = vec![];
         for id in graph_ids {
             let graphname: NamedOrBlankNode = match id.graphname()? {
                 GraphName::NamedNode(n) => NamedOrBlankNode::NamedNode(n),
                 _ => continue,
             };
 
             if !store.contains_named_graph(graphname.as_ref())? {
-                return Err(anyhow::anyhow!("Graph not found: {:?}", id));
+                failed_imports.push(FailedImport {
+                    ontology: id.clone(),
+                    error: "Graph not found".to_string(),
+                });
+                continue;
             }
 
             let mut count = 0;
@@ -881,12 +904,9 @@ impl OntoEnv {
                     ONTOLOGY,
                     graphname.as_ref(),
                 );
-                if !union.remove(to_remove) {
-                    error!("Failed to remove ontology declaration: {:?}", to_remove);
-                }
+                union.remove(to_remove);
             }
-
-
+            successful_imports.push(id.clone());
             info!("Added {} triples from graph: {:?}", count, id);
         }
         let first_id = graph_ids
@@ -896,15 +916,22 @@ impl OntoEnv {
 
         // Rewrite sh:prefixes
         // defaults to true if not specified
-        if let Some(true) = rewrite_sh_prefixes.or(Some(true)) {
+        if rewrite_sh_prefixes.unwrap_or(true) {
             transform::rewrite_sh_prefixes(&mut union, root_ontology);
         }
         // remove owl:imports
-        if let Some(true) = remove_owl_imports.or(Some(true)) {
-            transform::remove_owl_imports(&mut union)
+        if remove_owl_imports.unwrap_or(true) {
+            let to_remove: Vec<NamedNodeRef> = graph_ids.iter().map(|id| id.into()).collect();
+            println!("Removing owl:imports: {:?}", to_remove);
+            transform::remove_owl_imports(&mut union, Some(&to_remove));
         }
         transform::remove_ontology_declarations(&mut union, root_ontology);
-        Ok(union)
+        let failed_imports = if failed_imports.is_empty() {
+            None
+        } else {
+            Some(failed_imports)
+        };
+        Ok((union, successful_imports, failed_imports))
     }
 
     /// Returns a list of issues with the environment
diff --git a/lib/src/ontology.rs b/lib/src/ontology.rs
@@ -55,6 +55,18 @@ impl std::fmt::Display for GraphIdentifier {
     }
 }
 
+impl Into<NamedNode> for GraphIdentifier {
+    fn into(self) -> NamedNode {
+        self.name
+    }
+}
+
+impl<'a> Into<NamedNodeRef<'a>> for &'a GraphIdentifier {
+    fn into(self) -> NamedNodeRef<'a> {
+        (&self.name).into()
+    }
+}
+
 impl GraphIdentifier {
     pub fn new(name: NamedNodeRef) -> Self {
         // location is same as name
diff --git a/lib/src/transform.rs b/lib/src/transform.rs
@@ -1,5 +1,5 @@
 use crate::consts::{DECLARE, IMPORTS, ONTOLOGY, PREFIXES, TYPE};
-use oxigraph::model::{Dataset, Graph, Quad, QuadRef, SubjectRef, Triple, TripleRef};
+use oxigraph::model::{Dataset, Graph, Quad, QuadRef, SubjectRef, Triple, TripleRef, NamedNodeRef, TermRef};
 
 /// Rewrites all sh:prefixes in the graph to point to the provided root
 pub fn rewrite_sh_prefixes(graph: &mut Dataset, root: SubjectRef) {
@@ -68,13 +68,24 @@ pub fn rewrite_sh_prefixes_graph(graph: &mut Graph, root: SubjectRef) {
 
 /// Remove owl:imports statements from a graph. Can be helpful to do after computing the union of
 /// all imports so that downstream tools do not attempt to fetch these graph dependencies
-/// themselves
-pub fn remove_owl_imports(graph: &mut Dataset) {
-    // remove owl:imports
-    let mut to_remove: Vec<Quad> = vec![];
-    for quad in graph.quads_for_predicate(IMPORTS) {
-        to_remove.push(quad.into());
-    }
+/// themselves. If ontologies_to_remove is provided, only remove owl:imports to those ontologies
+pub fn remove_owl_imports(graph: &mut Dataset, ontologies_to_remove: Option<&[NamedNodeRef]>) {
+    let to_remove: Vec<Quad> = graph.quads_for_predicate(IMPORTS)
+        .filter_map(|quad| {
+            match quad.object {
+                TermRef::NamedNode(obj) => {
+                    if ontologies_to_remove.map_or(true, |ontologies| ontologies.contains(&obj)) {
+                        Some(quad.into())
+                    } else {
+                        None
+                    }
+                }
+                _ => None
+            }
+        })
+        .collect();
+
+    // Remove the collected quads
     for quad in to_remove {
         graph.remove(quad.as_ref());
     }
@@ -83,14 +94,25 @@ pub fn remove_owl_imports(graph: &mut Dataset) {
 /// Remove owl:imports statements from a graph. Can be helpful to do after computing the union of
 /// all imports so that downstream tools do not attempt to fetch these graph dependencies
 /// themselves
-pub fn remove_owl_imports_graph(graph: &mut Graph) {
-    // remove owl:imports
-    let mut to_remove: Vec<Triple> = vec![];
-    for triple in graph.triples_for_predicate(IMPORTS) {
-        to_remove.push(triple.into());
-    }
-    for triple in to_remove {
-        graph.remove(triple.as_ref());
+pub fn remove_owl_imports_graph(graph: &mut Graph, ontologies_to_remove: Option<&[NamedNodeRef]>) {
+    let to_remove: Vec<Triple> = graph.triples_for_predicate(IMPORTS)
+        .filter_map(|triple| {
+            match triple.object {
+                TermRef::NamedNode(obj) => {
+                    if ontologies_to_remove.map_or(true, |ontologies| ontologies.contains(&obj)) {
+                        Some(triple.into())
+                    } else {
+                        None
+                    }
+                }
+                _ => None
+            }
+        })
+        .collect();
+
+    // Remove the collected quads
+    for quad in to_remove {
+        graph.remove(quad.as_ref());
     }
 }
 
diff --git a/lib/src/util.rs b/lib/src/util.rs
@@ -1,6 +1,6 @@
 use anyhow::Result;
 
-use std::io::Read;
+use std::io::{Read, Seek};
 use std::path::Path;
 
 use reqwest::header::CONTENT_TYPE;
@@ -58,6 +58,42 @@ pub fn read_file(file: &Path) -> Result<OxigraphGraph> {
     Ok(graph)
 }
 
+fn read_format<T: Read + Seek>(mut original_content: BufReader<T>, format: Option<RdfFormat>) -> Result<OxigraphGraph> {
+    let format = format.unwrap_or(RdfFormat::Turtle);
+    for format in [
+        format,
+        RdfFormat::Turtle,
+        RdfFormat::RdfXml,
+        RdfFormat::NTriples,
+    ] {
+        let content = original_content.get_mut();
+        content.rewind()?;
+        let parser = RdfParser::from_format(format);
+        let mut graph = OxigraphGraph::new();
+        let parser = parser.for_reader(content);
+
+        // Process each quad from the parser
+        for quad in parser {
+            match quad {
+                Ok(q) => {
+                    let triple = Triple::new(q.subject, q.predicate, q.object);
+                    graph.insert(&triple);
+                }
+                Err(_) => {
+                    // Break the outer loop if an error occurs
+                    break;
+                }
+            }
+        }
+
+        // If we successfully processed quads and did not encounter an error
+        if !graph.is_empty() {
+            return Ok(graph);
+        }
+    }
+    Err(anyhow::anyhow!("Failed to parse graph"))
+}
+
 pub fn read_url(file: &str) -> Result<OxigraphGraph> {
     debug!("Reading url: {}", file);
 
@@ -83,39 +119,7 @@ pub fn read_url(file: &str) -> Result<OxigraphGraph> {
     });
 
     let content: BufReader<_> = BufReader::new(std::io::Cursor::new(resp.bytes()?));
-
-    // if content type is known, use it to parse the graph
-    if let Some(format) = content_type {
-        let parser = RdfParser::from_format(format);
-        let mut graph = OxigraphGraph::new();
-        let parser = parser.for_reader(content);
-        for quad in parser {
-            let quad = quad?;
-            let triple = Triple::new(quad.subject, quad.predicate, quad.object);
-            graph.insert(&triple);
-        }
-        return Ok(graph);
-    }
-
-    // if content  type is unknown, try all formats. Requires us to make a copy of the content
-    // since we can't rewind the reader
-    let content_vec: Vec<u8> = content.bytes().map(|b| b.unwrap()).collect();
-
-    for format in [RdfFormat::Turtle, RdfFormat::RdfXml, RdfFormat::NTriples] {
-        let vcontent = BufReader::new(std::io::Cursor::new(&content_vec));
-        let parser = RdfParser::from_format(format);
-        let mut graph = OxigraphGraph::new();
-
-        // TODO: if there's an error on parser.read_triples, try the next format
-        let parser = parser.for_reader(vcontent);
-        for quad in parser {
-            let quad = quad?;
-            let triple = Triple::new(quad.subject, quad.predicate, quad.object);
-            graph.insert(&triple);
-        }
-        return Ok(graph);
-    }
-    Err(anyhow::anyhow!("Failed to parse graph from {}", file))
+    read_format(content, content_type)
 }
 
 // return a "impl IntoIterator<Item = impl Into<Quad>>" for a graph. Iter through
diff --git a/python/src/lib.rs b/python/src/lib.rs
diff --git a/python/src/ontoenv.pyi b/python/src/ontoenv.pyi
