Add callbacks instead of printing to stdout when a data issue is found

This is useful for osrd-data, where such issues are usually reported
through the errdb mechanism, allowing more precise statistics.

Signed-off-by: Younes Khoudli <younes.khoudli@epita.fr>

Author: Khoyo

Cargo.lock

@@ -1,6 +1,6 @@
 [package]
 name = "liblrs"
-version = "0.4.3"
+version = "0.5.0"
 edition = "2024"
 description = "Library to manipulate linear referencing systems"
 license = "MIT"

Cargo.toml

@@ -1,7 +1,7 @@
 [package]
 name = "liblrs_python"
 description = "Python bindings for liblrs: a library to work with linear referencing systems"
-version = "0.4.3"
+version = "0.5.0"
 edition = "2024"
 license = "MIT"
 repository = "https://github.com/OpenRailAssociation/liblrs/"

python/Cargo.toml

@@ -1,6 +1,6 @@
 [project]
 name = "liblrs_python"
-version = "0.4.3"
+version = "0.5.0"
 requires-python = ">=3.12"
 dependencies = ["pip>=25.2"]
 

python/liblrs_python.pyi

@@ -27,6 +27,7 @@ fn liblrs_python(_py: Python, m: &Bound<'_, PyModule>) -> PyResult<()> {
     m.add_class::<AnchorOnLrm>()?;
     m.add_class::<SegmentOfTraversal>()?;
     m.add_class::<Builder>()?;
+    m.add_class::<DataIssueReporter>()?;
     Ok(())
 }
 
@@ -576,15 +577,22 @@ impl Builder {
     /// Read the topology from an OpenStreetMap source
     ///
     /// It reads the nodes, segments and traversals.
-    pub fn read_from_osm(
+    pub fn read_from_osm<'py>(
         &mut self,
         input_osm_file: PathBuf,
         lrm_tag: String,
         required: Vec<(String, String)>,
         to_reject: Vec<(String, String)>,
+        reporter: Bound<'py, DataIssueReporter>,
     ) {
-        self.inner
-            .read_from_osm(&input_osm_file, &lrm_tag, required, to_reject)
+        let mut reporter = PythonDataIssueReporter(reporter);
+        self.inner.read_from_osm(
+            &input_osm_file,
+            &lrm_tag,
+            required,
+            to_reject,
+            Some(&mut reporter),
+        )
     }
 
     /// Save the lrs to a file
@@ -656,3 +664,52 @@ impl Builder {
 }
 
 define_stub_info_gatherer!(stub_info);
+
+#[gen_stub_pyclass]
+#[pyclass(subclass)]
+struct DataIssueReporter {}
+
+#[pymethods]
+impl DataIssueReporter {
+    #[new]
+    fn new() -> Self {
+        Self {}
+    }
+}
+
+#[allow(unused_variables)]
+#[gen_stub_pymethods]
+impl liblrs::DataIssueReporter for DataIssueReporter {
+    fn report_ignoring_traversal_edges(
+        &mut self,
+        traversal_ref: &str,
+        ignored_count: usize,
+        total_count: usize,
+        first_node: i64,
+        last_node: i64,
+    ) {
+    }
+}
+
+struct PythonDataIssueReporter<'a>(Bound<'a, DataIssueReporter>);
+impl liblrs::DataIssueReporter for PythonDataIssueReporter<'_> {
+    fn report_ignoring_traversal_edges(
+        &mut self,
+        traversal_ref: &str,
+        ignored_count: usize,
+        total_count: usize,
+        first_node: i64,
+        last_node: i64,
+    ) {
+        let _ = self.0.call_method1(
+            "report_ignoring_traversal_edges",
+            (
+                traversal_ref,
+                ignored_count,
+                total_count,
+                first_node,
+                last_node,
+            ),
+        );
+    }
+}

python/pyproject.toml

@@ -13,7 +13,7 @@ use crate::lrs::Properties;
 use crate::lrs_ext::ExtLrs;
 use crate::lrs_generated::{self, *};
 use crate::osm_helpers::sort_edges;
-use crate::properties;
+use crate::{DataIssueReporter, properties};
 
 /// The linear position of an [`Anchor`] doesn’t always match the measured distance.
 /// For example if a road was transformed into a bypass, resulting in a longer road,
@@ -436,7 +436,10 @@ impl<'fbb> Builder<'fbb> {
         lrm_tag: &str,
         required: Vec<(String, String)>,
         to_reject: Vec<(String, String)>,
+        reporter: Option<&mut dyn DataIssueReporter>,
     ) {
+        let mut default_reporter = ();
+        let reporter = reporter.unwrap_or(&mut default_reporter);
         let mut reader = osm4routing::Reader::new().merge_ways().read_tag(lrm_tag);
 
         for (key, value) in required.iter() {
@@ -479,7 +482,7 @@ impl<'fbb> Builder<'fbb> {
 
         // Sort the traversals
         for (srv_ref, edges) in traversals.into_iter() {
-            let segments: Vec<_> = sort_edges(edges, &srv_ref)
+            let segments: Vec<_> = sort_edges(edges, &srv_ref, reporter)
                 .into_iter()
                 .map(|(edge, reversed)| SegmentOfTraversal {
                     segment_index: edges_map[&edge.id],

python/src/lib.rs

@@ -44,6 +44,7 @@ fn main() {
         &cli_args.lrm_tag,
         required,
         to_reject,
+        Some(&mut liblrs::LoggingDataIssueReporter),
     );
 
     builder.save(

python/uv.lock

@@ -21,6 +21,46 @@ pub mod lrs_ext;
 #[deny(missing_docs)]
 pub mod builder;
 
+pub trait DataIssueReporter {
+    fn report_ignoring_traversal_edges(
+        &mut self,
+        traversal_ref: &str,
+        ignored_count: usize,
+        total_count: usize,
+        first_node: i64,
+        last_node: i64,
+    );
+}
+
+pub struct LoggingDataIssueReporter;
+
+impl DataIssueReporter for LoggingDataIssueReporter {
+    fn report_ignoring_traversal_edges(
+        &mut self,
+        traversal_ref: &str,
+        ignored_count: usize,
+        total_count: usize,
+        first_node: i64,
+        last_node: i64,
+    ) {
+        println!(
+            "[WARN] on traversal {traversal_ref}, ignoring {ignored_count} edges out of {total_count}. Sorted from {first_node} to {last_node}"
+        );
+    }
+}
+
+impl DataIssueReporter for () {
+    fn report_ignoring_traversal_edges(
+        &mut self,
+        _traversal_ref: &str,
+        _ignored_count: usize,
+        _total_count: usize,
+        _first_node: i64,
+        _last_node: i64,
+    ) {
+    }
+}
+
 #[test]
 fn read_and_write_lrs() {
     use builder::*;