Merge pull request #20 from saezlab/feat/us3.2-graphblas-implementation

With this change, a user can choose if use Pronto or Graphblas as backend. Additionally, the load function has been unified to simplify the user experience.

Author: ecarrenolozano

notebooks/_graphblas_experiments.ipynb

@@ -12,6 +12,7 @@
 from functools import cached_property
 
 import pronto
+from charset_normalizer import from_path
 
 from ontograph.models import Ontology, CatalogOntologies
 from ontograph.downloader import (
@@ -155,6 +156,18 @@ def _extract_ontology_id(self, ontology: pronto.Ontology) -> str | None:
             )
             return None
 
+    def find_file_encoding(self, file: str | Path) -> str | None:
+        """Detect the encoding of a file.
+
+        Args:
+            file (str | Path): Path to the file whose encoding should be detected.
+
+        Returns:
+            str | None: The detected encoding, or None if it cannot be determined.
+        """
+        result = from_path(file).best()
+        return result.encoding
+
     def _load_ontology(
         self, path_file: Path
     ) -> tuple[pronto.Ontology, str | None]:
@@ -178,7 +191,9 @@ def _load_ontology(
 
         logger.debug(f'Parsing ontology file with Pronto: {path_file}')
         try:
-            ontology: pronto.Ontology = pronto.Ontology(path_file)
+            ontology: pronto.Ontology = pronto.Ontology(
+                path_file, encoding=self.find_file_encoding(path_file)
+            )
         except (TypeError, ValueError) as e:
             error_msg = f'Failed to load ontology from {path_file}: {str(e)}'
             logger.exception(error_msg)