Ajoute une fonction pour hasher le doi

eu_fact_force/ingestion/services.py

@@ -4,6 +4,7 @@
 Create a dedicated file for real pipeline steps.
 """
 
+import hashlib
 from pathlib import Path
 
 from eu_fact_force.ingestion.embedding import add_embeddings
@@ -12,6 +13,13 @@
 from .models import DocumentChunk, FileMetadata, SourceFile
 
 
+def hash_doi(doi: str) -> str:
+    """
+    Hash the DOI to a string of 128 bits.
+    """
+    return hashlib.sha256(doi.encode()).hexdigest()
+
+
 def fetch_file_and_metadata(doi: str) -> tuple[Path, list[str]]:
     """
     Simulate an API call to fetch a PDF and metadata.

tests/ingestion/test_services.py

@@ -0,0 +1,9 @@
+from eu_fact_force.ingestion.services import hash_doi
+
+
+def test_hash_doi():
+    """Test the hash_doi function."""
+    assert (
+        hash_doi("10.1234/example")
+        == "68b8f7c42b3c20b5b49680c9913c11520ec81f3022f0509564140d4ed3f70d78"
+    )