[SERVICE PUBLIC]Update service public definition

helpers/labels/urssaf.json

@@ -0,0 +1,35 @@
+{
+    "CDG PAM": "535104756",
+    "UR ile-de-France (116) Pole GE": "788617793",
+    "URSSAF ILE-DE-FRANCE": "788617793",
+    "URSSAF DE CORSE": "782993141",
+    "Champagne-Ardenne": "752855825",
+    "URSSAF DE PICARDIE": "753663277",
+    "Urssaf Centre Val de Loire": "795120039",
+    "Urssaf de Bourgogne": "794714733",
+    "Urssaf Normandie": "902097997",
+    "Urssaf Midi-Pyrénées-TGE": "535146500",
+    "Nord-Pas-de-Calais": "753673409",
+    "URSSAF DE LORRAINE": "753334481",
+    "URSSAF Alsace": "753570043",
+    "FRANCHE-COMTE": "795070101",
+    "CNTFS FRANCHE-COMTE": "795070101",
+    "URSSAF Pays de Loire": "535104756",
+    "URSSAF Bretagne": "753759570",
+    "URSSAF DU POITOU-CHARENTES": "753644152",
+    "URSSAF DU NORD - LILLE - TGE": "783662059",
+    "URSSAF RHONE ALPES": "794846501",
+    "URSSAF AQUITAINE": "788778777",
+    "URSSAF MIDI-PYRENEES": "535146500",
+    "Urssaf du Limousin": "753919521",
+    "CNTFS": "794846501",
+    "URSSAF D AUVERGNE": "535138218",
+    "URSSAF Languedoc-Roussillon": "753664127",
+    "URSSAF PACA": "794487231",
+    "CGSS DE GUADELOUPE": "314572025",
+    "CGSS DE MARTINIQUE": "314024969",
+    "CGSS DE LA GUYANE": "315190769",
+    "CGSS DE LA REUNION": "314635483",
+    "CGSS DE MAYOTTE": "479281719",
+    "CENTRE DE GESTION PAM": "314635483"
+}

tests/unit_tests/test_data_enrichment.py

@@ -0,0 +1,26 @@
+"""
+import pytest
+from dag_datalake_sirene.workflows.data_pipelines.elasticsearch.data_enrichment import (
+    is_service_public,
+)
+
+# Sample URSSAF SIREN numbers for testing
+urssaf_siren_numbers = {"123456789", "987654321"}
+
+
+@pytest.mark.parametrize(
+    "nature_juridique, siren, expected",
+    [
+        ("4711", "123456789", True),  # Valid prefix
+        ("72", "987654321", True),  # Valid prefix
+        ("1234", "320252489", True),  # BPI France
+        ("1234", "123456789", True),  # URSSAF SIREN
+        ("4120", "775663438", False),  # RATP should be excluded
+        ("1234", "111111111", False),  # Non-public SIREN
+        ("", "123456789", False),  # Empty nature juridique
+        (None, "123456789", False),  # None nature juridique
+    ],
+)
+def test_is_service_public(nature_juridique, siren, expected):
+    assert is_service_public(nature_juridique, siren) == expected
+"""

workflows/data_pipelines/elasticsearch/data_enrichment.py

@@ -34,6 +34,8 @@ def load_file(file_name: str):
 mapping_dep_to_reg = load_file("dep_to_reg.json")
 mapping_role_dirigeants = load_file("roles_dirigeants.json")
 mapping_commune_to_epci = load_file("epci.json")
+urssaf_data = load_file("urssaf.json")
+urssaf_siren_numbers = set(urssaf_data.values())
 
 
 # Nom complet
@@ -153,14 +155,39 @@ def is_ess(est_ess_france, ess_insee):
 
 # Service public
 def is_service_public(nature_juridique_unite_legale, siren):
-    if (
-        nature_juridique_unite_legale
-        and nature_juridique_unite_legale.startswith(("4", "71", "72", "73", "74"))
-    ) or siren == "320252489":
-        return True
-    else:
+    """
+    Determine if a given entity is classified as a public service.
+
+    Parameters:
+    - nature_juridique_unite_legale (str): The legal nature of the entity.
+    - siren (str): The SIREN number of the entity.
+
+    Returns:
+    - bool: True if the entity is classified as a public service,
+            False otherwise. Exceptions include:
+            - BPI France (SIREN: 320252489) and URSSAF are considered public.
+            - RATP (SIREN: 775663438) is explicitly excluded.
+    """
+
+    # Exclude RATP
+    if siren == "775663438":
         return False
 
+    # Define valid prefixes for public service
+    valid_prefixes = {"4", "71", "72", "73", "74"}
+
+    # Check if the entity is classified as a public service
+    is_public = (
+        (
+            nature_juridique_unite_legale
+            and nature_juridique_unite_legale.startswith(tuple(valid_prefixes))
+        )
+        or siren == "320252489"  # BPI France
+        or siren in urssaf_siren_numbers  # Check against URSSAF SIREN numbers
+    )
+
+    return is_public
+
 
 # Association
 def is_association(nature_juridique_unite_legale, identifiant_association):