fix: conserve l'identite d'un document re-extrait

Le remplacement par ID mettait a jour le document en le supprimant puis en le
recreant : la ligne changeait d'ID, cassant les liens qui la referencent (URL
d'edition, telechargement du PDF) juste apres un balayage de re-extraction. Le
document est desormais mis a jour sur place, seules les donnees derivees
(revenus, depenses) etant regenerees.

Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>

src/plesna_gerance/database/service.py

@@ -171,6 +171,7 @@ class DatabaseService:
         # Check for duplicates and preserve existing file paths if overwriting
         existing_pdf_path = None
         existing_json_path = None
+        reused: Document | None = None  # document mis a jour sur place
         if replace_document_id is not None:
             existing = self.session.get(Document, replace_document_id)
             if existing is None:
@@ -180,20 +181,28 @@ class DatabaseService:
             collision = self.check_duplicate(reference, doc_date)
             if collision is not None and collision.id != existing.id:
                 raise DuplicateDocumentError(reference, doc_date)
-            overwrite = True
+            existing_pdf_path = existing.pdf_path
+            existing_json_path = existing.json_path
+            # Les donnees derivees sont regenerees, mais la ligne document est
+            # conservee : son ID survit a la re-extraction, donc les liens qui
+            # la referencent (URL d'edition, PDF) restent valides.
+            existing.revenus.clear()
+            existing.depenses.clear()
+            self.session.flush()
+            reused = existing
         else:
             existing = self.check_duplicate(reference, doc_date)
-        if existing:
+            if existing:
-            if overwrite:
+                if overwrite:
-                # Preserve existing file paths for reuse
+                    # Preserve existing file paths for reuse
-                existing_pdf_path = existing.pdf_path
+                    existing_pdf_path = existing.pdf_path
-                existing_json_path = existing.json_path
+                    existing_json_path = existing.json_path
-                # Delete existing document (cascade will delete related data)
+                    # Delete existing document (cascade will delete related data)
-                # But DON'T delete files - we'll reuse or update them
+                    # But DON'T delete files - we'll reuse or update them
-                self.session.delete(existing)
+                    self.session.delete(existing)
-                self.session.flush()
+                    self.session.flush()
-            else:
+                else:
-                raise DuplicateDocumentError(reference, doc_date)
+                    raise DuplicateDocumentError(reference, doc_date)
 
         # Get or create immeuble
         immeuble = self.get_or_create_immeuble(
@@ -223,23 +232,29 @@ class DatabaseService:
             pdf_path = existing_pdf_path
             json_path = existing_json_path
 
-        # Create document
+        # Create document (ou mise a jour sur place lors d'un remplacement)
-        document = Document(
+        fields = {
-            reference=reference,
+            "reference": reference,
-            date=doc_date,
+            "date": doc_date,
-            type=doc_info.get("type"),
+            "type": doc_info.get("type"),
-            source_file=source_file,
+            "source_file": source_file,
-            immeuble_id=immeuble.id,
+            "immeuble_id": immeuble.id,
-            json_data=json.dumps(data, ensure_ascii=False, default=str),
+            "json_data": json.dumps(data, ensure_ascii=False, default=str),
-            editeur_nom=editeur_info.get("nom"),
+            "editeur_nom": editeur_info.get("nom"),
-            editeur_siret=editeur_info.get("siret"),
+            "editeur_siret": editeur_info.get("siret"),
-            solde_montant=self._normalize_amount(solde_info.get("montant")),
+            "solde_montant": self._normalize_amount(solde_info.get("montant")),
-            solde_type=solde_info.get("type"),
+            "solde_type": solde_info.get("type"),
-            solde_date_arrete=self._parse_date(solde_info.get("date_arrete")),
+            "solde_date_arrete": self._parse_date(solde_info.get("date_arrete")),
-            pdf_path=pdf_path,
+            "pdf_path": pdf_path,
-            json_path=json_path,
+            "json_path": json_path,
-        )
+        }
-        self.session.add(document)
+        if reused is not None:
+            document = reused
+            for key, value in fields.items():
+                setattr(document, key, value)
+        else:
+            document = Document(**fields)
+            self.session.add(document)
         self.session.flush()
 
         # Save files to storage

tests/test_reextraction.py

@@ -6,7 +6,7 @@ from datetime import date
 import pytest
 
 from plesna_gerance.database import storage
-from plesna_gerance.database.models import Depense, Document, Tag
+from plesna_gerance.database.models import Depense, Document, Revenu, Tag
 from plesna_gerance.database.service import DatabaseService, DuplicateDocumentError
 
 
@@ -36,6 +36,24 @@ def test_replace_document_met_a_jour_meme_si_la_reference_change(
     assert db_session.query(Document).count() == 1
     assert updated.reference == "REF001-CORRIGEE"
     assert updated.date == date(2024, 1, 15)
+    # L'ID survit au remplacement : les liens vers le document restent valides.
+    assert updated.id == original_id
+
+
+def test_replace_document_regenere_les_donnees_derivees(db_session, sample_data):
+    """Revenus et depenses sont remplaces, pas cumules."""
+    service = DatabaseService(db_session)
+    original = service.save_document(data=sample_data)
+
+    new_data = copy.deepcopy(sample_data)
+    new_data["recapitulatif_operations"][0]["montants"]["debit"] = 90.0
+    new_data["situation_locataires"][0]["lignes"][0]["loyers"] = 600.0
+    service.save_document(data=new_data, replace_document_id=original.id)
+
+    depenses = db_session.query(Depense).filter_by(document_id=original.id).all()
+    revenus = db_session.query(Revenu).filter_by(document_id=original.id).all()
+    assert len(depenses) == 1 and depenses[0].debit == 90.0
+    assert len(revenus) == 1 and revenus[0].loyers == 600.0
 
 
 def test_replace_document_conserve_le_pdf_et_reecrit_le_json(