web/app.py

@@ -5145,6 +5145,11 @@ def _make_progress_publisher(task_id: str, filename: str):
 
 
 class PrecedentUpdateRequest(BaseModel):
+    # case_number is the canonical identifier (e.g. "8027-25"). It is editable
+    # so a wrong identifier captured at upload (e.g. the full citation pasted
+    # into the field) can be corrected from the edit screen. update_case_law
+    # already whitelists it.
+    case_number: str | None = None
     case_name: str | None = None
     court: str | None = None
     decision_date: str | None = None
@@ -5488,6 +5493,7 @@ async def internal_decisions_upload(
     file: UploadFile = File(...),
     case_number: str = Form(...),
     case_name: str = Form(""),
+    citation: str = Form(""),
     court: str = Form(""),
     decision_date: str = Form(""),
     chair_name: str = Form(""),
@@ -5498,7 +5504,14 @@ async def internal_decisions_upload(
     is_binding: bool = Form(True),
     summary: str = Form(""),
 ):
-    """Upload a planning appeals-committee decision to the internal corpus."""
+    """Upload a planning appeals-committee decision to the internal corpus.
+
+    ``case_number`` is the canonical identifier (e.g. "8027-25"); ``citation``
+    is the full מראה-מקום (e.g. "ערר ... 8027/25 פלוני נ' הוועדה ..."). They
+    are distinct fields — previously the UI sent the citation as case_number,
+    leaving the identifier polluted and citation_formatted empty until the
+    metadata extractor ran. citation is stored as citation_formatted up-front
+    so it survives even if extraction is delayed."""
     if practice_area and practice_area not in _PRACTICE_AREAS:
         raise HTTPException(400, "practice_area לא תקין")
     if not case_number.strip():
@@ -5556,6 +5569,20 @@ async def internal_decisions_upload(
             # — precedent_library_upload and missing-precedent — already do
             # this; this path was missing it).
             case_law_id = result.get("case_law_id") if isinstance(result, dict) else None
+            # Persist the מראה-מקום the chair typed, up-front. The metadata
+            # extractor only fills citation_formatted when it is empty, so this
+            # preserves the user's exact citation rather than waiting on (or
+            # being overwritten by) extraction.
+            if case_law_id and citation.strip():
+                try:
+                    await db.update_case_law(
+                        UUID(case_law_id), citation_formatted=citation.strip()
+                    )
+                except Exception:
+                    logger.warning(
+                        "internal-decision %s: storing citation_formatted failed",
+                        case_number,
+                    )
             extraction_queued = True
             if case_law_id:
                 # Route to the correct company CEO. _get_company_id keys off