fix(crud): strip view-only join columns from order_by_li to prevent ambiguous account_id in WHERE

Sorting by join-derived columns (e.g. event_presenter_family_name on v_event_file)
caused MariaDB to expand the view's JOIN inline, making the unqualified account_id
clause from sql_and_qry_part ambiguous — resulting in a 400 SQL error. filter_order_by
now accepts raw_table_name and validates ORDER BY columns against the physical table
only; join-only columns are silently stripped. Also switches filter_order_by off the
global db connection to engine.connect() context managers. Updated all four call sites
in api_crud_v3.py and api_crud_v3_nested.py.

Docs: add order_by_li raw-table limitation and direct download link patterns to
GUIDE__AE_API_V3_for_Frontend.md; record fix in TODO__Agents.md.

Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>

app/lib_api_crud_v3.py

@@ -140,30 +140,37 @@ def apply_forced_account_filter(and_qry_dict: Optional[Dict], account: AccountCo
         
     return forced
 
-def filter_order_by(order_by_li: Any, model: Any, table_name: str = None) -> Optional[Dict[str, str]]:
+def filter_order_by(order_by_li: Any, model: Any, table_name: str = None, raw_table_name: str = None) -> Optional[Dict[str, str]]:
     """
     Sanitize Sorting Parameters.
-    
+
     Prevents SQL injection and logic errors by validating that requested sort columns
     actually exist in the Pydantic model and/or the database table.
+
+    When raw_table_name is provided (the physical table, not the view), columns are
+    validated against it instead of table_name. This prevents view-only join columns
+    (e.g. event_presenter_family_name on v_event_file) from reaching ORDER BY —
+    those columns cause MariaDB to expand the view's JOIN inline, making unqualified
+    WHERE references like `account_id` ambiguous across joined tables.
     """
     if not order_by_li or not isinstance(order_by_li, dict) or not model:
         return order_by_li
     if not hasattr(model, '__fields__'):
         return order_by_li
-    
+
     model_fields = set(model.__fields__.keys())
     model_fields.update({f.alias for f in model.__fields__.values() if f.alias})
     filtered = {k: v for k, v in order_by_li.items() if k in model_fields}
-    
-    if table_name and filtered:
-        from app.db_sql import db
+
+    check_table = raw_table_name or table_name
+    if check_table and filtered:
+        from app import lib_sql_core
         from sqlalchemy import text
         final_filtered = {}
         for column in filtered:
             try:
-                # Lightweight check to see if column exists in SQL
-                db.execute(text(f"SELECT `{column}` FROM `{table_name}` LIMIT 0"))
+                with lib_sql_core.engine.connect() as conn:
+                    conn.execute(text(f"SELECT `{column}` FROM `{check_table}` LIMIT 0"))
                 final_filtered[column] = filtered[column]
             except Exception:
                 pass

app/routers/api_crud_v3.py

@@ -226,11 +226,12 @@ async def get_obj_li(
 
     table_name = obj_cfg.get(f'tbl_{view}', obj_cfg.get('tbl_default', obj_cfg.get('tbl')))
     base_name = obj_cfg.get(f'mdl_{view}', obj_cfg.get('mdl_default', obj_cfg.get('mdl')))
+    raw_table_name = obj_cfg.get('tbl_update', obj_cfg.get('tbl'))
 
     if not table_name or not base_name:
         return mk_resp(data=False, status_code=500, response=response, status_message=f"Configuration error.")
 
-    order_by_li = filter_order_by(order_by_li, base_name, table_name)
+    order_by_li = filter_order_by(order_by_li, base_name, table_name, raw_table_name=raw_table_name)
     status_filter = get_supported_filters(base_name, status_filter)
 
     if not obj_cfg.get('public_read', False):
@@ -335,11 +336,12 @@ async def search_obj_li(
 
     table_name = obj_cfg.get(f'tbl_{view}', obj_cfg.get('tbl_default', obj_cfg.get('tbl')))
     base_name = obj_cfg.get(f'mdl_{view}', obj_cfg.get('mdl_default', obj_cfg.get('mdl')))
+    raw_table_name = obj_cfg.get('tbl_update', obj_cfg.get('tbl'))
 
     if not table_name or not base_name:
         return mk_resp(data=False, status_code=500, response=response, status_message=f"Configuration error.")
 
-    order_by_li = filter_order_by(order_by_li, base_name, table_name)
+    order_by_li = filter_order_by(order_by_li, base_name, table_name, raw_table_name=raw_table_name)
     status_filter = get_supported_filters(base_name, status_filter)
     searchable_fields = obj_cfg.get('searchable_fields')
 

app/routers/api_crud_v3_nested.py

@@ -84,6 +84,7 @@ async def get_child_obj_li(
     obj_cfg = obj_type_kv_li[obj_name]
     table_name = obj_cfg.get(f'tbl_{view}', obj_cfg.get('tbl_default', obj_cfg.get('tbl')))
     base_name = obj_cfg.get(f'mdl_{view}', obj_cfg.get('mdl_default', obj_cfg.get('mdl')))
+    raw_table_name = obj_cfg.get('tbl_update', obj_cfg.get('tbl'))
 
     # Log parent/child resolution details (use INFO so logs appear in production)
     log.info("nested.list start parent=%s parent_table=%s parent_id_random=%s child=%s table=%s allowed_parents=%s", parent_obj_type, parent_table, parent_obj_id, obj_name, table_name, obj_cfg.get('parent_types'))
@@ -91,7 +92,7 @@ async def get_child_obj_li(
     if not table_name or not base_name:
         return mk_resp(data=False, status_code=500, response=response, status_message=f"Configuration error.")
 
-    order_by_li = filter_order_by(order_by_li, base_name, table_name)
+    order_by_li = filter_order_by(order_by_li, base_name, table_name, raw_table_name=raw_table_name)
     status_filter = get_supported_filters(base_name, status_filter)
 
     resolved_parent_id = redis_lookup_id_random(record_id_random=parent_obj_id, table_name=parent_table)
@@ -187,11 +188,12 @@ async def search_child_obj_li(
     obj_cfg = obj_type_kv_li[obj_name]
     table_name = obj_cfg.get(f'tbl_{view}', obj_cfg.get('tbl_default', obj_cfg.get('tbl')))
     base_name = obj_cfg.get(f'mdl_{view}', obj_cfg.get('mdl_default', obj_cfg.get('mdl')))
+    raw_table_name = obj_cfg.get('tbl_update', obj_cfg.get('tbl'))
 
     if not table_name or not base_name:
         return mk_resp(data=False, status_code=500, response=response, status_message=f"Configuration error.")
 
-    order_by_li = filter_order_by(order_by_li, base_name, table_name)
+    order_by_li = filter_order_by(order_by_li, base_name, table_name, raw_table_name=raw_table_name)
     status_filter = get_supported_filters(base_name, status_filter)
     searchable_fields = obj_cfg.get('searchable_fields')
 

documentation/GUIDE__AE_API_V3_for_Frontend.md

@@ -96,6 +96,26 @@ The primary way to retrieve data.
 *   **Endpoint:** `POST /v3/crud/{obj_type}/search`
 *   **Security:** Automatically filters results to only show records belonging to your `x-account-id`. If no account context is provided, it will return **0 records** for private objects.
 
+#### Sorting with `order_by_li`
+
+Pass a JSON object as the `order_by_li` query parameter to sort results:
+
+```ts
+// ?order_by_li={"filename":"ASC","created_on":"DESC"}
+const params = new URLSearchParams({
+  order_by_li: JSON.stringify({ filename: 'ASC', created_on: 'DESC' })
+});
+```
+
+> [!IMPORTANT]
+> **`order_by_li` only accepts columns from the raw base table** — not view-only join columns.
+>
+> Some object types (e.g. `event_file`) have enriched views that JOIN other tables to expose convenience fields like `event_presenter_family_name`. These are available in search results when using `?view=alt`, but they **cannot** be used in `order_by_li`. Attempting to sort by them silently drops those sort keys (the query proceeds without them).
+>
+> If you need to sort by a joined field, sort client-side on the returned list.
+>
+> **Columns safe to sort on for `event_file`:** any field in the `event_file` table itself — `filename`, `title`, `extension`, `created_on`, `updated_on`, `sort`, `enable`, etc.
+
 ### C. POST Create / PATCH Update
 Modify data in the system.
 *   **Endpoints:**
@@ -281,6 +301,47 @@ Every Event File (`event_file`) **must** have a linked Hosted File (`hosted_file
     *   `hosted_file_size` (string - in bytes)
 2.  **Nested Hosted File Object:** A full `hosted_file` object will be nested under the `hosted_file` key. This object (`Hosted_File_Base` model) will contain all its standard fields, including `id` (random string ID), `hash_sha256`, `content_type`, `size`, etc.
 
+### Direct Download Links (Shareable / External)
+
+Event files can be downloaded without standard auth headers using one of two bypass mechanisms. This is useful for generating shareable links for staff or external recipients.
+
+- **Method:** `GET`
+- **Path:** `/v3/action/event_file/{event_file_id}/download`
+
+The endpoint also accepts `hosted_file_id` or `archive_content_id` directly — it resolves the chain automatically.
+
+#### Auth bypass options
+
+| Query param | Value | When to use |
+|---|---|---|
+| `?key=<account_id_random>` | Any valid account random ID | Staff sharing within a known account context |
+| `?site_key=<site_access_key>` | The site's `access_key` value | Public or semi-public distribution tied to a specific site |
+
+Either param replaces the need for `x-aether-api-key` / `x-account-id` headers, so the URL is self-contained and works in a plain browser tab or `<a href>` link.
+
+#### Optional params
+
+| Query param | Description |
+|---|---|
+| `filename` | Override the download filename (min 4 chars). Useful for giving files clean display names. |
+
+#### Building a shareable link
+
+```ts
+// Build a self-contained download URL for staff/external use
+function makeDownloadUrl(eventFileId: string, accountId: string, displayName?: string): string {
+  const base = `https://dev-api.oneskyit.com/v3/action/event_file/${eventFileId}/download`;
+  const params = new URLSearchParams({ key: accountId });
+  if (displayName) params.set('filename', displayName);
+  return `${base}?${params}`;
+}
+```
+
+The endpoint supports byte-range requests (`Range` header), so it works correctly for in-browser media streaming as well as direct file downloads.
+
+> [!NOTE]
+> The `?key=` bypass verifies only that the account ID exists — it does not confirm the file belongs to that account. It is appropriate for internal staff tools. For publicly distributed links, prefer `?site_key=` which ties access to a specific site's configured key.
+
 ---
 
 ## 6. Hosted File Actions: Convert & Clip (Frontend Notes)
@@ -301,18 +362,16 @@ These helper endpoints let the frontend request small server-side transformation
   - Required query params: `link_to_type`, `link_to_id`, `start_time`, `end_time` (format `HH:MM:SS`)
   - Optional query params: `filename_no_ext` (defaults to `automated_hosted_file_clip_video`), `reencode` (bool), `scale_down` (bool)
   - Auth: standard V3 headers
-  - Behavior: extracts a clip using `ffmpeg` and saves it as a new `hosted_file`. Defaults to stream-copying to be fast; set `reencode=true` to force H.264 or `scale_down=true` to resize. Returns 400 on failure.
-   - Behavior: extracts a clip using `ffmpeg` and saves it as a new `hosted_file`.
-     - Defaults to stream-copying to be fast; set `reencode=true` to force H.264 or `scale_down=true` to resize.
-     - For longer-running clips you can schedule the job in the background by adding `?background=true`. When scheduled the API returns `202 Accepted` and the clip runs asynchronously on the server; check the returned `hosted_file` record later via the standard V3 `hosted_file` endpoints.
-     - Returns 400 on synchronous failure; returns 202 when scheduled successfully.
+  - Behavior: extracts a clip using `ffmpeg` and saves it as a new `hosted_file`.
+    - Defaults to stream-copying (fast); set `reencode=true` to force H.264 or `scale_down=true` to resize.
+    - Add `?background=true` to schedule the clip asynchronously — returns `202 Accepted` immediately; poll the `hosted_file` record for completion.
+    - Returns 400 on synchronous failure; 202 when scheduled successfully.
 
 Frontend guidance:
 
 - Call these routes with the same `link_to_type` / `link_to_id` you plan to associate the resulting hosted_file with — the server resolves random IDs for you.
 - After a successful response, use the V3 `hosted_file` action endpoints (download/delete) to manage or retrieve the new file.
-- These endpoints run synchronously and can take time for large inputs; for heavy or batch workloads use a queued job pattern instead.
- - These endpoints may take time for large inputs. Prefer using `?background=true` to schedule work and receive a `202 Accepted` response for async processing. For heavy or batch workloads use a queued job pattern instead.
+- Prefer `?background=true` for large inputs to avoid request timeouts. For heavy or batch workloads use a queued job pattern instead.
 
 ---
 

documentation/TODO__Agents.md

@@ -23,6 +23,7 @@
 - [x] **[P4] Expose `pool_size` / `max_overflow` as env vars** — `create_ae_engine()` calls `settings.DB.get('pool_size', 10)` but `settings.DB` property doesn't include those keys, so they're always hardcoded 10/20. Add `AE_DB_POOL_SIZE` / `AE_DB_POOL_MAX_OVERFLOW` to `config.py`.
 
 ## 📋 Feature Tasks
+- [x] **`order_by_li` view-join ambiguity fix (June 2026):** Using view-only join columns (e.g. `event_presenter_family_name` from `v_event_file`) in `order_by_li` caused MariaDB error "Unknown column 'account_id' in WHERE" (HTTP 400). Root cause: `filter_order_by` validated columns against the view — which passes for join-derived fields — and `sql_and_qry_part` generates an unqualified `account_id =` clause that becomes ambiguous when MariaDB expands the view's JOIN inline. Fix: `filter_order_by` now accepts `raw_table_name` and validates ORDER BY columns against the physical table only. Join-only view columns are silently stripped. Updated all three call sites in `api_crud_v3.py` (×2) and `api_crud_v3_nested.py` (×2). **Follow-up (lower priority):** qualify `account_id` in `sql_and_qry_part` to fix the root ambiguity for any future JOIN-capable views.
 - [x] **Core Isolation:** Harden `apply_forced_account_filter` to Fail-Closed.
 - [x] **IDAA Baseline:** Remove `public_read` from Event, CMS, and Archive objects.
 - [x] **Detailed Feedback:** Implement descriptive 403 Forbidden reasons.