Made refiner, referencer and builder less strict

structure_builder.py

@@ -1,4 +1,5 @@
 import json
+import logging
 import traceback
 from typing import Annotated, Literal
 
@@ -180,6 +181,49 @@ class StructureBuilder:
                 "pending.event_ids must reference events from the current window"
             )
 
+    @staticmethod
+    def _keep_window_references(
+        result: _BuildResult,
+        window: Window[Event],
+    ) -> _BuildResult:
+        """Drop impossible image references and repair technical pending IDs."""
+        visual_ids = {
+            event.id
+            for event in window.past + window.present
+            if event.type == "vis"
+        }
+        elements: list[StructureElement] = []
+        for element in result.elements:
+            if isinstance(element, ImageElement) and element.event_id not in visual_ids:
+                logging.warning("Ignoring image outside the current window: id=%s", element.event_id)
+                continue
+            elements.append(element)
+
+        context = result.context
+        if context.pending is not None:
+            window_ids = {
+                event.id
+                for event in window.past + window.present + window.future
+            }
+            pending_ids = [
+                event_id
+                for event_id in context.pending.event_ids
+                if event_id in window_ids
+            ]
+            if pending_ids != context.pending.event_ids:
+                logging.warning(
+                    "Trimming pending IDs to the current window: %s -> %s",
+                    context.pending.event_ids,
+                    pending_ids,
+                )
+                pending = (
+                    context.pending.model_copy(update={"event_ids": pending_ids})
+                    if pending_ids else None
+                )
+                context = context.model_copy(update={"pending": pending})
+
+        return result.model_copy(update={"elements": elements, "context": context})
+
     def _build_window(self, window: Window[Event]) -> _BuildResult:
         """Build document elements from a single timeline window."""
         messages = [
@@ -211,6 +255,7 @@ class StructureBuilder:
 
             try:
                 result = _BuildResult.model_validate_json(response)
+                result = self._keep_window_references(result, window)
                 self._validate_result(result, window)
                 return result
             except (ValidationError, ValueError, TypeError):