Made refiner, referencer and builder less strict

video_references.py

@@ -1,4 +1,5 @@
 import json
+import logging
 import traceback
 from typing import Literal
 
@@ -99,6 +100,47 @@ class VideoReferenceBuilder:
                 raise ValueError("Duplicate reference in a single window")
             seen.add(key)
 
+    @staticmethod
+    def _keep_present_references(
+        result: _WindowResult,
+        window: Window[VideoReferenceEvent],
+    ) -> _WindowResult:
+        """Keep the usable part of references that cross a window boundary."""
+        present_positions = {
+            event.id: position
+            for position, event in enumerate(window.present)
+        }
+        references: list[UnresolvedReference] = []
+        seen: set[tuple[tuple[int, ...], str]] = set()
+
+        for reference in result.unresolved:
+            ids = sorted(
+                {event_id for event_id in reference.ids if event_id in present_positions},
+                key=present_positions.__getitem__,
+            )
+            if not ids:
+                logging.warning(
+                    "Ignoring reference outside present window: ids=%s",
+                    reference.ids,
+                )
+                continue
+            if ids != reference.ids:
+                logging.warning(
+                    "Trimming reference to present window: ids=%s -> %s",
+                    reference.ids,
+                    ids,
+                )
+
+            normalized = reference.model_copy(update={"ids": ids})
+            key = (tuple(ids), normalized.type)
+            if key in seen:
+                logging.warning("Ignoring duplicate reference: ids=%s", ids)
+                continue
+            seen.add(key)
+            references.append(normalized)
+
+        return result.model_copy(update={"unresolved": references})
+
     def _build_window(
         self,
         window: Window[VideoReferenceEvent],
@@ -135,6 +177,7 @@ class VideoReferenceBuilder:
 
             try:
                 result = _WindowResult.model_validate_json(response)
+                result = self._keep_present_references(result, window)
                 self._validate_result(result, window)
                 return result
             except (ValidationError, ValueError, TypeError):