Added video references search (codex)

main.py

@@ -14,6 +14,7 @@ import torch
 from asr import Asr, AsrRawResult
 from asr_filter import AsrFilter, AsrFilterResult
 from asr_eventizer import AsrEventizer
+from video_references import VideoReferenceBuilder
 from structure_builder import Structure, StructureBuilder
 from structure_refiner import StructureRefiner
 from markdown_builder import MarkdownBuilder
@@ -84,6 +85,21 @@ def setup_arguments() -> argparse.Namespace:
         type=str,
         default=ai_api_key
     )
+    parser.add_argument(
+        "--video-ref-ai-model",
+        type=str,
+        default="google/gemini-3.1-flash-lite"
+    )
+    parser.add_argument(
+        "--video-ref-ai-base-url",
+        type=str,
+        default="https://api.proxyapi.ru/v1"
+    )
+    parser.add_argument(
+        "--video-ref-ai-api-key",
+        type=str,
+        default=ai_api_key
+    )
     parser.add_argument(
         "--structure-ai-model",
         type=str,
@@ -215,9 +231,19 @@ def on_video_references(current_step: Step, input_data: dict | None) -> tuple[St
     if not os.path.isfile("video.mp4"):
         logging.info("No video, skipping")
         return (Step.REFERENCE_RESOLVER, None)
-    # NOT IMPLEMENTED
-    logging.warning("Video references are not implemented yet")
-    return (Step.REFERENCE_RESOLVER, None)
+    if input_data is None:
+        logging.error("Can't build video references without input_data")
+        return (None, None)
+    logging.info("Creating the agent")
+    agent = Agent(
+        model=ARGS.video_ref_ai_model,
+        base_url=ARGS.video_ref_ai_base_url,
+        api_key=ARGS.video_ref_ai_api_key
+    )
+    logging.info("Building video references...")
+    builder = VideoReferenceBuilder(agent, Windowizer())
+    result = builder.build(Timeline(**input_data))
+    return (Step.REFERENCE_RESOLVER, result.model_dump(mode="json"))
 
 def on_reference_resolver(current_step: Step, input_data: dict | None) -> tuple[Step | None, dict | None]:
     # do not filter if output file exists
@@ -230,7 +256,7 @@ def on_reference_resolver(current_step: Step, input_data: dict | None) -> tuple[
         return (Step.STRUCTURE_BUILDER, input_data)
     # NOT IMPLEMENTED
     logging.warning("Reference resolver is not implemented yet")
-    return (Step.STRUCTURE_BUILDER, input_data)
+    return (None, None)
 
 def on_structure_builder(current_step: Step, input_data: dict | None) -> tuple[Step | None, dict | None]:
     # don't if done