phase 3

detect/profiles/base.py

@@ -1,17 +1,20 @@
 """
 ContentTypeProfile protocol and config dataclasses.
 
-The pipeline graph is fixed — what varies per content type is configuration
-and hooks. Each profile provides stage configs, a brand dictionary,
-VLM prompt templates, and an aggregation strategy.
+Each profile defines the pipeline topology (as a JSONB blob), stage configs,
+brand dictionary, VLM prompt templates, and aggregation strategy.
+
+When profiles are persisted, the pipeline field is a JSONB column.
+For now, profiles are code-only and pipeline_config() returns a hardcoded value.
 """
 
 from __future__ import annotations
 
 from dataclasses import dataclass, field
-from typing import Protocol
+from typing import Any, Dict, Protocol
 
 from detect.models import BrandDetection, DetectionReport
+from core.schema.models.pipeline_config import PipelineConfig, StageRef, Edge
 
 
 @dataclass
@@ -64,9 +67,24 @@ class CropContext:
     position_hint: str = ""
 
 
+def pipeline_config_from_dict(data: Dict[str, Any]) -> PipelineConfig:
+    """Deserialize a PipelineConfig from a JSONB dict."""
+    stages = [StageRef(**s) for s in data.get("stages", [])]
+    edges = [Edge(**e) for e in data.get("edges", [])]
+    return PipelineConfig(
+        name=data.get("name", ""),
+        profile_name=data.get("profile_name", ""),
+        stages=stages,
+        edges=edges,
+        routing_rules=data.get("routing_rules", {}),
+    )
+
+
 class ContentTypeProfile(Protocol):
     name: str
+    pipeline: Dict[str, Any]  # JSONB blob — PipelineConfig shape
 
+    def pipeline_config(self) -> PipelineConfig: ...
     def frame_extraction_config(self) -> FrameExtractionConfig: ...
     def scene_filter_config(self) -> SceneFilterConfig: ...
     def region_analysis_config(self) -> RegionAnalysisConfig: ...