docs

ontology_platform/ont_platform/config.py

@@ -58,10 +58,11 @@ class Phase(IntEnum):
 
     BASE = 0  # OntoCast only, filesystem storage
     TRAFILATURA = 1
-    CRAWL4AI = 2
-    GUARDRAILS = 3
-    NEO4J_GRAPHRAG = 4
-    MULTI_AGENT = 5
+    CANDIDATE_REVIEW = 2
+    CRAWL4AI = 3
+    GUARDRAILS = 4
+    NEO4J_GRAPHRAG = 5
+    MULTI_AGENT = 6
 
 
 StorageBackend = Literal["filesystem", "fuseki", "neo4j"]
@@ -94,6 +95,10 @@ class PlatformSettings(BaseSettings):
             "regardless of any Neo4j/Fuseki credentials in the environment."
         ),
     )
+    database_url: str = Field(
+        default="sqlite:///./data/ontology_platform.db",
+        description="SQLAlchemy database URL for Phase 2 candidate/review storage.",
+    )
 
     # ─── Paths (mirror ONTOCAST_* but with platform defaults) ────────
     working_directory: Path = Field(
@@ -116,12 +121,24 @@ class PlatformSettings(BaseSettings):
         default="respect",
         description="robots.txt 준수 정책. Phase 2 Crawl4AI 통합에서 사용.",
     )
+    crawler_default_profile: Literal[
+        "fast_static",
+        "dynamic_page",
+        "full_capture",
+        "structured_extract",
+        "deep_discovery",
+    ] = Field(default="fast_static")
+    crawler_cache_policy: Literal["enabled", "disabled", "bypass"] = Field(default="enabled")
+    crawler_max_pages: int = Field(default=50, ge=1, le=50)
+    crawler_max_depth: int = Field(default=1, ge=0, le=5)
+    text2cypher_result_limit: int = Field(default=100, ge=1, le=1000)
+    graph_search_result_limit: int = Field(default=20, ge=1, le=100)
     daily_llm_call_limit: int = Field(default=10_000)
     daily_llm_token_limit: int = Field(default=10_000_000)
 
     # ─── Phase 0 enforcement ─────────────────────────────────────────
     @model_validator(mode="after")
-    def _enforce_phase_storage_consistency(self) -> "PlatformSettings":
+    def _enforce_phase_storage_consistency(self) -> PlatformSettings:
         """Phase 0 forces filesystem; later phases may opt into other backends.
 
         Anything other than 'filesystem' before Phase 4 is treated as a
@@ -130,7 +147,8 @@ class PlatformSettings(BaseSettings):
         """
         if self.phase < Phase.NEO4J_GRAPHRAG and self.storage_backend != "filesystem":
             raise ValueError(
-                f"storage_backend={self.storage_backend!r} requires Phase 4+, "
+                f"storage_backend={self.storage_backend!r} requires Phase 4+/Phase 5+ "
+                f"(Phase 5 in the current roadmap), "
                 f"but PHASE={int(self.phase)}. See docs/통합설계서.md §5."
             )
         # Ensure working directory exists for filesystem mode.
@@ -248,10 +266,10 @@ def load_settings() -> PlatformSettings:
 # Provider-grade LLM/embedding config helpers can be added in later phases.
 # For Phase 0, OntoCast's own ``LLMConfig`` is sufficient.
 __all__ = [
+    "LLMConfig",
     "Phase",
     "PlatformSettings",
     "StorageBackend",
     "build_ontocast_config",
     "load_settings",
-    "LLMConfig",
 ]