chore(refactor): legacy post-Wave-1 cleanup (#45, #46, #47, #51)

pyproject.toml

@@ -150,11 +150,11 @@ ignore = ["D100", "D101", "D102", "D103", "D104", "D105", "D107", "D203", "D212"
 # (Use of `assert` detected) is the only rule that meaningfully
 # diverges from the codebase-wide lint config for test files.
 "tests/**/*.py" = ["S101"]
-# Indexing helpers (index.py) and the langchain shim
-# (infrastructure.py) use function-local imports to break
-# circular dependency cycles between the bounded context and
-# the cross-context domain. PLC0415 is locally suppressed.
-"src/bodhi_rag/indexing/**/*.py" = ["PLC0415"]
+# Indexing application helpers (`index.py`, `delete.py`) use
+# function-local imports to break circular dependency cycles
+# between the bounded context and the cross-context domain.
+# PLC0415 is locally suppressed only for the application slice.
+"src/bodhi_rag/indexing/application/**/*.py" = ["PLC0415"]
 
 [tool.mypy]
 python_version = "3.11"

src/bodhi_rag/answering/application/synthesize.py

@@ -10,8 +10,8 @@
 from typing import TYPE_CHECKING
 
 if TYPE_CHECKING:
+    from bodhi_rag.domain.entities import RetrievedDocument
     from bodhi_rag.ports.llm import LLMMessage, LLMPort
-    from bodhi_rag.ports.vector_store import RetrievedDocument
 
 
 SYSTEM_MESSAGE = (

src/bodhi_rag/application/facade.py

@@ -18,41 +18,13 @@
 if TYPE_CHECKING:
     from bodhi_rag.answering.application.synthesize import SynthesizeAnswerUseCase
     from bodhi_rag.conversation.application.memory import ConversationMemoryUseCase
-    from bodhi_rag.domain.entities import ConversationTurn
+    from bodhi_rag.domain.entities import ConversationTurn, RetrievedDocument
     from bodhi_rag.domain.value_objects import ConversationId, DocumentId
     from bodhi_rag.indexing.application.delete import DeleteDocumentUseCase
     from bodhi_rag.indexing.application.index import IndexDocumentUseCase
-    from bodhi_rag.ports.vector_store import RetrievedDocument
     from bodhi_rag.retrieval.application.retrieve import RetrieveQueryUseCase
 
 
-RESERVED_PROVENANCE_KEYS = frozenset(
-    {
-        "source",
-        "source_path",
-        "filename",
-        "file_type",
-        "page_count",
-        "author",
-        "title",
-        "subject",
-    },
-)
-
-
-def _merge_document_metadata(
-    parser_metadata: dict[str, Any],
-    request_metadata: dict[str, Any],
-) -> dict[str, Any]:
-    merged = dict(parser_metadata)
-    for key, value in request_metadata.items():
-        if key in RESERVED_PROVENANCE_KEYS:
-            merged[f"user_{key}"] = value
-            continue
-        merged[key] = value
-    return merged
-
-
 def _citation_source_document(retrieved_document: RetrievedDocument) -> str:
     filename = retrieved_document.metadata.get("filename")
     if isinstance(filename, str) and filename:

src/bodhi_rag/indexing/__init__.py

@@ -8,9 +8,6 @@
 __all__ = [
     "IndexDocumentsRequest",
     "IndexDocumentsResponse",
-    "IndexingSettings",
-    "load_documents_from_directory",
-    "load_documents_from_file",
 ]
 
 _EXPORTS = {
@@ -22,23 +19,13 @@
         "bodhi_rag.application.models",
         "IndexDocumentsResponse",
     ),
-    "IndexingSettings": ("bodhi_rag.indexing.settings", "IndexingSettings"),
-    "load_documents_from_directory": (
-        "bodhi_rag.indexing.infrastructure",
-        "load_documents_from_directory",
-    ),
-    "load_documents_from_file": (
-        "bodhi_rag.indexing.infrastructure",
-        "load_documents_from_file",
-    ),
 }
 
 if TYPE_CHECKING:
     from bodhi_rag.application.models import (
         IndexDocumentsRequest,
         IndexDocumentsResponse,
     )
-    from bodhi_rag.indexing.settings import IndexingSettings
 
 
 def __getattr__(name: str) -> Any:

src/bodhi_rag/interfaces/__init__.py

@@ -1,3 +1,3 @@
 """Interface adapters for bodhi-rag platform entrypoints."""
 
-__all__ = ["api", "cli", "tui", "worker"]
+__all__ = ["api", "cli"]

src/bodhi_rag/ports/reranker.py

@@ -14,7 +14,7 @@
 from typing import TYPE_CHECKING, Protocol, runtime_checkable
 
 if TYPE_CHECKING:
-    from bodhi_rag.ports.vector_store import RetrievedDocument
+    from bodhi_rag.domain.entities import RetrievedDocument
 
 
 @runtime_checkable

src/bodhi_rag/ports/vector_store.py

@@ -9,24 +9,10 @@
 from typing import TYPE_CHECKING, Protocol
 
 if TYPE_CHECKING:
-    from bodhi_rag.domain.entities import Chunk
+    from bodhi_rag.domain.entities import Chunk, RetrievedDocument
     from bodhi_rag.domain.value_objects import DocumentId
 
 
-class RetrievedDocument(Protocol):
-    """
-    A document retrieved from the vector store.
-
-    This is a protocol to avoid coupling to specific implementations.
-    """
-
-    chunk_id: str
-    document_id: str
-    text: str
-    score: float
-    metadata: dict
-
-
 class VectorStorePort(Protocol):
     """
     Protocol for vector storage and retrieval.

src/bodhi_rag/retrieval/application/retrieve.py

@@ -12,9 +12,10 @@
 from typing import TYPE_CHECKING
 
 if TYPE_CHECKING:
+    from bodhi_rag.domain.entities import RetrievedDocument
     from bodhi_rag.ports.embedding import EmbeddingPort
     from bodhi_rag.ports.reranker import RerankerPort
-    from bodhi_rag.ports.vector_store import RetrievedDocument, VectorStorePort
+    from bodhi_rag.ports.vector_store import VectorStorePort
 
 
 class RetrieveQueryUseCase: