haiku.rag 0.4.3__py3-none-any.whl → 0.5.1__py3-none-any.whl

haiku/rag/chunker.py

@@ -1,6 +1,9 @@
 from typing import ClassVar
 
 import tiktoken
+from docling.chunking import HybridChunker  # type: ignore
+from docling_core.transforms.chunker.tokenizer.openai import OpenAITokenizer
+from docling_core.types.doc.document import DoclingDocument
 
 from haiku.rag.config import Config
 
@@ -8,9 +11,11 @@ from haiku.rag.config import Config
 class Chunker:
     """A class that chunks text into smaller pieces for embedding and retrieval.
 
+    Uses docling's structure-aware chunking to create semantically meaningful chunks
+    that respect document boundaries.
+
     Args:
         chunk_size: The maximum size of a chunk in tokens.
-        chunk_overlap: The number of tokens of overlap between chunks.
     """
 
     encoder: ClassVar[tiktoken.Encoding] = tiktoken.encoding_for_model("gpt-4o")
@@ -18,50 +23,29 @@ class Chunker:
     def __init__(
         self,
         chunk_size: int = Config.CHUNK_SIZE,
-        chunk_overlap: int = Config.CHUNK_OVERLAP,
     ):
         self.chunk_size = chunk_size
-        self.chunk_overlap = chunk_overlap
+        tokenizer = OpenAITokenizer(
+            tokenizer=tiktoken.encoding_for_model("gpt-4o"), max_tokens=chunk_size
+        )
+
+        self.chunker = HybridChunker(tokenizer=tokenizer)  # type: ignore
 
-    async def chunk(self, text: str) -> list[str]:
-        """Split the text into chunks based on token boundaries.
+    async def chunk(self, document: DoclingDocument) -> list[str]:
+        """Split the document into chunks using docling's structure-aware chunking.
 
         Args:
-            text: The text to be split into chunks.
+            document: The DoclingDocument to be split into chunks.
 
         Returns:
-            A list of text chunks with token-based boundaries and overlap.
+            A list of text chunks with semantic boundaries.
         """
-        if not text:
+        if document is None:
             return []
 
-        encoded_tokens = self.encoder.encode(text, disallowed_special=())
-
-        if self.chunk_size > len(encoded_tokens):
-            return [text]
-
-        chunks = []
-        i = 0
-        split_id_counter = 0
-        while i < len(encoded_tokens):
-            # Overlap
-            start_i = i
-            end_i = min(i + self.chunk_size, len(encoded_tokens))
-
-            chunk_tokens = encoded_tokens[start_i:end_i]
-            chunk_text = self.encoder.decode(chunk_tokens)
-
-            chunks.append(chunk_text)
-            split_id_counter += 1
-
-            # Exit loop if this was the last possible chunk
-            if end_i == len(encoded_tokens):
-                break
-
-            i += (
-                self.chunk_size - self.chunk_overlap
-            )  # Step forward, considering overlap
-        return chunks
+        # Chunk using docling's hybrid chunker
+        chunks = list(self.chunker.chunk(document))
+        return [self.chunker.contextualize(chunk) for chunk in chunks]
 
 
 chunker = Chunker()

haiku/rag/cli.py

@@ -1,4 +1,5 @@
 import asyncio
+import warnings
 from importlib.metadata import version
 from pathlib import Path
 
@@ -9,12 +10,14 @@ from haiku.rag.app import HaikuRAGApp
 from haiku.rag.config import Config
 from haiku.rag.utils import is_up_to_date
 
+if not Config.ENV == "development":
+    warnings.filterwarnings("ignore")
+
 cli = typer.Typer(
     context_settings={"help_option_names": ["-h", "--help"]}, no_args_is_help=True
 )
 
 console = Console()
-event_loop = asyncio.get_event_loop()
 
 
 async def check_version():
@@ -46,7 +49,7 @@ def main(
 ):
     """haiku.rag CLI - SQLite-based RAG system"""
     # Run version check before any command
-    event_loop.run_until_complete(check_version())
+    asyncio.run(check_version())
 
 
 @cli.command("list", help="List all stored documents")
@@ -58,7 +61,7 @@ def list_documents(
     ),
 ):
     app = HaikuRAGApp(db_path=db)
-    event_loop.run_until_complete(app.list_documents())
+    asyncio.run(app.list_documents())
 
 
 @cli.command("add", help="Add a document from text input")
@@ -73,7 +76,7 @@ def add_document_text(
     ),
 ):
     app = HaikuRAGApp(db_path=db)
-    event_loop.run_until_complete(app.add_document_from_text(text=text))
+    asyncio.run(app.add_document_from_text(text=text))
 
 
 @cli.command("add-src", help="Add a document from a file path or URL")
@@ -88,7 +91,7 @@ def add_document_src(
     ),
 ):
     app = HaikuRAGApp(db_path=db)
-    event_loop.run_until_complete(app.add_document_from_source(file_path=file_path))
+    asyncio.run(app.add_document_from_source(file_path=file_path))
 
 
 @cli.command("get", help="Get and display a document by its ID")
@@ -103,7 +106,7 @@ def get_document(
     ),
 ):
     app = HaikuRAGApp(db_path=db)
-    event_loop.run_until_complete(app.get_document(doc_id=doc_id))
+    asyncio.run(app.get_document(doc_id=doc_id))
 
 
 @cli.command("delete", help="Delete a document by its ID")
@@ -118,7 +121,7 @@ def delete_document(
     ),
 ):
     app = HaikuRAGApp(db_path=db)
-    event_loop.run_until_complete(app.delete_document(doc_id=doc_id))
+    asyncio.run(app.delete_document(doc_id=doc_id))
 
 
 @cli.command("search", help="Search for documents by a query")
@@ -144,7 +147,7 @@ def search(
     ),
 ):
     app = HaikuRAGApp(db_path=db)
-    event_loop.run_until_complete(app.search(query=query, limit=limit, k=k))
+    asyncio.run(app.search(query=query, limit=limit, k=k))
 
 
 @cli.command("ask", help="Ask a question using the QA agent")
@@ -159,7 +162,7 @@ def ask(
     ),
 ):
     app = HaikuRAGApp(db_path=db)
-    event_loop.run_until_complete(app.ask(question=question))
+    asyncio.run(app.ask(question=question))
 
 
 @cli.command("settings", help="Display current configuration settings")
@@ -180,7 +183,7 @@ def rebuild(
     ),
 ):
     app = HaikuRAGApp(db_path=db)
-    event_loop.run_until_complete(app.rebuild())
+    asyncio.run(app.rebuild())
 
 
 @cli.command(
@@ -216,7 +219,7 @@ def serve(
     elif sse:
         transport = "sse"
 
-    event_loop.run_until_complete(app.serve(transport=transport))
+    asyncio.run(app.serve(transport=transport))
 
 
 if __name__ == "__main__":

haiku/rag/client.py

@@ -16,6 +16,7 @@ from haiku.rag.store.models.chunk import Chunk
 from haiku.rag.store.models.document import Document
 from haiku.rag.store.repositories.chunk import ChunkRepository
 from haiku.rag.store.repositories.document import DocumentRepository
+from haiku.rag.utils import text_to_docling_document
 
 
 class HaikuRAG:
@@ -49,6 +50,24 @@ class HaikuRAG:
         self.close()
         return False
 
+    async def _create_document_with_docling(
+        self,
+        docling_document,
+        uri: str | None = None,
+        metadata: dict | None = None,
+        chunks: list[Chunk] | None = None,
+    ) -> Document:
+        """Create a new document from DoclingDocument."""
+        content = docling_document.export_to_markdown()
+        document = Document(
+            content=content,
+            uri=uri,
+            metadata=metadata or {},
+        )
+        return await self.document_repository._create_with_docling(
+            document, docling_document, chunks
+        )
+
     async def create_document(
         self,
         content: str,
@@ -67,12 +86,17 @@ class HaikuRAG:
         Returns:
             The created Document instance.
         """
+        # Convert content to DoclingDocument for processing
+        docling_document = text_to_docling_document(content)
+
         document = Document(
             content=content,
             uri=uri,
             metadata=metadata or {},
         )
-        return await self.document_repository.create(document, chunks)
+        return await self.document_repository._create_with_docling(
+            document, docling_document, chunks
+        )
 
     async def create_document_from_source(
         self, source: str | Path, metadata: dict = {}
@@ -101,16 +125,19 @@ class HaikuRAG:
         parsed_url = urlparse(source_str)
         if parsed_url.scheme in ("http", "https"):
             return await self._create_or_update_document_from_url(source_str, metadata)
-
-        # Handle as file path
-        source_path = Path(source) if isinstance(source, str) else source
+        elif parsed_url.scheme == "file":
+            # Handle file:// URI by converting to path
+            source_path = Path(parsed_url.path)
+        else:
+            # Handle as regular file path
+            source_path = Path(source) if isinstance(source, str) else source
         if source_path.suffix.lower() not in FileReader.extensions:
             raise ValueError(f"Unsupported file extension: {source_path.suffix}")
 
         if not source_path.exists():
             raise ValueError(f"File does not exist: {source_path}")
 
-        uri = source_path.as_uri()
+        uri = source_path.absolute().as_uri()
         md5_hash = hashlib.md5(source_path.read_bytes()).hexdigest()
 
         # Check if document already exists
@@ -119,7 +146,7 @@ class HaikuRAG:
             # MD5 unchanged, return existing document
             return existing_doc
 
-        content = FileReader.parse_file(source_path)
+        docling_document = FileReader.parse_file(source_path)
 
         # Get content type from file extension
         content_type, _ = mimetypes.guess_type(str(source_path))
@@ -131,13 +158,15 @@ class HaikuRAG:
 
         if existing_doc:
             # Update existing document
-            existing_doc.content = content
+            existing_doc.content = docling_document.export_to_markdown()
             existing_doc.metadata = metadata
-            return await self.update_document(existing_doc)
+            return await self.document_repository._update_with_docling(
+                existing_doc, docling_document
+            )
         else:
-            # Create new document
-            return await self.create_document(
-                content=content, uri=uri, metadata=metadata
+            # Create new document using DoclingDocument
+            return await self._create_document_with_docling(
+                docling_document=docling_document, uri=uri, metadata=metadata
             )
 
     async def _create_or_update_document_from_url(
@@ -193,18 +222,20 @@ class HaikuRAG:
                 temp_path = Path(temp_file.name)
 
                 # Parse the content using FileReader
-                content = FileReader.parse_file(temp_path)
+                docling_document = FileReader.parse_file(temp_path)
 
             # Merge metadata with contentType and md5
             metadata.update({"contentType": content_type, "md5": md5_hash})
 
             if existing_doc:
-                existing_doc.content = content
+                existing_doc.content = docling_document.export_to_markdown()
                 existing_doc.metadata = metadata
-                return await self.update_document(existing_doc)
+                return await self.document_repository._update_with_docling(
+                    existing_doc, docling_document
+                )
             else:
-                return await self.create_document(
-                    content=content, uri=url, metadata=metadata
+                return await self._create_document_with_docling(
+                    docling_document=docling_document, uri=url, metadata=metadata
                 )
 
     def _get_extension_from_content_type_or_url(
@@ -262,7 +293,12 @@ class HaikuRAG:
 
     async def update_document(self, document: Document) -> Document:
         """Update an existing document."""
-        return await self.document_repository.update(document)
+        # Convert content to DoclingDocument
+        docling_document = text_to_docling_document(document.content)
+
+        return await self.document_repository._update_with_docling(
+            document, docling_document
+        )
 
     async def delete_document(self, document_id: int) -> bool:
         """Delete a document by its ID."""
@@ -328,6 +364,13 @@ class HaikuRAG:
     async def rebuild_database(self) -> AsyncGenerator[int, None]:
         """Rebuild the database by deleting all chunks and re-indexing all documents.
 
+        For documents with URIs:
+        - Deletes the document and re-adds it from source if source exists
+        - Skips documents where source no longer exists
+
+        For documents without URIs:
+        - Re-creates chunks from existing content
+
         Yields:
             int: The ID of the document currently being processed
         """
@@ -343,9 +386,36 @@ class HaikuRAG:
         documents = await self.list_documents()
 
         for doc in documents:
-            if doc.id is not None:
+            assert doc.id is not None, "Document ID should not be None"
+            if doc.uri:
+                # Document has a URI - delete and try to re-add from source
+                try:
+                    # Delete the old document first
+                    await self.delete_document(doc.id)
+
+                    # Try to re-create from source (this creates the document with chunks)
+                    new_doc = await self.create_document_from_source(
+                        doc.uri, doc.metadata or {}
+                    )
+
+                    assert new_doc.id is not None, "New document ID should not be None"
+                    yield new_doc.id
+
+                except (FileNotFoundError, ValueError, OSError) as e:
+                    # Source doesn't exist or can't be accessed - document already deleted, skip
+                    print(f"Skipping document with URI {doc.uri}: {e}")
+                    continue
+                except Exception as e:
+                    # Unexpected error - log it and skip
+                    print(
+                        f"Unexpected error processing document with URI {doc.uri}: {e}"
+                    )
+                    continue
+            else:
+                # Document without URI - re-create chunks from existing content
+                docling_document = text_to_docling_document(doc.content)
                 await self.chunk_repository.create_chunks_for_document(
-                    doc.id, doc.content, commit=False
+                    doc.id, docling_document, commit=False
                 )
                 yield doc.id
 

haiku/rag/config.py

@@ -10,7 +10,7 @@ load_dotenv()
 
 
 class AppConfig(BaseModel):
-    ENV: str = "development"
+    ENV: str = "production"
 
     DEFAULT_DATA_DIR: Path = get_default_data_dir()
     MONITOR_DIRECTORIES: list[Path] = []
@@ -27,7 +27,6 @@ class AppConfig(BaseModel):
     QA_MODEL: str = "qwen3"
 
     CHUNK_SIZE: int = 256
-    CHUNK_OVERLAP: int = 32
 
     OLLAMA_BASE_URL: str = "http://localhost:11434"
 

haiku/rag/reader.py

@@ -2,6 +2,9 @@ from pathlib import Path
 from typing import ClassVar
 
 from docling.document_converter import DocumentConverter
+from docling_core.types.doc.document import DoclingDocument
+
+from haiku.rag.utils import text_to_docling_document
 
 
 class FileReader:
@@ -16,7 +19,8 @@ class FileReader:
         ".jpeg",
         ".jpg",
         ".md",
-        ".pdf.png",
+        ".pdf",
+        ".png",
         ".pptx",
         ".tiff",
         ".xlsx",
@@ -83,7 +87,7 @@ class FileReader:
     extensions: ClassVar[list[str]] = docling_extensions + text_extensions
 
     @staticmethod
-    def parse_file(path: Path) -> str:
+    def parse_file(path: Path) -> DoclingDocument:
         try:
             file_extension = path.suffix.lower()
 
@@ -91,7 +95,7 @@ class FileReader:
                 # Use docling for complex document formats
                 converter = DocumentConverter()
                 result = converter.convert(path)
-                return result.document.export_to_markdown()
+                return result.document
             elif file_extension in FileReader.text_extensions:
                 # Read plain text files directly
                 content = path.read_text(encoding="utf-8")
@@ -99,11 +103,13 @@ class FileReader:
                 # Wrap code files (but not plain txt) in markdown code blocks for better presentation
                 if file_extension in FileReader.code_markdown_identifier:
                     language = FileReader.code_markdown_identifier[file_extension]
-                    return f"```{language}\n{content}\n```"
+                    content = f"```{language}\n{content}\n```"
 
-                return content
+                # Convert text to DoclingDocument by wrapping as markdown
+                return text_to_docling_document(content, name=f"{path.stem}.md")
             else:
-                # Fallback: try to read as text
-                return path.read_text(encoding="utf-8")
+                # Fallback: try to read as text and convert to DoclingDocument
+                content = path.read_text(encoding="utf-8")
+                return text_to_docling_document(content, name=f"{path.stem}.md")
         except Exception:
             raise ValueError(f"Failed to parse file: {path}")

haiku/rag/store/repositories/chunk.py

@@ -1,6 +1,8 @@
 import json
 import re
 
+from docling_core.types.doc.document import DoclingDocument
+
 from haiku.rag.chunker import chunker
 from haiku.rag.embeddings import get_embedder
 from haiku.rag.store.models.chunk import Chunk
@@ -197,11 +199,11 @@ class ChunkRepository(BaseRepository[Chunk]):
         ]
 
     async def create_chunks_for_document(
-        self, document_id: int, content: str, commit: bool = True
+        self, document_id: int, document: DoclingDocument, commit: bool = True
     ) -> list[Chunk]:
-        """Create chunks and embeddings for a document."""
+        """Create chunks and embeddings for a document from DoclingDocument."""
         # Chunk the document content
-        chunk_texts = await chunker.chunk(content)
+        chunk_texts = await chunker.chunk(document)
         created_chunks = []
 
         # Create chunks with embeddings using the create method

haiku/rag/store/repositories/document.py

@@ -1,8 +1,11 @@
 import json
 from typing import TYPE_CHECKING
 
+from docling_core.types.doc.document import DoclingDocument
+
 from haiku.rag.store.models.document import Document
 from haiku.rag.store.repositories.base import BaseRepository
+from haiku.rag.utils import text_to_docling_document
 
 if TYPE_CHECKING:
     from haiku.rag.store.models.chunk import Chunk
@@ -20,8 +23,11 @@ class DocumentRepository(BaseRepository[Document]):
             chunk_repository = ChunkRepository(store)
         self.chunk_repository = chunk_repository
 
-    async def create(
-        self, entity: Document, chunks: list["Chunk"] | None = None
+    async def _create_with_docling(
+        self,
+        entity: Document,
+        docling_document: DoclingDocument,
+        chunks: list["Chunk"] | None = None,
     ) -> Document:
         """Create a document with its chunks and embeddings."""
         if self.store._connection is None:
@@ -62,9 +68,9 @@ class DocumentRepository(BaseRepository[Document]):
                     chunk.metadata["order"] = order
                     await self.chunk_repository.create(chunk, commit=False)
             else:
-                # Create chunks and embeddings using ChunkRepository
+                # Create chunks and embeddings using DoclingDocument
                 await self.chunk_repository.create_chunks_for_document(
-                    document_id, entity.content, commit=False
+                    document_id, docling_document, commit=False
                 )
 
             cursor.execute("COMMIT")
@@ -74,6 +80,13 @@ class DocumentRepository(BaseRepository[Document]):
             cursor.execute("ROLLBACK")
             raise
 
+    async def create(self, entity: Document) -> Document:
+        """Create a document with its chunks and embeddings."""
+        # Convert content to DoclingDocument
+        docling_document = text_to_docling_document(entity.content)
+
+        return await self._create_with_docling(entity, docling_document)
+
     async def get_by_id(self, entity_id: int) -> Document | None:
         """Get a document by its ID."""
         if self.store._connection is None:
@@ -134,7 +147,9 @@ class DocumentRepository(BaseRepository[Document]):
             updated_at=updated_at,
         )
 
-    async def update(self, entity: Document) -> Document:
+    async def _update_with_docling(
+        self, entity: Document, docling_document: DoclingDocument
+    ) -> Document:
         """Update an existing document and regenerate its chunks and embeddings."""
         if self.store._connection is None:
             raise ValueError("Store connection is not available")
@@ -163,10 +178,10 @@ class DocumentRepository(BaseRepository[Document]):
                 },
             )
 
-            # Delete existing chunks and regenerate using ChunkRepository
+            # Delete existing chunks and regenerate using DoclingDocument
             await self.chunk_repository.delete_by_document_id(entity.id, commit=False)
             await self.chunk_repository.create_chunks_for_document(
-                entity.id, entity.content, commit=False
+                entity.id, docling_document, commit=False
             )
 
             cursor.execute("COMMIT")
@@ -176,6 +191,13 @@ class DocumentRepository(BaseRepository[Document]):
             cursor.execute("ROLLBACK")
             raise
 
+    async def update(self, entity: Document) -> Document:
+        """Update an existing document and regenerate its chunks and embeddings."""
+        # Convert content to DoclingDocument
+        docling_document = text_to_docling_document(entity.content)
+
+        return await self._update_with_docling(entity, docling_document)
+
     async def delete(self, entity_id: int) -> bool:
         """Delete a document and all its associated chunks and embeddings."""
         # Delete chunks and embeddings first

haiku/rag/store/repositories/settings.py

@@ -63,7 +63,6 @@ class SettingsRepository:
             "EMBEDDINGS_MODEL",
             "EMBEDDINGS_VECTOR_DIM",
             "CHUNK_SIZE",
-            "CHUNK_OVERLAP",
         ]
 
         errors = []

haiku/rag/utils.py

@@ -1,8 +1,12 @@
 import sys
 from importlib import metadata
+from io import BytesIO
 from pathlib import Path
 
 import httpx
+from docling.document_converter import DocumentConverter
+from docling_core.types.doc.document import DoclingDocument
+from docling_core.types.io import DocumentStream
 from packaging.version import Version, parse
 
 
@@ -77,3 +81,20 @@ async def is_up_to_date() -> tuple[bool, Version, Version]:
             # If no network connection, do not raise alarms.
             pypi_version = running_version
     return running_version >= pypi_version, running_version, pypi_version
+
+
+def text_to_docling_document(text: str, name: str = "content.md") -> DoclingDocument:
+    """Convert text content to a DoclingDocument.
+
+    Args:
+        text: The text content to convert.
+        name: The name to use for the document stream (defaults to "content.md").
+
+    Returns:
+        A DoclingDocument created from the text content.
+    """
+    bytes_io = BytesIO(text.encode("utf-8"))
+    doc_stream = DocumentStream(name=name, stream=bytes_io)
+    converter = DocumentConverter()
+    result = converter.convert(doc_stream)
+    return result.document

{haiku_rag-0.4.3.dist-info → haiku_rag-0.5.1.dist-info}/METADATA

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: haiku.rag
-Version: 0.4.3
+Version: 0.5.1
 Summary: Retrieval Augmented Generation (RAG) with SQLite
 Author-email: Yiorgis Gozadinos <ggozadinos@gmail.com>
 License: MIT
@@ -17,7 +17,7 @@ Classifier: Programming Language :: Python :: 3.10
 Classifier: Programming Language :: Python :: 3.11
 Classifier: Programming Language :: Python :: 3.12
 Classifier: Typing :: Typed
-Requires-Python: >=3.10
+Requires-Python: >=3.11
 Requires-Dist: docling>=2.15.0
 Requires-Dist: fastmcp>=2.8.1
 Requires-Dist: httpx>=0.28.1

{haiku_rag-0.4.3.dist-info → haiku_rag-0.5.1.dist-info}/RECORD

@@ -1,14 +1,14 @@
 haiku/rag/__init__.py,sha256=47DEQpj8HBSa-_TImW-5JCeuQeRkm5NMpJWZG3hSuFU,0
 haiku/rag/app.py,sha256=FpLVyP1-zAq_XPmU8CPVLkuIAeuhBOGvMqhYS8RbN40,7649
-haiku/rag/chunker.py,sha256=MbCtP66OfTFoIBvqmVT9T9c87fozsYYzAQzJJEfPBVI,1812
-haiku/rag/cli.py,sha256=k7EhLkvTncxsdh5TYrg8BHLYh_lfyzupsWGj1dEEdqY,5992
-haiku/rag/client.py,sha256=MZNIpMm6MS3P6vjLqiCztT2dBOM7-bZOosX5IpbHJbI,12724
-haiku/rag/config.py,sha256=_Ss54kmfxVAJupExLKaYjYUlFxJgb7hEEdbG4-isapY,1662
+haiku/rag/chunker.py,sha256=PVe6ysv8UlacUd4Zb3_8RFWIaWDXnzBAy2VDJ4TaUsE,1555
+haiku/rag/cli.py,sha256=rk4uUwN_FdMC-rai9_R2sgXXMI3TIWKRtdWWHg_WoWM,5865
+haiku/rag/client.py,sha256=pFcrPkQo1h1zJ76jts-72goP_kGVtnJNfLuoT8qpsb8,15795
+haiku/rag/config.py,sha256=8mlQ8gYFxxq1q9gi9tjY9StjqhfhiHkO1FvS4b0et0E,1633
 haiku/rag/logging.py,sha256=zTTGpGq5tPdcd7RpCbd9EGw1IZlQDbYkrCg9t9pqRc4,580
 haiku/rag/mcp.py,sha256=tMN6fNX7ZtAER1R6DL1GkC9HZozTC4HzuQs199p7icI,4551
 haiku/rag/monitor.py,sha256=r386nkhdlsU8UECwIuVwnrSlgMk3vNIuUZGNIzkZuec,2770
-haiku/rag/reader.py,sha256=dLz3yyc5r8dzdqCc2VViC3fADpScw4lxXueKiu-cI7c,2915
-haiku/rag/utils.py,sha256=Ez_tvNlRO_D8c2CBZ83Hs9Gmzcqdq4cmw_V5GBdKy_8,2214
+haiku/rag/reader.py,sha256=qkPTMJuQ_o4sK-8zpDl9WFYe_MJ7aL_gUw6rczIpW-g,3274
+haiku/rag/utils.py,sha256=g-uNTG60iBLgkeHHuah6eVZEkX3NFLs-LZU1YnzJzLQ,2967
 haiku/rag/embeddings/__init__.py,sha256=yFBlxS0jBiVHl_rWz5kb43t6Ha132U1ZGdlIPfhzPdg,1491
 haiku/rag/embeddings/base.py,sha256=NTQvuzbZPu0LBo5wAu3qGyJ4xXUaRAt1fjBO0ygWn_Y,465
 haiku/rag/embeddings/ollama.py,sha256=y6-lp0XpbnyIjoOEdtSzMdEVkU5glOwnWQ1FkpUZnpI,370
@@ -31,13 +31,13 @@ haiku/rag/store/models/chunk.py,sha256=9-vIxW75-kMTelIhgVIMd_WhP-Drc1q65vjaWMP8w
 haiku/rag/store/models/document.py,sha256=TVXVY-nQs-1vCORQEs9rA7zOtndeGC4dgCoujLAS054,396
 haiku/rag/store/repositories/__init__.py,sha256=uIBhxjQh-4o3O-ck8b7BQ58qXQTuJdPvrDIHVhY5T1A,263
 haiku/rag/store/repositories/base.py,sha256=cm3VyQXhtxvRfk1uJHpA0fDSxMpYN-mjQmRiDiLsQ68,1008
-haiku/rag/store/repositories/chunk.py,sha256=UyvHhKb1ESZePoTp2GneAARdfKoocEdfPOwgWPPQ0v8,16878
-haiku/rag/store/repositories/document.py,sha256=fXIWevJaOe6x2cK4u9cQxiEGD0ntKQb9y3VRqklQypE,7920
-haiku/rag/store/repositories/settings.py,sha256=dme3_ulQdQvyF9daavSjAd-SjZ5hh0MJoxP7iXgap-A,2492
+haiku/rag/store/repositories/chunk.py,sha256=DIIdpHVemvxZOPHOLBL7pJGWY4VyNrUiQSWPWt24BYo,16974
+haiku/rag/store/repositories/document.py,sha256=ki8LiDukwU1469Yw51i0rQFvBzUQeYkFYWs3Ly83akc,8815
+haiku/rag/store/repositories/settings.py,sha256=qZLXvLsErnCWL0nBQQNfRnatHzCKhtUDLvUK9k-W_fU,2463
 haiku/rag/store/upgrades/__init__.py,sha256=kKS1YWT_P-CYKhKtokOLTIFNKf9jlfjFFr8lyIMeogM,100
 haiku/rag/store/upgrades/v0_3_4.py,sha256=GLogKZdZ40NX1vBHKdOJju7fFzNUCHoEnjSZg17Hm2U,663
-haiku_rag-0.4.3.dist-info/METADATA,sha256=T2ZHdGL_zd1eSfEjFolh3R_zJpuWmUhKsnNkYLKtT7E,4198
-haiku_rag-0.4.3.dist-info/WHEEL,sha256=qtCwoSJWgHk21S1Kb4ihdzI2rlJ1ZKaIurTj_ngOhyQ,87
-haiku_rag-0.4.3.dist-info/entry_points.txt,sha256=G1U3nAkNd5YDYd4v0tuYFbriz0i-JheCsFuT9kIoGCI,48
-haiku_rag-0.4.3.dist-info/licenses/LICENSE,sha256=eXZrWjSk9PwYFNK9yUczl3oPl95Z4V9UXH7bPN46iPo,1065
-haiku_rag-0.4.3.dist-info/RECORD,,
+haiku_rag-0.5.1.dist-info/METADATA,sha256=X4r-1CBCTef3_T9HWPgCHi5XumqOSF4tlHfUpxO533E,4198
+haiku_rag-0.5.1.dist-info/WHEEL,sha256=qtCwoSJWgHk21S1Kb4ihdzI2rlJ1ZKaIurTj_ngOhyQ,87
+haiku_rag-0.5.1.dist-info/entry_points.txt,sha256=G1U3nAkNd5YDYd4v0tuYFbriz0i-JheCsFuT9kIoGCI,48
+haiku_rag-0.5.1.dist-info/licenses/LICENSE,sha256=eXZrWjSk9PwYFNK9yUczl3oPl95Z4V9UXH7bPN46iPo,1065
+haiku_rag-0.5.1.dist-info/RECORD,,