haiku.rag 0.5.0__tar.gz → 0.5.2__tar.gz

haiku_rag-0.5.2/.python-version

@@ -0,0 +1 @@
+3.11

{haiku_rag-0.5.0 → haiku_rag-0.5.2}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: haiku.rag
-Version: 0.5.0
+Version: 0.5.2
 Summary: Retrieval Augmented Generation (RAG) with SQLite
 Author-email: Yiorgis Gozadinos <ggozadinos@gmail.com>
 License: MIT
@@ -17,12 +17,11 @@ 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
-Requires-Dist: mxbai-rerank>=0.1.6
-Requires-Dist: ollama>=0.5.1
+Requires-Dist: ollama>=0.5.3
 Requires-Dist: pydantic>=2.11.7
 Requires-Dist: python-dotenv>=1.1.0
 Requires-Dist: rich>=14.0.0
@@ -34,6 +33,8 @@ Provides-Extra: anthropic
 Requires-Dist: anthropic>=0.56.0; extra == 'anthropic'
 Provides-Extra: cohere
 Requires-Dist: cohere>=5.16.1; extra == 'cohere'
+Provides-Extra: mxbai
+Requires-Dist: mxbai-rerank>=0.1.6; extra == 'mxbai'
 Provides-Extra: openai
 Requires-Dist: openai>=1.0.0; extra == 'openai'
 Provides-Extra: voyageai

{haiku_rag-0.5.0 → haiku_rag-0.5.2}/docs/configuration.md

@@ -105,15 +105,37 @@ ANTHROPIC_API_KEY="your-api-key"
 
 ## Reranking
 
-Reranking is **enabled by default** and improves search quality by re-ordering the initial search results using specialized models. When enabled, the system retrieves more candidates (3x the requested limit) and then reranks them to return the most relevant results.
+Reranking improves search quality by re-ordering the initial search results using specialized models. When enabled, the system retrieves more candidates (3x the requested limit) and then reranks them to return the most relevant results.
 
-If you use the default reranked (running locally), it can slow down searching significantly. To disable reranking for faster searches:
+Reranking is **automatically enabled** by default using Ollama, or if you install the appropriate reranking provider package.
+
+### Disabling Reranking
+
+To disable reranking completely for faster searches:
 
 ```bash
-RERANK=false
+RERANK_PROVIDER=""
 ```
 
-### MixedBread AI (Default)
+### Ollama (Default)
+
+Ollama reranking uses LLMs with structured output to rank documents by relevance:
+
+```bash
+RERANK_PROVIDER="ollama"
+RERANK_MODEL="qwen3:1.7b"  # or any model that supports structured output
+OLLAMA_BASE_URL="http://localhost:11434"
+```
+
+### MixedBread AI
+
+For MxBAI reranking, install with mxbai extras:
+
+```bash
+uv pip install haiku.rag[mxbai]
+```
+
+Then configure:
 
 ```bash
 RERANK_PROVIDER="mxbai"

{haiku_rag-0.5.0 → haiku_rag-0.5.2}/pyproject.toml

@@ -1,11 +1,11 @@
 [project]
 name = "haiku.rag"
-version = "0.5.0"
+version = "0.5.2"
 description = "Retrieval Augmented Generation (RAG) with SQLite"
 authors = [{ name = "Yiorgis Gozadinos", email = "ggozadinos@gmail.com" }]
 license = { text = "MIT" }
 readme = { file = "README.md", content-type = "text/markdown" }
-requires-python = ">=3.10"
+requires-python = ">=3.11"
 keywords = ["RAG", "sqlite", "sqlite-vec", "ml", "mcp"]
 classifiers = [
     "Development Status :: 4 - Beta",
@@ -25,8 +25,7 @@ dependencies = [
     "docling>=2.15.0",
     "fastmcp>=2.8.1",
     "httpx>=0.28.1",
-    "mxbai-rerank>=0.1.6",
-    "ollama>=0.5.1",
+    "ollama>=0.5.3",
     "pydantic>=2.11.7",
     "python-dotenv>=1.1.0",
     "rich>=14.0.0",
@@ -41,6 +40,7 @@ voyageai = ["voyageai>=0.3.2"]
 openai = ["openai>=1.0.0"]
 anthropic = ["anthropic>=0.56.0"]
 cohere = ["cohere>=5.16.1"]
+mxbai = ["mxbai-rerank>=0.1.6"]
 
 [project.scripts]
 haiku-rag = "haiku.rag.cli:cli"

{haiku_rag-0.5.0 → haiku_rag-0.5.2}/src/haiku/rag/app.py

@@ -32,9 +32,9 @@ class HaikuRAGApp:
                 f"[b]Document with id [cyan]{doc.id}[/cyan] added successfully.[/b]"
             )
 
-    async def add_document_from_source(self, file_path: Path):
+    async def add_document_from_source(self, source: str):
         async with HaikuRAG(db_path=self.db_path) as self.client:
-            doc = await self.client.create_document_from_source(file_path)
+            doc = await self.client.create_document_from_source(source)
             self._rich_print_document(doc, truncate=True)
             self.console.print(
                 f"[b]Document with id [cyan]{doc.id}[/cyan] added successfully.[/b]"

{haiku_rag-0.5.0 → haiku_rag-0.5.2}/src/haiku/rag/chunker.py

@@ -1,11 +1,9 @@
-from io import BytesIO
 from typing import ClassVar
 
 import tiktoken
 from docling.chunking import HybridChunker  # type: ignore
-from docling.document_converter import DocumentConverter
 from docling_core.transforms.chunker.tokenizer.openai import OpenAITokenizer
-from docling_core.types.io import DocumentStream
+from docling_core.types.doc.document import DoclingDocument
 
 from haiku.rag.config import Config
 
@@ -33,27 +31,20 @@ class Chunker:
 
         self.chunker = HybridChunker(tokenizer=tokenizer)  # type: ignore
 
-    async def chunk(self, text: str) -> list[str]:
-        """Split the text into chunks using docling's structure-aware chunking.
+    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 semantic boundaries.
         """
-        if not text:
+        if document is None:
             return []
 
-        # Convert to docling document
-        bytes_io = BytesIO(text.encode("utf-8"))
-        doc_stream = DocumentStream(name="text.md", stream=bytes_io)
-        converter = DocumentConverter()
-        result = converter.convert(doc_stream)
-        doc = result.document
-
         # Chunk using docling's hybrid chunker
-        chunks = list(self.chunker.chunk(doc))
+        chunks = list(self.chunker.chunk(document))
         return [self.chunker.contextualize(chunk) for chunk in chunks]
 
 

{haiku_rag-0.5.0 → haiku_rag-0.5.2}/src/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")
@@ -73,12 +76,12 @@ 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")
 def add_document_src(
-    file_path: Path = typer.Argument(
+    source: str = typer.Argument(
         help="The file path or URL of the document to add",
     ),
     db: Path = typer.Option(
@@ -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(source=source))
 
 
 @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-0.5.0 → haiku_rag-0.5.2}/src/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."""
@@ -283,7 +319,7 @@ class HaikuRAG:
         return await self.document_repository.list_all(limit=limit, offset=offset)
 
     async def search(
-        self, query: str, limit: int = 5, k: int = 60, rerank=Config.RERANK
+        self, query: str, limit: int = 5, k: int = 60
     ) -> list[tuple[Chunk, float]]:
         """Search for relevant chunks using hybrid search (vector similarity + full-text search) with reranking.
 
@@ -295,8 +331,10 @@ class HaikuRAG:
         Returns:
             List of (chunk, score) tuples ordered by relevance.
         """
+        # Get reranker if available
+        reranker = get_reranker()
 
-        if not rerank:
+        if reranker is None:
             return await self.chunk_repository.search_chunks_hybrid(query, limit, k)
 
         # Get more initial results (3X) for reranking
@@ -304,7 +342,6 @@ class HaikuRAG:
             query, limit * 3, k
         )
         # Apply reranking
-        reranker = get_reranker()
         chunks = [chunk for chunk, _ in search_results]
         reranked_results = await reranker.rerank(query, chunks, top_n=limit)
 
@@ -328,6 +365,13 @@ 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 +387,36 @@ 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-0.5.0 → haiku_rag-0.5.2}/src/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] = []
@@ -19,9 +19,8 @@ class AppConfig(BaseModel):
     EMBEDDINGS_MODEL: str = "mxbai-embed-large"
     EMBEDDINGS_VECTOR_DIM: int = 1024
 
-    RERANK: bool = True
-    RERANK_PROVIDER: str = "mxbai"
-    RERANK_MODEL: str = "mixedbread-ai/mxbai-rerank-base-v2"
+    RERANK_PROVIDER: str = "ollama"
+    RERANK_MODEL: str = "qwen3"
 
     QA_PROVIDER: str = "ollama"
     QA_MODEL: str = "qwen3"

{haiku_rag-0.5.0 → haiku_rag-0.5.2}/src/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:
@@ -84,7 +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()
 
@@ -92,7 +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")
@@ -100,11 +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-0.5.2/src/haiku/rag/reranking/__init__.py

@@ -0,0 +1,40 @@
+from haiku.rag.config import Config
+from haiku.rag.reranking.base import RerankerBase
+
+_reranker: RerankerBase | None = None
+
+
+def get_reranker() -> RerankerBase | None:
+    """
+    Factory function to get the appropriate reranker based on the configuration.
+    Returns None if if reranking is disabled.
+    """
+    global _reranker
+    if _reranker is not None:
+        return _reranker
+
+    if Config.RERANK_PROVIDER == "mxbai":
+        try:
+            from haiku.rag.reranking.mxbai import MxBAIReranker
+
+            _reranker = MxBAIReranker()
+            return _reranker
+        except ImportError:
+            return None
+
+    if Config.RERANK_PROVIDER == "cohere":
+        try:
+            from haiku.rag.reranking.cohere import CohereReranker
+
+            _reranker = CohereReranker()
+            return _reranker
+        except ImportError:
+            return None
+
+    if Config.RERANK_PROVIDER == "ollama":
+        from haiku.rag.reranking.ollama import OllamaReranker
+
+        _reranker = OllamaReranker()
+        return _reranker
+
+    return None