ragit 0.8__py3-none-any.whl → 0.8.2__py3-none-any.whl

ragit/core/experiment/results.py

@@ -0,0 +1,131 @@
+#
+# Copyright RODMENA LIMITED 2025
+# SPDX-License-Identifier: Apache-2.0
+#
+"""
+Ragit experiment results.
+"""
+
+from collections.abc import Iterator
+from dataclasses import asdict, dataclass, field
+from typing import Any
+
+
+@dataclass
+class EvaluationResult:
+    """
+    Result from evaluating a single RAG configuration.
+
+    Parameters
+    ----------
+    pattern_name : str
+        Name of the RAG pattern (e.g., "Pattern_1").
+    indexing_params : dict[str, Any]
+        Hyperparameters used during indexing (chunk_size, overlap, etc.).
+    inference_params : dict[str, Any]
+        Hyperparameters used during inference (num_chunks, llm_model, etc.).
+    scores : dict[str, dict]
+        Evaluation scores (answer_correctness, context_relevance, faithfulness).
+    execution_time : float
+        Time taken for evaluation in seconds.
+    final_score : float
+        Combined score for optimization ranking.
+    """
+
+    pattern_name: str
+    indexing_params: dict[str, Any]
+    inference_params: dict[str, Any]
+    scores: dict[str, dict[str, float]]
+    execution_time: float
+    final_score: float
+
+    def to_dict(self) -> dict[str, Any]:
+        """Convert to dictionary."""
+        return asdict(self)
+
+    def __repr__(self) -> str:
+        return (
+            f"EvaluationResult(name={self.pattern_name}, score={self.final_score:.3f}, time={self.execution_time:.1f}s)"
+        )
+
+
+@dataclass
+class ExperimentResults:
+    """
+    Collection of evaluation results from an optimization experiment.
+
+    Attributes
+    ----------
+    evaluations : list[EvaluationResult]
+        All evaluation results.
+    """
+
+    evaluations: list[EvaluationResult] = field(default_factory=list)
+
+    def __len__(self) -> int:
+        return len(self.evaluations)
+
+    def __iter__(self) -> Iterator[EvaluationResult]:
+        yield from self.evaluations
+
+    def __bool__(self) -> bool:
+        return bool(self.evaluations)
+
+    def add(self, result: EvaluationResult) -> None:
+        """Add an evaluation result."""
+        self.evaluations.append(result)
+
+    def is_cached(
+        self,
+        indexing_params: dict[str, Any],
+        inference_params: dict[str, Any],
+    ) -> float | None:
+        """
+        Check if this configuration was already evaluated.
+
+        Returns
+        -------
+        float or None
+            Final score if cached, None otherwise.
+        """
+        for ev in self.evaluations:
+            if ev.indexing_params == indexing_params and ev.inference_params == inference_params:
+                return ev.final_score
+        return None
+
+    @property
+    def scores(self) -> list[float]:
+        """All final scores."""
+        return [ev.final_score for ev in self.evaluations]
+
+    def sorted(self, reverse: bool = True) -> list[EvaluationResult]:
+        """
+        Get results sorted by final score.
+
+        Parameters
+        ----------
+        reverse : bool
+            If True (default), best scores first.
+
+        Returns
+        -------
+        list[EvaluationResult]
+            Sorted results.
+        """
+        return sorted(self.evaluations, key=lambda x: x.final_score, reverse=reverse)
+
+    def get_best(self, k: int = 1) -> list[EvaluationResult]:
+        """
+        Get k best results.
+
+        Parameters
+        ----------
+        k : int
+            Number of results to return.
+
+        Returns
+        -------
+        list[EvaluationResult]
+            Top k results by score.
+        """
+        return self.sorted()[:k]

ragit/loaders.py

@@ -0,0 +1,245 @@
+#
+# Copyright RODMENA LIMITED 2025
+# SPDX-License-Identifier: Apache-2.0
+#
+"""
+Document loading and chunking utilities.
+
+Provides simple functions to load documents from files and chunk text.
+"""
+
+import re
+from pathlib import Path
+from typing import Any
+
+from ragit.core.experiment.experiment import Chunk, Document
+
+
+def load_text(path: str | Path) -> Document:
+    """
+    Load a single text file as a Document.
+
+    Parameters
+    ----------
+    path : str or Path
+        Path to the text file (.txt, .md, .rst, etc.)
+
+    Returns
+    -------
+    Document
+        Document with file content and metadata.
+
+    Examples
+    --------
+    >>> doc = load_text("docs/tutorial.rst")
+    >>> print(doc.id, len(doc.content))
+    """
+    path = Path(path)
+    content = path.read_text(encoding="utf-8")
+    return Document(id=path.stem, content=content, metadata={"source": str(path), "filename": path.name})
+
+
+def load_directory(path: str | Path, pattern: str = "*.txt", recursive: bool = False) -> list[Document]:
+    """
+    Load all matching files from a directory as Documents.
+
+    Parameters
+    ----------
+    path : str or Path
+        Directory path.
+    pattern : str
+        Glob pattern for files (default: "*.txt").
+    recursive : bool
+        If True, search recursively (default: False).
+
+    Returns
+    -------
+    list[Document]
+        List of loaded documents.
+
+    Examples
+    --------
+    >>> docs = load_directory("docs/", "*.rst")
+    >>> docs = load_directory("docs/", "**/*.md", recursive=True)
+    """
+    path = Path(path)
+    glob_method = path.rglob if recursive else path.glob
+    documents = []
+
+    for file_path in sorted(glob_method(pattern)):
+        if file_path.is_file():
+            documents.append(load_text(file_path))
+
+    return documents
+
+
+def chunk_text(
+    text: str,
+    chunk_size: int = 512,
+    chunk_overlap: int = 50,
+    doc_id: str = "doc",
+    metadata: dict[str, Any] | None = None,
+) -> list[Chunk]:
+    """
+    Split text into overlapping chunks.
+
+    Parameters
+    ----------
+    text : str
+        Text to chunk.
+    chunk_size : int
+        Maximum characters per chunk (default: 512).
+    chunk_overlap : int
+        Overlap between chunks (default: 50).
+    doc_id : str
+        Document ID for the chunks (default: "doc").
+    metadata : dict, optional
+        Metadata to attach to each chunk (default: None).
+
+    Returns
+    -------
+    list[Chunk]
+        List of text chunks.
+
+    Examples
+    --------
+    >>> chunks = chunk_text("Long document...", chunk_size=256, chunk_overlap=50)
+    """
+    if chunk_overlap >= chunk_size:
+        raise ValueError("chunk_overlap must be less than chunk_size")
+
+    chunks = []
+    start = 0
+    chunk_idx = 0
+    chunk_metadata = metadata or {}
+
+    while start < len(text):
+        end = start + chunk_size
+        chunk_content = text[start:end].strip()
+
+        if chunk_content:
+            chunks.append(
+                Chunk(content=chunk_content, doc_id=doc_id, chunk_index=chunk_idx, metadata=chunk_metadata.copy())
+            )
+            chunk_idx += 1
+
+        start = end - chunk_overlap
+        if start >= len(text) - chunk_overlap:
+            break
+
+    return chunks
+
+
+def chunk_document(doc: Document, chunk_size: int = 512, chunk_overlap: int = 50) -> list[Chunk]:
+    """
+    Split a Document into overlapping chunks.
+
+    Parameters
+    ----------
+    doc : Document
+        Document to chunk.
+    chunk_size : int
+        Maximum characters per chunk.
+    chunk_overlap : int
+        Overlap between chunks.
+
+    Returns
+    -------
+    list[Chunk]
+        List of chunks from the document.
+    """
+    return chunk_text(doc.content, chunk_size, chunk_overlap, doc.id, metadata=doc.metadata)
+
+
+def chunk_by_separator(
+    text: str, separator: str = "\n\n", doc_id: str = "doc", metadata: dict[str, Any] | None = None
+) -> list[Chunk]:
+    """
+    Split text by a separator (e.g., paragraphs, sections).
+
+    Parameters
+    ----------
+    text : str
+        Text to split.
+    separator : str
+        Separator string (default: double newline for paragraphs).
+    doc_id : str
+        Document ID for the chunks.
+    metadata : dict, optional
+        Metadata to attach to each chunk (default: None).
+
+    Returns
+    -------
+    list[Chunk]
+        List of chunks.
+
+    Examples
+    --------
+    >>> chunks = chunk_by_separator(text, separator="\\n---\\n")
+    """
+    parts = text.split(separator)
+    chunks = []
+    chunk_metadata = metadata or {}
+
+    for idx, part in enumerate(parts):
+        content = part.strip()
+        if content:
+            chunks.append(Chunk(content=content, doc_id=doc_id, chunk_index=idx, metadata=chunk_metadata.copy()))
+
+    return chunks
+
+
+def chunk_rst_sections(text: str, doc_id: str = "doc", metadata: dict[str, Any] | None = None) -> list[Chunk]:
+    """
+    Split RST document by section headers.
+
+    Parameters
+    ----------
+    text : str
+        RST document text.
+    doc_id : str
+        Document ID for the chunks.
+    metadata : dict, optional
+        Metadata to attach to each chunk (default: None).
+
+    Returns
+    -------
+    list[Chunk]
+        List of section chunks.
+    """
+    # Match RST section headers (title followed by underline of =, -, ~, etc.)
+    pattern = r"\n([^\n]+)\n([=\-~`\'\"^_*+#]+)\n"
+    chunk_metadata = metadata or {}
+
+    # Find all section positions
+    matches = list(re.finditer(pattern, text))
+
+    if not matches:
+        # No sections found, return whole text as one chunk
+        return (
+            [Chunk(content=text.strip(), doc_id=doc_id, chunk_index=0, metadata=chunk_metadata.copy())]
+            if text.strip()
+            else []
+        )
+
+    chunks = []
+
+    # Handle content before first section
+    first_pos = matches[0].start()
+    if first_pos > 0:
+        pre_content = text[:first_pos].strip()
+        if pre_content:
+            chunks.append(Chunk(content=pre_content, doc_id=doc_id, chunk_index=0, metadata=chunk_metadata.copy()))
+
+    # Extract each section
+    for i, match in enumerate(matches):
+        start = match.start()
+        end = matches[i + 1].start() if i + 1 < len(matches) else len(text)
+
+        section_content = text[start:end].strip()
+        if section_content:
+            chunks.append(
+                Chunk(content=section_content, doc_id=doc_id, chunk_index=len(chunks), metadata=chunk_metadata.copy())
+            )
+
+    return chunks

ragit/providers/__init__.py

@@ -0,0 +1,47 @@
+#
+# Copyright RODMENA LIMITED 2025
+# SPDX-License-Identifier: Apache-2.0
+#
+"""
+Ragit Providers - LLM and Embedding providers for RAG optimization.
+
+Supported providers:
+- OllamaProvider: Connect to local or remote Ollama servers
+- FunctionProvider: Wrap custom embedding/LLM functions
+- SentenceTransformersProvider: Offline embedding (requires ragit[transformers])
+
+Base classes for implementing custom providers:
+- BaseLLMProvider: Abstract base for LLM providers
+- BaseEmbeddingProvider: Abstract base for embedding providers
+"""
+
+from ragit.providers.base import (
+    BaseEmbeddingProvider,
+    BaseLLMProvider,
+    EmbeddingResponse,
+    LLMResponse,
+)
+from ragit.providers.function_adapter import FunctionProvider
+from ragit.providers.ollama import OllamaProvider
+
+__all__ = [
+    # Base classes
+    "BaseLLMProvider",
+    "BaseEmbeddingProvider",
+    "LLMResponse",
+    "EmbeddingResponse",
+    # Built-in providers
+    "OllamaProvider",
+    "FunctionProvider",
+]
+
+# Conditionally export SentenceTransformersProvider if available
+try:
+    from ragit.providers.sentence_transformers import (
+        SentenceTransformersProvider as SentenceTransformersProvider,
+    )
+
+    __all__ += ["SentenceTransformersProvider"]
+except ImportError:
+    # sentence-transformers not installed, SentenceTransformersProvider not available
+    pass

ragit/providers/base.py

@@ -0,0 +1,147 @@
+#
+# Copyright RODMENA LIMITED 2025
+# SPDX-License-Identifier: Apache-2.0
+#
+"""
+Base provider interfaces for LLM and Embedding providers.
+
+These abstract classes define the interface that all providers must implement,
+making it easy to add new providers (Gemini, Claude, OpenAI, etc.)
+"""
+
+from abc import ABC, abstractmethod
+from dataclasses import dataclass
+
+
+@dataclass
+class LLMResponse:
+    """Response from an LLM call."""
+
+    text: str
+    model: str
+    provider: str
+    usage: dict[str, int] | None = None
+
+
+@dataclass(frozen=True)
+class EmbeddingResponse:
+    """Response from an embedding call (immutable)."""
+
+    embedding: tuple[float, ...]
+    model: str
+    provider: str
+    dimensions: int
+
+
+class BaseLLMProvider(ABC):
+    """
+    Abstract base class for LLM providers.
+
+    Implement this to add support for new LLM providers like Gemini, Claude, etc.
+    """
+
+    @property
+    @abstractmethod
+    def provider_name(self) -> str:
+        """Return the provider name (e.g., 'ollama', 'gemini', 'claude')."""
+        pass
+
+    @abstractmethod
+    def generate(
+        self,
+        prompt: str,
+        model: str,
+        system_prompt: str | None = None,
+        temperature: float = 0.7,
+        max_tokens: int | None = None,
+    ) -> LLMResponse:
+        """
+        Generate text from the LLM.
+
+        Parameters
+        ----------
+        prompt : str
+            The user prompt/query.
+        model : str
+            Model identifier (e.g., 'llama3', 'qwen3-vl:235b-instruct-cloud').
+        system_prompt : str, optional
+            System prompt for context/instructions.
+        temperature : float
+            Sampling temperature (0.0 to 1.0).
+        max_tokens : int, optional
+            Maximum tokens to generate.
+
+        Returns
+        -------
+        LLMResponse
+            The generated response.
+        """
+        pass
+
+    @abstractmethod
+    def is_available(self) -> bool:
+        """Check if the provider is available and configured."""
+        pass
+
+
+class BaseEmbeddingProvider(ABC):
+    """
+    Abstract base class for embedding providers.
+
+    Implement this to add support for new embedding providers.
+    """
+
+    @property
+    @abstractmethod
+    def provider_name(self) -> str:
+        """Return the provider name."""
+        pass
+
+    @property
+    @abstractmethod
+    def dimensions(self) -> int:
+        """Return the embedding dimensions for the current model."""
+        pass
+
+    @abstractmethod
+    def embed(self, text: str, model: str) -> EmbeddingResponse:
+        """
+        Generate embedding for text.
+
+        Parameters
+        ----------
+        text : str
+            Text to embed.
+        model : str
+            Model identifier (e.g., 'nomic-embed-text').
+
+        Returns
+        -------
+        EmbeddingResponse
+            The embedding response.
+        """
+        pass
+
+    @abstractmethod
+    def embed_batch(self, texts: list[str], model: str) -> list[EmbeddingResponse]:
+        """
+        Generate embeddings for multiple texts.
+
+        Parameters
+        ----------
+        texts : list[str]
+            Texts to embed.
+        model : str
+            Model identifier.
+
+        Returns
+        -------
+        list[EmbeddingResponse]
+            List of embedding responses.
+        """
+        pass
+
+    @abstractmethod
+    def is_available(self) -> bool:
+        """Check if the provider is available and configured."""
+        pass