rnsr 0.1.0__py3-none-any.whl

rnsr/models.py

@@ -0,0 +1,167 @@
+"""
+RNSR Data Models
+
+Pydantic models for all data structures in the pipeline.
+"""
+
+from __future__ import annotations
+
+from typing import Any, Literal
+from uuid import uuid4
+
+from pydantic import BaseModel, Field
+
+
+# =============================================================================
+# Ingestion Models
+# =============================================================================
+
+
+class BoundingBox(BaseModel):
+    """Bounding box for a text element on a page."""
+
+    x0: float
+    y0: float
+    x1: float
+    y1: float
+
+    @property
+    def width(self) -> float:
+        return self.x1 - self.x0
+
+    @property
+    def height(self) -> float:
+        return self.y1 - self.y0
+
+    @property
+    def center(self) -> tuple[float, float]:
+        return ((self.x0 + self.x1) / 2, (self.y0 + self.y1) / 2)
+
+
+class SpanInfo(BaseModel):
+    """Information about a single text span from PyMuPDF."""
+
+    text: str
+    font_size: float
+    font_name: str
+    is_bold: bool = False
+    is_italic: bool = False
+    bbox: BoundingBox
+    page_num: int
+
+
+class FontAnalysis(BaseModel):
+    """Results of font histogram analysis."""
+
+    body_size: float
+    header_threshold: float
+    size_histogram: dict[float, int]
+    span_count: int
+    unique_sizes: int
+
+
+class ClassifiedSpan(SpanInfo):
+    """A span with its classification (header level or body)."""
+
+    role: Literal["header", "body", "caption", "footnote"] = "body"
+    header_level: int = 0  # 0 = not a header, 1-3 = H1-H3
+
+
+class DocumentNode(BaseModel):
+    """A node in the document tree structure."""
+
+    id: str = Field(default_factory=lambda: str(uuid4())[:8])
+    level: int  # 0 = root, 1 = H1, 2 = H2, 3 = H3
+    header: str = ""
+    content: str = ""  # Full text content
+    page_num: int | None = None
+    bbox: BoundingBox | None = None
+    children: list[DocumentNode] = Field(default_factory=list)
+
+    @property
+    def child_ids(self) -> list[str]:
+        return [child.id for child in self.children]
+
+
+class DocumentTree(BaseModel):
+    """Complete document tree structure."""
+
+    id: str = Field(default_factory=lambda: f"doc_{str(uuid4())[:8]}")
+    title: str = ""
+    root: DocumentNode
+    total_nodes: int = 0
+    ingestion_tier: Literal[1, 2, 3] = 1
+    ingestion_method: IngestionMethod | None = None
+
+
+# =============================================================================
+# Indexing Models
+# =============================================================================
+
+
+class SkeletonNode(BaseModel):
+    """A lightweight node for the skeleton index."""
+
+    node_id: str
+    parent_id: str | None
+    level: int
+    header: str
+    summary: str  # 50-100 words max - this goes in vector store
+    child_ids: list[str]
+    page_num: int | None = None
+    metadata: dict[str, Any] = Field(default_factory=dict)
+
+
+class IngestionResult(BaseModel):
+    """Result of document ingestion including metadata."""
+
+    tree: DocumentTree
+    tier_used: Literal[1, 2, 3]
+    method: IngestionMethod
+    warnings: list[str] = Field(default_factory=list)
+    stats: dict[str, Any] = Field(default_factory=dict)
+
+
+# =============================================================================
+# Agent Models
+# =============================================================================
+
+
+class StoredVariable(BaseModel):
+    """Metadata about a stored variable in the VariableStore."""
+
+    pointer: str  # e.g., "$LIABILITY_CLAUSE"
+    source_node_id: str
+    content_hash: str
+    char_count: int
+    created_at: str
+
+
+class TraceEntry(BaseModel):
+    """A single entry in the retrieval trace log."""
+
+    timestamp: str
+    node_type: Literal["decomposition", "navigation", "variable_stitching", "synthesis"]
+    action: str
+    details: dict[str, Any] = Field(default_factory=dict)
+
+
+class RetrievalTrace(BaseModel):
+    """Complete trace of agent's retrieval process."""
+
+    query: str
+    total_steps: int
+    nodes_visited: list[str]
+    nodes_rejected: list[dict[str, str]]
+    variables_stored: list[str]
+    final_path: str
+    entries: list[TraceEntry] = Field(default_factory=list)
+
+# Define the type alias for all valid ingestion methods
+IngestionMethod = Literal[
+    "font_histogram",
+    "semantic_splitter", 
+    "ocr",
+    "xy_cut",
+    "hierarchical_clustering",
+]

rnsr/py.typed

@@ -0,0 +1,2 @@
+# Marker file for PEP 561
+# This package supports type checking