feat: MCP refactor Phases 1-7 — enterprise engineering drawing extraction stack

docstrange/__init__.py

@@ -1,5 +1,9 @@
 """
 Document Data Extractor - Extract structured data from any document into LLM-ready formats.
+
+For engineering drawing extraction use EngineeringDrawingPipeline directly — it is the
+dedicated entry point for PDFs and images containing title blocks, dimensions, GD&T, BOM,
+notes, and revision history.
 """
 
 from .extractor import DocumentExtractor
@@ -8,13 +12,56 @@
 from .exceptions import ConversionError, UnsupportedFormatError
 from .config import InternalConfig
 
+# Engineering drawing extraction surface
+from .pipelines.engineering import EngineeringDrawingPipeline
+from .schemas.engineering import (
+    BBoxSchema,
+    ExtractionElement,
+    EngineeringDrawingResult,
+    DimensionElement,
+    TitleBlockField,
+    NoteElement,
+    GDTElement,
+    BOMRow,
+    RevisionEntry,
+)
+from .extractors import (
+    BaseExtractor,
+    TitleBlockExtractor,
+    DimensionExtractor,
+    NoteExtractor,
+    GDTExtractor,
+    BOMExtractor,
+    RevisionExtractor,
+)
+
 __version__ = "1.1.5"
 __all__ = [
-    "DocumentExtractor", 
-    "ConversionResult", 
+    # Generic document extraction
+    "DocumentExtractor",
+    "ConversionResult",
     "GPUConversionResult",
     "CloudConversionResult",
-    "ConversionError", 
-    "UnsupportedFormatError", 
-    "InternalConfig"
-] 
+    "ConversionError",
+    "UnsupportedFormatError",
+    "InternalConfig",
+    # Engineering drawing extraction
+    "EngineeringDrawingPipeline",
+    "EngineeringDrawingResult",
+    "BBoxSchema",
+    "ExtractionElement",
+    "DimensionElement",
+    "TitleBlockField",
+    "NoteElement",
+    "GDTElement",
+    "BOMRow",
+    "RevisionEntry",
+    # Individual extractors (for custom pipelines)
+    "BaseExtractor",
+    "TitleBlockExtractor",
+    "DimensionExtractor",
+    "NoteExtractor",
+    "GDTExtractor",
+    "BOMExtractor",
+    "RevisionExtractor",
+]

docstrange/api/main.py

@@ -0,0 +1,11 @@
+"""FastAPI application entry point.
+
+Run with:
+    uvicorn docstrange.api.main:app --reload --port 8000
+
+Then visit http://localhost:8000/docs for the interactive API explorer.
+"""
+
+from .routes import create_app
+
+app = create_app()

docstrange/api/models.py

@@ -0,0 +1,86 @@
+"""Pydantic response models for the DocStrange Engineering API.
+
+These are thin wrappers / re-exports that let FastAPI generate accurate
+OpenAPI schemas for every endpoint without duplicating the core schemas.
+"""
+
+from typing import Any, Dict, List, Optional
+
+from pydantic import BaseModel
+
+from ..schemas.engineering import (
+    BBoxSchema,
+    BOMRow,
+    DimensionElement,
+    EngineeringDrawingResult,
+    ExtractionMetadata,
+    GDTElement,
+    NoteElement,
+    RevisionEntry,
+    TitleBlockField,
+)
+
+__all__ = [
+    "HealthResponse",
+    "FullExtractionResponse",
+    "OverlayBBox",
+    "OverlayAnnotation",
+    "OverlaySummary",
+    "OverlayImageSize",
+    "OverlayResponse",
+    # Re-exported schema types used as list element response models
+    "TitleBlockField",
+    "DimensionElement",
+    "NoteElement",
+    "GDTElement",
+    "BOMRow",
+    "RevisionEntry",
+]
+
+
+class HealthResponse(BaseModel):
+    status: str
+    service: str
+    version: str
+
+
+class FullExtractionResponse(EngineeringDrawingResult):
+    """EngineeringDrawingResult extended with an optional overlay payload."""
+    overlay_json: Optional[Dict[str, Any]] = None
+
+
+class OverlayBBox(BaseModel):
+    x: float
+    y: float
+    width: float
+    height: float
+
+
+class OverlayAnnotation(BaseModel):
+    change_id: str
+    type: str
+    text: str
+    page: int
+    confidence: float
+    bbox: OverlayBBox
+    bbox_normalized: OverlayBBox
+    color: str
+    label: str
+
+
+class OverlaySummary(BaseModel):
+    by_type: Dict[str, int]
+    total: int
+
+
+class OverlayImageSize(BaseModel):
+    width: int
+    height: int
+
+
+class OverlayResponse(BaseModel):
+    image_size: OverlayImageSize
+    page_filter: Optional[int] = None
+    total_annotations: int
+    summary: OverlaySummary
+    annotations: List[OverlayAnnotation]