Changes committed

docengine/app/schemas/document.py

@@ -0,0 +1,134 @@
+from __future__ import annotations
+
+import uuid
+from datetime import datetime
+from typing import Any
+
+from pydantic import Field
+
+from app.schemas.common import BaseSchema
+
+
+class DocumentUploadResponse(BaseSchema):
+    """Response after document upload."""
+
+    id: uuid.UUID
+    filename: str
+    original_filename: str
+    content_type: str
+    file_size: int
+    checksum: str
+    status: str
+    created_at: datetime
+
+
+class TextBlockResponse(BaseSchema):
+    """Extracted text block."""
+
+    id: uuid.UUID
+    text: str
+    x: float
+    y: float
+    width: float
+    height: float
+    confidence: float | None
+    font_family: str | None
+    font_size: float | None
+    font_color: str | None
+    font_style: str | None
+    block_type: str
+    sequence: int
+
+
+class DocumentImageResponse(BaseSchema):
+    """Extracted document image."""
+
+    id: uuid.UUID
+    x: float
+    y: float
+    width: float
+    height: float
+    image_path: str
+    image_type: str
+
+
+class DocumentTableResponse(BaseSchema):
+    """Extracted document table."""
+
+    id: uuid.UUID
+    x: float
+    y: float
+    width: float
+    height: float
+    rows: int
+    columns: int
+    data: dict[str, Any] | None
+
+
+class DocumentPageResponse(BaseSchema):
+    """Document page with extracted content."""
+
+    id: uuid.UUID
+    page_number: int
+    width: float
+    height: float
+    image_path: str | None
+    text_content: str | None
+    text_blocks: list[TextBlockResponse] = Field(default_factory=list)
+    images: list[DocumentImageResponse] = Field(default_factory=list)
+    tables: list[DocumentTableResponse] = Field(default_factory=list)
+
+
+class DocumentResponse(BaseSchema):
+    """Full document response."""
+
+    id: uuid.UUID
+    filename: str
+    original_filename: str
+    content_type: str
+    file_size: int
+    checksum: str
+    storage_path: str
+    status: str
+    page_count: int | None
+    is_scanned: bool | None
+    document_metadata: dict[str, Any] | None
+    error_message: str | None
+    uploaded_by: uuid.UUID | None
+    pages: list[DocumentPageResponse] = Field(default_factory=list)
+    created_at: datetime
+    updated_at: datetime
+
+
+class DocumentListResponse(BaseSchema):
+    """Minimal document response for lists."""
+
+    id: uuid.UUID
+    original_filename: str
+    content_type: str
+    file_size: int
+    status: str
+    page_count: int | None
+    is_scanned: bool | None
+    created_at: datetime
+
+
+class TemplateMatchResponse(BaseSchema):
+    """Template match result."""
+
+    id: uuid.UUID
+    document_id: uuid.UUID
+    format_id: uuid.UUID
+    confidence_score: float
+    match_details: dict[str, Any] | None
+    selected: bool
+    template_name: str | None = None
+    created_at: datetime
+
+
+class TemplateMatchRequest(BaseSchema):
+    """Request to match a document against templates."""
+
+    document_id: uuid.UUID
+    min_confidence: float = Field(default=0.5, ge=0.0, le=1.0)
+    max_results: int = Field(default=5, ge=1, le=20)