Spaces:
Running on Zero
Running on Zero
File size: 1,787 Bytes
53d9817 | 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 | from __future__ import annotations
import uuid
from typing import Literal
from pydantic import BaseModel, Field
class BBox(BaseModel):
"""Bounding box of an element on a page. Coordinates in points (PDF units)."""
x0: float
y0: float
x1: float
y1: float
class TableData(BaseModel):
"""Structured representation of a table."""
headers: list[str]
rows: list[list[str]]
class Element(BaseModel):
"""A single unit of content extracted from a document."""
element_id: str = Field(default_factory=lambda: str(uuid.uuid4()))
type: Literal["heading", "paragraph", "table", "figure", "list_item", "code_block"]
level: int | None = None # heading level: 1, 2, 3 etc. None for non-headings
text: str | None = None
page: int | None = None # 1-indexed page number
bbox: BBox | None = None
parent_id: str | None = (
None # links to a parent element (e.g. heading this paragraph belongs to)
)
data: TableData | None = None # only for type="table"
markdown_repr: str | None = None # pre-rendered markdown string of this element
confidence: float | None = None # 0.0 to 1.0, used for tables and OCR output
vlm_description: str | None = None # Phase 8 — optional VLM-generated caption for figures
class DocumentMetadata(BaseModel):
"""Facts about the source file — not its content."""
file_name: str
file_type: str
page_count: int | None = None
has_scanned_pages: bool = False
class Document(BaseModel):
"""The top-level output object. This is what parse() returns."""
schema_version: str = "1.0"
doc_id: str = Field(default_factory=lambda: str(uuid.uuid4()))
metadata: DocumentMetadata
content_tree: list[Element] = Field(default_factory=list)
|