| |
| |
| |
| """Canonical 11-class layout taxonomy.""" |
| from __future__ import annotations |
|
|
| from ax_doclayout.errors import TaxonomyError |
|
|
| TAXONOMY_ID = "ax-doclayout-11/1" |
| RESERVED_ID = 0 |
|
|
| CLASS_NAMES: dict[int, str] = { |
| 1: "caption", |
| 2: "footnote", |
| 3: "formula", |
| 4: "list_item", |
| 5: "page_footer", |
| 6: "page_header", |
| 7: "picture", |
| 8: "section_header", |
| 9: "table", |
| 10: "text", |
| 11: "title", |
| } |
|
|
| NAME_TO_ID: dict[str, int] = {name: class_id for class_id, name in CLASS_NAMES.items()} |
|
|
|
|
| def public_name(class_id: int) -> str: |
| if class_id == RESERVED_ID: |
| raise TaxonomyError("class_id_reserved") |
| try: |
| return CLASS_NAMES[class_id] |
| except KeyError as exc: |
| raise TaxonomyError("class_id_unknown") from exc |
|
|
|
|
| def public_id(class_name: str) -> int: |
| try: |
| return NAME_TO_ID[class_name] |
| except KeyError as exc: |
| raise TaxonomyError("class_name_unknown") from exc |
|
|
|
|
| def to_internal(class_id: int) -> int: |
| """Map public 1–11 to zero-based indices 0–10.""" |
| if class_id not in CLASS_NAMES: |
| raise TaxonomyError("class_id_unknown") |
| return class_id - 1 |
|
|
|
|
| def from_internal(index: int) -> int: |
| class_id = index + 1 |
| if class_id not in CLASS_NAMES: |
| raise TaxonomyError("class_id_unknown") |
| return class_id |
|
|
