| """Path resolution: prefer vendored data (HF Spaces), fall back to sibling repos (local dev).""" |
| from pathlib import Path |
|
|
| APP_DIR = Path(__file__).resolve().parent |
| VENDORED = APP_DIR / "data" |
| REPO_ROOT = APP_DIR.parent |
| CODE_DIR = REPO_ROOT / "code" |
|
|
| EACL_REPO = CODE_DIR / "EACL2026-historical-languages" |
| TUTORIAL_REPO = CODE_DIR / "lrec2026-llm-as-annotator-tutorial" |
|
|
|
|
| def _pick(vendored: Path, fallback: Path) -> Path: |
| """Return whichever exists; vendored wins on HF Spaces, fallback wins locally.""" |
| return vendored if vendored.exists() else fallback |
|
|
|
|
| TAGSETS_DIR = _pick(VENDORED / "tagsets", EACL_REPO / "data" / "tagsets") |
| CORPUS_DIR = _pick(VENDORED / "corpus", EACL_REPO / "data" / "corpus") |
| REF_PROMPTS_DIR = _pick(VENDORED / "prompts", EACL_REPO / "data" / "prompt-templates") |
|
|
| TUTORIAL_PROMPTS_DIR = _pick(VENDORED / "prompts", TUTORIAL_REPO / "notebooks" / "prompts") |
| TUTORIAL_SCHEMAS_DIR = _pick(VENDORED / "schemas", TUTORIAL_REPO / "schemas") |
| TUTORIAL_HANDOUTS_DIR = _pick(VENDORED / "handouts", TUTORIAL_REPO / "handouts") |
|
|
| LANGUAGES = { |
| "GRC": "Ancient Greek", |
| "HYE": "Old Armenian", |
| "KAT": "Old Georgian", |
| "SYC": "Syriac", |
| } |
|
|
|
|
| def tagset_file(lang_code: str) -> Path: |
| return TAGSETS_DIR / f"{lang_code}_tagset.txt" |
|
|
|
|
| def corpus_file(lang_code: str, split: str = "train") -> Path: |
| """split in {'train', 'test_in-domain', 'test_out-of-domain'}. Vendored data has only 'train'.""" |
| suffix = "train" if split == "train" else "test" |
| return CORPUS_DIR / split / f"{lang_code}_{suffix}.tsv" |
|
|
|
|
| def read_text(path: Path) -> str: |
| return path.read_text(encoding="utf-8") |
|
|