fast-antx

Fast implementation of the antx text annotation tool. Transfer annotations from one text to another using regex patterns and an optimized diff algorithm.

Installation

Requires Python 3.8+

From GitHub

pip install "git+https://github.com/OpenPecha/fast-antx.git"

From source

git clone https://github.com/OpenPecha/fast-antx.git
cd fast-antx
# optional but recommended
python -m venv .venv && source .venv/bin/activate
pip install -e .

For development:

pip install -e ".[dev]"
pre-commit install
pytest

Usage

Basic example with custom patterns:

from fast_antx.core import transfer

source = "[1a]Hello\n[1a.1]World"
target = "Hello\nWorld"
patterns = [
    ["pages", r"(\[\d+[ab]\])"],
    ["lines", r"(\[\d+\.\d\])"],
]

# output: "txt" | "yaml" | "diff"
annotated = transfer(source, patterns, target, "txt")
print(annotated)

Using built-in HFML tag patterns:

from fast_antx.core import transfer
from fast_antx.ann_patterns import HFML_ANN_PATTERN

layer = "<...>"  # text containing HFML-like tags
base = "..."     # target text to receive transferred annotations

result = transfer(layer, HFML_ANN_PATTERN, base, "txt")

Notes:

• On first use, a small node-dmp-cli binary is downloaded automatically to ~/.antx/bin/ to speed up diff computation.
• See tests/ for more examples (e.g., tests/test_annotation_transfer.py).

Contributing

Contributions are welcome!

• Open an issue for bugs or feature requests.
• Fork the repo and create a feature branch.
• Set up dev environment: pip install -e ".[dev]" then pre-commit install.
• Run tests with pytest and include tests for new features.
• Submit a pull request to main.

License (MIT)

This project is licensed under the MIT License. See the LICENSE file for details.