danbooru_to_ja.py

import argparse
import re
from pathlib import Path


def load_json_dict(path: str):
    import json
    from pathlib import Path
    dict = {}
    if not Path(path).exists(): return dict
    try:
        with open(path, encoding='utf-8') as f:
            dict = json.load(f)
    except Exception:
        print(f"Failed to open dictionary file: {path}")
        return dict
    return dict


danbooru_ja_dict = load_json_dict('e621_ja_dict.json') | load_json_dict('danbooru_ja_dict.json')


def danbooru_tags_to_jas(tags: list[str]):
    from rapidfuzz.process import extractOne
    from rapidfuzz.utils import default_process
    A1111_SPECIAL_SYNTAX_RE = re.compile(r"\s*<[^>]+>\s*")
    keys = list(danbooru_ja_dict.keys())
    jas = []
    for tag in tags:
        tag = str(tag).strip()
        if A1111_SPECIAL_SYNTAX_RE.fullmatch(tag): continue
        s = default_process(str(tag))
        e1 = extractOne(s, keys, processor=default_process, score_cutoff=90.0)
        if e1:
            jas.extend(danbooru_ja_dict[e1[0]].copy())
    return jas


def danbooru_to_ja(input_tag, input_file, output_file, is_append):
    if input_file and Path(input_file).exists():
        try:
            with open(input_file, 'r', encoding='utf-8') as f:
                input_tag = f.read()
        except Exception:
            print(f"Failed to open input file: {input_file}")
    tags = [tag.strip() for tag in input_tag.split(",")] if input_tag else []
    ja_tags = danbooru_tags_to_jas(tags)
    output_tags = tags + ja_tags if is_append else ja_tags
    output_tag = ", ".join(output_tags)
    if output_file:
        try:
            with open(output_file, mode='w', encoding="utf-8") as f:
                f.write(output_tag)
        except Exception:
            print(f"Failed to write output file: {output_file}")
    else:
        print(output_tag)
    return output_tag


if __name__ == "__main__":
    parser = argparse.ArgumentParser()
    parser.add_argument("--tags", default=None, type=str, required=False, help="Input tags.")
    parser.add_argument("--file", default=None, type=str, required=False, help="Input tags from a text file.")
    parser.add_argument("--out", default=None, type=str, help="Output to text file.")
    parser.add_argument("--append", default=False, type=bool, help="Whether the output contains the input tags or not.")

    args = parser.parse_args()
    assert (args.tags, args.file) != (None, None), "Must provide --tags or --file!"

    danbooru_to_ja(args.tags, args.file, args.out, args.append)


# Usage:
# python danbooru_to_ja.py --tags "1girl, oomuro sakurako, solo, sitting, starry sky"
# python danbooru_to_ja.py --file inputtag.txt
# python danbooru_to_ja.py --file inputtag.txt --append True
# Datasets: https://huggingface.co/datasets/p1atdev/danbooru-ja-tag-pair-20240715
# Datasets: https://github.com/ponapon280/danbooru-e621-converter