OBIMD.py

import json
import os
from datasets import DatasetInfo, BuilderConfig, GeneratorBasedBuilder, SplitGenerator, \
    Split, Features, Value, Sequence, Image
import zipfile

class OBIMDConfig(BuilderConfig):
    def __init__(self, **kwargs):
        super().__init__(version="1.0.0", **kwargs)


class OBIMDDataset(GeneratorBasedBuilder):
    BUILDER_CONFIGS = [
        OBIMDConfig(name="default", description="Oracle bone OBIMD dataset."),
    ]

    def _info(self):
        return DatasetInfo(
            description="Paired images of oracle bone rubbings and facsimiles with annotations.",
            features=Features({
                "Facsimile": Image(),
                "Rubbing": Image(),
                "RubbingName": Value("string"),
                "RecordUtilSentenceGroupVoList": Sequence({
                    "GroupCategory": Value("string"),
                    "RecordUtilOracleCharVoList": Sequence({
                        "Position": Value("string"),
                        "OrderNumber": Value("int32"),
                        "SeatFont": Value("int32"),
                        "Mark": Value("int32"),
                        "Label": Value("string"),
                        "SubLabel": Value("string"),
                    }),
                }),
            }),
            supervised_keys=None,
        )

    def _split_generators(self, dl_manager):
        if os.path.exists(self.base_path):
            data_dir = self.base_path
        else:
            data_dir = dl_manager.download_and_extract(self.config.data_dir or ".")

        return [
            SplitGenerator(
                name=Split.TRAIN,
                gen_kwargs={
                    "filepath": os.path.join(data_dir, "data.json"),
                    "base_path": data_dir,
                },
            ),
        ]

    def _generate_examples(self, filepath, base_path):
        with open(filepath, encoding="utf-8") as f:
            data = json.load(f)
        with zipfile.ZipFile(os.path.join(base_path, "rubbing.zip"), 'r') as rubbing_ref:
            with zipfile.ZipFile(os.path.join(base_path, "facsimile.zip"), 'r') as facsimile_ref:
                for idx, item in enumerate(data):
                    
                    with rubbing_ref.open(item["Rubbing"]) as file:
                        rubbing_content = file.read()
                    with facsimile_ref.open(item["Facsimile"]) as file:
                        facsimile_content = file.read()
                        
                    yield idx, {
                        "Facsimile": facsimile_content,
                        "Rubbing": rubbing_content,
                        "RubbingName": item["RubbingName"],
                        "RecordUtilSentenceGroupVoList": [
                            {
                                "GroupCategory": group["GroupCategory"],
                                "RecordUtilOracleCharVoList": [
                                    {
                                        "Position": char["Position"],
                                        "OrderNumber": char["OrderNumber"],
                                        "SeatFont": char["SeatFont"],
                                        "Mark": char["Mark"],
                                        "Label": char["Label"],
                                        "SubLabel": char["SubLabel"],
                                    }
                                    for char in group["RecordUtilOracleCharVoList"]
                                ],
                            }
                            for group in item["RecordUtilSentenceGroupVoList"]
                        ],
                    }