wsd_plwordnet_glex.py

import json
import datasets

_DESCRIPTION = "definitions, usage examples and lemma candidates datasets from PlWn"

class WsdPlwordnetGlexConfig(datasets.BuilderConfig):
    """BuilderConfig for wsd_plwordnet_glex"""

    def __init__(self, url, data_dir, **kwargs):
        """
        BuilderConfig for wsd_plwordnet_glex.
        """
        super().__init__(version=datasets.Version("1.0.0"), **kwargs)
        self.url = url
        self.data_dir = data_dir

class WsdPlwordnetGlex(datasets.GeneratorBasedBuilder):
    """WSD PLWordnet GLEX dataset"""

    BUILDER_CONFIGS = [
        WsdPlwordnetGlexConfig(
            data_dir="data/definitions.jsonl",
            name="definitions",
            url = 'definitions.jsonl',
            description="Definitions from PLWordnet",
        ),
        WsdPlwordnetGlexConfig(
            data_dir="data/usage_examples.jsonl",
            name="usage_examples",
            description="Usage examples from PLWordnet",
            url = 'usage_examples.jsonl',

        ),
        WsdPlwordnetGlexConfig(
            data_dir="data/lemma_candidates.json",
            name="lemma_candidates",
            url = 'lemma_candidates.json',
            description="Lemma candidates from PLWordnet",

        ),
    ]

    def _info(self):
        features = datasets.Features()
        if self.config.name == "usage_examples":
            features = features = datasets.Features(
                {

                    "text": datasets.Value("string"),
                    "tokens": datasets.features.Sequence(
                        dict(
                            {
                                "index": datasets.Value("int32"),
                                "position": datasets.features.Sequence(
                                    length=2,
                                    feature=datasets.Value("int32"),
                                ),
                                "orth": datasets.Value("string"),
                                "lemma": datasets.Value("string"),
                                "pos": datasets.Value("string"),
                                "ctag": datasets.Value("string"),
                            }
                        ),
                    ),
                    "phrases": datasets.features.Sequence(datasets.Value("string")),
                    "wsd": datasets.features.Sequence(
                        dict({
                            "index": datasets.Value("int32"),
                            "pl_sense": datasets.Value("string"),
                            "plWN_lex_id": datasets.Value("string"),
                            "plWN_syn_id": datasets.Value("string"),
                            "plWN_lex_legacy_id": datasets.Value("string"),
                            "plWN_syn_legacy_id": datasets.Value("string"),
                            "PWN_syn_id": datasets.Value("string"),
                            "bn_syn_id": datasets.Value("string"),
                            "mapping_relation": datasets.Value("string"),
                            
                        })
                    ),
                    'context_file': datasets.Value('string'),
                }
            )
        if self.config.name == "definitions":
            features = datasets.Features(
                {
                    "pl_sense": datasets.Value("string"),
                    "text": datasets.Value("string"),
                    "tokens": datasets.features.Sequence(
                        dict(
                            {
                                "index": datasets.Value("int32"),
                                "position": datasets.features.Sequence(
                                    length=2,
                                    feature=datasets.Value("int32"),
                                ),
                                "orth": datasets.Value("string"),
                                "lemma": datasets.Value("string"),
                                "pos": datasets.Value("string"),
                                "ctag": datasets.Value("string"),
                            }
                        ),
                    ),
                    "phrases": datasets.features.Sequence(datasets.Value("string")),
                    "plWN_lex_id": datasets.Value("string"),
                    "plWN_syn_id": datasets.Value("string"),
                    "wsd_indices": datasets.features.Sequence(datasets.Value("int32")),

                }
            )
        if self.config.name == "lemma_candidates":
            features = datasets.Features(
                {
                    "lemma": datasets.Value("string"),
                    "candidates": datasets.features.Sequence(datasets.Value("string")),
                }
            )
        return datasets.DatasetInfo(
            description=_DESCRIPTION,
            features=features,
            )
    
    def _split_generators(self, dl_manager):
        dl_dir = dl_manager.download_and_extract(self.config.data_dir)
        return [
            datasets.SplitGenerator(
                name=datasets.Split.TRAIN,
                gen_kwargs={"path": dl_dir},
            ),
        ]

    def _generate_examples(self, path):
        """Generate examples."""

        with open(path, encoding="utf-8") as f:
            if self.config.name == "usage_examples":
                for k,line in enumerate(f):
                    data = json.loads(line)
                    yield k, {
                        "text": data["text"],
                        "tokens": data["tokens"],
                        "phrases": data["phrases"],
                        "wsd": data["wsd"],
                        'context_file': data['context_file'],
                    }
        
            if self.config.name == "definitions":
                for k,line in enumerate(f):
                    data = json.loads(line)
                    yield k, {
                        "pl_sense": data["pl_sense"],
                        "text": data["text"],
                        "tokens": data["tokens"],
                        "phrases": data["phrases"],
                        "plWN_lex_id": data["plWN_lex_id"],
                        "plWN_syn_id": data["plWN_syn_id"],
                        "wsd_indices": data["wsd_indices"],
                    }

            if self.config.name == "lemma_candidates":
                data = json.loads(f.read())
                processed = ((ind,{"lemma": k, "candidates": v}) for ind,(k,v) in enumerate(data.items()))
                yield from processed