Multi_Legal_Pile.py

"""Multi Legal Pile"""

import json

import datasets
from huggingface_hub.file_download import hf_hub_url

try:
    import lzma as xz
except ImportError:
    import pylzma as xz

datasets.logging.set_verbosity_info()
logger = datasets.logging.get_logger(__name__)

_DESCRIPTION = """
"""

_CITATION = """
"""

_URL = "https://huggingface.co/datasets/joelito/Multi_Legal_Pile"

# Specific Data
_DATA_NAME = {
    "bg": {
        "legislation": ["bulgaria_marcell"],
    },
    "cs": {
        "caselaw": ["czechia_constitutional_court", "czechia_supreme_administrative_court", "czechia_supreme_court"],
    },
    "da": {
        "caselaw": ["denmark_ddsc"],
        "legislation": ["denmark_ddsc"],
    },
    "de": {
        "caselaw": ["germany_openlegaldata"],
        "legislation": ["germany_openlegaldata", "switzerland_lexfind"],
    },
    "en": {
        "legislation": ["switzerland_lexfind"],
    },
    "fr": {
        "caselaw": ["belgium_jurportal", "france_cass", "luxembourg_judoc"],
        "legislation": ["belgium_ejustice", "switzerland_lexfind"],
    },
    "hu": {
        "caselaw": ["hungary_marcell"],
    },
    "it": {
        "legislation": ["switzerland_lexfind"],
    },
    "nl": {
        "legislation": ["belgium_ejustice"],
    },
    "pl": {
        "legislation": ["poland_marcell"],
    },
    "pt": {
        "caselaw": ["brazil_ruling_br"],
    },
    "ro": {
        "legislation": ["romania_marcell"],
    },
    "sk": {
        "legislation": ["slovakia_marcell"],
    },
    "sl": {
        "legislation": ["slovenia_marcell"],
    },
}

_PILE_OF_LAW_NAMES = {
    "caselaw": {
        "courtlistener_opinions": {
            "train": [
                "train.courtlisteneropinions.0",
                "train.courtlisteneropinions.1",
                "train.courtlisteneropinions.2",
                "train.courtlisteneropinions.3",
                "train.courtlisteneropinions.4",
                "train.courtlisteneropinions.5",
                "train.courtlisteneropinions.6",
                "train.courtlisteneropinions.7",
                "train.courtlisteneropinions.8",
                "train.courtlisteneropinions.9",
            ],
            "validation": [
                "validation.courtlisteneropinions.0",
                "validation.courtlisteneropinions.1",
                "validation.courtlisteneropinions.2",
                "validation.courtlisteneropinions.3"
            ]
        },
        "courtlistener_docket_entry_documents": {
            "train": [
                "train.courtlistenerdocketentries.0",
                "train.courtlistenerdocketentries.1",
                "train.courtlistenerdocketentries.2",
                "train.courtlistenerdocketentries.3"
            ],
            "validation": [
                "validation.courtlistenerdocketentries.0",
                "validation.courtlistenerdocketentries.1"
            ]
        },
        "scotus_filings": {
            "train": ["train.scotus_docket_entries"],
            "validation": ["validation.scotus_docket_entries"]
        },
        "echr": {
            "train": ["train.echr"],
            "validation": ["validation.echr"]
        },
        "canadian_decisions": {
            "train": ["train.canadian_decisions"],
            "validation": ["validation.canadian_decisions"]
        },
    },
    "contracts": {
        "atticus_contracts": {
            "train": [
                "train.atticus_contracts.0",
                "train.atticus_contracts.1",
                "train.atticus_contracts.2",
                "train.atticus_contracts.3",
                "train.atticus_contracts.4"
            ],
            "validation": [
                "validation.atticus_contracts.0",
                "validation.atticus_contracts.1"
            ]
        },
        "edgar": {
            "train": ["train.edgar"],
            "validation": ["validation.edgar"]
        },
        "cfpb_creditcard_contracts": {
            "train": ["train.cfpb_cc"],
            "validation": ["validation.cfpb_cc"]
        },
    },
    "legislation": {
        "uscode": {
            "train": ["train.uscode"],
            "validation": ["validation.uscode"]
        },
        "state_codes": {
            "train": ["train.state_code"],
            "validation": ["validation.state_code"]
        },
        "us_bills": {
            "train": ["train.us_bills"],
            "validation": ["validation.us_bills"]
        },
        "federal_register": {
            "train": ["train.federal_register"],
            "validation": ["validation.federal_register"]
        },
        "cfr": {
            "train": ["train.cfr"],
            "validation": ["validation.cfr"]
        },
        "constitutions": {
            "train": ["train.constitutions"],
            "validation": ["validation.constitutions"]
        },
    },
    "other": {  # analyses
        "oig": {
            "train": ["train.oig"],
            "validation": ["validation.oig"]
        },
        "olc_memos": {
            "train": ["train.olcmemos"],
            "validation": ["validation.olcmemos"]
        },
    },
}

_LANGUAGES = ["bg", "cs", "da", "de", "el", "en", "es", "et", "fi", "fr", "ga", "hr",
              "hu", "it", "lt", "lv", "mt", "nl", "pl", "pt", "ro", "sk", "sl", "sv"]

_TYPES = ["caselaw", "contracts", "legislation", "other"]

_JURISDICTONS = ["Austria", "Belgium", "Bulgaria", "Croatia", "Czechia", "Denmark", "Estonia", "Finland",
                 "France", "Germany", "Greece", "Hungary", "Ireland", "Italy", "Latvia", "Lithuania", "Luxembourg",
                 "Malta", "Netherlands", "Poland", "Portugal", "Romania", "Slovakia", "Slovenia", "Spain", "Sweden",
                 "EU", "Switzerland", "UK", "US", "Canada", "N/A"]


class MultiLegalPileConfig(datasets.BuilderConfig):
    """BuilderConfig for MultiLegalPile."""

    def __init__(self, name: str, **kwargs):
        """BuilderConfig for MultiLegalPile.
        Args:
            name: combination of language and type with _
            language:       One of bg,cs,da,de,el,en,es,et,fi,fr,ga,hr,hu,it,lt,lv,mt,nl,pl,pt,ro,sk,sl,sv or all
            type:           One of caselaw,contracts,legislation,other or all
          **kwargs: keyword arguments forwarded to super.
        """
        super(MultiLegalPileConfig, self).__init__(**kwargs)
        self.name = name
        self.language = name.split("_")[0]
        self.type = name.split("_")[1]


class MultiLegalPile(datasets.GeneratorBasedBuilder):
    """MultiLegalPile: A Corpus Covering Large-Scale Multilingual Legal Data"""

    BUILDER_CONFIGS = [MultiLegalPileConfig(f"{language}_{type}")
                       for type in _TYPES + ["all"]
                       for language in _LANGUAGES + ["all"]]

    def _info(self):
        return datasets.DatasetInfo(
            description=_DESCRIPTION,
            features=datasets.Features(
                {
                    "language": datasets.Value("string"),  # one of _LANGUAGES
                    "type": datasets.Value("string"),  # one of _TYPES
                    "jurisdiction": datasets.Value("string"),  # one of _JURISDICTONS
                    "text": datasets.Value("string"),
                }
            ),
            supervised_keys=None,
            homepage=_URL,
            citation=_CITATION,
        )

    def _split_generators(self, dl_manager):
        def download_url(dataset, file_name):
            url = hf_hub_url(repo_id=dataset, filename=f"data/{file_name}.jsonl.xz", repo_type="dataset")
            return dl_manager.download(url)

        data_infos = []
        languages = _LANGUAGES if self.config.language == "all" else [self.config.language]
        types = _TYPES if self.config.type == "all" else [self.config.type]

        for language in languages:
            for type in types:
                info = {"language": language, "type": type}
                # add native MultiLegalPile data
                if language in _DATA_NAME and type in _DATA_NAME[language]:  # if we actually have something there
                    for data_name in _DATA_NAME[language][type]:
                        info["filepath"] = download_url("joelito/Multi_Legal_Pile", f"{language}/{type}/{data_name}")
                        data_infos.append(info.copy())

                # add eurlex_resources data
                info["jurisdiction"] = "EU"
                if type == 'caselaw':
                    info["filepath"] = download_url("joelito/eurlex_resources", f"{language}/caselaw")
                    data_infos.append(info.copy())
                elif type == 'legislation':
                    for resource_type in ["decision", "directive", "regulation", "recommendation", "proposal"]:
                        info["filepath"] = download_url("joelito/eurlex_resources", f"{language}/{resource_type}")
                        data_infos.append(info.copy())
                elif type == 'contracts':
                    # international agreements
                    info["filepath"] = download_url("joelito/eurlex_resources", f"{language}/intagr")
                    data_infos.append(info.copy())

                # add mc4_legal data
                if type == 'other':
                    info["jurisdiction"] = "N/A"
                    if language in ["de", "en", "es"]:  # here we need to chunk because the files are too large
                        for idx in [0, 1]:
                            info["filepath"] = download_url("joelito/mc4_legal", f"{language}_{idx}")
                            data_infos.append(info.copy())
                    elif language in ["hr"]:
                        pass  # nothing there for Croatian
                    else:
                        info["filepath"] = download_url("joelito/mc4_legal", f"{language}")
                        data_infos.append(info.copy())

                # add pile-of-law data
                if language == 'en':
                    info["jurisdiction"] = "US"
                    for name, train_val_files in _PILE_OF_LAW_NAMES[type].items():
                        if name == 'canadian_decisions':
                            info["jurisdiction"] = "Canada"
                        elif name in ['echr']:
                            info["jurisdiction"] = "EU"
                        elif name in ['constitutions']:
                            info["jurisdiction"] = "N/A"
                        for split in ["train", "validation"]:
                            for file_name in train_val_files[split]:
                                info["filepath"] = download_url("pile-of-law/pile-of-law", file_name)
                                data_infos.append(info.copy())

        return [datasets.SplitGenerator(name=datasets.Split.TRAIN, gen_kwargs={"data_infos": data_infos})]

    def _generate_examples(self, data_infos):
        """This function returns the examples in the raw (text) form by iterating on all the files."""
        id_ = 0
        for data_info in data_infos:
            logger.info("Generating examples from = %s", data_info["filepath"])
            try:
                with xz.open(open(data_info["filepath"], "rb"), "rt", encoding="utf-8") as f:
                    for line in f:
                        if line:
                            example = json.loads(line)
                            if example is not None and isinstance(example, dict):
                                yield id_, {
                                    # uses the default values for language, type and jurisdiction if it is not set
                                    "language": example.get("language", data_info["language"]),
                                    "type": example.get("type", data_info["type"]),
                                    "jurisdiction": example.get("jurisdiction", data_info["jurisdiction"]),
                                    "text": example.get("text", ""),
                                }
                                id_ += 1
            except Exception:
                logger.exception("Error while processing file %s", data_info["filepath"])