xor-tydi.py

# coding=utf-8
# Copyright 2020 The HuggingFace Datasets Authors and the current dataset script contributor.
#
# Licensed under the Apache License, Version 2.0 (the "License");
# you may not use this file except in compliance with the License.
# You may obtain a copy of the License at
#
#     http://www.apache.org/licenses/LICENSE-2.0
#
# Unless required by applicable law or agreed to in writing, software
# distributed under the License is distributed on an "AS IS" BASIS,
# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
# See the License for the specific language governing permissions and
# limitations under the License.

# Lint as: python3
"""Wikipedia NQ dataset."""

import json
import random

random.seed(42)

import datasets
RANGE = (0, 1000)

_CITATION = """
@inproceedings{xorqa,
    title   = {{XOR} {QA}: Cross-lingual Open-Retrieval Question Answering},
    author  = {Akari Asai and Jungo Kasai and Jonathan H. Clark and Kenton Lee and Eunsol Choi and Hannaneh Hajishirzi},
    booktitle={NAACL-HLT},
    year    = {2021}
}
"""

_DESCRIPTION = "dataset load script for Wikipedia NQ"

base = "/home/czhang/src/task-sparse/tevatron/hgf_datasets/xor-tydi"
_DATASET_URLS = {
    'targetQ': {
        'train': f'https://huggingface.co./datasets/crystina-z/xor-tydi/resolve/main/train/targetL_dpr_train_data.json',
        'dev': f'https://huggingface.co./datasets/crystina-z/xor-tydi/resolve/main/dev/xor_dev_retrieve_eng_span_v1_1.jsonl',
        'test': f'https://huggingface.co./datasets/crystina-z/xor-tydi/resolve/main/test/xor_test_retrieve_eng_span_q_only_v1_1.jsonl',
    },
    'engQ': {
        'train': f'https://huggingface.co./datasets/crystina-z/xor-tydi/resolve/main/train/EN_dpr_train_data.json',
    }
}


class XORTyDi(datasets.GeneratorBasedBuilder):
    VERSION = datasets.Version("0.0.1")

    BUILDER_CONFIGS = [
        datasets.BuilderConfig(
            version=VERSION,
            name="targetQ",
            description="XOR-TyDI train/dev/test datasets of English Span Task"),
        datasets.BuilderConfig(
            version=VERSION,
            name="engQ",
            description="XOR-TyDI train/dev/test datasets of Full Task"), 
    ]

    def _info(self):
        features = datasets.Features({
            'query_id': datasets.Value('string'),
            'query': datasets.Value('string'),
            'answers': [datasets.Value('string')],
            'lang': datasets.Value('string'),
            'positive_passages': [
                {'docid': datasets.Value('string'), 'text': datasets.Value('string'),
                 'title': datasets.Value('string')}
            ],
            'negative_passages': [
                {'docid': datasets.Value('string'), 'text': datasets.Value('string'),
                 'title': datasets.Value('string')}
            ],
        })
        return datasets.DatasetInfo(
            # This is the description that will appear on the datasets page.
            description=_DESCRIPTION,
            # This defines the different columns of the dataset and their types
            features=features,  # Here we define them above because they are different between the two configurations
            supervised_keys=None,
            # Homepage of the dataset for documentation
            homepage="",
            # License for the dataset if available
            license="",
            # Citation for the dataset
            citation=_CITATION,
        )

    def _split_generators(self, dl_manager):
        group = self.config.name
        if self.config.data_files:
            downloaded_files = self.config.data_files
        else:
            downloaded_files = dl_manager.download_and_extract(_DATASET_URLS[group])
        splits = [
            datasets.SplitGenerator(
                name=split,
                gen_kwargs={
                    "files": [downloaded_files[split]] if isinstance(downloaded_files[split], str) else downloaded_files[split],
                },
            ) for split in downloaded_files
        ]
        return splits
    
    def _generate_examples(self, files):
        assert len(files) == 1
        filepath = files[0]


        def process_doc_text(doc):
            if isinstance(doc["text"], list):
                assert len(doc["text"]) == 1
                return doc['text'][0].strip()
            else:
                assert isinstance(doc["text"], str)
                return doc['text'].strip()

        
        # prepare doc
        def get_doc2docid(all_data):
            doc2docid = {}
            # with open(filepath, encoding="utf-8") as f:
                # all_data = json.load(f)
            for i, data in enumerate(all_data):
                positive_ctxs = data["positive_ctxs"]
                hard_negative_ctxs = data["hard_negative_ctxs"]
                ctxs = positive_ctxs + hard_negative_ctxs

                for doc in ctxs:
                    text = process_doc_text(doc)
                    if text not in doc2docid:
                        doc2docid[text] = len(doc2docid)
            return doc2docid
                
        def process_train_entry(data, _id, doc2docid):
            positive_ctxs = data["positive_ctxs"]
            hard_negative_ctxs = data["hard_negative_ctxs"]
            # each ctx: {'title':... , 'text': ....}

            def process_ctx(ctxs, tag):
                processed = []
                for i, doc in enumerate(ctxs):
                    text = process_doc_text(doc)
                    processed.append({
                        "title": doc["title"],
                        "text": text,
                        # 'docid': f'{tag}-{i}-{random.randint(*RANGE)}'
                        'docid': doc2docid[text]
                    }) 
                return processed

            return _id, {
                "query_id": _id,
                "query": data["question"],
                "answers": data.get("answers", []),
                "lang": "",
                "positive_passages": process_ctx(positive_ctxs, "pos"),
                "negative_passages": process_ctx(hard_negative_ctxs, "neg"),
            }

        def process_dev_test_entry(data):
            return data["id"], {
                "query_id": data["id"],
                "query": data["question"],
                "answers": data.get("answers", []),
                "lang": data["lang"],
                "positive_passages": [],
                "negative_passages": [],
            }


        try:
            with open(filepath, encoding="utf-8") as f:
                all_data = json.load(f)
                doc2docid = get_doc2docid(all_data)
                for i, data in enumerate(all_data):
                    yield process_train_entry(data, i, doc2docid)

        # if filepath.endswith(".jsonl"): <-- doesn't work
        except Exception as e:
            with open(filepath, encoding="utf-8") as f:
                for line in f:
                    data = json.loads(line)

                    if "id" in data and "query_id" not in data:
                        yield process_dev_test_entry(data)