|
| 1 | +from typing import Iterator |
| 2 | + |
| 3 | +import json |
| 4 | + |
| 5 | +from dataset_reader.base_reader import BaseReader, Record, Query |
| 6 | + |
| 7 | + |
| 8 | +class JSONReader(BaseReader): |
| 9 | + def __init__(self, path): |
| 10 | + self.path = path |
| 11 | + |
| 12 | + def read_queries(self) -> Iterator[Query]: |
| 13 | + with open(self.path, "r") as json_fp: |
| 14 | + for json_line in json_fp: |
| 15 | + line = json.loads(json_line) |
| 16 | + yield Query( |
| 17 | + vector=line, meta_conditions=None, expected_result=None, |
| 18 | + ) |
| 19 | + |
| 20 | + def read_data(self) -> Iterator[Record]: |
| 21 | + with open(self.path, "r") as json_fp: |
| 22 | + for idx, json_line in enumerate(json_fp): |
| 23 | + line = json.loads(json_line) |
| 24 | + yield Record(id=idx, vector=line, metadata=None) |
| 25 | + |
| 26 | + |
| 27 | +if __name__ == "__main__": |
| 28 | + import os |
| 29 | + from benchmark.settings import DATASET_DIR |
| 30 | + |
| 31 | + test_path = os.path.join(DATASET_DIR, "random-100", "vectors.jsonl") |
| 32 | + record = next(JSONReader(test_path).read_data()) |
| 33 | + print(record, end="\n\n") |
| 34 | + |
| 35 | + query = next(JSONReader(test_path).read_queries()) |
| 36 | + print(query) |
0 commit comments