Datasets:

Languages:
English
Size:
n>1T
ArXiv:
License:

JSON ERROR in loading files of v1_6-sample using load_dataset

#22
by sakurapeng - opened

Error:

raise JSONDecodeError("Extra data", s, end)
json.decoder.JSONDecodeError: Extra data: line 2 column 1 (char 1469)

File: v1_5r2_sample-0097.json.gz (there is no problem loading the previous files

Test code: (no problem to load directly using gzip

import gzip
import json
from datasets import load_dataset

def read_gz_file(path):
    with gzip.open(path, 'rt') as f:
        for line in f:
            yield json.loads(line)

def load(path, data_file):
    ds = load_dataset(path, data_files=data_file, split="train", streaming=True)
    for example_i, example in enumerate(ds):
        print(example_i)
        print(example)
        break

def main():
    #path = "/data/dolma/v1_6-sample/v1_5r2_sample-0097.json.gz"  #
    #for data in read_gz_file(path):
    #    print(data) 
    #    break 
    path = "/data/dolma/v1_6-sample"
    data_file = ["v1_5r2_sample-0097.json.gz"]
    load(path, data_file)

if __name__ == "__main__":
    main()

Given that certain rows within this file exhibit varying key-value types—wherein "created" keys hold integer values while others store string values

You can use the following code to revise the "v1_5r2_sample-0097.json.gz"

import gzip
import json

def read_gz_file(path):
    """
    Read a gzipped JSON file line by line, yielding each JSON object.
    """
    with gzip.open(path, 'rt') as f:
        for line in f:
            yield json.loads(line)

def check_data_types(data):
    """
    Check and print inconsistencies in data types within the data,
    specifically looking at the 'created' field to identify type changes.
    """
    # Assuming 'created' field might have inconsistent types
    expected_type = None
    for i, item in enumerate(data):
        # Get the type of the 'created' field
        current_type = type(item.get("created"))
        if expected_type is None:
            expected_type = current_type
        elif current_type != expected_type:
            print(f"Inconsistency in type: At line {i+1}, 'created' field type changed from {expected_type} to {current_type}.")

def replace_missing_with_zero(data):
    """
    Replace missing 'created' values with 0.
    """
    for item in data:
        if item.get('created') is None:
            item['created'] = 0
    return data

def write_to_gz_file(data, path):
    """
    Write data to a JSON file, and then gzip compress it. The data is written line by line as JSON strings.
    """
    with open(path, 'w', encoding='utf-8') as f:
        for item in data:
            f.write(json.dumps(item) + '\n')


if __name__ == "__main__":
    # Read data from a gzipped file
    path = "./data/dolma_data/v1_6-sample/v1_5r2_sample-0097.json.gz"  
    data = read_gz_file(path)
    
    # Temporary output path for the processed data before compression
    temp_output_path = "./data/dolma_data/v1_6-sample/v1_5r2_sample-0097_v2_processed.json"
    
    # Replace missing 'created' values with 0
    processed_data = replace_missing_with_zero(data)
    
    # Write the processed data to a new file
    write_to_gz_file(processed_data, temp_output_path)
    
    # gzip your file

    # Attempt to read the processed and supposedly gzipped file
    new_path = "./data/dolma_data/v1_6-sample/v1_5r2_sample-0097_v2_processed.json.gz"
    data = read_gz_file(new_path)
    
    # Check data types and handle exceptions
    try:
        check_data_types(data)
    except json.JSONDecodeError as e:
        print(f"JSON parsing error: {e}")
    except Exception as e:
        print(f"Error processing data: {e}")
sakurapeng changed discussion status to closed

Sign up or log in to comment