parquet2json / app.py
TheDrummer's picture
Update app.py
409fb0b verified
raw
history blame
807 Bytes
import gradio as gr
import pandas as pd
import json
import jsonlines
import tempfile
import os
def convert_parquet_to_jsonl(file):
df = pd.read_parquet(file.name)
output_file = tempfile.NamedTemporaryFile(delete=False, suffix='.json')
df.to_json(output_file.name, orient='records')
return output_file.name
def convert_file(file):
if file.name.endswith('.parquet'):
return convert_parquet_to_jsonl(file)
else:
return "Unsupported file format. Please upload a .parquet or .json file."
iface = gr.Interface(
fn=convert_file,
inputs=gr.File(label="Upload .parquet file"),
outputs=gr.File(label="Download converted .json file"),
title="Parquet to JSON Converter",
description="Upload a .parquet file to convert it to .json format."
)
iface.launch()