Indian_CIVICS.py

from datasets import Dataset, Features, Value, Sequence
import json
import os

# --- 1. Configuration ---
JSON_FILE_PATH = "Indian_CIVICS_Dataset.json"
OUTPUT_DIR = "indian_civics_dataset_hf"

# --- 2. Define the Dataset Schema (Features) ---
# Defines the column names and their strict data types for the Hugging Face Dataset.

DATASET_FEATURES = Features(
    {
        "ID": Value(dtype="string"),
        "Topic": Value(dtype="string"),
        "Sub-Topic": Value(dtype="string"),
        "Statement": Value(dtype="string"),  # Original Indian Language text (Hindi/Telugu)
        "Statement - Translation": Value(dtype="string"), # English translation
        "Data Source": Value(dtype="string"),
        "Data Producer Organization": Value(dtype="string"),
        "Organization Type": Value(dtype="string"),
        "Language": Value(dtype="string"),
        "State/Region": Value(dtype="string"),
        "Link": Value(dtype="string"), # Matches the 'Link' key in your JSON
        # Sequence is used for a list of strings (e.g., ["Tag 1", "Tag 2"])
        "Consensus Value Annotation": Sequence(feature=Value(dtype="string")),
    }
)

# --- 3. Data Utility Functions ---

def load_data_from_json(file_path):
    """Loads a list of dictionaries from a JSON file."""
    if not os.path.exists(file_path):
        raise FileNotFoundError(f"Error: JSON file not found at {file_path}")
    # Use 'utf-8' encoding for correct handling of Hindi and Telugu characters
    with open(file_path, 'r', encoding='utf-8') as f:
        data_list = json.load(f)
    return data_list

def convert_to_hf_format(data_list):
    """Converts a list of dictionaries into a dictionary of lists (HF format)."""
    if not data_list:
        return {}
    
    # Initialize the dictionary of lists based on the first item's keys
    keys = data_list[0].keys()
    hf_data = {key: [] for key in keys}
    
    # Populate the lists
    for item in data_list:
        for key in keys:
            hf_data[key].append(item.get(key)) 
            
    return hf_data

# --- 4. Main Execution ---

if __name__ == "__main__":
    try:
        print(f"Loading data from: {JSON_FILE_PATH}...")
        
        # Load the data (List of Dicts)
        statement_list = load_data_from_json(JSON_FILE_PATH)
        print(f"✅ Loaded {len(statement_list)} entries.")
        
        # Transform the data structure (Dict of Lists)
        hf_sample_data = convert_to_hf_format(statement_list)
        
        # --- Create and Save the Dataset ---
        dataset = Dataset.from_dict(
            hf_sample_data, 
            features=DATASET_FEATURES # Apply the defined schema
        )

        os.makedirs(OUTPUT_DIR, exist_ok=True) 
        dataset.save_to_disk(OUTPUT_DIR)

        print("-" * 50)
        print(f"✅ Dataset successfully created.")
        print(f"Total examples: {len(dataset)}")
        print(f"Dataset saved locally to: ./{OUTPUT_DIR}")
        print("-" * 50)
        print("--- Sample Example (Index 0) ---")
        print(dataset[0])
        print("-" * 50)

    except FileNotFoundError as e:
        print(f"\nFATAL ERROR: {e}")
        print(f"Please make sure the file '{JSON_FILE_PATH}' is in the same directory as '{os.path.basename(__file__)}'.")
    except Exception as e:
        print(f"\nAn unexpected error occurred during dataset creation: {e}")