LLMValues / 4-HSLS /parse_human_response.py
yijiexu's picture
Initial push, ignore notebooks
2a1ec59 verified
Raw History Blame Contribute Delete
3.79 kB
import pandas as pd
import json
import sys
def generate_student_answers(record_file, data_df, output_file):
"""
Generates an answer JSONL file from a record file and a student data DataFrame.
Args:
record_file (str): The path to the record_file.jsonl.
data_df (pd.DataFrame): The DataFrame containing student data.
output_file (str): The name for the output JSONL file.
"""
# Step 1: Read all valid question_ids from the record file
print(f"Reading question IDs from '{record_file}'...")
try:
with open(record_file, 'r', encoding='utf-8') as f:
# Use a set for efficient lookup
valid_question_ids = {json.loads(line)['question_id'] for line in f}
print(f"Found {len(valid_question_ids)} valid question IDs.")
except FileNotFoundError:
print(f"Error: The file '{record_file}' was not found. Please check the path.")
sys.exit(1)
except (json.JSONDecodeError, KeyError):
print(f"Error: The file '{record_file}' is not a valid JSONL or is missing 'question_id' keys.")
sys.exit(1)
# Step 2: Identify the columns to process in the DataFrame
student_id_col = 'STU_ID'
if student_id_col not in data_df.columns:
print(f"Error: The required student identifier column '{student_id_col}' was not found in the DataFrame.")
sys.exit(1)
# Get the intersection of DataFrame columns and valid question IDs
columns_to_process = [col for col in data_df.columns if col in valid_question_ids]
print(f"Will process {len(columns_to_process)} question columns found in the DataFrame.")
# Step 3: Iterate through each row of the DataFrame and write to the JSONL file
print(f"Generating answer file '{output_file}'...")
try:
with open(output_file, 'w', encoding='utf-8') as f:
# Using to_dict('records') is an efficient way to iterate over rows
for record in data_df.to_dict('records'):
student_answer = {
# Add the student ID
'student_id': record.get(student_id_col)
}
# Add the answers for all relevant questions
for col_name in columns_to_process:
student_answer[col_name] = record.get(col_name)
# Write the student's complete answer dictionary to the file
f.write(json.dumps(student_answer, ensure_ascii=False) + '\n')
except IOError:
print(f"Error: Could not write to file '{output_file}'. Check permissions or path.")
sys.exit(1)
print(f"\n✅ Success! Processing is complete.")
print(f"Generated records for {len(data_df)} students in '{output_file}'.")
if __name__ == '__main__':
# --- Configuration ---
# 1. Record file (codebook) path
record_filename = "record_file.jsonl"
# 2. Raw student data file path
data_filename = "hsls_17_student_pets_sr_v1_0.csv"
# 3. Desired name for the final output file
output_filename = "student_answers.jsonl"
# --- Load Data ---
print(f"Loading data from '{data_filename}'...")
try:
# Using the CSV reader for your specified file
data = pd.read_csv(data_filename)
print(f"Successfully loaded {len(data)} rows and {len(data.columns)} columns.")
except FileNotFoundError:
print(f"Error: The data file '{data_filename}' was not found. Make sure it's in the same directory.")
sys.exit(1)
except Exception as e:
print(f"An error occurred while loading the data: {e}")
sys.exit(1)
# --- Run the main function ---
generate_student_answers(record_filename, data, output_filename)