src/data_collection/conversion.py

import pandas as pd
import csv

# INPUT AND OUTPUT FILE NAMES
input_file = "AI-MODEL-FINGERPRINTING/src/data_collection/claude_raw.txt"
output_file = "AI-MODEL-FINGERPRINTING/src/data_collection/claude_responses.csv"
#input_file = "AI-MODEL-FINGERPRINTING/src/data_collection/openai_raw.txt"
#output_file = "AI-MODEL-FINGERPRINTING/src/data_collection/openai_responses.csv"


# Read complete text file
with open(input_file, "r", encoding="utf-8") as f:
    content = f.read()

# Split each response block
blocks = content.split("===END===")

rows = []

for block in blocks:
    block = block.strip()

    if not block:
        continue

    lines = block.split("\n")

    prompt_id = ""
    category = ""
    model = ""
    prompt = ""
    response = ""

    response_started = False
    response_lines = []

    for line in lines:

        if line.startswith("PROMPT_ID:"):
            prompt_id = line.replace("PROMPT_ID:", "").strip()

        elif line.startswith("CATEGORY:"):
            category = line.replace("CATEGORY:", "").strip()

        elif line.startswith("MODEL:"):
            model = line.replace("MODEL:", "").strip()

        elif line.startswith("PROMPT:"):
            prompt = line.replace("PROMPT:", "").strip()

        elif line.startswith("RESPONSE:"):
            response_started = True

        elif response_started:
            response_lines.append(line.strip())

    response = " ".join(response_lines)

    rows.append([
        prompt_id,
        category,
        model,
        prompt,
        response
    ])

# Create dataframe
df = pd.DataFrame(
    rows,
    columns=[
        "prompt_id",
        "category",
        "model",
        "prompt",
        "response"
    ]
)

# Save proper CSV
df.to_csv(
    output_file,
    index=False,
    encoding="utf-8",
    
)

print("CSV file created successfully")
print("Saved as:", output_file)