import pandas as pd

def extract_data(path_file, path_save):
    names = []
    acc_replies = []
    acc_sources = []
    with open(path_file) as fp:
        lines = fp.readlines()

    for line in lines:
        if "test_RAG" in line:
            names += [line.strip()]
        elif "Accuracy replies:" in line:
            line = line.replace("Accuracy replies: ", "")
            acc_replies += [float(line.strip())]
        elif "Accuracy sources:" in line:
            line = line.replace("Accuracy sources: ", "")
            acc_sources += [float(line.strip())]

    df = pd.DataFrame(data = {'Name': names, 
                            'Accuracy replies': acc_replies, 
                            'Accuracy sources': acc_sources})

    df.to_csv(path_save, index=False)


if __name__ == "__main__":
    path_file = "out_QA_Nicolas.txt"
    path_save = "results_QA_Nicolas.csv"
    # path_file = "out_QA_generated.txt"
    # path_save = "results_QA_generated.csv"
    extract_data(path_file, path_save)