classify.py

from regex_helper import classify_with_regex
from bert_helper import classify_with_bert
from llm_helper import classify_with_llm

def classify(logs):
    labels = []
    for source, log_msg in logs:
        label = classify_log(source, log_msg)
        labels.append(label)
    return labels


def classify_log(source, log_msg):
    if source == "LegacyCRM":
        label = classify_with_llm(log_msg)
    else:
        label = classify_with_regex(log_msg)
        if not label:
            label = classify_with_bert(log_msg)
    return label

def classify_csv(input_file):
    import pandas as pd
    df = pd.read_csv(input_file)

    # Perform classification
    df["target_label"] = classify(list(zip(df["source"], df["log_message"])))

    # Save the modified file
    output_file = "testing/output.csv"
    df.to_csv(output_file, index=False)

    return output_file

if __name__ == '__main__':
    classify_csv("testing/test.csv")