import os
import pandas as pd
import numpy as np


compare_to_gt = True
ground_truth = pd.read_csv("data/zipfiles.csv", sep="\t")
results = pd.read_csv("data/results.csv", sep="\t")
verbose = 0

eval_readme = []
eval_training = []
eval_evaluating = []
eval_licensing = []
eval_weights = []
eval_dependencies = []
full_results = []
for (index1, row1), (index2, row2) in zip(ground_truth.iterrows(), results.iterrows()):
    if (pd.isna(row1["training"])):
        continue

    print(f"\nEvaluating {index1+1} out of {len(ground_truth.index)} papers...")
    print(f'Paper title - "{row1["title"]}" ({row1["year"]})')
    print(f'Repository link - {row1["url"]}')
    if ((not(pd.isna(row1["dependencies"]))) & (row2["pred_dependencies"] is not None)):
        eval_dependencies.append(row2["pred_dependencies"] == row1["dependencies"])
        if (row2["pred_dependencies"] != row1["dependencies"]):
            print(f"Dependencies acc. - {row2['pred_dependencies']} (GT:{row1['dependencies']})")
    if ((not(pd.isna(row1["training"]))) & (row2["pred_dependencies"] is not None)):
        eval_training.append(row1["training"] == row2["pred_training"])
        if (row1["training"] != row2["pred_training"]):
            print(f"Training acc. -{row2['pred_training']} (GT:{row1['training']})")
    if ((not(pd.isna(row1["evaluation"]))) & (row2["pred_dependencies"] is not None)):
        eval_evaluating.append(row1["evaluation"] == row2["pred_evaluation"])
        if (row1["evaluation"] != row2["pred_evaluation"]):
            print(f"Evaluating acc. - {row2['pred_evaluation']} (GT:{row1['evaluation']})")
    if ((not(pd.isna(row1["weights"]))) & (row2["pred_dependencies"] is not None)):
        eval_weights.append(row1["weights"] == row2["pred_weights"])
        if (row1["weights"] != row2["pred_weights"]):
            print(f"Weights acc. - {row2['pred_weights']} (GT:{row1['weights']})")
    if ((not(pd.isna(row1["readme"]))) & (row2["pred_dependencies"] is not None)):
        eval_readme.append(row1["readme"] == row2["pred_readme"])
        if (row1["readme"] != row2["pred_readme"]):
            print(f"README acc. - {row2['pred_readme']} (GT:{row1['readme']})")
    if ((not(pd.isna(row1["license"]))) & (row2["pred_dependencies"] is not None)):
        eval_licensing.append(("No" if row1["license"] == "No" else "Yes") == row2["pred_license"])
        if (("No" if row1["license"] == "No" else "Yes") != row2["pred_license"]):
            print(f"LICENSE acc. - {row2['pred_license']} (GT:{row1['license']})")


print("\nSummary:")
print(f"Dependencies acc. - {int(100 * np.mean(eval_dependencies))}%")
print(f"Training acc. - {int(100 * np.mean(eval_training))}%")
print(f"Evaluating acc. - {int(100 * np.mean(eval_evaluating))}%")
print(f"Weights acc. - {int(100 * np.mean(eval_weights))}%")
print(f"README acc. - {int(100 * np.mean(eval_readme))}%")
print(f"LICENSE acc. - {int(100 * np.mean(eval_licensing))}%")