src/main.py

#main.py
import pandas as pd
from analyze import analyze_csv
from plan import generate_cleaning_plan
from execute import execute_plan
from insight import generate_insights
from visual_insight import generate_visual_plan
from run_viz_code import run_visualizations
import time
from report import ReportBuilder
import re

start_time = time.time()




input_path = "adult.csv"
output_path = "output.csv"

# Step 1: Analyze
analysis = analyze_csv(input_path)
print("Analysis Complete")

# Step 2: Plan
cleaning_plan, explanation = generate_cleaning_plan(analysis)
print("\nCleaning Plan:")
for step in cleaning_plan:
    if "colum" in step and "column" not in step:
        step["column"] = step.pop()
    print(step)

print("\nExplanation:")
print(explanation)

# Step 3: Execute
original_df = pd.read_csv(input_path)
cleaned_df = execute_plan(original_df, cleaning_plan)
cleaned_df.to_csv(output_path, index=False)
print(f"\nOutput saved to {output_path}")

# Step 4: Re-analyze cleaned file
cleaned_analysis = analyze_csv(output_path)
print("\nCleaned Data EDA Summary:")
for col, info in cleaned_analysis["columns"].items():
    print(f"{col} → type: {info['dtype']}, missing: {info['missing_pct']:.1f}%, unique: {info['unique_vals']}")

# Step 5: LLM derives human-style insights
print("\nData Insights:")
insights = generate_insights(cleaned_analysis["columns"])
print(insights)

# Step 6: Visualization Plan
print("\nGenerating visual plan from LLM...")
visual_plan = generate_visual_plan(cleaned_analysis["columns"])
run_visualizations(cleaned_df, visual_plan)
print("Visuals saved to charts/")

# 🧾 Build report
report = ReportBuilder("final_report.pdf")

report.add_title("Smart Data Cleaning & EDA Report")
report.add_section("Cleaning Summary", explanation)

report.add_section("EDA Summary", insights)

for viz in visual_plan:
    image_path = re.findall(r"plt\.savefig\(['\"](.*?)['\"]\)", viz["code"])
    if image_path:
        report.add_plot(image_path[0], viz["description"])

report.save()
print("PDF report saved to final_report.pdf")

end_time = time.time()
duration = end_time - start_time
print(f"\nTotal Time: {duration:.2f} seconds")