1"""
2CSV analysis: reads all .csv files, produces summary statistics and a report.
3"""
4import modal
5
6app = modal.App("csv-analysis")
7image = modal.Image.debian_slim().pip_install("pandas")
8
9@app.function(image=image)
10def analyze():
11    import os
12    import pandas as pd
13
14    csv_files = sorted(f for f in os.listdir(".") if f.endswith(".csv"))
15    if not csv_files:
16        print("No CSV files found")
17        return
18
19    all_frames = []
20    for fname in csv_files:
21        df = pd.read_csv(fname)
22        df["_source"] = fname
23        all_frames.append(df)
24
25    combined = pd.concat(all_frames, ignore_index=True)
26
27    # summary stats
28    report = ["# csv analysis report\n"]
29    for fname in csv_files:
30        subset = combined[combined["_source"] == fname]
31        report.append(f"## {fname}")
32        report.append(f"- rows: {len(subset)}")
33        report.append(f"- columns: {', '.join(c for c in subset.columns if c != '_source')}")
34        for col in subset.select_dtypes(include="number").columns:
35            report.append(f"- {col}: sum={subset[col].sum():.2f}, mean={subset[col].mean():.2f}")
36        report.append("")
37
38    # pivot table if quantity and product columns exist
39    if "product" in combined.columns and "quantity" in combined.columns:
40        pivot = combined.pivot_table(index="product", values="quantity", aggfunc="sum")
41        report.append("## totals by product\n")
42        for product, row in pivot.iterrows():
43            report.append(f"- {product}: {int(row['quantity'])} units")
44        report.append("")
45
46    if "product" in combined.columns and "quantity" in combined.columns and "price" in combined.columns:
47        combined["revenue"] = combined["quantity"] * combined["price"]
48        total = combined["revenue"].sum()
49        report.append(f"## total revenue: ${total:,.2f}\n")
50
51    with open("report.md", "w") as f:
52        f.write("\n".join(report))
53
54    # also export cleaned combined csv
55    combined.drop(columns=["_source"]).to_csv("combined.csv", index=False)
56
57    print(f"Done: analyzed {len(csv_files)} files, {len(combined)} total rows")