Create app.py

app.py

@@ -0,0 +1,99 @@
+"""
+03 · Data Detective   (LOCAL — no token)   ·   JSON + bind, mixed output types
+=============================================================================
+
+Upload a CSV; a single **file** reference fans out to four analysts that render
+different port types on the canvas — a preview table, summary statistics, a
+missing-value report, and a correlation heatmap.
+
+Graph:
+                       ┌─▶ preview        ─▶ [Preview]        (dataframe)
+    [CSV file] ─▶──────┼─▶ summary_stats  ─▶ [Statistics]     (dataframe)
+                       ├─▶ missing_report ─▶ [Missing values] (json)
+                       └─▶ correlation    ─▶ [Heatmap]        (image)
+
+Because functions exchange values as JSON, each analyst reads the CSV path
+itself and returns a JSON-safe payload: a `{headers, data}` table, a dict, or a
+base64 image. Port types (`dataframe` / `json` / `image`) tell the canvas how to
+render each result.
+
+Try it with the bundled ../../assets/sample_data.csv
+
+Run it:
+  python apps/03_data_detective/app.py
+"""
+
+import base64
+import io
+import os
+
+import gradio as gr
+import matplotlib
+matplotlib.use("Agg")
+import matplotlib.pyplot as plt
+import pandas as pd
+
+
+def _read(file) -> pd.DataFrame:
+    if isinstance(file, dict):                       # {path|url} from a file component
+        file = file.get("path") or file.get("url")
+    return pd.read_csv(file)
+
+
+def _table(df: pd.DataFrame) -> dict:
+    """A JSON-safe {headers, data} payload for a `dataframe` port."""
+    df = df.astype(object).where(pd.notna(df), None)
+    data = [[(x.item() if hasattr(x, "item") else x) for x in row] for row in df.values.tolist()]
+    return {"headers": [str(c) for c in df.columns], "data": data}
+
+
+def preview(file: str) -> dict:
+    return _table(_read(file).head(25))
+
+
+def summary_stats(file: str) -> dict:
+    desc = _read(file).describe(include="all").transpose().round(3)
+    desc.insert(0, "column", desc.index)
+    return _table(desc.reset_index(drop=True))
+
+
+def missing_report(file: str) -> dict:
+    df = _read(file)
+    na = df.isna().sum()
+    return {
+        "rows": int(len(df)),
+        "columns": int(df.shape[1]),
+        "total_missing": int(na.sum()),
+        "missing_by_column": {c: int(v) for c, v in na.items() if v > 0} or "none 🎉",
+        "dtypes": {c: str(t) for c, t in df.dtypes.items()},
+    }
+
+
+def correlation(file: str) -> str:
+    """A correlation heatmap as a base64 `data:image/png` string."""
+    corr = _read(file).corr(numeric_only=True)
+    fig, ax = plt.subplots(figsize=(1.1 * len(corr) + 2, 1.1 * len(corr) + 1.5))
+    im = ax.imshow(corr.values, cmap="RdBu", vmin=-1, vmax=1)
+    ax.set_xticks(range(len(corr)), corr.columns, rotation=45, ha="right")
+    ax.set_yticks(range(len(corr)), corr.columns)
+    for i in range(len(corr)):
+        for j in range(len(corr)):
+            v = corr.values[i, j]
+            ax.text(j, i, f"{v:.2f}", ha="center", va="center",
+                    color="white" if abs(v) > 0.5 else "black", fontsize=8)
+    ax.set_title("Correlation matrix")
+    fig.colorbar(im, ax=ax, shrink=0.8)
+    buf = io.BytesIO()
+    fig.savefig(buf, format="png", bbox_inches="tight", dpi=110)
+    plt.close(fig)
+    return "data:image/png;base64," + base64.b64encode(buf.getvalue()).decode()
+
+
+BIND = {"preview": preview, "summary_stats": summary_stats,
+        "missing_report": missing_report, "correlation": correlation}
+
+WORKFLOW = os.path.join(os.path.dirname(os.path.abspath(__file__)), "workflow.json")
+demo = gr.Workflow(WORKFLOW, bind=BIND)
+
+if __name__ == "__main__":
+    demo.launch()