Improve html output

download_digital.py

@@ -76,6 +76,7 @@ def main(
     tag: str = "bez_data_2",
     dry_run: bool = False,
     grouped: bool = False,
+    html: bool = False,
 ) -> pd.DataFrame:
     r = requests.get(url)
     soup = BeautifulSoup(r.text, "html.parser")
@@ -106,11 +107,51 @@ def main(
     data["Digitale Befragung"] = pd.Series(tot_col_data, index=tot_col_index)
     df = pd.DataFrame(data=data)
     if grouped:
-        df = df.groupby("Bundesland")[["Digitale Befragung"]].sum()
+        df = df.groupby("Bundesland", as_index=False)[["Digitale Befragung"]].sum()
-    if dry_run:
+
-        df.loc["Total"] = df.sum(numeric_only=True)
+    if dry_run or html:
-        with pd.option_context("display.max_rows", None):
+        if html:
-            print(df.astype({"Digitale Befragung": "Int32"}))
+            print("<link rel='stylesheet' href='style.css'>")
+            print("<script src='sorttable.js'></script>")
+
+        def _print_as_html(df: pd.DataFrame):
+            df = df.astype({"Digitale Befragung": "Int32"})
+            with pd.option_context("display.max_rows", None):
+                if html:
+                    table = df.to_html(
+                        index_names=False,
+                        justify="left",
+                        index=False,
+                        classes="sortable dataframe",
+                    )
+
+                    tfoot = [
+                        "  <tfoot>",
+                        "      <td>Gesamt</td>",
+                    ]
+                    for i in range(len(df.columns) - 2):
+                        tfoot.append("      <td/>")
+                    tfoot.extend(
+                        [
+                            f"      <td>{df['Digitale Befragung'].sum()}</td>",
+                            "    </tr>",
+                            "  </tfoot>",
+                        ]
+                    )
+                    tfoot = "\n".join(tfoot)
+                    idx = table.index("</table>")
+                    print(table[: idx - 1])
+                    print(tfoot)
+                    print(table[idx:])
+                else:
+                    print(df)
+
+        _print_as_html(df)
+        if html:
+            _print_as_html(
+                df.groupby("Bundesland", as_index=False)[["Digitale Befragung"]].sum()
+            )
+
     else:
         filename = f"data/{datetime.today().strftime('%Y-%m-%d')}_data.ods"
         if Path(filename).exists():