Fix fallback

regression.py

@@ -15,8 +15,7 @@ from wsgi import create_fig, create_plot_df, get_tables, plot
 
 def create_dfs(url: str = "https://beschaeftigtenbefragung.verdi.de/"):
     try:
-        curr_datetime = datetime.datetime.now()
+        df, df_state, curr_datetime = get_tables(url)
-        df, df_state = get_tables(url)
 
         df = df.sort_values(
             ["Digitale Befragung", "Bundesland", "Bezirk"],

wsgi.py

@@ -46,13 +46,13 @@ app.config.from_mapping(config)
 cache = Cache(app)
 
 
-def get_tables(url: str) -> tuple[pd.DataFrame, pd.DataFrame]:
+def get_tables(url: str) -> tuple[pd.DataFrame, pd.DataFrame, datetime.datetime]:
     bez_data = get_bez_data(["bez_data_0", "bez_data_2"], url)
 
     df = construct_dataframe(bez_data=bez_data[0], special_tag="stud")
     df_state = construct_dataframe(bez_data=bez_data[1])
 
-    return df, df_state
+    return df, df_state, datetime.datetime.now()
 
 
 def create_plot_df(
@@ -113,7 +113,10 @@ def plot(
     fix_lims: bool = True,
     max_shading_date=None,
 ) -> str:
-    fig = plt.figure(dpi=300)
+    fig = plt.figure(dpi=300, figsize=(8.5, 5))
+
+    target_time = pd.Timestamp("2023-10-01")
+    plt.axvline(x=target_time, color="tab:green", linestyle=":")
 
     if fix_lims:
         for total_target in total_targets:
@@ -173,7 +176,7 @@ def plot(
         idx = np.argmin(nearest_target)
 
         ceil_val = max(max_val, total_targets[idx])
-        plt.ylim(0, ceil_val * 1.025)
+        plt.ylim(0, ceil_val * 1.04)
     plt.legend()
 
     # use timezone offset to center tick labels
@@ -202,14 +205,14 @@ def plot(
 
     # fill weekends
     if max_shading_date is None:
-        max_shading_date = df.index.max() + datetime.timedelta(days=3)
+        max_shading_date = df.index.max() + datetime.timedelta(days=4)
     days = pd.date_range(start="2023-08-14", end=max_shading_date)
     for idx, day in enumerate(days[:-1]):
         if day.weekday() >= 5:
             plt.gca().axvspan(days[idx], days[idx + 1], alpha=0.2, color="gray")
 
     # reset xlim
-    plt.xlim(xlim)
+    plt.xlim((xlim[0], pd.Timestamp("2023-10-02")))
 
     plt.tight_layout()
 
@@ -224,7 +227,7 @@ def create_fig(
 ):
     curr_datetime = datetime.datetime.now()
     try:
-        df, df_state = get_tables(url)
+        df, df_state, curr_datetime = get_tables(url)
 
         df = df.sort_values(
             ["Digitale Befragung", "Bundesland", "Bezirk"],
@@ -251,7 +254,7 @@ def create_fig(
                 {"Digitale Befragung": "Int32"}
             )
 
-        plot_df = create_plot_df(curr_datetime)
+        plot_df = create_plot_df(curr_datetime, df_state)
         annotate_current = False
         timestamp = Markup(f'<font color="red">{key} 10:00:00</font>')
 
@@ -291,11 +294,19 @@ def _print_as_html(
     dropna: bool = True,
 ) -> list[str]:
     df = df.astype({"Digitale Befragung": "Int32"})
-    missing_df = df[["Digitale Befragung"]].isna().join(df[["Landesbezirk"]]).groupby("Landesbezirk").sum()
+    missing_df = (
+        df[["Digitale Befragung"]]
+        .isna()
+        .join(df[["Landesbezirk"]])
+        .groupby("Landesbezirk")
+        .sum()
+    )
     total = df_state["Digitale Befragung"].sum() if df_state is not None else None
 
     if df_state is not None:
-        for idx, row in missing_df.loc[missing_df["Digitale Befragung"] == 1].iterrows():
+        for idx, row in missing_df.loc[
+            missing_df["Digitale Befragung"] == 1
+        ].iterrows():
             df_tmp = df.loc[df["Landesbezirk"] == idx]
             df_state_tmp = df_state.loc[df_state["Landesbezirk"] == idx]
             missing_idx = df_tmp.loc[df_tmp.isna().any(axis=1)].iloc[0].name
@@ -334,7 +345,7 @@ def _print_as_html(
         )
         if total and (diff := total - df["Digitale Befragung"].sum()):
             tfoot.append("    <tr>")
-            num_missing = missing_df['Digitale Befragung'].sum()
+            num_missing = missing_df["Digitale Befragung"].sum()
             tfoot.append(
                 f"      <td>Weitere Bezirke ({num_missing})</td>"
                 if num_missing
@@ -385,9 +396,7 @@ def state_dashboard(state: str):
 
     output_str = []
     output_str = _print_as_html(df_state, output_str, dropna=False)
-    output_str = _print_as_html(
+    output_str = _print_as_html(df, output_str, df_state=df_state, dropna=False)
-        df, output_str, df_state=df_state, dropna=False
-    )
 
     return render_template(
         "base.html",
@@ -425,5 +434,13 @@ def dashboard():
     )
 
 
+@app.route("/total")
+@cache.cached(timeout=60)
+def total_result(url: str = "https://beschaeftigtenbefragung.verdi.de/"):
+    df, df_state, curr_datetime = get_tables(url)
+    total = df_state["Digitale Befragung"].sum().item()
+    return f"{total}"
+
+
 if __name__ == "__main__":
     app.run()