Update app.py

app.py

@@ -8,6 +8,7 @@ from plotly.subplots import make_subplots
 from google.oauth2.service_account import Credentials
 import gspread
 
+# 爬取 Booking.com 台南飯店數據
 def scrape_booking_hotel():
     url = "https://www.booking.com/searchresults.zh-tw.html"
     headers = {
@@ -36,18 +37,21 @@ def scrape_booking_hotel():
         for hotel in hotel_cards:
             try:
                 name = hotel.find('div', {'data-testid': 'title', 'class': 'f6431b446c'}).text.strip() or "無資料"
+                
                 price = hotel.find('span', {'data-testid': 'price-and-discounted-price', 'class': 'f6431b446c'}).text.strip() or "無資料"
                 price = price.replace('TWD', '').replace(' ', '').replace(',', '').strip()
+                price = float(price) if price.replace('.', '', 1).isdigit() else None  # 價格轉為浮點數
 
                 rating_container = hotel.find('div', {'class': 'a3b8729ab1'})
                 rating = rating_container.find('div', {'class': 'ac4a7896c7'}).text.strip() if rating_container else "無評分"
+                rating = float(rating) if rating.replace('.', '', 1).isdigit() else None  # 評分轉為浮點數
 
                 description_elem = hotel.find('div', {'data-testid': 'recommended-units'})
                 if description_elem:
                     room_type = description_elem.find('h4', {'class': 'abf093bdfe'}).text.strip() if description_elem.find('h4', {'class': 'abf093bdfe'}) else ""
                     bed_info = description_elem.find('div', {'class': 'abf093bdfe'}).text.strip() if description_elem.find('div', {'class': 'abf093bdfe'}) else ""
-                    cancellation = "可免費取消" if description_elem.find('strong', text='可免費取消') else ""
-                    payment = "無需訂金" if description_elem.find('strong', text='無需訂金') else ""
+                    cancellation = "可免費取消" if description_elem.find('strong', string='可免費取消') else ""
+                    payment = "無需訂金" if description_elem.find('strong', string='無需訂金') else ""
                     description = f"{room_type} | {bed_info} | {cancellation} | {payment}".strip(' |')
                 else:
                     description = "無說明"
@@ -71,6 +75,7 @@ def scrape_booking_hotel():
         print(f"請求發生錯誤: {e}")
         return pd.DataFrame()
 
+# 散佈圖
 def create_price_rating_scatter(df):
     fig = px.scatter(
         df,
@@ -86,6 +91,7 @@ def create_price_rating_scatter(df):
     fig.update_layout(height=600, showlegend=True, title_x=0.5, title_font_size=20)
     return fig
 
+# 價格分佈圖
 def create_price_distribution(df):
     fig = go.Figure()
     fig.add_trace(go.Histogram(x=df['價格'], name='價格分布', nbinsx=10, marker_color='rgb(55, 83, 109)'))
@@ -93,21 +99,7 @@ def create_price_distribution(df):
     fig.update_layout(title_text='台南飯店價格分布', title_x=0.5, title_font_size=20, xaxis_title='價格 (TWD)', yaxis_title='數量', height=500, bargap=0.2, showlegend=True)
     return fig
 
-def create_rating_box_by_price_range(df):
-    fig = px.box(df, x='價格區間', y='評分', title='不同價格區間的評分分布', labels={'價格區間': '價格類型', '評分': '評分 (0-10)'}, color='價格區間')
-    fig.update_layout(title_x=0.5, title_font_size=20, height=500, showlegend=False)
-    return fig
-
-def create_hotel_comparison(df):
-    fig = make_subplots(specs=[[{"secondary_y": True}]])
-    df_sorted = df.sort_values('評分', ascending=True)
-    fig.add_trace(go.Bar(x=df_sorted['飯店名稱'], y=df_sorted['評分'], name="評分", marker_color='rgb(55, 83, 109)'))
-    fig.add_trace(go.Scatter(x=df_sorted['飯店名稱'], y=df_sorted['價格'], name="價格", marker_color='rgb(26, 118, 255)'), secondary_y=True)
-    fig.update_layout(title_text='台南飯店評分與價格比較', title_x=0.5, title_font_size=20, height=700, showlegend=True, xaxis_tickangle=45)
-    fig.update_yaxes(title_text="評分", secondary_y=False)
-    fig.update_yaxes(title_text="價格 (TWD)", secondary_y=True)
-    return fig
-
+# 更新 Google Sheet
 def update_google_sheet(df):
     scope = ['https://www.googleapis.com/auth/spreadsheets']
     creds = Credentials.from_service_account_file("realtime-441511-f5708eabdf26.json", scopes=scope)
@@ -117,33 +109,30 @@ def update_google_sheet(df):
     worksheet.update([df.columns.values.tolist()] + df.astype(str).values.tolist())
     st.success("Data updated to Google Sheet successfully!")
 
+# 主函數
 def main():
     st.set_page_config(page_title="Booking.com Hotel Analysis")
     st.title("Booking.com Hotel Analysis")
 
     df = scrape_booking_hotel()
 
+    # 顯示資料表格
     st.subheader("Hotel Data")
     st.dataframe(df)
 
+    # 顯示散佈圖
     st.subheader("Price vs Rating Scatter Plot")
     scatter_fig = create_price_rating_scatter(df)
     st.plotly_chart(scatter_fig)
 
+    # 顯示價格分布圖
     st.subheader("Price Distribution")
     dist_fig = create_price_distribution(df)
     st.plotly_chart(dist_fig)
 
-    st.subheader("Rating by Price Range")
-    box_fig = create_rating_box_by_price_range(df)
-    st.plotly_chart(box_fig)
-
-    st.subheader("Hotel Comparison")
-    comparison_fig = create_hotel_comparison(df)
-    st.plotly_chart(comparison_fig)
-
+    # Google Sheet 更新
     if st.button("Update Google Sheet"):
         update_google_sheet(df)
 
 if __name__ == "__main__":
-    main()
+    main()