first files

united-utilities-scarper.py

@@ -0,0 +1,72 @@
+import asyncio
+from playwright.async_api import async_playwright
+
+UU_EMAIL = "email@gmail.com"
+UU_PASSWORD = "password"
+
+async def download_uu_bill_live():
+    async with async_playwright() as p:
+        # Opens visible browser with 1.5 second step delay
+        browser = await p.chromium.launch(headless=False, slow_mo=1500)
+        page = await browser.new_page()
+
+        print("Opening United Utilities homepage...")
+        await page.goto("https://unitedutilities.com")
+
+        # Handle cookie banner (confirmed working)
+        try:
+            await page.click("button:has-text('Accept All')", timeout=5000)
+            print("Cookies accepted.")
+        except Exception:
+            pass 
+
+        # NEW STEP: Click the header login button based on your HTML code
+        print("Clicking 'LOG IN / REGISTER' button in the header...")
+        await page.click("a.c-navigation__actions-btn:has-text('LOG IN / REGISTER')")
+        
+        # Wait for the login portal page to fully transition and load
+        print("Waiting for login portal page to load...")
+        await page.wait_for_url("https://myaccount.unitedutilities.com/**", timeout=15000)
+
+        print("Typing credentials live...")
+        # Targets the unique ID 'login-email' from your HTML snippet
+        await page.fill("#login-email", UU_EMAIL)
+        
+        # Targets the likely matching password ID structure
+        await page.fill("#PasswordMasking", UU_PASSWORD)
+
+        
+        print("Clicking submit login form...")
+        await page.click("#login-btn")
+
+        
+        print("Waiting for dashboard to load...")
+        await page.wait_for_url("https://myaccount.unitedutilities.com/Secure/Dashboard", timeout=30000)
+
+
+        print("Navigating to billing section...")
+        await page.click("text=View bills and payments")
+
+        print("Clicking 'View statement' button...")
+        await page.click("a[href*='type=viewStatement']")
+
+        # NEW STEP: Wait for the dynamic bills table to finish loading
+        print("Waiting for bills table to load...")
+        await page.wait_for_selector("table#viewBill tbody tr")
+
+        # Trigger and catch the file download pipeline
+        async with page.expect_download() as download_info:
+            print("Clicking the newest bill download link in the table...")
+            # Targets the very first row's download link inside the table body
+            await page.locator("table#viewBill tbody tr").first.locator("a[href*='DownloadBill.aspx']").click()
+        
+        download = await download_info.value
+        save_path = "./united_utilities_bill.pdf"
+        await download.save_as(save_path)
+        print(f"Success! Newest bill saved to {save_path}")
+
+        print("Finishing up, closing browser in 5 seconds...")
+        await asyncio.sleep(5)
+        await browser.close()
+
+asyncio.run(download_uu_bill_live())