feat(webscrape-bank): added code to webscrape my bank, api is next

webscrape-bank/src/extract.py

@@ -0,0 +1,99 @@
+from playwright.sync_api import Playwright, sync_playwright, Page
+from dotenv import load_dotenv
+import os
+from time import sleep
+
+load_dotenv()
+
+BANK_USER = os.getenv("BANK_USER")
+BANK_PASS = os.getenv("BANK_PASS")
+LOGIN_LINK = os.getenv("LOGIN_LINK")
+ACCOUNT_LINK = os.getenv("ACCOUNT_LINK")
+
+def main(playwright: Playwright) -> None:
+    browser = playwright.chromium.launch(headless=False)
+    context = browser.new_context()
+
+    page = context.new_page()
+    page.goto(LOGIN_LINK)
+    page.locator("#username").click()
+    page.locator("#username").fill(BANK_USER)
+    page.locator("#password").click()
+    page.locator("#password").fill(BANK_PASS)
+    page.get_by_role("button", name="Login", exact=True).click()
+
+    # Wait a little bit
+    page.wait_for_load_state()
+    page.get_by_role("button", name="Text").click()
+    page.get_by_test_id("text-field").click()
+
+    # Bank prompts 2FA (hopefully only once)
+    two_fac = input("2FA Code: ")
+
+    page.get_by_test_id("text-field").fill(two_fac)
+    page.get_by_test_id("private-button").click()
+
+    # Wait for everything to load...
+    page.wait_for_timeout(5000)
+
+    # We should be at the home page here
+    download_file(page)
+
+
+def switchingForever(page: Page):
+    print("Switching to not get timed out!")
+    page.goto(ACCOUNT_LINK)
+
+    count = 0
+
+    while (count < 12):
+        page.wait_for_timeout(5000)
+        page.get_by_role("link", name="Cards").hover()
+        page.get_by_role("link", name="Activate Card").click(timeout=0)
+        sleep(300)
+        page.get_by_role("link", name="My Accounts").click(timeout=0)
+        count = count + 1
+        print("Count updated: " + str(count))
+
+    download_file(page)
+
+
+def download_file(page: Page):
+    print("Going to download a new QFX File")
+    page.goto(ACCOUNT_LINK)
+
+    page.locator('iframe[title="Next Gen Home Page"]').content_frame.get_by_test_id(
+        "com.ncr.dbk.olb.widgets.my-accounts"
+    ).get_by_test_id("feature-link").click()
+
+    page.wait_for_timeout(5000)
+
+    page.locator("#ncr-la-chat-win-notif-close path").click()
+
+    page.locator(
+        'iframe[title="NextGen account history page"]'
+    ).content_frame.get_by_role("button", name="Download Transactions Button").click()
+
+    with page.expect_download(timeout=0) as download_info:
+        page.mouse.wheel(0, 70)
+        
+        page.locator(
+            'iframe[title="NextGen account history page"]'
+        ).content_frame.get_by_role("menuitem", name="Export QFX").dblclick()
+        
+
+    download = download_info.value
+
+    download.save_as("./test.qfx")
+
+    print("Downloaded!")
+
+    switchingForever(page)
+
+    # context.close()
+    # browser.close()
+
+
+if __name__ == "__main__":
+    with sync_playwright() as playwright:
+        main(playwright)