From 97c833186602b6815a4c0a64b392cc7dae141bb9 Mon Sep 17 00:00:00 2001
From: devaine <devaine@devdoes.work>
Date: Thu, 29 Jan 2026 15:15:27 -0600
Subject: [PATCH] feat(diff): adding file management and diff to extract files

---
 .gitignore                    |  1 +
 webscrape-bank/run.sh         | 32 ++++++++++++++++++++++
 webscrape-bank/src/extract.py | 10 +++----
 webscrape-bank/src/main.py    | 50 +++++++++++++++++++++++++++++++++++
 4 files changed, 86 insertions(+), 7 deletions(-)
 create mode 100755 webscrape-bank/run.sh

diff --git a/.gitignore b/.gitignore
index d71c11c..c3d38e3 100644
--- a/.gitignore
+++ b/.gitignore
@@ -6,3 +6,4 @@ notes
 
 # Others
 node_modules/
+**/__pycache__/
diff --git a/webscrape-bank/run.sh b/webscrape-bank/run.sh
new file mode 100755
index 0000000..31e8b3e
--- /dev/null
+++ b/webscrape-bank/run.sh
@@ -0,0 +1,32 @@
+#!/bin/bash
+
+PIP_PACKAGES=("python-dotenv" "pytest-playwright")
+
+if [ "$1" == "del" ]; then
+	rm -rf .venv
+	echo ".venv/ removed, restart the program again"
+fi
+
+if [ -d .venv ]; then
+	echo ".venv/ found!"
+
+	source .venv/bin/activate
+
+	python src/main.py
+
+else
+	echo ".venv/ not found!"
+	echo "Creating new venv/"
+
+	python -m venv .venv
+	source .venv/bin/activate
+
+	for i in  "${PIP_PACKAGES[@]}"
+	do
+		pip install $i 
+	done
+
+	playwright install
+
+	python src/main.py
+fi
diff --git a/webscrape-bank/src/extract.py b/webscrape-bank/src/extract.py
index be7c735..3ab746f 100644
--- a/webscrape-bank/src/extract.py
+++ b/webscrape-bank/src/extract.py
@@ -2,6 +2,7 @@ from playwright.sync_api import Playwright, sync_playwright, Page
 from dotenv import load_dotenv
 import os
 from time import sleep
+from datetime import datetime
 
 load_dotenv()
 
@@ -76,23 +77,18 @@ def download_file(page: Page):
 
     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")
-
+    # Download
+    download.save_as("./qfx/download-" + str(datetime.now().isoformat()) + ".qfx")
     print("Downloaded!")
 
     switchingForever(page)
 
-    # context.close()
-    # browser.close()
-
 
 if __name__ == "__main__":
     with sync_playwright() as playwright:
diff --git a/webscrape-bank/src/main.py b/webscrape-bank/src/main.py
index e69de29..3b53bd3 100644
--- a/webscrape-bank/src/main.py
+++ b/webscrape-bank/src/main.py
@@ -0,0 +1,50 @@
+from playwright.sync_api import sync_playwright
+from threading import Thread
+import os
+import extract
+#from datetime import datetime
+import difflib
+
+# Playwright in the background
+def playwright():
+    with sync_playwright() as playwright:
+        extract.main(playwright)
+
+
+def revise():
+    # Removes the OLDEST file on the list ()
+    file_count = 0
+    for file in os.scandir("./qfx"):
+        print(file_count)
+        if file_count > 1:
+            #print(os.listdir("./qfx"))
+            print("Removed: " + os.listdir("./qfx")[0])
+            os.remove("./qfx/" + os.listdir("./qfx")[0])
+
+
+        if file.is_file():
+            file_count += 1
+
+    # Get the last two files
+    oldest_file = open("./qfx/" + os.listdir("./qfx")[0])
+    newest_file = open("./qfx/" + os.listdir("./qfx")[-1])
+
+    # Differs the two files
+    diff = difflib.ndiff(oldest_file.readlines(), newest_file.readlines())
+
+    # Grabs only changes
+    # Thanks to: https://stackoverflow.com/a/15864920
+    changes = [l for l in diff if l.startswith("+ ") or l.startswith('- ')]
+    print("RESULT:")
+    for change in changes:
+        print(change[2:])
+
+def main():
+    revise_thread = Thread(target=revise())
+    #pw_thread = Thread(target=playwright())
+    #pw_thread.start()
+    revise_thread.start()
+
+
+if __name__ == "__main__":
+    main()