Update system_stats.py

system_stats.py

@@ -31,4 +31,26 @@ def get_stats(file_path):
 
         # Save the DataFrame to a CSV file
         data.to_csv(file_path, index=False)
+
+
+
+file_list = ['doc/page_1.txt', 'doc/page_1.txt', 'doc/page_2.table', 'doc/page3_img.table', 'doc/page3_img2.table', 'doc/page3_img3.table']
+
+page_dict = {}
+
+for file_path in file_list:
+    # Extracting page number from the file name
+    page_number = file_path.split('/')[-1].split('_')[1].split('.')[0]
+    
+    # Checking if the file is a text file
+    if file_path.endswith('.txt'):
+        # Opening the text file and reading its content
+        with open(file_path, 'r') as file:
+            content = file.read()
+        
+        # Adding the content to the dictionary with page number as key
+        page_dict[page_number] = content
+
+print(page_dict)
+