add to gitignore

.gitignore

@@ -1,3 +1,7 @@
 images*
 test.py
 1.jpg
+test-files.tbkpf
+".DS_Store"
+.DS_Store
+compile*

TODO

@@ -1,2 +1 @@
-improve accuracy
-
+improve accuracy

fixer.py

@@ -0,0 +1,30 @@
+import cv2
+import pytesseract
+import os
+import numpy
+directory = 'images'
+
+# iterate over files in
+# that directory
+pattern = 'review' # Replace with your target substring
+matching_files = [f for f in os.listdir(directory) if pattern in f]
+
+for files in os.listdir(directory):
+    if files.__contains__(pattern):
+        f = os.path.join(directory, files)
+        print(f)
+        image = cv2.imread(f, 0)
+        thresh = 255 - cv2.threshold(image, 0, 255, cv2.THRESH_BINARY_INV + cv2.THRESH_OTSU)[1]
+        #bottom 2 locations
+        ROI1 = thresh[1383:1447,77:154] 
+        ROI2 = thresh[1383:1447,950:1027]
+        ROI = cv2.hconcat([ROI1, ROI2])
+        cv2.imshow('Review', ROI)
+        cv2.waitKey(1)
+        manpage = input("please input the page number")
+        if os.path.isfile(f'images/{manpage}.jpg') == True:
+            rng = numpy.random.default_rng().random()
+            os.rename (f'images/{manpage}.jpg', f'images/{manpage}.review-{rng}.jpg')
+            os.rename(f'{f}', f'images/{manpage}.jpg')
+        elif os.path.isfile(f'images/{manpage}.jpg') == False:
+            os.rename(f'{f}', f'images/{manpage}.jpg')

main.py

@@ -1,9 +1,9 @@
 import cv2
 import pytesseract
 import os
+import numpy
 directory = 'images'
- 
-k = 1
+
 # iterate over files in
 # that directory
 for filename in os.listdir(directory):
@@ -20,9 +20,21 @@ for filename in os.listdir(directory):
         print(data)
         if os.path.isfile(f'images/{data}.jpg') == False:
             os.rename(f'{f}', f'images/{data}.jpg')
-        else:
-            cv2.imshow(ROI)
-            cv2.waitKey(50)
-            os.rename(f'{f}', f'images/{data}.{k}.jpg')
-            k = k + 1
+        elif os.path.isfile(f'images/{data}.jpg') == True:
+            cv2.imshow(data, ROI)
+            cv2.waitKey(1)
+            manpage = input("please input the number on the page (if the title is correct, enter nothing)")
+            #check if the new name already exists
+            if manpage == "\n":
+                rng = numpy.random.default_rng().random()
+                os.rename (f'images/{data}.jpg', f'images/{data}.review-{rng}.jpg')
+                os.rename(f'{f}', f'images/{data}.jpg')
+            elif manpage != "\n":
+                if os.path.isfile(f'images/{manpage}.jpg') == True:
+                    rng = numpy.random.default_rng().random()
+                    os.rename (f'images/{manpage}.jpg', f'images/{manpage}.review-{rng}.jpg')
+                    os.rename(f'{f}', f'images/{manpage}.jpg')
+                elif os.path.isfile(f'images/{manpage}.jpg') == False:
+                    os.rename(f'{f}', f'images/{manpage}.jpg')
+            cv2.destroyAllWindows()
         cv2.destroyAllWindows()