From e6c7549bff687383478ee9b2f3fe5d76c564fe49 Mon Sep 17 00:00:00 2001
From: dev_alex <dev_alex@192.168.178.37>
Date: Sun, 30 Jul 2023 22:31:24 +0200
Subject: [PATCH] Finalize Script + Add export to mount fn

---
 ocr_scan.py | 41 +++++++++++++++++++++++++++++++++--------
 1 file changed, 33 insertions(+), 8 deletions(-)

diff --git a/ocr_scan.py b/ocr_scan.py
index 62b51e5..ec45b6a 100644
--- a/ocr_scan.py
+++ b/ocr_scan.py
@@ -5,13 +5,14 @@ from configparser import ConfigParser
 import shutil
 
 
-def scan_document_folder():
+def scan_document_folder() -> None:
+    """Scan Input Odner für OCR Analyse und Verarbeitung der Bilder
+    """
     # Read Config, Setup
     ocr_config = ConfigParser()
     ocr_config.read("ocr_config.ini")
     ocr_settings = ocr_config['ocr']
-    path_to_tesseract = ocr_settings["path_to_tesseract"]
-    # TODO Checken warum hier keine Umlaute erkannt werden -> Sprachmodell wechseln /config
+    path_to_tesseract = ocr_settings["path_to_tesseract"]    
     pytesseract.tesseract_cmd = path_to_tesseract
 
     #Get the file names in the directory
@@ -32,18 +33,42 @@ def scan_document_folder():
             target_path = os.path.join(os.path.dirname(os.path.abspath(__file__)) , ocr_settings["path_to_output"] , file,"")            
             os.makedirs(os.path.dirname(target_path), exist_ok=True)            
             print(f"Creating files for : {file}")
-            
+
             original_file = os.path.join(os.path.dirname(os.path.abspath(__file__)) , ocr_settings["path_to_output"] , file,"")
             shutil.copy2(source_path,original_file)
-            
+            os.remove(source_path)
+
             extracted_file = os.path.join(os.path.dirname(os.path.abspath(__file__)) , ocr_settings["path_to_output"] , file, file + ".txt")
-            #print(text)
             with open(extracted_file, "w") as text_file:
                 text_file.write(text)
-            
+
             pdf_file = os.path.join(os.path.dirname(os.path.abspath(__file__)) , ocr_settings["path_to_output"] , file, file + ".pdf")
             img.save(pdf_file, "PDF" ,resolution=100.0, save_all=True)
 
 
+def export_to_mount(symlinks=False, ignore=None) -> None:    
+    """Export zu einem externen Folder(z.b. NAS Mount)
+
+    Args:
+        symlinks (bool, optional): Config, ob shutil symlinks berücksichtigen soll. Defaults to False.
+        ignore (str, optional): Config, ob shutil gewisse Files ignorieren soll. Defaults to None.
+    """
+    # Read Config, Setup
+    ocr_config = ConfigParser()
+    ocr_config.read("ocr_config.ini")
+    ocr_settings = ocr_config['ocr']
+    
+    src = os.path.join(os.path.dirname(os.path.abspath(__file__)) , ocr_settings["path_to_output"])
+    mount_path = ocr_settings["path_to_mount"]
+    for item in os.listdir(src):
+        s = os.path.join(src, item)
+        d = os.path.join(mount_path, item)
+        if os.path.isdir(s):
+            shutil.copytree(s, d, symlinks, ignore)
+        else:
+            shutil.copy2(s, d)
+
+
 if __name__ == "__main__":
-    scan_document_folder()
\ No newline at end of file
+    scan_document_folder()
+    export_to_mount()
\ No newline at end of file