rework unneeded checks and iterations

src/KSG_anomaly_detection/preparation.py

@@ -1,5 +1,6 @@
 import os
 import re
+import sys
 import traceback
 from pathlib import Path
 from pprint import pprint
@@ -31,7 +32,12 @@ class Preparation:
         #     / Path(self.folder_path).parents[1].name
         #     / Path(self.folder_path).parent.name
         # )  # Pfad zu Fileserver/KI/...
-        self.original_data_path = Path(config.STORING_PATH).joinpath(*self.path.parts[-4:-1])
+
+        # the original path's last component is the product type name instead of the package
+        self.original_data_base_path = Path(config.STORING_PATH).joinpath(
+            *self.path.parts[-4:-1]
+        )
+        self.original_data_path = self.original_data_base_path / self.path.name
 
         print(f"[{self.visper}] {Path(self.folder_path).name} Vorbereitung gestartet...")
 
@@ -105,16 +111,17 @@ class Preparation:
             #     copytree(src, dst, dirs_exist_ok=True)
 
             src = self.path
-            dst = self.original_data_path / self.path.name
+            # dst = self.original_data_base_path / self.path.name
-            copytree(src, dst, dirs_exist_ok=True)
+            copytree(src, self.original_data_path, dirs_exist_ok=True)
 
             # ?? Soll hier der Pfad mit den kopierten Originaldaten zurückgegeben werden?
             # Das ist aktuell nicht der Fall, da die letzten Pfadkomponenten als absolute Pfade das
             # Ergebnis komplett überschreiben
+            # Zudem wird der Basispfad (also Produktebene, nicht Päckchen zurückgegeben)!
             # return Path(config.STORING_PATH) / self.visper / Path(
             #     self.folder_path
             # ).parent.parent / Path(self.folder_path).parent, None
-            return self.original_data_path, None
+            return self.original_data_base_path, None
 
         except FileExistsError:
             return (
@@ -214,10 +221,19 @@ class Preparation:
         self, current_folder_to_inspect, checkimg_folders, folder_type
     ):
         for folder_to_inspect in checkimg_folders:
+            print(
+                f"\n-------------------------------\n{folder_to_inspect=}\n,{current_folder_to_inspect=}\n"
+            )
+            check_1 = Path(folder_to_inspect).parts[-2]
+            check_2 = Path(current_folder_to_inspect).parts[-2]
+            print(f"{check_1=},\n{check_2=}\n------------------------------------")
+
             if folder_to_inspect.is_dir() and (
                 Path(folder_to_inspect).parts[-2] == Path(current_folder_to_inspect).parts[-2]
             ):
-                relative_path = current_folder_to_inspect.relative_to(self.original_data_path)
+                relative_path = current_folder_to_inspect.relative_to(
+                    self.original_data_base_path
+                )
                 relative_path = Path(*relative_path.parts[1:])
                 return relative_path
 
@@ -243,7 +259,7 @@ class Preparation:
 
         # find all checkimg folders within the folder
         checkimg_folders = [
-            p for p in self.original_data_path.rglob("checkimg") if p.is_dir()
+            p for p in self.original_data_base_path.rglob("checkimg") if p.is_dir()
         ]
 
         # iterate through all 'checkimg' folders recursively
@@ -293,8 +309,10 @@ class Preparation:
 
         # create folder name in our temp folder "Backup" and store it
         # therefore, first extract the name of the current folder from the whole path
-        folder_name = Path(self.folder_path).name
+        # folder_name = self.path.name
-        new_folder_path = Path(config.CURRENT_PATH_RGB) / folder_name
+        # print("folder name: ", folder_name)
+        new_folder_path = Path(config.CURRENT_PATH_RGB) / self.path.name
+        # print("new_folder_path: ", new_folder_path)
 
         try:
             new_folder_path.mkdir(parents=True, exist_ok=False)
@@ -307,21 +325,35 @@ class Preparation:
             return None, f"Fehlermeldung: {e}"
 
         # find all checkimg folders within the folder
-        checkimg_folders = [
+        # ?? Hier gewinnen wir wieder alle Verzeichnisse oberhalb der Paketebene, d.h.
+        # ?? unabhängig vom Päckchen
+        # Annahme: Wir wollen tatsächlich nur auf Päckchenebene arbeiten
+        checkimg_folders = tuple(
             p for p in self.original_data_path.rglob("checkimg") if p.is_dir()
-        ]
+        )
+        # print(f">>> {checkimg_folders=}")
+
+        # sys.exit(0)
 
         # iterate through all 'checkimg' folders recursively
-        for current_folder_to_inspect in checkimg_folders:
+        for checkimg_folder in checkimg_folders:
+            # ?? Das scheint unnötig, da wir nun nur für das Päckchen relevante checkimg
+            # ?? Ordner durchsuchen bei jeder Iteration. Damit können wir direkt auf den
+            # ?? Pfaden arbeiten.
             # identify the path starting from self.folder_path until the checkimg folder
-            relative_path = self.extract_folder_path_within_one_AOI_folder(
+            # relative_path = self.extract_folder_path_within_one_AOI_folder(
-                current_folder_to_inspect, checkimg_folders, "checkimg"
+            #     checkimg_folder, checkimg_folders, "checkimg"
-            )
+            # )
+            # save_path_rgb = new_folder_path / relative_path
+            # save_path_rgb.mkdir(parents=True, exist_ok=True)
 
-            save_path_rgb = new_folder_path / relative_path
+            relative_path = checkimg_folder.parts[-2:]
+            save_path_rgb = new_folder_path.joinpath(*relative_path)
             save_path_rgb.mkdir(parents=True, exist_ok=True)
+            # print(f">>>> {relative_path=}")
+            # print(f">>>> {save_path_rgb=}")
 
-            for file_path in current_folder_to_inspect.glob("R_NG*_*.jpg"):
+            for file_path in checkimg_folder.glob("R_NG*_*.jpg"):
                 # find match according to pattern defined at the very beginning
                 match = pattern.match(file_path.name)
                 if not match:
@@ -331,14 +363,14 @@ class Preparation:
 
                 # find all three images belonging together
                 r_path = file_path
-                g_path = current_folder_to_inspect / f"G_NG{num1}_{num2}.jpg"
+                g_path = checkimg_folder / f"G_NG{num1}_{num2}.jpg"
-                b_path = current_folder_to_inspect / f"B_NG{num1}_{num2}.jpg"
+                b_path = checkimg_folder / f"B_NG{num1}_{num2}.jpg"
 
                 # open all three images and combine them to RGB
                 r = pyvips.Image.new_from_file(r_path, access="sequential")
                 g = pyvips.Image.new_from_file(g_path, access="sequential")
                 b = pyvips.Image.new_from_file(b_path, access="sequential")
-                rgb_image = r.bandjoin([g, b])
+                rgb_image = r.bandjoin([g, b])  # type: ignore
                 rgb_image = rgb_image.copy(interpretation="srgb")
                 filename = f"RGB_NG{num1}_{num2}.png"
                 rgb_image.write_to_file(save_path_rgb / filename)