feat(websocket): frontend websocket

polylan_submitter/submissions/tasks.py

@@ -5,25 +5,28 @@ from PIL import Image
 import pytesseract
 
 
-def _process_ocr_file(file) -> bool:
-    if file.ocr_status != OCRStatus.TO_PROCESS:
+def _process_ocr_file(obj, file, crop: ImageCrop | None = None) -> bool:
+    if obj.ocr_status != OCRStatus.TO_PROCESS:
         return False
 
-    file.ocr_status = OCRStatus.PROCESSING
-    file.save(update_fields=["ocr_status"])
+    obj.ocr_status = OCRStatus.PROCESSING
+    obj.save(update_fields=["ocr_status"])
 
     try:
-        file.ocr_raw_data = pytesseract.image_to_string(Image.open(file.file))
+        image = Image.open(file)
 
-    except Exception:
-        file.ocr_status = OCRStatus.ERROR
-        file.save(update_fields=["ocr_status"])
+        if crop:
+            image = image.crop((crop.x1, crop.y1, crop.x2, crop.y2))
 
-        return False
+        obj.ocr_raw_data = pytesseract.image_to_string(image)
+        obj.ocr_status = OCRStatus.PROCESSED
+
+    except Exception as e:
+        obj.ocr_status = OCRStatus.ERROR
+        raise e
 
     finally:
-        file.ocr_status = OCRStatus.PROCESSED
-        file.save(update_fields=["ocr_status", "ocr_raw_data"])
+        obj.save(update_fields=["ocr_status", "ocr_raw_data"])
 
     return True
 
@@ -42,7 +45,7 @@ def process_submission_file(file_id: str):
     if not file:
         return
 
-    _process_ocr_file(file)
+    _process_ocr_file(file, file.file)
 
     process_crops.enqueue(file_id)
 
@@ -65,4 +68,4 @@ def process_crop(crop_id: str):
     if not crop:
         return
 
-    _process_ocr_file(crop)
+    _process_ocr_file(crop, crop.submission_file.file, crop=crop)