Merge pull request #60 from gmbrax/fix/Importing-duplicate-images

Fix the bug that collides the hash if identical images

src/pilgrim/models/photo.py

@@ -4,6 +4,7 @@ from pathlib import Path
 
 from sqlalchemy import Column, Integer, String, ForeignKey, DateTime
 from sqlalchemy.orm import relationship
+from sqlalchemy.sql.schema import Index
 
 from pilgrim.models.photo_in_entry import photo_entry_association
 from ..database import Base
@@ -24,6 +25,9 @@ class Photo(Base):
     )
 
     fk_travel_diary_id = Column(Integer, ForeignKey("travel_diaries.id"),nullable=False)
+    __table_args__ = (
+        Index('idx_photo_hash_diary', 'hash', 'fk_travel_diary_id'),
+    )
 
     def __init__(self, filepath, name, photo_hash, addition_date=None, caption=None, entries=None, fk_travel_diary_id=None, **kw: Any):
         super().__init__(**kw)

src/pilgrim/service/photo_service.py

@@ -14,8 +14,9 @@ class PhotoService:
     def __init__(self, session):
         self.session = session
 
-    def _hash_file(self, filepath: Path) -> str:
-        """Calculate hash of a file using SHA3-384."""
+    @staticmethod
+    def hash_file(filepath: Path) -> str:
+        """Calculate the hash of a file using SHA3-384."""
         hash_func = hashlib.new('sha3_384')
         with open(filepath, 'rb') as f:
             while chunk := f.read(8192):
@@ -64,10 +65,18 @@ class PhotoService:
 
         return dest_path
 
+    def check_photo_by_hash(self, photohash:str, traveldiaryid:int):
+        photo = (self.session.query(Photo).filter(Photo.photo_hash == photohash,Photo.fk_travel_diary_id == traveldiaryid)
+                 .first())
+        return photo
+
     def create(self, filepath: Path, name: str, travel_diary_id: int, caption=None, addition_date=None) -> Photo | None:
         travel_diary = self.session.query(TravelDiary).filter(TravelDiary.id == travel_diary_id).first()
         if not travel_diary:
             return None
+        photo_hash = self.hash_file(filepath)
+        if self.check_photo_by_hash(photo_hash, travel_diary_id):
+            return None
 
         # Copy photo to diary's images directory
         copied_path = self._copy_photo_to_diary(filepath, travel_diary)
@@ -79,8 +88,6 @@ class PhotoService:
             except ValueError:
                 addition_date = None
 
-        # Calculate hash from the copied file
-        photo_hash = self._hash_file(copied_path)
         
         new_photo = Photo(
             filepath=str(copied_path),  # Store the path to the copied file
@@ -118,7 +125,7 @@ class PhotoService:
                         old_path.unlink()
                     original.filepath = str(new_path)
                     # Update hash based on the new copied file
-                    original.photo_hash = self._hash_file(new_path)
+                    original.photo_hash = self.hash_file(new_path)
             
             original.name = photo_dst.name
             original.addition_date = photo_dst.addition_date

src/pilgrim/ui/screens/modals/add_photo_modal.py

@@ -63,10 +63,16 @@ class AddPhotoModal(Screen):
 
     async def _async_create_photo(self, photo_data: dict):
         """Creates a new photo asynchronously using PhotoService"""
+
+
         try:
             service_manager = self.app.service_manager
             photo_service = service_manager.get_photo_service()
 
+            if photo_service.check_photo_by_hash(photo_service.hash_file(photo_data["filepath"]),self.diary_id):
+                self.notify("Photo already exists in database", severity="error")
+                return
+
             new_photo = photo_service.create(
                 filepath=Path(photo_data["filepath"]),
                 name=photo_data["name"],