Major update. Support for 15 LLMs, World Flora Online taxonomy validation, geolocation, 2 OCR methods, significant UI changes, stability improvements, consistent JSON parsing

app.py

@@ -249,6 +249,126 @@ def load_gallery(converted_files, uploaded_file):
             file_path_small = save_uploaded_file(st.session_state['dir_uploaded_images_small'], uploaded_file, img)
             st.session_state['input_list_small'].append(file_path_small)
 
+@st.cache_data
+def handle_image_upload_and_gallery_hf(uploaded_files):
+    if uploaded_files:
+        # Clear input image gallery and input list
+        clear_image_uploads()
+
+        for uploaded_file in uploaded_files:
+            # Determine the file type
+            if uploaded_file.name.lower().endswith('.pdf'):
+                # Handle PDF files
+                file_path = save_uploaded_file(st.session_state['dir_uploaded_images'], uploaded_file)
+                # Convert each page of the PDF to an image
+                n_pages = convert_pdf_to_jpg(file_path, st.session_state['dir_uploaded_images'], dpi=200)#st.session_state.config['leafmachine']['project']['dir_images_local'])
+                # Update the input list for each page image
+                converted_files = os.listdir(st.session_state['dir_uploaded_images'])
+                for file_name in converted_files:   
+                    if file_name.split('.')[1].lower() in ['jpg','jpeg']:
+                        jpg_file_path = os.path.join(st.session_state['dir_uploaded_images'], file_name)
+                        st.session_state['input_list'].append(jpg_file_path)
+
+                        # Optionally, create a thumbnail for the gallery
+                        img = Image.open(jpg_file_path)
+                        img.thumbnail((GALLERY_IMAGE_SIZE, GALLERY_IMAGE_SIZE), Image.Resampling.LANCZOS)
+                        try:
+                            file_path_small = save_uploaded_file(st.session_state['dir_uploaded_images_small'], file_name, img)
+                        except:
+                            file_path_small = save_uploaded_file_local(st.session_state['dir_uploaded_images_small'],st.session_state['dir_uploaded_images_small'], file_name, img)
+                        st.session_state['input_list_small'].append(file_path_small)
+                
+            else:
+                # Handle JPG/JPEG files (existing process)
+                file_path = save_uploaded_file(st.session_state['dir_uploaded_images'], uploaded_file)
+                st.session_state['input_list'].append(file_path)
+                img = Image.open(file_path)
+                img.thumbnail((GALLERY_IMAGE_SIZE, GALLERY_IMAGE_SIZE), Image.Resampling.LANCZOS)
+                file_path_small = save_uploaded_file(st.session_state['dir_uploaded_images_small'], uploaded_file, img)
+                st.session_state['input_list_small'].append(file_path_small)
+
+        # After processing all files
+        st.info(f"Processing images from {st.session_state.config['leafmachine']['project']['dir_images_local']}")
+    
+    if st.session_state['input_list_small']:
+        if len(st.session_state['input_list_small']) > MAX_GALLERY_IMAGES:
+            # Only take the first 100 images from the list
+            images_to_display = st.session_state['input_list_small'][:MAX_GALLERY_IMAGES]
+        else:
+            # If there are less than 100 images, take them all
+            images_to_display = st.session_state['input_list_small']
+        show_gallery_small_hf(images_to_display)
+
+
+@st.cache_data
+def handle_image_upload_and_gallery():
+    st.session_state['view_local_gallery'] = st.toggle("View Image Gallery",)
+    
+    if st.session_state['view_local_gallery'] and st.session_state['input_list_small'] and (st.session_state['dir_images_local_TEMP'] == st.session_state.config['leafmachine']['project']['dir_images_local']):
+        if MAX_GALLERY_IMAGES <= st.session_state['processing_add_on']:
+            info_txt = f"Showing {MAX_GALLERY_IMAGES} out of {st.session_state['processing_add_on']} images"
+        else:
+            info_txt = f"Showing {st.session_state['processing_add_on']} out of {st.session_state['processing_add_on']} images"
+        st.info(info_txt)
+        try:
+            show_gallery_small()
+        except:
+            pass
+
+    elif not st.session_state['view_local_gallery'] and st.session_state['input_list_small'] and (st.session_state['dir_images_local_TEMP'] == st.session_state.config['leafmachine']['project']['dir_images_local']):
+        pass
+    elif not st.session_state['view_local_gallery'] and not st.session_state['input_list_small'] and (st.session_state['dir_images_local_TEMP'] == st.session_state.config['leafmachine']['project']['dir_images_local']):
+        pass
+    # elif st.session_state['input_list_small'] and (st.session_state['dir_images_local_TEMP'] != st.session_state.config['leafmachine']['project']['dir_images_local']):
+    elif (st.session_state['dir_images_local_TEMP'] != st.session_state.config['leafmachine']['project']['dir_images_local']):
+        has_pdf = False
+        clear_image_uploads()
+
+        for input_file in os.listdir(st.session_state.config['leafmachine']['project']['dir_images_local']):
+            if input_file.split('.')[1].lower() in ['jpg','jpeg']:
+                pass
+            elif input_file.split('.')[1].lower() in ['pdf',]:
+                has_pdf = True
+                # Handle PDF files
+                file_path = save_uploaded_file_local(st.session_state.config['leafmachine']['project']['dir_images_local'], st.session_state['dir_uploaded_images'], input_file)
+                # Convert each page of the PDF to an image
+                n_pages = convert_pdf_to_jpg(file_path, st.session_state['dir_uploaded_images'], dpi=200)#st.session_state.config['leafmachine']['project']['dir_images_local'])
+                
+                
+                # pdf_files_pattern = os.path.join(st.session_state['dir_uploaded_images'], '*.pdf')
+                # for pdf_file in glob.glob(pdf_files_pattern):
+                #     os.remove(pdf_file)
+
+                # # Update the input list for each page image
+                # converted_files = os.listdir(st.session_state['dir_uploaded_images'])
+                # for file_name in converted_files:   
+                #     if file_name.lower().endswith('.jpg'):
+                #         jpg_file_path = os.path.join(st.session_state['dir_uploaded_images'], file_name)
+                #         st.session_state['input_list'].append(jpg_file_path)
+
+                #         # Optionally, create a thumbnail for the gallery
+                #         img = Image.open(jpg_file_path)
+                #         img.thumbnail((GALLERY_IMAGE_SIZE, GALLERY_IMAGE_SIZE), Image.Resampling.LANCZOS)
+                #         file_path_small = save_uploaded_file_local(st.session_state['dir_uploaded_images'], st.session_state['dir_uploaded_images_small'], file_name, img)
+                #         st.session_state['input_list_small'].append(file_path_small)
+                
+                # st.session_state.config['leafmachine']['project']['dir_images_local'] = st.session_state['dir_uploaded_images']
+                
+            else:
+                pass
+                # st.warning("Inputs must be '.PDF' or '.jpg' or '.jpeg'")
+        if has_pdf:
+            st.session_state.config['leafmachine']['project']['dir_images_local'] = st.session_state['dir_uploaded_images']
+
+        dir_images_local = st.session_state.config['leafmachine']['project']['dir_images_local']
+        count_n_imgs = list_jpg_files(dir_images_local)
+        st.session_state['processing_add_on'] = count_n_imgs
+        # print(st.session_state['processing_add_on'])
+        st.session_state['dir_images_local_TEMP'] = st.session_state.config['leafmachine']['project']['dir_images_local']
+        print("rerun")
+        st.rerun()
+
+
 def content_input_images(col_left, col_right):
     st.write('---')
     # col1, col2 = st.columns([2,8])
@@ -272,120 +392,10 @@ def content_input_images(col_left, col_right):
     
     with col_right:
         if st.session_state.is_hf:
-            if uploaded_files:
-                # Clear input image gallery and input list
-                clear_image_uploads()
-
-                for uploaded_file in uploaded_files:
-                    # Determine the file type
-                    if uploaded_file.name.lower().endswith('.pdf'):
-                        # Handle PDF files
-                        file_path = save_uploaded_file(st.session_state['dir_uploaded_images'], uploaded_file)
-                        # Convert each page of the PDF to an image
-                        n_pages = convert_pdf_to_jpg(file_path, st.session_state['dir_uploaded_images'], dpi=200)#st.session_state.config['leafmachine']['project']['dir_images_local'])
-                        # Update the input list for each page image
-                        converted_files = os.listdir(st.session_state['dir_uploaded_images'])
-                        for file_name in converted_files:   
-                            if file_name.split('.')[1].lower() in ['jpg','jpeg']:
-                                jpg_file_path = os.path.join(st.session_state['dir_uploaded_images'], file_name)
-                                st.session_state['input_list'].append(jpg_file_path)
-
-                                # Optionally, create a thumbnail for the gallery
-                                img = Image.open(jpg_file_path)
-                                img.thumbnail((GALLERY_IMAGE_SIZE, GALLERY_IMAGE_SIZE), Image.Resampling.LANCZOS)
-                                try:
-                                    file_path_small = save_uploaded_file(st.session_state['dir_uploaded_images_small'], file_name, img)
-                                except:
-                                    file_path_small = save_uploaded_file_local(st.session_state['dir_uploaded_images_small'],st.session_state['dir_uploaded_images_small'], file_name, img)
-                                st.session_state['input_list_small'].append(file_path_small)
-                        
-                    else:
-                        # Handle JPG/JPEG files (existing process)
-                        file_path = save_uploaded_file(st.session_state['dir_uploaded_images'], uploaded_file)
-                        st.session_state['input_list'].append(file_path)
-                        img = Image.open(file_path)
-                        img.thumbnail((GALLERY_IMAGE_SIZE, GALLERY_IMAGE_SIZE), Image.Resampling.LANCZOS)
-                        file_path_small = save_uploaded_file(st.session_state['dir_uploaded_images_small'], uploaded_file, img)
-                        st.session_state['input_list_small'].append(file_path_small)
-
-                # After processing all files
-                st.info(f"Processing images from {st.session_state.config['leafmachine']['project']['dir_images_local']}")
-            
-            if st.session_state['input_list_small']:
-                if len(st.session_state['input_list_small']) > MAX_GALLERY_IMAGES:
-                    # Only take the first 100 images from the list
-                    images_to_display = st.session_state['input_list_small'][:MAX_GALLERY_IMAGES]
-                else:
-                    # If there are less than 100 images, take them all
-                    images_to_display = st.session_state['input_list_small']
-                show_gallery_small_hf(images_to_display)
+            handle_image_upload_and_gallery_hf(uploaded_files)
 
         else:
-            st.session_state['view_local_gallery'] = st.toggle("View Image Gallery",)
-            
-            if st.session_state['view_local_gallery'] and st.session_state['input_list_small'] and (st.session_state['dir_images_local_TEMP'] == st.session_state.config['leafmachine']['project']['dir_images_local']):
-                if MAX_GALLERY_IMAGES <= st.session_state['processing_add_on']:
-                    info_txt = f"Showing {MAX_GALLERY_IMAGES} out of {st.session_state['processing_add_on']} images"
-                else:
-                    info_txt = f"Showing {st.session_state['processing_add_on']} out of {st.session_state['processing_add_on']} images"
-                st.info(info_txt)
-                try:
-                    show_gallery_small()
-                except:
-                    pass
-
-            elif not st.session_state['view_local_gallery'] and st.session_state['input_list_small'] and (st.session_state['dir_images_local_TEMP'] == st.session_state.config['leafmachine']['project']['dir_images_local']):
-                pass
-            elif not st.session_state['view_local_gallery'] and not st.session_state['input_list_small'] and (st.session_state['dir_images_local_TEMP'] == st.session_state.config['leafmachine']['project']['dir_images_local']):
-                pass
-            # elif st.session_state['input_list_small'] and (st.session_state['dir_images_local_TEMP'] != st.session_state.config['leafmachine']['project']['dir_images_local']):
-            elif (st.session_state['dir_images_local_TEMP'] != st.session_state.config['leafmachine']['project']['dir_images_local']):
-                has_pdf = False
-                clear_image_uploads()
-
-                for input_file in os.listdir(st.session_state.config['leafmachine']['project']['dir_images_local']):
-                    if input_file.split('.')[1].lower() in ['jpg','jpeg']:
-                        pass
-                    elif input_file.split('.')[1].lower() in ['pdf',]:
-                        has_pdf = True
-                        # Handle PDF files
-                        file_path = save_uploaded_file_local(st.session_state.config['leafmachine']['project']['dir_images_local'], st.session_state['dir_uploaded_images'], input_file)
-                        # Convert each page of the PDF to an image
-                        n_pages = convert_pdf_to_jpg(file_path, st.session_state['dir_uploaded_images'], dpi=200)#st.session_state.config['leafmachine']['project']['dir_images_local'])
-                        
-                        
-                        # pdf_files_pattern = os.path.join(st.session_state['dir_uploaded_images'], '*.pdf')
-                        # for pdf_file in glob.glob(pdf_files_pattern):
-                        #     os.remove(pdf_file)
-
-                        # # Update the input list for each page image
-                        # converted_files = os.listdir(st.session_state['dir_uploaded_images'])
-                        # for file_name in converted_files:   
-                        #     if file_name.lower().endswith('.jpg'):
-                        #         jpg_file_path = os.path.join(st.session_state['dir_uploaded_images'], file_name)
-                        #         st.session_state['input_list'].append(jpg_file_path)
-
-                        #         # Optionally, create a thumbnail for the gallery
-                        #         img = Image.open(jpg_file_path)
-                        #         img.thumbnail((GALLERY_IMAGE_SIZE, GALLERY_IMAGE_SIZE), Image.Resampling.LANCZOS)
-                        #         file_path_small = save_uploaded_file_local(st.session_state['dir_uploaded_images'], st.session_state['dir_uploaded_images_small'], file_name, img)
-                        #         st.session_state['input_list_small'].append(file_path_small)
-                        
-                        # st.session_state.config['leafmachine']['project']['dir_images_local'] = st.session_state['dir_uploaded_images']
-                        
-                    else:
-                        pass
-                        # st.warning("Inputs must be '.PDF' or '.jpg' or '.jpeg'")
-                if has_pdf:
-                    st.session_state.config['leafmachine']['project']['dir_images_local'] = st.session_state['dir_uploaded_images']
-
-                dir_images_local = st.session_state.config['leafmachine']['project']['dir_images_local']
-                count_n_imgs = list_jpg_files(dir_images_local)
-                st.session_state['processing_add_on'] = count_n_imgs
-                # print(st.session_state['processing_add_on'])
-                st.session_state['dir_images_local_TEMP'] = st.session_state.config['leafmachine']['project']['dir_images_local']
-                print("rerun")
-                st.rerun()
+            handle_image_upload_and_gallery()
 
 def list_jpg_files(directory_path):
     jpg_count = 0