Update app.py

app.py

@@ -1833,255 +1833,316 @@ def extract_images_from_pdf(pdf_path: Path):
     except Exception as e:
         raise RuntimeError(f"❌ Error in extract_images_from_pdf: {str(e)}")
 
-def similarity_matching(input_json_path: str, project_folder:str) -> str:
-
-    logger.info("🔍 Running similarity matching...")
+def similarity_matching(input_json_path: str, project_folder: str) -> str:
+    logger.info("🔍 Running similarity matching…")
     os.makedirs(project_folder, exist_ok=True)
-    # ============================== #
-    #         DEFINE PATHS           #
-    # ============================== #
-    #image_dirs = [backdrop_images_path, sprite_images_path]
-    print(" --------------------------------------- input_json_path ---------------------------------------", input_json_path)
+
+    # ----------------------------------------
+    # CHANGED: define normalized base-paths so startswith() checks work
+    backdrop_base_path = os.path.normpath(str(BACKDROP_DIR))
+    sprite_base_path   = os.path.normpath(str(SPRITE_DIR))
+    # ----------------------------------------
+
     project_json_path = os.path.join(project_folder, "project.json")
-    
-    # ============================== #
-    #      READ SPRITE METADATA      #
-    # ============================== #
+
+    # ==============================
+    #    READ SPRITE METADATA
+    # ==============================
     with open(input_json_path, 'r') as f:
         sprites_data = json.load(f)
 
     sprite_ids, texts, sprite_base64 = [], [], []
     for sid, sprite in sprites_data.items():
         sprite_ids.append(sid)
-        texts.append(
-            "This is " + sprite.get("description", sprite.get("name", "")))
+        texts.append("This is " + sprite.get("description", sprite.get("name", "")))
         sprite_base64.append(sprite["base64"])
 
-    # ========================================= #
-    #  Walk folders to collect all image paths  #
-    # ========================================= #
+    # =========================================
+    #  Build the list of all candidate images
+    # =========================================
     folder_image_paths = [
-    #backdrops
-    BACKDROP_DIR / "badroom3.sb3" / "8cc0b88d53345b3e337e8f028a32a4e7.png",
-    BACKDROP_DIR / "baseball2.sb3" / "7be1f5b3e682813dac1f297e52ff7dca.png",
-    BACKDROP_DIR / "beach_malibu.sb3" / "050615fe992a00d6af0e664e497ebf53.png",
-    BACKDROP_DIR / "castle2.sb3" / "951765ee7f7370f120c9df20b577c22f.png",
-    BACKDROP_DIR / "hall.sb3" / "ea86ca30b346f27ca5faf1254f6a31e3.png",
-    BACKDROP_DIR / "jungle.sb3" / "f4f908da19e2753f3ed679d7b37650ca.png",
-    #sprite
-    SPRITE_DIR / "Batter.sprite3" / "baseball_sprite_motion_1.png",
-    SPRITE_DIR / "Bear.sprite3" / "bear_motion_2.png",
-    SPRITE_DIR / "Beetle.sprite3" / "46d0dfd4ae7e9bfe3a6a2e35a4905eae.png",
-    SPRITE_DIR / "cat" / "cat_motion_1.png",
-    SPRITE_DIR / "Centaur.sprite3" / "2373556e776cad3ba4d6ee04fc34550b.png",
-    SPRITE_DIR / "Crab.sprite3" / "bear_element.png",
-    SPRITE_DIR / "Soccer Ball.sprite3" / "cat_football.png",
+        # backdrops
+        BACKDROP_DIR / "badroom3.sb3"   / "8cc0b88d53345b3e337e8f028a32a4e7.png",
+        BACKDROP_DIR / "baseball2.sb3"  / "7be1f5b3e682813dac1f297e52ff7dca.png",
+        BACKDROP_DIR / "beach_malibu.sb3" / "050615fe992a00d6af0e664e497ebf53.png",
+        BACKDROP_DIR / "castle2.sb3"    / "951765ee7f7370f120c9df20b577c22f.png",
+        BACKDROP_DIR / "hall.sb3"       / "ea86ca30b346f27ca5faf1254f6a31e3.png",
+        BACKDROP_DIR / "jungle.sb3"     / "f4f908da19e2753f3ed679d7b37650ca.png",
+        # sprites
+        SPRITE_DIR   / "Batter.sprite3" / "baseball_sprite_motion_1.png",
+        SPRITE_DIR   / "Bear.sprite3"   / "bear_motion_2.png",
+        SPRITE_DIR   / "Beetle.sprite3" / "46d0dfd4ae7e9bfe3a6a2e35a4905eae.png",
+        SPRITE_DIR   / "cat"            / "cat_motion_1.png",
+        SPRITE_DIR   / "Centaur.sprite3" / "2373556e776cad3ba4d6ee04fc34550b.png",
+        SPRITE_DIR   / "Crab.sprite3"   / "bear_element.png",
+        SPRITE_DIR   / "Soccer Ball.sprite3" / "cat_football.png",
     ]
-    print(" --------------------------------------- folder_image_paths ---------------------------------------\n", folder_image_paths)
-    # ============================== #
-    #      DECODE SPRITE IMAGES      #
-    # ============================== #
-    # temp_dir = tempfile.mkdtemp()
-    # sprite_image_paths = []
-    # for idx, b64 in enumerate(sprite_base64):
-    #     image_data = base64.b64decode(b64.split(",")[-1])
-    #     img = Image.open(BytesIO(image_data)).convert("RGB")
-    #     temp_path = os.path.join(temp_dir, f"sprite_{idx}.png")
-    #     img.save(temp_path)
-    #     sprite_image_paths.append(temp_path)
-    # print(f"\n\n\nSPRITE IMAGE PATHS: \n{sprite_image_paths}")
-
-    # # ============================== #
-    # #      EMBED SPRITE IMAGES       #
-    # # ============================== #
-    # sprite_features = clip_embd.embed_image(sprite_base64)
-    
-    # # ============================== #
-    # #     COMPUTE SIMILARITIES       #
-    # # ============================== #
-    # with open(f"{BLOCKS_DIR}/embeddings.json", "r") as f:
-    #     embedding_json = json.load(f)
-    #     # print(f"\n\n EMBEDDING JSON: {embedding_json}")
-        
-    # img_matrix = np.array([img["embeddings"] for img in embedding_json])
-    # sprite_matrix = np.array(sprite_features)
-    
-    # similarity = np.matmul(sprite_matrix, img_matrix.T)
+    folder_image_paths = [os.path.normpath(str(p)) for p in folder_image_paths]
+    # =========================================
 
-    # most_similar_indices = np.argmax(similarity, axis=1)
-        # ========================================= #
-    #  Load reference embeddings from JSON file #
-    # ========================================= #
+    # -----------------------------------------
+    #  Load reference embeddings from JSON
+    # -----------------------------------------
     with open(f"{BLOCKS_DIR}/embeddings.json", "r") as f:
         embedding_json = json.load(f)
     img_matrix = np.array([img["embeddings"] for img in embedding_json])
 
-    # ============================== #
-    #      EMBED SPRITE IMAGES       #
-    # ============================== #
+    # =========================================
+    #  Decode & embed each sprite image
+    # =========================================
     sprite_features = []
     for b64 in sprite_base64:
-        # strip data URI if present
         if "," in b64:
             b64 = b64.split(",", 1)[1]
         img_bytes = base64.b64decode(b64)
         pil_img = Image.open(BytesIO(img_bytes)).convert("RGB")
-        
         buf = BytesIO()
         pil_img.save(buf, format="PNG")
         buf.seek(0)
-    
-        # 4) pass the buffer inside a list
-        feats_list = clip_embd.embed_image([buf])
-        feats = feats_list[0]
+        feats = clip_embd.embed_image([buf])[0]
         sprite_features.append(feats)
-
-    # stack into (n_sprites, dim)
     sprite_matrix = np.vstack(sprite_features)
 
-    # ============================== #
-    #     COMPUTE SIMILARITIES       #
-    # ============================== #
+    # =========================================
+    #  Compute similarities & pick best match
+    # =========================================
     similarity = np.matmul(sprite_matrix, img_matrix.T)
     most_similar_indices = np.argmax(similarity, axis=1)
-    print(f"")
-    # ============= Match and copy ===============
-    project_data = []
+
+    # =========================================
+    #  Copy matched sprite assets + collect data
+    # =========================================
+    project_data   = []
     copied_folders = set()
 
-    # =============================================================== #
-    #   Loop through most similar images from Sprites folder          #
-    #   → Copy sprite assets (excluding matched image + sprite.json)  #
-    #   → Load sprite.json and append its data to project_data        #
-    # =============================================================== #
     for sprite_idx, matched_idx in enumerate(most_similar_indices):
         matched_image_path = folder_image_paths[matched_idx]
-        matched_image_path = os.path.normpath(matched_image_path)
-        print(" --------------------------------------1- matched_image_path ---------------------------------------",matched_image_path)
-        matched_folder = os.path.dirname(matched_image_path)
-        #folder_name = os.path.basename(matched_folder)
-        print(" --------------------------------------1- matched_folder ---------------------------------------",matched_folder)
+        matched_folder     = os.path.dirname(matched_image_path)
+
+        # CHANGED: use our new normalized sprite_base_path
+        if not matched_folder.startswith(sprite_base_path):
+            continue
+
         if matched_folder in copied_folders:
             continue
         copied_folders.add(matched_folder)
-        logger.info(f"Matched image path: {matched_image_path}")
+        logger.info(f"Matched sprite: {matched_image_path}")
 
         sprite_json_path = os.path.join(matched_folder, 'sprite.json')
-        print(" --------------------------------------- sprite_json_path ---------------------------------------",sprite_json_path)
         if not os.path.exists(sprite_json_path):
-            logger.warning(f"sprite.json not found in: {matched_folder}")
+            logger.warning(f"No sprite.json in {matched_folder}")
             continue
 
         with open(sprite_json_path, 'r') as f:
-            sprite_data = json.load(f)
-            # print(f"SPRITE DATA: \n{sprite_data}")
-        # # Copy only non-matched files
-        # for fname in os.listdir(matched_folder):
-        #     fpath = os.path.join(matched_folder, fname)
-        #     if os.path.isfile(fpath) and fname not in {os.path.basename(matched_image_path), 'sprite.json'}:
-        #         shutil.copy2(fpath, os.path.join(project_folder, fname))
-        #         # logger.info(f"Copied Sprite asset: {fname}")
-        project_data.append(sprite_data)
-        print(" --------------------------------------1- project_data ---------------------------------------",project_data)
+            sprite_info = json.load(f)
+        # copy all non‐matched files
         for fname in os.listdir(matched_folder):
-            fpath = os.path.join(matched_folder, fname)
-            dest_path = os.path.join(project_folder, fname)
-            if os.path.isfile(fpath) and fname not in {os.path.basename(matched_image_path), 'sprite.json'}:
-                shutil.copy2(fpath, dest_path)
-                logger.info(f"🟢 Copied Sprite Asset: {fpath} → {dest_path}")
-        
-    # ================================================================== #
-    #   Loop through most similar images from Backdrops folder           #
-    #   → Copy Backdrop assets (excluding matched image + project.json)  #
-    #   → Load project.json and append its data to project_data          #
-    # ================================================================== #
-    backdrop_data = []  # for backdrop-related entries
+            if fname in (os.path.basename(matched_image_path), 'sprite.json'):
+                continue
+            shutil.copy2(os.path.join(matched_folder, fname),
+                         os.path.join(project_folder, fname))
+        project_data.append(sprite_info)
+
+    # =========================================
+    #  Copy matched backdrop assets + collect
+    # =========================================
+    backdrop_data = []
 
     for backdrop_idx, matched_idx in enumerate(most_similar_indices):
-        matched_image_path = os.path.normpath(folder_image_paths[matched_idx])
-        print(" --------------------------------------2- matched_image_path ---------------------------------------",matched_image_path)
-        # Check if the match is from the Backdrops folder
-        if matched_image_path.startswith(os.path.normpath(backdrop_images_path)):
-            matched_folder = os.path.dirname(matched_image_path)
-            print(" --------------------------------------2- matched_folder ---------------------------------------",matched_folder)
-            folder_name = os.path.basename(matched_folder)
-
-            logger.info(f"Backdrop matched image: {matched_image_path}")
-
-            # Copy only non-matched files
-            # for fname in os.listdir(matched_folder):
-            #     fpath = os.path.join(matched_folder, fname)
-            #     if os.path.isfile(fpath) and fname not in {os.path.basename(matched_image_path), 'project.json'}:
-            #         shutil.copy2(fpath, os.path.join(project_folder, fname))
-            #         # logger.info(f"Copied Backdrop asset: {fname}")
-            for fname in os.listdir(matched_folder):
-                fpath = os.path.join(matched_folder, fname)
-                dest_path = os.path.join(project_folder, fname)
-                if os.path.isfile(fpath) and fname not in {os.path.basename(matched_image_path), 'project.json'}:
-                    shutil.copy2(fpath, dest_path)
-                    logger.info(f"🟡 Copied Backdrop Asset: {fpath} → {dest_path}")
-
-
-            # Append backdrop's project.json
-            backdrop_json_path = os.path.join(matched_folder, 'project.json')
-            print(" --------------------------------------2- backdrop_json_path ---------------------------------------",backdrop_json_path)
-            if os.path.exists(backdrop_json_path):
-                with open(backdrop_json_path, 'r') as f:
-                    backdrop_json_data = json.load(f)
-                    # print(f"SPRITE DATA: \n{backdrop_json_data}")
-                if "targets" in backdrop_json_data:
-                    for target in backdrop_json_data["targets"]:
-                        if target.get("isStage") == True:
-                            backdrop_data.append(target)
-            else:
-                logger.warning(f"project.json not found in: {matched_folder}")
+        matched_image_path = folder_image_paths[matched_idx]
+        matched_folder     = os.path.dirname(matched_image_path)
 
-    # Merge JSON structure
+        # CHANGED: use our new normalized backdrop_base_path
+        if not matched_folder.startswith(backdrop_base_path):
+            continue
+
+        logger.info(f"Matched backdrop: {matched_image_path}")
+        # copy non‐matched files
+        for fname in os.listdir(matched_folder):
+            if fname in (os.path.basename(matched_image_path), 'project.json'):
+                continue
+            shutil.copy2(os.path.join(matched_folder, fname),
+                         os.path.join(project_folder, fname))
+
+        # append the stage‐target from its project.json
+        pj = os.path.join(matched_folder, 'project.json')
+        if os.path.exists(pj):
+            with open(pj, 'r') as f:
+                bd_json = json.load(f)
+            for tgt in bd_json.get("targets", []):
+                if tgt.get("isStage"):
+                    backdrop_data.append(tgt)
+        else:
+            logger.warning(f"No project.json in {matched_folder}")
+
+    # =========================================
+    #  Merge into final Scratch project.json
+    # =========================================
     final_project = {
-        "targets": [],
-        "monitors": [],
-        "extensions": [],
+        "targets": [], "monitors": [], "extensions": [],
         "meta": {
             "semver": "3.0.0",
             "vm": "11.3.0",
-            "agent": "Mozilla/5.0 (Windows NT 10.0; Win64; x64) AppleWebKit/537.36 (KHTML, like Gecko) Chrome/138.0.0.0 Safari/537.36"
+            "agent": "Mozilla/5.0 (Windows NT 10.0; Win64; x64) AppleWebKit/537.36 (KHTML, like Gecko) Chrome/138.0.0.0 Safari/537.36
         }
     }
+    # sprites first
+    for spr in project_data:
+        if not spr.get("isStage", False):
+            final_project["targets"].append(spr)
 
-    for sprite in project_data:
-        if not sprite.get("isStage", False):
-            final_project["targets"].append(sprite)
-
+    # then backdrop as the Stage
     if backdrop_data:
         all_costumes, sounds = [], []
-        for idx, bd in enumerate(backdrop_data):
+        for i, bd in enumerate(backdrop_data):
             all_costumes.extend(bd.get("costumes", []))
-            if idx == 0 and "sounds" in bd:
-                sounds = bd["sounds"]
+            if i == 0:
+                sounds = bd.get("sounds", [])
         final_project["targets"].append({
             "isStage": True,
             "name": "Stage",
-            "variables": {},
-            "lists": {},
-            "broadcasts": {},
-            "blocks": {},
-            "comments": {},
+            "variables": {}, "lists": {}, "broadcasts": {},
+            "blocks": {}, "comments": {},
             "currentCostume": 1 if len(all_costumes) > 1 else 0,
             "costumes": all_costumes,
             "sounds": sounds,
-            "volume": 100,
-            "layerOrder": 0,
-            "tempo": 60,
-            "videoTransparency": 50,
-            "videoState": "on",
-            "textToSpeechLanguage": None
+            "volume": 100, "layerOrder": 0,
+            "tempo": 60, "videoTransparency": 50,
+            "videoState": "on", "textToSpeechLanguage": None
         })
 
     with open(project_json_path, 'w') as f:
         json.dump(final_project, f, indent=2)
 
-    # logger.info(f"🎉 Final project saved: {project_json_path}")
     return project_json_path
+    # for sprite_idx, matched_idx in enumerate(most_similar_indices):
+    #     matched_image_path = folder_image_paths[matched_idx]
+    #     matched_image_path = os.path.normpath(matched_image_path)
+    #     print(" --------------------------------------1- matched_image_path ---------------------------------------",matched_image_path)
+    #     matched_folder = os.path.dirname(matched_image_path)
+    #     #folder_name = os.path.basename(matched_folder)
+    #     print(" --------------------------------------1- matched_folder ---------------------------------------",matched_folder)
+    #     if matched_folder in copied_folders:
+    #         continue
+    #     copied_folders.add(matched_folder)
+    #     logger.info(f"Matched image path: {matched_image_path}")
+
+    #     sprite_json_path = os.path.join(matched_folder, 'sprite.json')
+    #     print(" --------------------------------------- sprite_json_path ---------------------------------------",sprite_json_path)
+    #     if not os.path.exists(sprite_json_path):
+    #         logger.warning(f"sprite.json not found in: {matched_folder}")
+    #         continue
+
+    #     with open(sprite_json_path, 'r') as f:
+    #         sprite_data = json.load(f)
+    #         # print(f"SPRITE DATA: \n{sprite_data}")
+    #     # # Copy only non-matched files
+    #     # for fname in os.listdir(matched_folder):
+    #     #     fpath = os.path.join(matched_folder, fname)
+    #     #     if os.path.isfile(fpath) and fname not in {os.path.basename(matched_image_path), 'sprite.json'}:
+    #     #         shutil.copy2(fpath, os.path.join(project_folder, fname))
+    #     #         # logger.info(f"Copied Sprite asset: {fname}")
+    #     project_data.append(sprite_data)
+    #     print(" --------------------------------------1- project_data ---------------------------------------",project_data)
+    #     for fname in os.listdir(matched_folder):
+    #         fpath = os.path.join(matched_folder, fname)
+    #         dest_path = os.path.join(project_folder, fname)
+    #         if os.path.isfile(fpath) and fname not in {os.path.basename(matched_image_path), 'sprite.json'}:
+    #             shutil.copy2(fpath, dest_path)
+    #             logger.info(f"🟢 Copied Sprite Asset: {fpath} → {dest_path}")
+        
+    # # ================================================================== #
+    # #   Loop through most similar images from Backdrops folder           #
+    # #   → Copy Backdrop assets (excluding matched image + project.json)  #
+    # #   → Load project.json and append its data to project_data          #
+    # # ================================================================== #
+    # backdrop_data = []  # for backdrop-related entries
+
+    # for backdrop_idx, matched_idx in enumerate(most_similar_indices):
+    #     matched_image_path = os.path.normpath(folder_image_paths[matched_idx])
+    #     print(" --------------------------------------2- matched_image_path ---------------------------------------",matched_image_path)
+    #     # Check if the match is from the Backdrops folder
+    #     if matched_image_path.startswith(os.path.normpath(backdrop_images_path)):
+    #         matched_folder = os.path.dirname(matched_image_path)
+    #         print(" --------------------------------------2- matched_folder ---------------------------------------",matched_folder)
+    #         folder_name = os.path.basename(matched_folder)
+
+    #         logger.info(f"Backdrop matched image: {matched_image_path}")
+
+    #         # Copy only non-matched files
+    #         # for fname in os.listdir(matched_folder):
+    #         #     fpath = os.path.join(matched_folder, fname)
+    #         #     if os.path.isfile(fpath) and fname not in {os.path.basename(matched_image_path), 'project.json'}:
+    #         #         shutil.copy2(fpath, os.path.join(project_folder, fname))
+    #         #         # logger.info(f"Copied Backdrop asset: {fname}")
+    #         for fname in os.listdir(matched_folder):
+    #             fpath = os.path.join(matched_folder, fname)
+    #             dest_path = os.path.join(project_folder, fname)
+    #             if os.path.isfile(fpath) and fname not in {os.path.basename(matched_image_path), 'project.json'}:
+    #                 shutil.copy2(fpath, dest_path)
+    #                 logger.info(f"🟡 Copied Backdrop Asset: {fpath} → {dest_path}")
+
+
+    #         # Append backdrop's project.json
+    #         backdrop_json_path = os.path.join(matched_folder, 'project.json')
+    #         print(" --------------------------------------2- backdrop_json_path ---------------------------------------",backdrop_json_path)
+    #         if os.path.exists(backdrop_json_path):
+    #             with open(backdrop_json_path, 'r') as f:
+    #                 backdrop_json_data = json.load(f)
+    #                 # print(f"SPRITE DATA: \n{backdrop_json_data}")
+    #             if "targets" in backdrop_json_data:
+    #                 for target in backdrop_json_data["targets"]:
+    #                     if target.get("isStage") == True:
+    #                         backdrop_data.append(target)
+    #         else:
+    #             logger.warning(f"project.json not found in: {matched_folder}")
+
+    # # Merge JSON structure
+    # final_project = {
+    #     "targets": [],
+    #     "monitors": [],
+    #     "extensions": [],
+    #     "meta": {
+    #         "semver": "3.0.0",
+    #         "vm": "11.3.0",
+    #         "agent": "Mozilla/5.0 (Windows NT 10.0; Win64; x64) AppleWebKit/537.36 (KHTML, like Gecko) Chrome/138.0.0.0 Safari/537.36"
+    #     }
+    # }
+
+    # for sprite in project_data:
+    #     if not sprite.get("isStage", False):
+    #         final_project["targets"].append(sprite)
+
+    # if backdrop_data:
+    #     all_costumes, sounds = [], []
+    #     for idx, bd in enumerate(backdrop_data):
+    #         all_costumes.extend(bd.get("costumes", []))
+    #         if idx == 0 and "sounds" in bd:
+    #             sounds = bd["sounds"]
+    #     final_project["targets"].append({
+    #         "isStage": True,
+    #         "name": "Stage",
+    #         "variables": {},
+    #         "lists": {},
+    #         "broadcasts": {},
+    #         "blocks": {},
+    #         "comments": {},
+    #         "currentCostume": 1 if len(all_costumes) > 1 else 0,
+    #         "costumes": all_costumes,
+    #         "sounds": sounds,
+    #         "volume": 100,
+    #         "layerOrder": 0,
+    #         "tempo": 60,
+    #         "videoTransparency": 50,
+    #         "videoState": "on",
+    #         "textToSpeechLanguage": None
+    #     })
+
+    # with open(project_json_path, 'w') as f:
+    #     json.dump(final_project, f, indent=2)
+
+    # # logger.info(f"🎉 Final project saved: {project_json_path}")
+    # return project_json_path
 
 def delay_for_tpm_node(state: GameState):
     logger.info("--- Running DelayForTPMNode ---")