Update app.py

app.py

@@ -26,6 +26,7 @@ from utils.block_relation_builder import block_builder#, variable_adder_main
 from langchain.chat_models import ChatOpenAI
 from langchain_openai import ChatOpenAI
 from pydantic import Field, SecretStr
+from difflib import get_close_matches
 
 os.environ["OPENROUTER_API_KEY"] = os.getenv("OPENROUTER_API_KEY", "default_key_or_placeholder")
 class ChatOpenRouter(ChatOpenAI):
@@ -740,6 +741,68 @@ def clean_base64_for_model(raw_b64):
         # Return both prefixed and clean reduced versions
         return f"data:image/jpeg;base64,{reduced_b64}"
     return f"data:image/jpeg;base64,{clean_b64}"
+
+SCRATCH_OPCODES = [
+    'motion_movesteps', 'motion_turnright', 'motion_turnleft', 'motion_goto',
+    'motion_gotoxy', 'motion_glideto', 'motion_glidesecstoxy', 'motion_pointindirection',
+    'motion_pointtowards', 'motion_changexby', 'motion_setx', 'motion_changeyby',
+    'motion_sety', 'motion_ifonedgebounce', 'motion_setrotationstyle', 'looks_sayforsecs',
+    'looks_say', 'looks_thinkforsecs', 'looks_think', 'looks_switchcostumeto',
+    'looks_nextcostume', 'looks_switchbackdropto', 'looks_switchbackdroptowait',
+    'looks_nextbackdrop', 'looks_changesizeby', 'looks_setsizeto', 'looks_changeeffectby',
+    'looks_seteffectto', 'looks_cleargraphiceffects', 'looks_show', 'looks_hide',
+    'looks_gotofrontback', 'looks_goforwardbackwardlayers', 'sound_playuntildone',
+    'sound_play', 'sound_stopallsounds', 'sound_changevolumeby', 'sound_setvolumeto',
+    'event_broadcast', 'event_broadcastandwait', 'control_wait', 'control_wait_until',
+    'control_stop', 'control_create_clone_of', 'control_delete_this_clone',
+    'data_setvariableto', 'data_changevariableby', 'data_addtolist', 'data_deleteoflist',
+    'data_insertatlist', 'data_replaceitemoflist', 'data_showvariable', 'data_hidevariable',
+    'data_showlist', 'data_hidelist', 'sensing_askandwait', 'sensing_resettimer',
+    'sensing_setdragmode', 'procedures_call', 'operator_lt', 'operator_equals',
+    'operator_gt', 'operator_and', 'operator_or', 'operator_not', 'operator_contains',
+    'sensing_touchingobject', 'sensing_touchingcolor', 'sensing_coloristouchingcolor',
+    'sensing_keypressed', 'sensing_mousedown', 'data_listcontainsitem', 'control_repeat',
+    'control_forever', 'control_if', 'control_if_else', 'control_repeat_until',
+    'motion_xposition', 'motion_yposition', 'motion_direction', 'looks_costumenumbername',
+    'looks_size', 'looks_backdropnumbername', 'sound_volume', 'sensing_distanceto',
+    'sensing_answer', 'sensing_mousex', 'sensing_mousey', 'sensing_loudness',
+    'sensing_timer', 'sensing_of', 'sensing_current', 'sensing_dayssince2000',
+    'sensing_username', 'operator_add', 'operator_subtract', 'operator_multiply',
+    'operator_divide', 'operator_random', 'operator_join', 'operator_letterof',
+    'operator_length', 'operator_mod', 'operator_round', 'operator_mathop',
+    'data_variable', 'data_list', 'data_itemoflist', 'data_lengthoflist',
+    'data_itemnumoflist', 'event_whenflagclicked', 'event_whenkeypressed',
+    'event_whenthisspriteclicked', 'event_whenbackdropswitchesto', 'event_whengreaterthan',
+    'event_whenbroadcastreceived', 'control_start_as_clone', 'procedures_definition'
+]
+
+def validate_and_fix_opcodes(opcode_counts):
+    """
+    Ensures all opcodes are valid. If an opcode is invalid, replace with closest match.
+    """
+    corrected_list = []
+    for item in opcode_counts:
+        opcode = item.get("opcode")
+        count = item.get("count", 1)
+
+        if opcode not in SCRATCH_OPCODES:
+            # Find closest match (case-sensitive)
+            match = get_close_matches(opcode, SCRATCH_OPCODES, n=1, cutoff=0.6)
+            if match:
+                print(f"Opcode '{opcode}' not found. Replacing with '{match[0]}'")
+                opcode = match[0]
+            else:
+                print(f"Opcode '{opcode}' not recognized and no close match found. Skipping.")
+                continue
+        
+        corrected_list.append({"opcode": opcode, "count": count})
+
+    # Merge duplicates after correction
+    merged = {}
+    for item in corrected_list:
+        merged[item["opcode"]] = merged.get(item["opcode"], 0) + item["count"]
+
+    return [{"opcode": k, "count": v} for k, v in merged.items()]
     
 def format_scratch_pseudo_code(code_string):
     """
@@ -1796,7 +1859,7 @@ Example output:
 
             # Directly use the 'opcode_counts' list from the LLM's output
             plan["opcode_counts"] = llm_json.get("opcode_counts", [])
-            
+            plan["opcode_counts"] = validate_and_fix_opcodes(plan["opcode_counts"])
             # Optionally, you can remove the individual category lists from the plan
             # if they are no longer needed after the LLM provides the consolidated list.
             # for key in ["motion", "control", "operator", "sensing", "looks", "sounds", "events", "data"]: