fix: filtering logics

src/models/ai_engine.py

@@ -296,9 +296,19 @@ def is_valid_item(nm, price_str, cnt_str=None):
             
     price_raw_clean = str(price_str).strip()
     if len(re.findall(r'[a-zA-Z]', price_raw_clean)) > 3: return False
-    if re.match(r'^[A-Z0-9\s\-:]{5,}$', nm_str) and not re.search(r'[a-z]', nm_str):
-        if not re.search(r'(?i)(nasi|mie|ayam|ikan|teh|kopi|rice|chicken|coffee|tea|water|juice|beer|wine|pizza|burger|fries|soda|cake|spaghetti|alfredo|schnitzel|cordon|squash|poffertjes)', nm_str):
+
+    # Reject only obvious transaction codes / serial IDs:
+    # must be all-caps, contain digits, AND contain a separator like / or - or be >= 8 digits
+    if re.match(r'^[A-Z0-9\s\-:/\.]{5,}$', nm_str) and not re.search(r'[a-z]', nm_str):
+        digit_count = len(re.findall(r'\d', nm_str))
+        has_separator = bool(re.search(r'[/\-:]', nm_str))
+        # Only reject if it looks like a code: has digits AND a separator, OR is >50% digits
+        word_count = len(nm_str.split())
+        if digit_count > 0 and has_separator:
+            return False
+        if word_count <= 2 and digit_count > 0 and digit_count >= len(nm_str.replace(' ', '')) * 0.4:
             return False
+
     return True
 
 # ===============================
@@ -352,9 +362,16 @@ def parse_cord_to_schema(cord_json):
     if isinstance(menu, dict): menu = [menu]
 
     for item in menu:
-        entries = item if isinstance(item, list) else [item]
+        if not isinstance(item, dict):
+            continue
+        entries = [item]
         for entry in entries:
-            nm = entry.get("nm", "Unknown")
+            if not isinstance(entry, dict):
+                continue
+            nm = entry.get("nm", None)
+            # Skip items with no name at all — don't accept "Unknown" fallback
+            if not nm or str(nm).strip() in ("", "Unknown"):
+                continue
             price_raw = entry.get("price", "0")
             cnt_raw = entry.get("cnt", "1")
             if not is_valid_item(nm, price_raw, cnt_raw): continue
@@ -366,6 +383,8 @@ def parse_cord_to_schema(cord_json):
     sub_total_node = cord_json.get("sub_total", {})
     if isinstance(sub_total_node, list):
         sub_total_node = sub_total_node[0] if sub_total_node else {}
+    if not isinstance(sub_total_node, dict):
+        sub_total_node = {}
     subtotal = clean_price(sub_total_node.get("subtotal_price", "0"))
     tax_amount = clean_price(sub_total_node.get("tax_price", "0"))
     service_charge = clean_price(sub_total_node.get("service_price", "0"))
@@ -374,6 +393,8 @@ def parse_cord_to_schema(cord_json):
     total_node = cord_json.get("total", {})
     if isinstance(total_node, list):
         total_node = total_node[0] if total_node else {}
+    if not isinstance(total_node, dict):
+        total_node = {}
     total_amount = clean_price(total_node.get("total_price", "0"))
 
     items_sum = sum(i["item_price"] for i in items)

src/pages/1_AI_OCR.py

@@ -57,6 +57,7 @@ for k, v in {
     "tess_json": None,
     "preprocess_steps": [],
     "img_arr": None,
+    "last_success_msg": None,
 }.items():
     if k not in st.session_state:
         st.session_state[k] = v
@@ -136,7 +137,7 @@ if img_arr is not None:
                     st.session_state.preprocess_steps = steps
                     st.session_state.donut_base_parsed = parsed
                     st.session_state.donut_base_json = jout
-                    st.toast("✅ Donut base extraction complete!", icon="🍩")
+                    st.session_state.last_success_msg = "✅ Donut Base extraction complete!"
                 except Exception as e:
                     st.error(f"Donut base failed: {e}")
     with b3:
@@ -147,7 +148,7 @@ if img_arr is not None:
                     st.session_state.preprocess_steps = steps
                     st.session_state.donut_v6_parsed = parsed
                     st.session_state.donut_v6_json = jout
-                    st.toast("✅ Donut v6 extraction complete!", icon="🍩")
+                    st.session_state.last_success_msg = "✅ Donut v6 extraction complete!"
                 except Exception as e:
                     st.error(f"Donut v6 failed: {e}")
     with b4:
@@ -158,10 +159,14 @@ if img_arr is not None:
                     st.session_state.preprocess_steps = steps
                     st.session_state.tess_parsed = parsed
                     st.session_state.tess_json = jout
-                    st.toast("✅ Tesseract extraction complete!", icon="🔤")
+                    st.session_state.last_success_msg = "✅ Tesseract extraction complete!"
                 except Exception as e:
                     st.error(f"Tesseract failed: {e}")
 
+    if st.session_state.get("last_success_msg"):
+        st.success(st.session_state.last_success_msg)
+        st.session_state.last_success_msg = None
+
     if st.session_state.preprocess_steps:
         st.subheader("Preprocessing Pipeline Visualization")
         cols = st.columns(min(len(st.session_state.preprocess_steps), 6))