Fix malformed coords, stronger repetition penalty, better validation

app.py

@@ -349,6 +349,15 @@ def get_model():
 # GCODE PROCESSING
 # ============================================================================
 
+def is_valid_coord(s: str) -> bool:
+    """Check if a string is a valid coordinate number."""
+    try:
+        v = float(s)
+        return -1000 < v < 1000  # Reasonable bounds
+    except (ValueError, TypeError):
+        return False
+
+
 def clean_gcode(gcode: str) -> str:
     """Clean up generated gcode - fix formatting, remove garbage."""
     
@@ -380,22 +389,32 @@ def clean_gcode(gcode: str) -> str:
             continue
         if line.startswith("Workarea:") or line.startswith("Algorithm:"):
             continue
+        
+        # Skip lines with mixed axis prefixes: Y-X-288 or X-Y-100
+        if re.search(r'X-Y-|Y-X-|X-X-|Y-Y-', line):
+            continue
             
-        # Fix malformed coordinates like X-X-X-100 or X-361.X-390
-        line = re.sub(r'X-X-X-', 'X-', line)
-        line = re.sub(r'X-X-', 'X-', line)
-        line = re.sub(r'X-\d+\.X-', 'X-', line)
-        line = re.sub(r'Y-Y-Y-', 'Y-', line)
-        line = re.sub(r'Y-Y-', 'Y-', line)
-        line = re.sub(r'Y-\d+\.Y-', 'Y-', line)
+        # Fix double negatives: X--411 -> X-411
+        line = re.sub(r'X--(\d)', r'X-\1', line)
+        line = re.sub(r'Y--(\d)', r'Y-\1', line)
         
         # Fix missing spaces: G1X -> G1 X
         line = re.sub(r'(G[01])X', r'\1 X', line)
         line = re.sub(r'(G[01])Y', r'\1 Y', line)
         
-        # Extract coordinates to check for stuck positions
+        # Validate coordinates - extract and check
         x_match = re.search(r'X([-\d.]+)', line)
         y_match = re.search(r'Y([-\d.]+)', line)
+        
+        # If line has X or Y, validate them
+        if x_match:
+            if not is_valid_coord(x_match.group(1)):
+                continue  # Skip malformed line
+        if y_match:
+            if not is_valid_coord(y_match.group(1)):
+                continue  # Skip malformed line
+        
+        # Check for stuck coordinates (repeated positions)
         if x_match and y_match:
             try:
                 coord = (round(float(x_match.group(1)), 1), round(float(y_match.group(1)), 1))
@@ -423,7 +442,7 @@ def center_and_scale_gcode(gcode: str) -> str:
     """Center the drawing on the workplane and scale to fill 80% of it."""
     lines = gcode.split("\n")
     
-    # Extract all coordinates
+    # Extract all valid coordinates (filter outliers)
     coords = []
     for line in lines:
         x_match = re.search(r"X([-\d.]+)", line, re.IGNORECASE)
@@ -432,7 +451,9 @@ def center_and_scale_gcode(gcode: str) -> str:
             try:
                 x = float(x_match.group(1))
                 y = float(y_match.group(1))
-                coords.append((x, y))
+                # Only include reasonable coordinates
+                if -1000 < x < 1000 and -1000 < y < 1000:
+                    coords.append((x, y))
             except ValueError:
                 pass
     
@@ -731,10 +752,10 @@ def generate(prompt: str, temperature: float, max_tokens: int, num_steps: int, g
                     next_logits[:, pad_id] = float('-inf')
                 next_logits[:, 1] = float('-inf')  # <unk>
                 
-                # Repetition penalty
+                # Repetition penalty - stronger to prevent garbage
                 if recent_tokens:
-                    for token_id in set(recent_tokens[-30:]):
-                        next_logits[:, token_id] *= 0.7
+                    for token_id in set(recent_tokens[-50:]):
+                        next_logits[:, token_id] *= 0.5  # Stronger penalty
                 
                 # Top-k + Top-p sampling
                 top_k = 50