Fix normalize_json.py to format number arrays with multiple numbers per line

- Updated format_json_value to format number arrays with multiple numbers per line
- Numbers wrap at printWidth (150) to match Prettier style
- Matches the formatting style in book-sets.json

Author: vikahaze

data/book-sets.json

@@ -24,7 +24,6 @@
       338,
       345,
       374,
-      
       392,
       605,
       643,
@@ -629,6 +628,7 @@
       1768,
       1798,
       1920,
+      1925,
       1926,
       1929,
       1957,
@@ -657,11 +657,23 @@
       2888,
       2894,
       2942,
+      3100,
       3110,
+      3133,
+      3164,
       3190,
+      3197,
+      3228,
+      3291,
+      3320,
+      3351,
+      3380,
       3381,
+      3413,
       3424,
       3432,
+      3444,
+      3471,
       3512,
       3522,
       3602,
@@ -2130,7 +2142,6 @@
       1921,
       1922,
       1923,
-      1925,
       1927,
       1928,
       1930,
@@ -3018,7 +3029,6 @@
       3097,
       3098,
       3099,
-      3100,
       3101,
       3102,
       3105,
@@ -3042,7 +3052,6 @@
       3130,
       3131,
       3132,
-      3133,
       3134,
       3136,
       3137,
@@ -3066,7 +3075,6 @@
       3161,
       3162,
       3163,
-      3164,
       3165,
       3168,
       3169,
@@ -3090,7 +3098,6 @@
       3194,
       3195,
       3196,
-      3197,
       3200,
       3201,
       3202,
@@ -3114,7 +3121,6 @@
       3225,
       3226,
       3227,
-      3228,
       3229,
       3232,
       3233,
@@ -3162,7 +3168,6 @@
       3288,
       3289,
       3290,
-      3291,
       3292,
       3295,
       3296,
@@ -3186,7 +3191,6 @@
       3317,
       3318,
       3319,
-      3320,
       3321,
       3324,
       3325,
@@ -3210,7 +3214,6 @@
       3348,
       3349,
       3350,
-      3351,
       3352,
       3354,
       3355,
@@ -3234,7 +3237,6 @@
       3377,
       3378,
       3379,
-      3380,
       3382,
       3386,
       3387,
@@ -3258,7 +3260,6 @@
       3410,
       3411,
       3412,
-      3413,
       3414,
       3417,
       3418,
@@ -3282,7 +3283,6 @@
       3441,
       3442,
       3443,
-      3444,
       3445,
       3446,
       3447,
@@ -3306,7 +3306,6 @@
       3468,
       3469,
       3470,
-      3471,
       3472,
       3473,
       3474,

scripts/normalize_json.py

@@ -14,7 +14,7 @@
 from collections import OrderedDict
 
 
-def format_json_value(value, indent_level=0):
+def format_json_value(value, indent_level=0, print_width=150):
     """Format a JSON value with custom formatting following Prettier style."""
     indent = "  " * indent_level
     next_indent = "  " * (indent_level + 1)
@@ -25,19 +25,43 @@ def format_json_value(value, indent_level=0):
         items = []
         for k, v in sorted(value.items()):
             formatted_key = json.dumps(k, ensure_ascii=False)
-            formatted_value = format_json_value(v, indent_level + 1)
+            formatted_value = format_json_value(v, indent_level + 1, print_width)
             items.append(f"{next_indent}{formatted_key}: {formatted_value}")
         return "{\n" + ",\n".join(items) + "\n" + indent + "}"
     elif isinstance(value, list):
         if not value:
             return "[]"
-        # Check if it's a list of numbers (for problems arrays) - format one per line
+        # Check if it's a list of numbers (for problems arrays) - format multiple per line
         if value and isinstance(value[0], (int, float)):
-            items = [f"{next_indent}{json.dumps(item, ensure_ascii=False)}" for item in value]
-            return "[\n" + ",\n".join(items) + "\n" + indent + "]"
+            # Calculate available width (print_width minus indentation and brackets)
+            available_width = print_width - len(next_indent) - 2
+            lines = []
+            current_line = []
+            current_length = 0
+            
+            for i, item in enumerate(value):
+                item_str = str(item)
+                # Add comma and space length (2) if not first item on line
+                item_length = len(item_str) + (2 if current_line else 0)
+                
+                if current_length + item_length > available_width and current_line:
+                    # Start a new line
+                    lines.append(", ".join(str(x) for x in current_line))
+                    current_line = [item]
+                    current_length = len(item_str)
+                else:
+                    current_line.append(item)
+                    current_length += item_length
+            
+            if current_line:
+                lines.append(", ".join(str(x) for x in current_line))
+            
+            # Format with proper indentation
+            formatted_lines = [f"{next_indent}{line}" for line in lines]
+            return "[\n" + ",\n".join(formatted_lines) + "\n" + indent + "]"
         else:
             # Format other arrays - check if they fit on one line
-            formatted_items = [format_json_value(item, indent_level + 1) for item in value]
+            formatted_items = [format_json_value(item, indent_level + 1, print_width) for item in value]
             total_length = sum(len(str(item)) for item in formatted_items)
             if total_length < 100 and len(value) <= 5:
                 # Short arrays on one line
@@ -74,7 +98,7 @@ def sort_json_by_numeric_keys(input_file, output_file):
         items = []
         for key, value in sorted_data.items():
             formatted_key = json.dumps(key, ensure_ascii=False)
-            formatted_value = format_json_value(value, 1)
+            formatted_value = format_json_value(value, 1, print_width=150)
             items.append(f'  {formatted_key}: {formatted_value}')
         lines.append(",\n".join(items))
         lines.append("}")