Refine texture metrics and update default weights for premium pattern detection

tests/test_i18n.py

@@ -0,0 +1,73 @@
+import pytest
+from pathlib import Path
+import re
+
+# Base directory for language files
+LANG_DIR = Path(__file__).resolve().parent.parent / "app" / "lang"
+
+def get_structure(file_path: Path):
+    """
+    Returns a list of (line_number, key/header) for a TOML file.
+    Only captures keys and section headers, ignoring the values.
+    """
+    structure = []
+    # Regex to capture "key" = or [header]
+    key_pattern = re.compile(r'^\s*"?([^"\s=]+)"?\s*=')
+    header_pattern = re.compile(r'^\s*\[([^\]]+)\]')
+    
+    with open(file_path, "r", encoding="utf-8") as f:
+        for i, line in enumerate(f, 1):
+            line = line.strip()
+            if not line:
+                structure.append((i, "<empty>"))
+                continue
+            
+            # Check for header [section]
+            header_match = header_pattern.match(line)
+            if header_match:
+                structure.append((i, f"[{header_match.group(1)}]"))
+                continue
+                
+            # Check for key "name" =
+            key_match = key_pattern.match(line)
+            if key_match:
+                structure.append((i, key_match.group(1)))
+                continue
+                
+            # Comments or anything else
+            structure.append((i, "<other/comment>"))
+            
+    return structure
+
+def test_i18n_files_exist():
+    assert LANG_DIR.exists(), f"Language directory {LANG_DIR} not found"
+    en_file = LANG_DIR / "en.toml"
+    assert en_file.exists(), "English language file (en.toml) must exist as baseline"
+
+def test_i18n_synchronization():
+    """
+    Ensures all language files have the same keys/headers on the same lines
+    as the baseline en.toml.
+    """
+    en_path = LANG_DIR / "en.toml"
+    en_structure = get_structure(en_path)
+    
+    other_files = list(LANG_DIR.glob("*.toml"))
+    other_files.remove(en_path)
+    
+    for lang_file in other_files:
+        lang_name = lang_file.name
+        lang_structure = get_structure(lang_file)
+        
+        # Check line count
+        assert len(lang_structure) == len(en_structure), \
+            f"{lang_name} length mismatch: expected {len(en_structure)} lines, got {len(lang_structure)}"
+            
+        # Check line-by-line sync
+        for (en_line, en_key), (lang_line, lang_key) in zip(en_structure, lang_structure):
+            assert en_key == lang_key, \
+                f"Sync error at {lang_name}:{lang_line}. Expected '{en_key}', found '{lang_key}'"
+
+if __name__ == "__main__":
+    # Allow running directly as a script
+    pytest.main([__file__])