chore: fix gitignore — remove mt5_batch_config.json from tracking

set_comparator.py

@@ -4,19 +4,38 @@ from pathlib import Path
 
 def parse_set_file(file_bytes, filename):
     """Parse a .set file and return dict of {param: value} and ordered param list"""
-    text      = file_bytes.decode('utf-16')
+    # MT5 .set files can be UTF-16 LE (with BOM) or plain UTF-8 depending on MT5 version
+    if file_bytes[:2] == b'\xff\xfe':
+        text = file_bytes[2:].decode('utf-16-le')
+    elif file_bytes[:2] == b'\xfe\xff':
+        text = file_bytes[2:].decode('utf-16-be')
+    else:
+        # No BOM — try UTF-8 first, fall back to latin-1
+        try:
+            text = file_bytes.decode('utf-8')
+        except Exception:
+            text = file_bytes.decode('latin-1')
     params    = {}
     raw_lines = {}  # preserve full line for export
     order     = []
 
+    # Detect format: Ultimate Breakout .set files use || delimiter
+    # Other EA formats (plain key=value with comma-suffixed metadata) are not fully supported
+    is_ubs_format = any('||' in line for line in text.splitlines() if '=' in line)
+
+    all_raw_lines = []  # all lines in order for round-trip export
     for line in text.splitlines():
-        line = line.strip()
-        if not line or line.startswith(';'):
+        stripped = line.strip()
+        all_raw_lines.append(line)
+        if not stripped or stripped.startswith(';'):
             continue
-        if '=' not in line:
+        if '=' not in stripped:
             continue
-        key, _, rest = line.partition('=')
+        key, _, rest = stripped.partition('=')
         key = key.strip()
+        # Skip optimisation metadata lines (ParamName,F / ParamName,1 etc.)
+        if ',' in key:
+            continue
         # Value is first field before ||
         parts = rest.split('||')
         value = parts[0].strip()
@@ -24,7 +43,10 @@ def parse_set_file(file_bytes, filename):
         raw_lines[key] = rest  # preserve everything after =
         order.append(key)
 
-    return params, raw_lines, order
+    # Store full original lines for round-trip (preserves metadata lines)
+    raw_lines['__all_lines__'] = all_raw_lines
+
+    return params, raw_lines, order, is_ubs_format
 
 def build_comparison_df(files_data):
     """
@@ -58,7 +80,15 @@ def export_set_file(filename, params_edited, raw_lines, order, original_bytes):
     Returns bytes (utf-16 encoded)
     """
     # Get original header comments
-    text  = original_bytes.decode('utf-16')
+    if original_bytes[:2] == b'\xff\xfe':
+        text = original_bytes[2:].decode('utf-16-le')
+    elif original_bytes[:2] == b'\xfe\xff':
+        text = original_bytes[2:].decode('utf-16-be')
+    else:
+        try:
+            text = original_bytes.decode('utf-8')
+        except Exception:
+            text = original_bytes.decode('latin-1')
     lines = text.splitlines()
     header_lines = []
     for line in lines:
@@ -67,19 +97,43 @@ def export_set_file(filename, params_edited, raw_lines, order, original_bytes):
         else:
             break
 
-    output_lines = header_lines.copy()
-
-    for key in order:
-        if key not in params_edited:
-            continue
-        new_value = params_edited[key]
-        rest      = raw_lines.get(key, new_value)
-        parts     = rest.split('||')
-        parts[0]  = str(new_value)
-        output_lines.append(f"{key}={'||'.join(parts)}")
-
-    content = '\r\n'.join(output_lines) + '\r\n'
-    return content.encode('utf-16')
+    # Use the full original line list for round-trip, only updating real param values
+    all_lines = raw_lines.get('__all_lines__', [])
+    if all_lines:
+        output_lines = []
+        for line in all_lines:
+            stripped = line.strip()
+            if not stripped or stripped.startswith(';') or '=' not in stripped:
+                output_lines.append(line)
+                continue
+            key, _, rest = stripped.partition('=')
+            key = key.strip()
+            if ',' in key or key not in params_edited:
+                # Metadata line or unknown — preserve as-is
+                output_lines.append(line)
+            else:
+                new_value = params_edited[key]
+                parts     = rest.split('||')
+                parts[0]  = str(new_value)
+                output_lines.append(f"{key}={'||'.join(parts)}")
+        content = '\r\n'.join(output_lines) + '\r\n'
+    else:
+        # Fallback: rebuild from order (original behaviour)
+        output_lines = header_lines.copy()
+        for key in order:
+            if key not in params_edited:
+                continue
+            new_value = params_edited[key]
+            rest      = raw_lines.get(key, new_value)
+            parts     = rest.split('||')
+            parts[0]  = str(new_value)
+            output_lines.append(f"{key}={'||'.join(parts)}")
+        content = '\r\n'.join(output_lines) + '\r\n'
+    # Preserve original encoding on export
+    if original_bytes[:2] in (b'\xff\xfe', b'\xfe\xff'):
+        return b'\xff\xfe' + content.encode('utf-16-le')
+    else:
+        return content.encode('utf-8')
 
 def create_zip(files_export):
     """