fix: count filename lengths in bytes instead of characters

config.default.toml

@@ -42,8 +42,8 @@ template = '''
 '''
 # Filename template for generated HTML files.
 filename_template = "{{ published_formatted }}\t[{{ origin_title }}]\t{{ title }}.html"
-# Maximum allowed filename length before truncating.
-max_filename_length = 200
+# Maximum allowed filename length (in bytes assuming UTF-8 encoding) before truncating. Depending on your filesystem filename's limits it may be possible to increase the value, ask Wikipedia for details.
+max_filename_length = 250
 
 [time]
 # Which timezone to use when writing date and time.

main.py

@@ -114,9 +114,14 @@ def escape_filename(filename):
 
 def truncate_filename(config, filename):
     max_filename_length = config.max_filename_length
-    suffix = Path(filename).suffix
-    max_basename_length = max_filename_length - len(suffix)
-    return filename[:max_basename_length] + '…' + suffix if len(filename) > max_filename_length else filename
+    filename_utf8 = filename.encode("utf-8")
+    if len(filename_utf8) <= max_filename_length:
+        return filename
+    else:
+        suffix = Path(filename).suffix
+        max_basename_length = max_filename_length - len(suffix.encode("utf-8"))
+        cutoff = len(filename.encode('utf-8')[:max_basename_length].decode('utf-8', errors="ignore"))
+        return filename[:cutoff] + '…' + suffix
 
 def get_html_path(config, item_json):
     folder_directory = config.html_root / escape_filename(item_json["folder"])
@@ -124,8 +129,7 @@ def get_html_path(config, item_json):
 
     html_name = truncate_filename(config, escape_filename(config.item_filename_template.render(item_json)))
 
-    html_path = folder_directory / html_name
-    return html_path
+    return folder_directory / html_name
 
 def format_datetime(config, timestamp):
     return datetime.fromtimestamp(timestamp, config.timezone).strftime(config.time_format)