truncate forum topics over 255 characters

archive.py

@@ -80,6 +80,28 @@ def http_get_json(site_url: str, path: str) -> dict:
         log.warning("Unable to decode JSON response from %r", path)
         raise
 
+# ----- Helper: Truncate Filename -----
+def truncate_filename(filename: str, max_length: int = 255) -> str:
+    """
+    Truncates the file name to a maximum length (default 255 characters).
+    It preserves the file extension.
+    """
+    if len(filename) <= max_length:
+        return filename
+
+    # Split into stem and suffix(s)
+    p = Path(filename)
+    stem = p.stem
+    suffix = "".join(p.suffixes)
+    # Calculate the maximum allowed length for the stem after accounting for the suffix
+    max_stem_length = max_length - len(suffix)
+    if max_stem_length <= 0:
+        # In the unlikely event that the suffix itself is longer than max_length,
+        # simply return a truncated version of the entire filename.
+        return filename[:max_length]
+    truncated_stem = stem[:max_stem_length]
+    return truncated_stem + suffix
+
 # ----- Data Models -----
 
 @dataclass(frozen=True)
@@ -101,6 +123,8 @@ class Post:
         """Save the raw JSON post to disk if not already archived."""
         idstr = str(self.id).zfill(10)
         filename = f"{idstr}-{self.raw.get('username', 'anonymous')}-{self.raw.get('topic_slug', 'unknown')}.json"
+        # Truncate file name if necessary.
+        filename = truncate_filename(filename)
         folder_name = self.get_created_at().strftime('%Y-%m-%B')
         full_path = dir / folder_name / filename
 
@@ -141,9 +165,12 @@ class Topic:
         """
         Save the rendered Markdown topic to disk.
         Filename built from creation date, slug, and id.
+        Truncate the filename if it is too long for the operating system.
         """
         date_str = str(self.get_created_at().date())
         filename = f"{date_str}-{self.slug}-id{self.id}.md"
+        # Truncate if necessary
+        filename = truncate_filename(filename)
         folder_name = self.get_created_at().strftime('%Y-%m-%B')
         full_path = dir / folder_name / filename
         full_path.parent.mkdir(parents=True, exist_ok=True)