Using substring containment (p in f) risks incorrect matches (e.g., language code 'en' matching an unrelated path segment). Parse the path segments and require an exact directory match after doc/distrib/html to avoid accidental misclassification. Use something like:

parts = f.split('/')
if len(parts) > 3:
candidate = parts[3]
if candidate in lang_dirs:
lang_match = candidate
else:
print(f"No exact language directory match for {f}")
continue

Deletions or renames in doc/distrib/html are not propagated; stale files under src/ will persist. Consider comparing the set of expected html files per language and removing (or flagging) files in src/ that no longer exist at source. Add logic to detect removed originals and either delete or open an issue/PR note.

Using git add . may include unrelated files (e.g., transient changes, generated artifacts) leading to noisy automated PRs. Restrict the add to only the mapped destination files (tracked during copy) or to patterns (e.g., git add src/**/*.html) to minimize risk.