Strip unsupported symbols from captions

movmaker.py

@@ -547,6 +547,25 @@ def parse_data_file(path: Path | None) -> MovieData:
     return data
 
 
+def strip_unsupported_caption_chars(text: str) -> str:
+    """Remove characters that bundled caption fonts cannot render reliably.
+
+    FFmpeg drawtext does not provide dependable emoji fallback for our bundled
+    caption fonts, so emoji/symbol glyphs become unknown-character boxes in the
+    final video. Keep normal letters, numbers, punctuation, spacing, and marks;
+    drop symbols such as emoji.
+    """
+    cleaned = []
+    for ch in text:
+        cat = unicodedata.category(ch)
+        if cat.startswith("C"):
+            continue
+        if cat.startswith("S"):
+            continue
+        cleaned.append(ch)
+    return re.sub(r"\s+", " ", "".join(cleaned)).strip()
+
+
 def drawtext_escape(text: str) -> str:
     # Escape for ffmpeg drawtext text=... without surrounding quotes.
     return (
@@ -842,7 +861,11 @@ def build_video(
             f"if(lt(t\\,{start + caption_fade:.3f})\\,(t-{start:.3f})/{caption_fade:.3f}\\,"
             f"if(lt(t\\,{end - caption_fade:.3f})\\,1\\,({end:.3f}-t)/{caption_fade:.3f}))"
         ) if caption_fade > 0 else "1"
-        caption_lines = [line.strip() for line in caption.split("|") if line.strip()]
+        caption_lines = []
+        for line in caption.split("|"):
+            clean_line = strip_unsupported_caption_chars(line.strip())
+            if clean_line:
+                caption_lines.append(clean_line)
         line_gap = int(caption_size * 1.2)
         block_height = line_gap * (len(caption_lines) - 1)
         for line_idx, caption_line in enumerate(caption_lines):