fix(blocks): strip chapter metadata before MoviePy opens files

MoviePy 2.x crashes with IndexError when parsing video/audio files
with embedded chapter metadata (moviepy#2419). Add strip_chapters_inplace()
which runs `ffmpeg -map_chapters -1 -codec copy` to remove chapters
without re-encoding before any VideoFileClip/AudioFileClip call.

Applied to all video blocks: clip, concat, text_overlay, narration,
add_audio, loop, and duration.

Co-Authored-By: Claude Opus 4.5 <noreply@anthropic.com>

autogpt_platform/backend/backend/blocks/video/_utils.py

@@ -1,6 +1,12 @@
 """Shared utilities for video blocks."""
 
+from __future__ import annotations
+
+import logging
 import os
+import subprocess
+
+logger = logging.getLogger(__name__)
 
 
 def get_video_codecs(output_path: str) -> tuple[str, str]:
@@ -32,3 +38,46 @@ def get_video_codecs(output_path: str) -> tuple[str, str]:
     }
 
     return codec_map.get(ext, ("libx264", "aac"))
+
+
+def strip_chapters_inplace(video_path: str) -> None:
+    """Strip chapter metadata from a media file in-place using ffmpeg.
+
+    MoviePy 2.x crashes with IndexError when parsing files with embedded
+    chapter metadata (https://github.com/Zulko/moviepy/issues/2419).
+    This strips chapters without re-encoding.
+
+    Args:
+        video_path: Absolute path to the media file to strip chapters from.
+    """
+    base, ext = os.path.splitext(video_path)
+    tmp_path = base + ".tmp" + ext
+    try:
+        result = subprocess.run(
+            [
+                "ffmpeg",
+                "-y",
+                "-i",
+                video_path,
+                "-map_chapters",
+                "-1",
+                "-codec",
+                "copy",
+                tmp_path,
+            ],
+            capture_output=True,
+            text=True,
+        )
+        if result.returncode != 0:
+            logger.warning(
+                "ffmpeg chapter strip failed (rc=%d): %s",
+                result.returncode,
+                result.stderr,
+            )
+            return
+        os.replace(tmp_path, video_path)
+    except FileNotFoundError:
+        logger.warning("ffmpeg not found; skipping chapter strip")
+    finally:
+        if os.path.exists(tmp_path):
+            os.unlink(tmp_path)

autogpt_platform/backend/backend/blocks/video/add_audio.py

@@ -6,6 +6,7 @@ import tempfile
 from moviepy.audio.io.AudioFileClip import AudioFileClip
 from moviepy.video.io.VideoFileClip import VideoFileClip
 
+from backend.blocks.video._utils import strip_chapters_inplace
 from backend.data.block import (
     Block,
     BlockCategory,
@@ -76,6 +77,7 @@ class AddAudioToVideoBlock(Block):
         audio_abspath = os.path.join(abs_temp_dir, local_audio_path)
 
         # 2) Load video + audio with moviepy
+        strip_chapters_inplace(video_abspath)
         video_clip = VideoFileClip(video_abspath)
         audio_clip = AudioFileClip(audio_abspath)
         # Optionally scale volume

autogpt_platform/backend/backend/blocks/video/clip.py

@@ -5,7 +5,7 @@ from typing import Literal
 
 from moviepy.video.io.VideoFileClip import VideoFileClip
 
-from backend.blocks.video._utils import get_video_codecs
+from backend.blocks.video._utils import get_video_codecs, strip_chapters_inplace
 from backend.data.block import (
     Block,
     BlockCategory,
@@ -89,6 +89,7 @@ class VideoClipBlock(Block):
         clip = None
         subclip = None
         try:
+            strip_chapters_inplace(video_abspath)
             clip = VideoFileClip(video_abspath)
             subclip = clip.subclipped(start_time, end_time)
             video_codec, audio_codec = get_video_codecs(output_abspath)

autogpt_platform/backend/backend/blocks/video/concat.py

@@ -6,7 +6,7 @@ from moviepy import concatenate_videoclips
 from moviepy.video.fx import CrossFadeIn, CrossFadeOut, FadeIn, FadeOut
 from moviepy.video.io.VideoFileClip import VideoFileClip
 
-from backend.blocks.video._utils import get_video_codecs
+from backend.blocks.video._utils import get_video_codecs, strip_chapters_inplace
 from backend.data.block import (
     Block,
     BlockCategory,
@@ -53,8 +53,13 @@ class VideoConcatBlock(Block):
             categories={BlockCategory.MULTIMEDIA},
             input_schema=self.Input,
             output_schema=self.Output,
-            test_input={"videos": ["/tmp/a.mp4", "/tmp/b.mp4"]},
-            test_output=[("video_out", str), ("total_duration", float)],
+            test_input={
+                "videos": ["/tmp/a.mp4", "/tmp/b.mp4"],
+            },
+            test_output=[
+                ("video_out", str),
+                ("total_duration", float),
+            ],
             test_mock={
                 "_concat_videos": lambda *args: 20.0,
                 "_store_input_video": lambda *args, **kwargs: "test.mp4",
@@ -89,13 +94,18 @@ class VideoConcatBlock(Block):
         transition: str,
         transition_duration: int,
     ) -> float:
-        """Concatenate videos. Extracted for testability."""
+        """Concatenate videos. Extracted for testability.
+
+        Returns:
+            Total duration of the concatenated video.
+        """
         clips = []
         faded_clips = []
         final = None
         try:
             # Load clips
             for v in video_abspaths:
+                strip_chapters_inplace(v)
                 clips.append(VideoFileClip(v))
 
             if transition == "crossfade":

autogpt_platform/backend/backend/blocks/video/duration.py

@@ -3,6 +3,7 @@
 from moviepy.audio.io.AudioFileClip import AudioFileClip
 from moviepy.video.io.VideoFileClip import VideoFileClip
 
+from backend.blocks.video._utils import strip_chapters_inplace
 from backend.data.block import (
     Block,
     BlockCategory,
@@ -59,7 +60,8 @@ class MediaDurationBlock(Block):
             execution_context.graph_exec_id, local_media_path
         )
 
-        # 2) Load the clip
+        # 2) Strip chapters to avoid MoviePy crash, then load the clip
+        strip_chapters_inplace(media_abspath)
         if input_data.is_video:
             clip = VideoFileClip(media_abspath)
         else:

autogpt_platform/backend/backend/blocks/video/loop.py

@@ -6,6 +6,7 @@ from typing import Optional
 from moviepy.video.fx.Loop import Loop
 from moviepy.video.io.VideoFileClip import VideoFileClip
 
+from backend.blocks.video._utils import strip_chapters_inplace
 from backend.data.block import (
     Block,
     BlockCategory,
@@ -71,6 +72,7 @@ class LoopVideoBlock(Block):
         input_abspath = get_exec_file_path(graph_exec_id, local_video_path)
 
         # 2) Load the clip
+        strip_chapters_inplace(input_abspath)
         clip = VideoFileClip(input_abspath)
 
         # 3) Apply the loop effect

autogpt_platform/backend/backend/blocks/video/narration.py

@@ -14,7 +14,7 @@ from backend.blocks.elevenlabs._auth import (
     ElevenLabsCredentials,
     ElevenLabsCredentialsInput,
 )
-from backend.blocks.video._utils import get_video_codecs
+from backend.blocks.video._utils import get_video_codecs, strip_chapters_inplace
 from backend.data.block import (
     Block,
     BlockCategory,
@@ -150,6 +150,7 @@ class VideoNarrationBlock(Block):
         original = None
 
         try:
+            strip_chapters_inplace(video_abspath)
             video = VideoFileClip(video_abspath)
             narration_original = AudioFileClip(audio_abspath)
             narration_scaled = narration_original.with_volume_scaled(narration_volume)

autogpt_platform/backend/backend/blocks/video/text_overlay.py

@@ -6,7 +6,7 @@ from typing import Literal
 from moviepy import CompositeVideoClip, TextClip
 from moviepy.video.io.VideoFileClip import VideoFileClip
 
-from backend.blocks.video._utils import get_video_codecs
+from backend.blocks.video._utils import get_video_codecs, strip_chapters_inplace
 from backend.data.block import (
     Block,
     BlockCategory,
@@ -117,6 +117,7 @@ class VideoTextOverlayBlock(Block):
         final = None
         txt_clip = None
         try:
+            strip_chapters_inplace(video_abspath)
             video = VideoFileClip(video_abspath)
 
             txt_clip = TextClip(