ATH-MaaS/Pixelle-Video · error · Exception

Video analysis failed: {error_msg}

Error message

Video analysis failed: {error_msg}

What it means

After executing the video-understanding workflow via kit.execute, the analyzer checks result.status and raises a generic Exception if it is not "completed", carrying the workflow's own message. This surfaces upstream workflow/kit failures (API errors, content moderation, workflow misconfiguration) as a Python exception.

Source

Thrown at pixelle_video/services/video_analysis.py:155

            kit = await self.core._get_or_create_comfykit()
            
            # Determine what to pass to ComfyKit based on source
            if workflow_info["source"] == "runninghub" and "workflow_id" in workflow_info:
                # RunningHub: pass workflow_id
                workflow_input = workflow_info["workflow_id"]
                logger.info(f"Executing RunningHub workflow: {workflow_input}")
            else:
                # Selfhost: pass file path
                workflow_input = workflow_info["path"]
                logger.info(f"Executing selfhost workflow: {workflow_input}")
            
            result = await kit.execute(workflow_input, workflow_params)
            
            # 6. Extract description from result
            if result.status != "completed":
                error_msg = result.msg or "Unknown error"
                logger.error(f"Video analysis failed: {error_msg}")
                raise Exception(f"Video analysis failed: {error_msg}")
            
            # Extract text description from result
            # Video understanding workflow returns text in result.texts array
            description = None
            
            # Format 1: Direct texts array (most common for video understanding)
            if result.texts and len(result.texts) > 0:
                description = result.texts[0]
                logger.debug(f"Found description in result.texts: {description[:100]}...")
            
            # Format 2: Selfhost outputs (direct text in outputs)
            # Format: {'6': {'text': ['description text']}}
            elif result.outputs:
                for node_id, node_output in result.outputs.items():
                    if 'text' in node_output:
                        text_list = node_output['text']
                        if text_list and len(text_list) > 0:
                            description = text_list[0]

View on GitHub (pinned to 848b054e4f)

Solutions

  1. Read error_msg in the exception — it is the workflow backend's own message
  2. Verify workflow service credentials and that the workflow named by resolve_workflow_path loads correctly
  3. Retry on transient statuses if the message indicates timeout/rate limits
  4. Validate the video meets backend limits (size, duration, codec) before calling

Example fix

// before
result = await kit.execute(workflow_input, workflow_params)
description = result.texts[0]  # may raise or be empty
// after
result = await kit.execute(workflow_input, workflow_params)
if result.status != 'completed':
    logger.error(result.msg)
    raise RuntimeError(f'workflow failed: {result.msg}')
Defensive patterns

Strategy: try-catch

Validate before calling

def validate_workflow_inputs(video_path: str, workflow) -> None:
    assert Path(video_path).is_file(), 'video must exist'
    assert workflow is None or Path(workflow).is_file(), 'workflow json must exist'

# plus pre-call credential check
assert os.getenv('WORKFLOW_API_KEY'), 'workflow credentials missing'

Try / catch

result = None
for attempt in range(3):
    try:
        description = await analyzer(video_path=vp)
        break
    except Exception as e:
        logger.warning(f'analysis attempt {attempt} failed: {e}')
        if attempt == 2:
            raise

Prevention

When it happens

Trigger: kit.execute returns a result whose status != 'completed' with result.msg describing why — e.g. invalid API credentials for the workflow backend, workflow JSON resolving to a broken graph, input video rejected/too large, or the workflow runner timing out.

Common situations: Expired or missing workflow-service API keys; the analyse_video workflow JSON edited/broken; video exceeding platform limits (duration/size); transient service outages.

Related errors

AI-assisted analysis of ATH-MaaS/Pixelle-Video@848b054e4f (2026-08-30). Data as JSON: /api/errors/ede7f5b0e033be59. Report an issue: GitHub.