Some jobs report UNSUPPORTED_FILE_TYPE Exception but by Status is Good According to API

[adreichert]

Describe the bug
I have a directory containing several hundred HTML files. When I parse them with theload_data in the Python llama_parse package, the code fails with the UNSUPPORTED_FILE_TYPE exception.

• Each time, it happens with different files
• If I check the status quickly with web UI within a few seconds, I see "failed" and the error. But If I refresh the page, it lists success. The results seems correct.
• One can also see successful job status using the API

lp = llama_parse.LlamaParse(
        api_key=lp,
        result_type='markdown',
        verbose=False,
        use_vendor_multimodal_model=True,
        vendor_multimodal_model_name='openai-gpt4o',
       [... other options..]
    )
lp.load_data(files)  <-- files is a list of paths


File ".../llama_parse/base.py", line 654, in _get_job_result
    raise Exception(exception_str)
Exception: Job ID: 3e3b9c18-8059-4c4d-bf84-a138d73e2208 failed with status: ERROR, Error code: UNSUPPORTED_FILE_TYPE, Error message: Unsupported file type.
[...]
% curl -X 'GET' \
  'https://api.cloud.llamaindex.ai/api/parsing/job/3e3b9c18-8059-4c4d-bf84-a138d73e2208' \
  -H 'accept: application/json' \
  -H "Authorization: Bearer $LLAMA_CLOUD_API_KEY"
{"id":"3e3b9c18-8059-4c4d-bf84-a138d73e2208","status":"SUCCESS"}%      

Job ID
3e3b9c18-8059-4c4d-bf84-a138d73e2208

Client:
Please remove untested options:

• Python Library

[adreichert]

One more thing. I parse these files a few days ago without issue. I don't believe the contents have changed, but it is possible.

[adreichert]

The workaround was to keep retrying while relying on cached results. I ran it a few more times. Each time, a different file would cause the error and then be marked as successful. After this, the process completed.

[edanweis]

any updates on this? I'm getting the same problem intermittently via REST api