docling-project / docling-project/docling

Docling hangs with enrich formula and document timeout

Open
#2,478 5 comments 0 reactions 0 assignees View on GitHub
bug
Dominant language
Python
Stars
66.4k
Forks
4.8k
Avg merge
3d 4h
Merged PRs (30d)
95

Description

### Bug

Docling hangs with some PDF documents with `do_formula_enrichment=True` after the execution was aborted with by the document timeout:

```
2025-10-16 13:07:08,746 - INFO - Loading plugin 'docling_defaults'
2025-10-16 13:07:08,759 - INFO - Registered ocr engines: ['easyocr', 'ocrmac', 'rapidocr', 'tesserocr', 'tesseract']
2025-10-16 13:07:12,402 - INFO - paths: [WindowsPath('C:/Users/lb/AppData/Local/Temp/tmpzs2tfimf/1-s2.0-S0963869523002086-main.pdf')]
2025-10-16 13:07:12,404 - INFO - detected formats: []
2025-10-16 13:07:13,251 - INFO - Going to convert document batch...
2025-10-16 13:07:13,252 - INFO - Initializing pipeline for StandardPdfPipeline with options hash 16773cf0d1ad6514cb4bd739bcb17c35
2025-10-16 13:07:13,437 - INFO - Loading plugin 'docling_defaults'
2025-10-16 13:07:13,440 - INFO - Registered picture descriptions: ['vlm', 'api']
2025-10-16 13:07:13,761 - INFO - Accelerator device: 'cpu'
2025-10-16 13:07:15,603 - INFO - Accelerator device: 'cpu'
2025-10-16 13:07:17,409 - INFO - Accelerator device: 'cpu'
2025-10-16 13:07:17,812 - INFO - Accelerator device: 'cpu'
2025-10-16 13:07:19,348 - INFO - Processing document 1-s2.0-S0963869523002086-main.pdf
2025-10-16 13:08:31,509 - WARNING - Document processing time (72.160 seconds) exceeded the specified timeout of 60.000 seconds
2025-10-16 13:08:31,509 - INFO - Filtered out 1 uninitialized pages due to timeout or processing failures
-- HANGS HERE --
```

Example PDF: https://www.ndt.net/article/ndt-e-intl/papers/1-s2.0-S0963869523002086-main.pdf

### Steps to reproduce

```shell
docling --from pdf --to md --document-timeout 60 --enrich-formula -vv "https://www.ndt.net/article/ndt-e-intl/papers/1-s2.0-S0963869523002086-main.pdf"
```

### Docling version

Docling version: 2.53.0
Docling Core version: 2.48.2
Docling IBM Models version: 3.9.1
Docling Parse version: 4.5.0
Python: cpython-313 (3.13.7)
Platform: Windows-11-10.0.26100-SP0

### Python version

Python 3.13.7

Contributor guide

Open the contributing guide

Research direction

Start by reproducing the documented CLI command with the example PDF, --document-timeout 60, and --enrich-formula. Trace what happens after the timeout warning and the uninitialized page is filtered; done means the command exits without hanging while preserving timeout behavior.

Written by the indexing model from the issue text.

Assessment

Tech stack
python
Domain
backend
Issue type
Bug
Difficulty
4/5
Estimated time
3-5 days
Activity status
Stale
Clarity
Mostly clear
Newbie friendliness
38/100

Get new issues in your inbox

A short digest of beginner-friendly GitHub issues.