docling-project / docling-project/docling-serve

Migration Bug: Page breaks md_page_break_placeholder in PPTX/XLSX files break after migrating from v1alpha to v1 (v1.5.1 → v1.9.x)

Open
#449 2 comments 0 reactions 0 assignees View on GitHub
Dominant language
Python
Stars
1.8k
Forks
340
Avg merge
6d 11h
Merged PRs (30d)
8

Description

### Description
After migrating documents from `v1alpha` to `v1` (specifically from v1.5.1 to v1.9.x), page breaks in `.pptx` (PowerPoint) and `.xlsx` (Excel) files are no longer rendered correctly. This causes critical layout issues, such as:
- Missing page breaks in Excel sheets.
- Incorrect slide/page divisions in PowerPoint presentations.
- Content overflow or misalignment in rendered markdown output.

### Steps to Reproduce
1. Create a `.pptx` or `.xlsx` file in `v1alpha` with explicit page breaks (e.g., manual page breaks in Excel, section breaks in PowerPoint).
2. Migrate the file to `v1` using the official [v1_migration.md](https://github.com/docling-project/docling-serve/blob/main/docs/v1_migration.md).
3. Set the `md_page_break_placeholder` configuration.
4. Check the docling output `md_content`:
- **For `.xlsx`**: Page breaks are mislocated across multiple sheets; content of the second sheet runs into the first sheet.
- **For `.pptx`**: Page breaks are missing or mislocated; slide boundaries are ignored.

### Expected Behaviour
- Page breaks should be preserved and functional after migration.
- Excel sheets should paginate exactly as in `v1alpha`.
- PowerPoint slides should retain their original structure in the markdown output.

### Actual Behaviour
- **Excel (`.xlsx`)**:
- Manual page breaks are mislocated.
- In rendered markdown, content of the second sheet runs into the first sheet.
- **PowerPoint (`.pptx`)**:
- Slide boundaries are ignored; content spills across slides.
- Page breaks are missing or incorrectly positioned in the markdown output.

### Environment
- **Source Version**: `v1alpha`
- **Target Version**: `v1` (v1.5.1 → v1.9.x)
- **File Formats Affected**: `.pptx`, `.xlsx`
- **Other Formats**: `.pdf` works correctly (no issues observed).
- **Configuration**: `md_page_break_placeholder` is set.

### Additional Context
- The issue occurs **only** for `.pptx` and `.xlsx` files.
- Migration logs show no errors, suggesting silent data loss.
- **Workaround**: Using earlier versions of `v1` (e.g., `v1.3.x`, `v1.4.x` or `v1.5.0`) resolves the issue.
- The problem is specifically observed in the `md_content` output after migration.

**Labels:**
`bug`, `migration`, `v1`, `pptx`, `xlsx`, `md_page_break_placeholder`

Contributor guide

Open the contributing guide

Research direction

Start with docs/v1_migration.md and trace how md_page_break_placeholder and md_content are handled for migrated .pptx and .xlsx files between the reported v1 versions. Reproduce the issue with files containing explicit page or slide breaks, then compare the output with v1alpha or the working v1.3.x–v1.5.0 versions; done means sheet and slide boundaries are preserved in markdown.

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
35/100

Get new issues in your inbox

A short digest of beginner-friendly GitHub issues.