"page_numbers": [],Cannot extract the number of pages from the doc file.

[z-x-x136]

Question

...

{
"chunk_id": 3,
"text": "工艺模块",
"metadata": {
"source": "

预报工艺模块",
"page_numbers": [],
"headings": []
}

[dosubot]

Docling can't extract page numbers from DOC files because Word documents don't store page number metadata in their file format—so the provenance for page numbers is always empty for DOC/DOCX inputs. If page numbers are only present as images or visual elements (like a screenshot of "第 1 页 共 14 页"), Docling won't extract them either, since it doesn't run OCR on embedded images in DOC files and has no built-in pipeline for parsing visual pagination indicators from images source.

If you need page numbers for downstream tasks, the recommended workaround is to export or print the DOC file to PDF, then process the PDF with Docling. Docling can extract and preserve page numbers from PDFs in the provenance metadata source.

If you absolutely need to extract page numbers from images inside DOC files, you'll need to use an external OCR tool to process those images and manually associate the extracted page numbers with your document content source. Docling does not provide built-in support for this scenario.

To reply, just mention @dosu.

---

^{How did I do? Good | Irrelevant | Incorrect | Verbose | Hallucination | Report 🐛 | Other}

   

[Ritinikhil]

Hi @z-x-x136 , DOC/DOCX files don’t store page numbers as metadata, so Docling leaves page_numbers empty. If you need page numbers, please export/print the DOC to PDF and run Docling on the PDF—page numbers will be preserved there. If page numbers only exist as images inside the DOC, you’d need an OCR pass to extract them, which Docling doesn’t handle natively.