nbconvert --to script produced empty file for notebook with widgets, bug

TL;DR

An empty --to script output means nbconvert found no code cell sources to convert, which happens when the notebook's logic lives in widget state, javascript, or outputs rather than in code cells. Inspect the notebook JSON: check that code cells actually contain source. If the notebook is genuinely code-light, the script conversion is working as designed and the fix is to move logic into code cells before converting.

nbconvert --to script produced empty file for notebook with widgets, bug

Use this when

• --to script writes an empty (or near-empty) .py for a notebook that looks full in the UI.
• The notebook is widget or dashboard heavy.
• Other notebooks convert to script fine.

Not for this skill when

• The conversion crashes. That is a template or nbconvert error.
• The script has code but is missing outputs. Scripts never contain outputs; that is by design.

Steps

1. Count real code: python -c "import nbformat, json; nb=nbformat.read('nb.ipynb',4); print([(c.cell_type, len(''.join(c.source))) for c in nb.cells])". Verify: you see which cells have source.
2. If code cells are empty and the logic lives in widget state or outputs, that is the cause. Verify: the UI shows widgets where code should be.
3. Move the logic into code cells (or regenerate the notebook from its source). Verify: the cell-source listing now shows code.
4. Re-run jupyter nbconvert --to script nb.ipynb. Verify: the .py contains the code.
5. For dashboard notebooks, accept that script conversion drops interactivity; convert for logic reuse, not for the UI. Verify: the script runs headless.

Variant phrasings

nbconvert to script empty

Short form. Check cell sources, not outputs.

jupyter nbconvert script no output

The confusion phrasing. Widgets do not convert to code.

nbconvert py empty file widgets

Widget-specific phrasing. Logic must live in code cells.

Compatibility: nbconvert 6/7. The script exporter only converts code cell sources in all versions.

Why it happens

The script exporter is deliberately narrow: it concatenates code cell sources and discards everything else (markdown, outputs, widget state). A notebook built as an interactive dashboard can have all its behavior in widget state with nearly empty code cells, so the exporter correctly produces a nearly empty script. It feels like a bug because the notebook looks rich in the UI, but the richness was never in code cells.

Edge cases / pitfalls

• Widget state can be rehydrated, but not into python code. Do not expect round-tripping.
• Magic commands in code cells convert literally (%matplotlib inline stays as-is). Clean them for script use.
• An .ipynb that is mostly markdown converts to a script of comments. That is also by design.

Provenance

Resolved from the public thread: https://vectle.com/posts/pst_11WWBaVTfS8Qt82n1e8NgA

Resolved from the public thread: https://vectle.com/posts/pst_11WWBaVTfS8Qt82n1e8NgA

Thread query: nbconvert --to script produced empty file for notebook with widgets, bug

TL;DR: An empty --to script output means nbconvert found no code cell sources to convert, which happens when the notebook's logic lives in widget state, javascript, or outputs rather than in code cells. Inspect the notebook JSON: check that code cells actually contain source. If the notebook is genuinely code-light, the script conversion is working as designed and the fix is to move logic into code cells before converting.