nbctl extract¶

Extract outputs (images, graphs, data) from notebook cells.

Description¶

The extract command extracts all outputs from your Jupyter notebook and saves them as separate files. This includes plots, images, data tables, CSV files, JSON data, and text outputs.

Use this command to: - Save visualizations for reports or publications - Extract data tables and analysis results - Archive notebook outputs separately - Share results without sharing the notebook - Include figures in documentation

Usage¶

nbctl extract NOTEBOOK [OPTIONS]

Arguments¶

Argument	Description	Required
`NOTEBOOK`	Path to the Jupyter notebook file	Yes

Options¶

Option	Short	Type	Default	Description
`--output`	`-o`	PATH	`outputs/`	Output directory for extracted files
`--data`		Flag	False	Extract only data outputs
`--images`		Flag	False	Extract only image outputs
`--all`		Flag	False	Extract all without prompting

Output Structure¶

Extracted files are organized into subdirectories:

outputs/
├── data/
│   ├── cell_0_output_0_data_0.json
│   ├── cell_1_output_0_data_1.html      # DataFrame
│   ├── cell_2_output_0_data_2.csv
│   └── cell_3_output_0_data_3.txt
└── images/
    ├── cell_4_output_0_img_0.png        # Matplotlib plot
    ├── cell_5_output_0_img_1.svg        # Vector graphic
    └── cell_6_output_0_img_2.jpeg

File Naming Convention¶

Format: cell_{cell_idx}_output_{output_idx}_{type}_{counter}.{ext}

cell_idx: Cell number in notebook (0-indexed)
output_idx: Output number within cell (0-indexed)
type: data or img
counter: Sequential counter for type
ext: File extension based on content type

Extracted Types¶

Image Outputs¶

Format	Extension	Description
PNG	`.png`	Raster images, plots
JPEG	`.jpeg`	Compressed images
SVG	`.svg`	Vector graphics, scalable plots

Common sources: - Matplotlib/Seaborn plots - PIL/Pillow images - Plotly visualizations - Image display outputs

Data Outputs¶

Format	Extension	Description
JSON	`.json`	JSON data structures
HTML	`.html`	HTML tables (pandas DataFrames)
CSV	`.csv`	Comma-separated values
Text	`.txt`	Plain text outputs

Common sources: - Pandas DataFrame displays - JSON API responses - Data summaries - Model evaluation metrics

Interactive Mode¶

When run without --data, --images, or --all, the command prompts:

What would you like to extract?
1. Both data and images
2. Only data outputs
3. Only image outputs
4. All outputs without prompting

Enter choice (1-4):

Output Summary¶

Extraction Summary:

Images extracted:
cell_4_output_0_img_0.png
cell_5_output_0_img_1.svg
cell_6_output_0_img_2.jpeg

Data extracted:
cell_0_output_0_data_0.json
cell_1_output_0_data_1.html
cell_2_output_0_data_2.csv

Total: 3 images, 3 data files
Saved to: outputs/

Exit Codes¶

Code	Meaning
0	Success, outputs extracted
1	File not found or invalid notebook
2	No outputs to extract
3	Output directory creation failed

Notes¶

Read-only operation: Never modifies the source notebook
Base64 decoding: Automatically decodes embedded images
DataFrame handling: Saves pandas DataFrames as HTML tables
Overwrites existing: Files in output directory may be overwritten
Empty cells skipped: Cells without outputs are ignored

Common Use Cases¶

Extract All Plots¶

nbctl extract analysis.ipynb --images

Extract Data Tables¶

nbctl extract results.ipynb --data

Custom Output Location¶

nbctl extract notebook.ipynb -o ./figures/

Extract Everything¶

nbctl extract notebook.ipynb --all

export - Export entire notebook to different formats
run - Execute notebook to generate outputs
clean - Remove outputs from notebook