Graphed Excel

Python script to visualize dependencies between cells in Excel spreadsheets.

Meant as a tool to visualize and understand the complexity of Excel spreadsheets.

Will generate a graph of the dependencies between cells in an Excel spreadsheet. Data extracted with openpyxl (https://foss.heptapod.net/openpyxl/openpyxl), the graph is generated with the networkx library (https://networkx.org/) and is visualized using matplotlib.

This is a simple tool and maybe even naïve in its approach - it was hacked together in two evenings and would benefit from some refactoring and more features. It is meant as a starting point for further development.

Definitions

Single-cell references in a formula sitting in cell A3 like =A1+A2 is considered a dependency between the node A3 and the nodes A2 and A1.

Loading

graph TD
    A1 --> A3
    A2 --> A3
    A3["=A1 + A2"]

A range defined in a formula like =SUM(B1:B3) is kept as a single node in the graph, but all the containing cells are expanded as dependencies of the range node.

So when a cell, C1 contains =SUM(B1:B3) the graph will look like this:

Loading

graph TD
    R -->B1
    R -->B2
    R -->B3
    R["B1:B3"]
    C1 --> R

    C1["C1=SUM(B1:B3)"]

The way the graph is built is by iterating over all cells in the spreadsheet and extracting the references in the formula of each cell. The references are then added as edges in the graph.

A cell within a range is considered a dependency of the range itself, but not of the other cells in the range.

Installation

python -m venv venv
source venv/bin/activate
pip install -e .

Usage

python -m graphedexcel <path_to_excel_file> [--verbose] [--no-visualize] [--keep-direction] [--open-image]

Depending on the size of the spreadsheet you might want to adjust the plot configuration in the code to to make the graph more readable (remove labels, decrease widths and sizes etc)

In graph_visualizer.py you will find three configuration for small, medium and large graphs. You can adjust the configuration to your needs.

Arguments

--verbose will dump formula cell contents during (more quiet)

--no-visualize will skip the visualization step and only print the summary (faster)

--keep-direction will keep the direction of the graph as it is in the excel file, otherwise it will be simplified to an undirected graph (slower)

--open-image will open the generated image in the default image viewer (only on Windows)

Sample output

The following is the output of running the script on the provided docs/Book1.xlsx file.

===  Dependency Graph Summary ===
Cell/Node count                70
Dependency count              100


===  Most connected nodes     ===
Range Madness!A2:A11           22
Range Madness!B2:B11           11
Range Madness!F1               10
Main Sheet!B5                   4
Main Sheet!B22                  4
Detached !A2:A4                 4
Range Madness!B2                4
Range Madness!B3                4
Range Madness!B4                4
Range Madness!B5                4

===  Most used functions      ===
SUM                             4
POWER                           1

Visualizing the graph of dependencies.
This might take a while...

Graph visualization saved to images/.\Book1.xlsx.png

Sample plot

More in /images folder.

Tests

pytest test_cell_reference_extraction.py

Contribute

Feel free to contribute by opening an issue or a pull request.

You can help with the following, that I have thought of so far:

• Add more tests
• Improve the code
• Add more features
• Improve the visualization and the ease of configuration
• Add more examples
• Add more documentation
• Package the script for easier installation and use with PyPi