Feature request: support mid_value in color_scale for Excel-like diverging gradients

[MarekOzana]

First of all, thank you for building this library. Development seems to be moving fast, and I really appreciate the work — I’ve been enjoying streamlit-pivot a lot.

I’d like to suggest a small enhancement to color_scale:

Could it support an optional numeric mid_value (for example 0)?

My main use case is a PnL column where I want a continuous diverging scale like in Excel:

• more negative values = darker red
• 0 = white
• more positive values = darker blue

What I’m looking for is not a threshold/banded rule, but a smooth gradient anchored at a meaningful midpoint.

Example:

rule = {
    "type": "color_scale",
    "apply_to": ["pnl"],
    "min_color": "#ff0000",
    "mid_color": "#ffffff",
    "max_color": "#0000ff",
    "mid_value": 0,
}

[mayagbarnes]

Hi @MarekOzana, thanks so much for your submission and for using this new streamlit component! 🎈

I added a commit (1e2edfc) - which hopefully addresses your request. I'll likely release another update this coming week.

Also, if you wouldn't mind sharing, I'd love to understand more about your use case(s):

• What are you building?
  • Is this a personal analysis tool, an internal dashboard for a team, or something you're shipping to end users?
• What does the underlying data look like?
  • Roughly how many rows/columns, what are the main dimensions and measures, and how often does it refresh?
• How are you/your audience interacting with it?
  • Mostly viewing pre-configured layouts, or actively dragging fields / filtering / drilling down?
• How did you discover streamlit-pivot, and what drew you to using it over the alternatives?
  • Are you transitioning from another framework?
• Any feature gaps or friction so far?
  • Anything you expected to work that didn't, features you've been working around, or things you'd love to see next?

No pressure to answer all of these — just trying to gather as much feedback as possible to help me understand where users are finding value and prioritize efforts.
Thanks again!

[MarekOzana]

Thanks again for implementing this feature — really appreciated.

To answer your questions:

What am I building?
An internal app for portfolio vs. index risk exposure analysis at an asset management company.

What does the data look like?
Position-level rows with measures such as DTS, duration, yield, monthly PnL, rating, currency, and contribution fields. Portfolio positions contribute positive values, while index positions contribute negative values, which gives a net exposure view. That is also why filtering by position type (Portfolio vs. Index) would be very helpful.

How do we use it?
Quite actively — dragging dimensions into rows/columns and selecting exposure fields to analyze. Typical dimensions are currency, rating, duration buckets, sectors, and issuer tickers. Typical values are duration contribution, DTS contribution, average return, and similar measures.

How did I discover streamlit-pivot?
I was missing pivot-table functionality in Streamlit and was otherwise falling back to Excel, so this library immediately caught my attention.

Feature gaps / ideas
A few things that would help a lot for this workflow:

• a larger and ideally resizable column selection field, since it feels a bit cramped now ([screenshot])
• filtering on a column/dimension, similar to Excel ([screenshot])
• support for placing values on rows as well, not only on columns ([same screenshot])
• a clear Apache 2.0 LICENSE file in the repo, not only a mention in the README

Overall, I think this is a really impressive product. The quality is already strong, and the speed of development has been remarkable. It’s clear that a lot of thought and effort is going into it, and I really appreciate that.
/Marek