fix: chart aggregation Function code + SemanticModel .platform path

[13w13]

Summary

Two bug fixes for issues encountered when building dashboards with multiple aggregation types.

1. Chart aggregation Function code (chart.py)

y_axis_var_aggregation_type parameter was accepted but ignored — "Function": 0 (Sum) was hardcoded in both the Y-axis projection and the sort definition. All charts defaulted to Sum regardless of the aggregation type passed.

Fix: Added _AGGREGATION_FUNCTION_CODES lookup dict mapping aggregation type strings to their PBI SQExpr Function codes, and replaced both hardcoded 0 values.

_AGGREGATION_FUNCTION_CODES = {
    "Sum": 0, "Min": 1, "Max": 2, "Count": 3,
    "LongCount": 4, "DistinctCount": 5, "CountDistinct": 5,
    "Average": 6, "Avg": 6,
}

Backwards compatible — defaults to 0 (Sum) if an unknown aggregation type is passed.

2. SemanticModel .platform displayName (dashboard.py)

Dashboard.create() used self.platform_file_path (Report .platform) instead of self.sm_platform_file_path when updating the SemanticModel .platform file. This left the SemanticModel displayName as "blank_template", which can cause Power BI Desktop errors when the folder name doesn't match.

Fix: Changed to self.sm_platform_file_path and used a separate variable name to avoid confusion.

Testing

Tested with a multi-page dashboard using 4 CSV datasets and DAX measures. Charts now correctly render with DistinctCount, Count, and Average aggregations.

[13w13]

Hi Russell! I'm 13w — I work in data/information management and recently started using powerbpy to generate Power BI dashboards programmatically. Really cool project, exactly what I was looking for.

I've been building multi-page dashboards with various aggregation types and ran into these two bugs, so I figured I'd contribute the fixes back. I've forked the repo and plan to keep using and developing on my side — I code with LLM assistance (not a developer by trade), so I can iterate pretty fast on features and fixes.

Happy to submit more PRs for other issues I've spotted (lazy imports, f-string fix, encoding). Let me know if you have any preferences on how you'd like contributions structured!

[Russell-Shean]

Hi 13w,

Glad you found the project and have been using it! It's exciting to see the tool being used in the real world! I'm currently traveling, so it may be a few more days before I'm able to approve the pull requests. Please contribute any bug fixes and new features back to the main branch, that would be super appreciated!

I'm curious to learn more about how you've been using the project and your development process. If you are open to I'd be happy to schedule a quick zoom call to chat. I can also talk a bit more about my plans for project. I think using AI is great to speed up development, but I think it will require me to develop more robust testing for this project, (something I should do anyways 😅 ). I think it might be helpful to talk a bit about that before I start integrating significant amounts of llm generated code.

If you preferred to maintain anonymity, no worries; I'll try to put all of that in writing too. I just thought at this stage, without a ton of contributors, a chat might be quicker!

[13w13]

Hey, happy traveling! Sure, let's have a quick zoom call :) Let's plan by email: contact.talon651@passinbox.com
cheers