EcoLogic

Code candidates and resource comparison

EcoLogic explores software-energy analysis through two related implementations: a Python/Streamlit optimizer and a React/TypeScript/D3 dashboard with FastAPI analysis services. The optimizer compares candidate refactors and reports measured runtime alongside modeled energy/carbon values. The dashboard emphasizes code-analysis views and telemetry presentation.

Both repositories remain discoverable under one canonical project identity, with their distinct analysis/presentation paths explicit.

Code to comparison

1. Source / workspace
2. Candidate refactors
3. Predict & profile
4. Compare candidates
5. Report

Code analysis, runtime evidence and energy-model output remain identifiable stages of the workflow.

What the project does

• Accept source text or workspace input such as a folder, ZIP or repository for analysis.
• Generate candidate changes through built-in heuristics and an optional assistant workflow.
• Score candidates through energy/runtime models and run measured profiling where the supported environment allows it.
• Present candidate rounds, Pareto comparisons, code context and downloadable reports.

Engineering decisions

Static analysis extracts source structure and features for candidate comparison. The FastAPI implementation includes AST analysis, with Tree-sitter/GraphSAGE-oriented work in the available code. Supported profiling runs supply timing information; feature-based models estimate resource impact.

Candidate ranking and Pareto views expose tradeoffs across dimensions. Original/refactored code can be compared through runtime and modeled estimates, which remain separately labeled. Streamlit reports and optional PDF export preserve a generated result report. React/D3 provides another interactive view of analysis data. The report is an output artifact, while the measured/model-derived fields communicate how each comparison was produced.