GSoC'25 week 05 update by Om Santosh Suneri

[omsuneri]

Week 05 Progress Report by Om Santosh Suneri

Project: AI-powered Debugger for Music Blocks
Mentors: Walter Bender Sumit Srivastava
Assisting Mentors: Devin Ulibarri
Reporting Period: 2025-06-29 - 2025-07-05

---

Goal for This Week

Migrating Music Blocks JSON-to-Text Converter to Streamlit App

---

This Week’s Achievements

Introduction

This week, I focused on one of the key deliverables of the project — integrating the JSON-to-Text Representation Converter into a Streamlit application. This marks a major step in our plan to create a seamless debugging experience for Music Blocks users. The converter, which was originally built using JavaScript, is now fully functional in Python via Streamlit and ready for integration with the AI-powered debugger.

What I Did

Goal

The previous tool was a client-side JavaScript converter that parsed JSON representations of Music Blocks projects and produced a structured, readable tree-view text format. The aim this week was to translate this logic into Python and build a Streamlit interface to make the tool backend-friendly and easily integrable with the AI Debugger app.

Migration from JavaScript to Python

Converting the JavaScript-based logic to Python wasn’t a simple one-to-one translation. It involved rethinking data structures, managing recursion differently, and carefully ensuring that each Music Blocks "block" type was accurately represented in the output.

Key technical components of the migration included:

• Parsing the block structure:

  • Each block in the JSON is structured like [block_id, block_type, ..., connections]
  • The Python version uses dictionaries (block_map) and recursion to follow nested or sequential connections (clamp and stack logic).

• Handling specific block types:

  • Each block type (like start, setmasterbpm2, newnote, repeat, etc.) has a distinct logic for representation.
  • For example, the setmasterbpm2 block may include a divider block to represent beat values (like 4/4 = 1.00), which must be parsed recursively.

• Redacting base64-encoded content:

  • Just like in the JS version, the Python converter checks for base64 strings (e.g., audio/image data) and replaces them with "data" to keep the output clean and safe.

• Maintaining tree formatting:

  • I replicated the ├── and │ styled tree structure for visual clarity.
  • Indentation is handled dynamically based on the depth of recursion.

Enhancements Added

While rewriting, I also took the opportunity to extend the support to more block types that weren’t handled earlier — for example:

• arc
• incrementOne
• pitch
• text
• settransposition

This ensures that even newer or more complex Music Blocks projects are parsed correctly and comprehensively.

The Streamlit App

Once the backend logic was ready, I integrated it with a user-friendly Streamlit interface. The app consists of:

• A text area for JSON input.
• A convert button to trigger the parsing.
• A cleanly formatted output section with scrollable, monospaced text.
• Error handling for invalid or empty JSON.

---

Why It Matters

Music Blocks is used in educational environments. One of the biggest challenges new users face is understanding how blocks connect and function under the hood. By converting the visual block code into a readable text format, this tool:

• Makes debugging more accessible for students.
• Helps educators explain project logic in class.
• Provides an exportable, printable format of block logic.

Foundational Component for the Debugger

This converter will play a crucial role in the AI-powered Music Blocks Debugger. By giving a structured, simplified text representation of the project:

• The LLM (Large Language Model) will better understand the project logic.
• It enables embedding, chunk retrieval, and semantic search for debugging.
• Users will be able to see both their visual project and a clean text summary on the same platform.

Seamless Integration Ahead

Now that the converter is in Streamlit (and Python), integrating it into the AI Debugger system becomes straightforward:

• No need to mix JavaScript and Python — the backend stays unified.
• Users can input JSON and debug in the same interface.
• It aligns with the vector database and LLM pipeline we’re building.

📸 Preview

Here’s a quick preview of the app:

Final Thoughts

Rewriting an entire logic-heavy JavaScript app into Python was both challenging and rewarding. It made me deeply understand how each block works in Music Blocks and how a simple but well-structured parser can bring clarity to even the most complex visual projects.

---

Next Week’s Roadmap

• Integrate the Streamlit-based converter directly into the AI Debugger interface.
• Improve the understanding of actions in the project by enhancing the LLM prompts.

Resources & References

• Repository: JSON to Text representation
• Repository: AI-powered Debugger for Music Blocks
• Debugger Streamlit App: Music Blocks Debugger
• Directory for Projects: Embedding Project Set

Acknowledgments

Thank you to my mentors, the Sugar Labs community, and fellow GSoC contributors for ongoing support.

---

[omsuneri]

@pikurasa update by me for this week are here !!

[github-actions]

🎉 All Checks Passed!

Status: ✅ Ready to merge

✅ Completed Workflows

Workflow	Status	Details
🔨 Continuous Integration	✅ Passed	Build completed successfully
📝 Code Linting	✅ Passed	All formatting and style checks passed

---

🚀 This PR is ready for review and can be safely merged to main branch!

Great work! Your code meets all quality standards. 👏

[pikurasa]

Helpful details. Good work!