🧩 API Reference

---

The Jimi API reference is generated from the project source files using MkDocs, MkDocs Material, mkdocstrings, and Google-style Python docstrings. These pages document the application entry point, Gemini provider wrappers, runtime configuration, and SQLite-backed exception logging layer.

The API documentation is intended for developers who need to understand, extend, troubleshoot, or validate the Jimi codebase.

📚 API Modules

Module	API Page	Purpose
app.py	App	Streamlit application entry point, session-state management, workflow routing, UI rendering, and user interaction logic.
gemini.py	Gemini	Google Gemini wrapper classes for text, images, embeddings, audio, files, and storage workflows.
config.py	Configuration	Runtime constants, environment values, local paths, model settings, logging paths, and validation helpers.
boogr.py	Logging	Structured exception wrapper and SQLite logger used across the application.

🧭 How the API Reference Is Built

The API reference is built directly from source comments. Each Python module uses Google-style docstrings that mkdocstrings can parse into structured documentation.

Python source files
      │
      ▼
Google-style docstrings
      │
      ▼
mkdocstrings
      │
      ▼
MkDocs Material API pages
      │
      ▼
GitHub Pages documentation site

This approach keeps the source code and documentation synchronized. When a method signature, purpose, argument, return value, or exception behavior changes, the corresponding docstring should be updated in the same change.

🧱 Source Modules

app.py

The app.py module is the Streamlit application layer. It owns interface construction, page configuration, sidebar controls, session-state initialization, file upload handling, workflow routing, result display, and user-facing fallback behavior.

Use the App API page when reviewing:

• Streamlit page setup.
• Session-state defaults.
• Sidebar model and mode controls.
• Chat, document, semantic search, prompt, and data workflows.
• UI helpers and rendering functions.
• Application-level exception logging.
• Recoverable fallback behavior.

gemini.py

The gemini.py module is the provider integration layer for Google Gemini and related Google services. It keeps Google GenAI SDK request construction, configuration, response extraction, file operations, embeddings, audio workflows, and Google Cloud Storage operations outside the Streamlit UI.

Primary wrapper classes include:

Class	Purpose
Gemini	Base configuration and shared attribute state.
Chat	Text generation, structured history, tool use, grounding, and streaming output.
Images	Image generation, image analysis, image editing, image extraction, and grounding metadata.
Embeddings	Text embedding generation for semantic search and similarity workflows.
TTS	Text-to-speech generation and WAV output handling.
Transcription	Audio-to-text transcription.
Translation	Spoken-language translation from audio files.
Files	Gemini file upload, retrieval, listing, deletion, summarization, and search workflows.
VectorStores	Google Cloud Storage-backed collection and object management.

config.py

The config.py module centralizes application settings. It keeps runtime paths, model constants, API key lookups, logging paths, and validation helpers in one place.

Use the Configuration API page when reviewing:

• API key configuration.
• Runtime directories.
• Logging database paths.
• Model constants.
• File-system helpers.
• Input validation helpers.
• Environment-driven defaults.

boogr.py

The boogr.py module provides structured exception handling and durable SQLite logging. Application and provider failures are wrapped in an Error object, enriched with module/cause/method metadata, and written through Logger.

Use the Logging API page when reviewing:

• Exception metadata fields.
• SQLite table creation.
• Log-write behavior.
• Traceback capture.
• Application-wide diagnostic pattern.