Replace Scala server with C/C++ server

[scholarsmate]

Replace Scala server with C/C++ server

Summary

The current gRPC server is implemented in Scala (Pekko), which adds a JVM runtime dependency and a third build/test/package system. We want to replace the Scala middleware with a cross-platform C/C++ gRPC server that directly bridges to the core native library, reducing complexity and deployment friction.

Motivation

• Reduce language/toolchain count (Scala/JVM stack removal).
• Eliminate JRE requirement on target systems.
• Simplify packaging and distribution (single native server per platform).
• Improve operational consistency with the native core library.

Goals

• Provide a fully functional C/C++ gRPC server implementing all RPCs in proto/omega_edit.proto.
• Preserve existing behavior and API compatibility with clients.
• Support cross-platform builds (Linux/macOS/Windows).
• Replace Tika-based file type detection with Apache-2.0 compatible libraries.
• Maintain current event streaming semantics for sessions and viewports.

Non-Goals

• Redesign of the gRPC API.
• Changes to core library public APIs.
• Client protocol changes.
• New feature work unrelated to the server migration.

Dependencies and Licensing

All new dependencies must be compatible with Apache-2.0.

Proposed replacements:

• Content type detection: libmagic (BSD-2-Clause) using the magic database.
• Language detection: cld3 (Apache-2.0).
• Encoding/BOM: continue using existing core BOM detection.

Proposed Implementation Plan

1. Scaffold C/C++ gRPC server

   • Create server/cpp module with CMake build.
   • Generate gRPC stubs from proto/omega_edit.proto.
   • Add CI build target for the new server.

2. Implement RPC handlers

   • Map all service methods to the native core APIs:
     • Sessions, changes, viewports, search, counts, metadata.

3. Event streaming

   • Use session and viewport callbacks from core (omega_session_event_cbk_t, omega_viewport_event_cbk_t).
   • Implement subscription management and backpressure-safe queues.

4. Replace Tika functionality

   • Add C++ wrappers for file type and language detection.
   • Integrate in GetContentType and GetLanguage handlers.

5. Packaging and distribution

   • Produce platform-specific native server binaries.
   • Update packages/server to bundle the native server instead of the Scala server.

6. Parity validation

   • Run existing client tests against both servers.
   • Add explicit gRPC parity test suite if needed.

Acceptance Criteria

• All gRPC endpoints pass existing test suites.
• Node client can use the new server without code changes.
• CI builds the native server on Linux/macOS/Windows.
• No JVM runtime dependency for the server package.
• All dependencies verified Apache-2.0 compatible.

Risks

• Feature parity with the Scala server event stream semantics.
• Threading/concurrency differences between Scala and C++.
• Packaging complexity for gRPC C++ runtime on Windows.

Notes

• Keep Scala server temporarily for parity comparison until C++ server is production-ready.