Code base Indexing in C#/.Net Not Extracting Valuable Code Segments

[Source-GuyCoder]

App Version

3.23.16

API Provider

Ollama

Model Used

nomic-embed-text

Roo Code Task Links (Optional)

No response

🔁 Steps to Reproduce

OS: Windows
Repo Code Language: C#/.Net
Vector DB: QDrant Local

Example File:

///
/// Description of file
///

using namespace1;
using namespace2;
using namespace3;
using namespace4;
....
using namespaceN;

namespace SomeNameSpace
{
    public static class SomeClass
   {
     <bunch of static functions>
   }
}

The code base indexing feature doesn't index any actual valuable code, rather when searching for the file path in the vector DB, all the code chunks are simply the using statements (multiple points for random selection of singular using statements)

Roo also indicates that indexing is complete (green dot).

💥 Outcome Summary

Expected: Other chunks in the file for actual functions, etc... within that file
Actual: Only code segments extracted is using directives.

📄 Relevant Logs or Errors (Optional)

[roomote]

I'm investigating this issue with C#/.NET code indexing. The problem appears to be that the indexer is only capturing using statements instead of the actual code content (classes, methods, etc.). I'll analyze the indexing logic and provide a fix.

[vadash]

Can be part of #5238

Repo to check embeddings on https://github.com/openai/openai-dotnet

Current version of roocode should get < 100 embeddings

[daniel-lxs]

Closing as duplicate of #5238