configloader

AI Config Loader (ai/configloader)

The configloader package provides a unified, simple YAML configuration file loader, designed for AI modules, supporting caching and path fallback mechanisms.

Architecture

classDiagram
    class Loader {
        -baseDir string
        -cache sync.Map
        +Load(subPath, target)
        +LoadCached(subPath, factory)
        +LoadDir(subDir, factory)
        -ReadFileWithFallback(path)
    }

• Loader: Core struct holding a base directory (baseDir) and a thread-safe cache (sync.Map).
• Caching Mechanism: Loaded configuration objects are cached in memory to avoid repeated disk reads and YAML parsing.
• Fallback Mechanism: Considering differences between development and deployment environments, the loader supports automatic path fallback lookup.

Algorithm

flowchart TD
    Start(LoadCached Request) --> Check{In Cache?}
    Check -- Yes --> Return[Return Cached Object]
    Check -- No --> Read{Read File}
    Read -- Try baseDir --> Found1{Found?}
    Found1 -- Yes --> Parse[YAML Unmarshal]
    Found1 -- No --> Fallback{Try ExecDir}
    Fallback -- Found --> Parse
    Fallback -- Not Found --> Error[Return Error]
    Parse --> Store[Store in Cache]
    Store --> Return

1. Request Config: Call LoadCached(path, factory).
2. Check Cache: Check if memory cache already has the config object. Return directly if yes.
3. Read File:
   • First try reading baseDir/path.
   • If failed (e.g., in test or binary running environment), try looking up relative to executable file.
4. Parse: Use yaml.v3 to Unmarshal file content into target struct created by factory function.
5. Cache Fill: Store successfully parsed object in cache for next use.

Methods

Use Case

Used to load Agent Prompt templates, model configurations, tool definitions and other static resource files, ensuring resources can be found correctly in different running environments.