package config

import "strings"

type API struct {
    Addr               string  `yaml:"addr"`
    ModelsDir          string  `yaml:"models_dir"`
    CacheDir           string  `yaml:"cache_dir"`
    // DefaultModel: whisper model id for OpenAI /v1/audio/transcriptions (maps whisper-1).
    DefaultModel       string  `yaml:"default_model"`
    Language           string  `yaml:"language"`
    Transcript         Transcript `yaml:"transcript"`
    DefaultSpeakers    bool    `yaml:"default_speakers"`
    Prompt             string  `yaml:"prompt"`
    Threads            uint    `yaml:"threads"`
    MaxContext         uint    `yaml:"max_context"`
    BeamSize           uint    `yaml:"beam_size"`
    EntropyThold       float64 `yaml:"entropy_thold"`
    VAD                VAD     `yaml:"vad"`
    Debug              bool    `yaml:"debug"`
    SpeedUp            bool    `yaml:"speedup"`
    Translate          bool    `yaml:"translate"`
    DefaultPunctuation bool `yaml:"default_punctuation"`
    // DefaultAsync: STT via API enqueues to cache/waiting and returns taskID (use ?async=0 for sync).
    DefaultAsync bool `yaml:"default_async"`
    // Garbage: artifact substrings removed from transcript text and words (default includes *выбая*).
    Garbage []string `yaml:"garbage"`
}

func (a API) WithDefaults() API {
    a.VAD = a.VAD.WithDefaults()
    a.Transcript = a.Transcript.WithDefaults()
    if strings.TrimSpace(a.Language) == "" {
        a.Language = "ru"
    }
    return a
}

// GarbagePatterns returns garbage filter list (never empty unless explicitly set to [] in YAML).
func (a API) GarbagePatterns() []string {
    return a.garbagePatterns()
}