fix zombie leak, add quiet-hours toggle, improve query reply, configurable router threshold, JS dashboard

This commit is contained in:
kami
2026-07-03 00:42:35 +02:00
parent 612583d59a
commit e00cb07658
26 changed files with 3652 additions and 15 deletions
+131
View File
@@ -0,0 +1,131 @@
// Package main is mavsttd — maven's stt module process.
//
// Per spec: stt/tts are restart-free, key-free, fail-independent modules —
// separate processes from core, reachable over the worker boundary
// (internal/worker). Core dials mavsttd's unix socket and ships audio bytes
// for transcription; mavsttd ships text back.
//
// With -model <path>: loads a whisper.cpp ggml model (e.g. ggml-small.bin)
// for real transcription. Without -model: serves the stub transcriber
// (deterministic no-model floor) so the loop is exercisable end-to-end
// without weights.
//
// Module topology:
//
// $ mavsttd -socket /run/user/$UID/maven/stt.sock
//
// The daemon's config points at this socket:
//
// "stt": { "socket": "/run/user/1000/maven/stt.sock" }
//
// Both processes are same-user on the box ⇒ the 0600 socket floor (same
// unix user) is sufficient today; the wg / mTLS cuts in internal/auth are
// for the NETWORK radius (client↔core), not the local module radius.
package main
import (
"context"
"errors"
"flag"
"fmt"
"log"
"net"
"os"
"os/signal"
"syscall"
"github.com/kami/maven/internal/worker"
)
func main() {
if err := run(os.Args[1:]); err != nil {
fmt.Fprintln(os.Stderr, "mavsttd:", err)
os.Exit(1)
}
}
func run(args []string) error {
sock := flag.String("socket", defaultSocket("stt.sock"), "unix socket path")
model := flag.String("model", "", "path to whisper ggml model file")
flag.CommandLine.Parse(args)
ctx, stop := signal.NotifyContext(context.Background(), syscall.SIGINT, syscall.SIGTERM, syscall.SIGHUP)
defer stop()
var t worker.Transcriber
if *model != "" {
w, err := newWhisperHandler(*model)
if err != nil {
return fmt.Errorf("whisper: %w", err)
}
t = w
defer func() {
log.Printf("mavsttd: closing whisper model")
w.Close()
}()
log.Printf("mavsttd: loaded whisper model from %s", *model)
} else {
log.Printf("mavsttd: no model specified, using stub handler")
t = &stubHandler{}
}
srv := worker.NewServer(*sock, t)
if err := srv.Listen(); err != nil {
return err
}
defer srv.Close()
log.Printf("mavsttd: worker listening on %s", srv.Path())
errCh := make(chan error, 1)
go func() { errCh <- srv.Serve() }()
select {
case <-ctx.Done():
log.Printf("mavsttd: shutdown signal received")
srv.Close()
return nil
case err := <-errCh:
if err != nil && !errors.Is(err, net.ErrClosed) {
return err
}
return nil
}
}
// stubHandler — worker.Transcriber that delegates to the package Stub. Tiny
// now; the production swap replaces this whole struct with a faster-whisper
// / vosk-backed struct (the same Worker.Transcriber interface).
type stubHandler struct{}
func (h *stubHandler) Transcribe(ctx context.Context, req worker.TranscribeReq) (worker.TranscribeResp, error) {
// delegate to the same deterministic Stub the daemon could have wired
// in-process; mavsttd is the "separate process" equivalent.
_ = req
// hash for variation; same approach as stt.Stub.
if len(req.Audio.Bytes) == 0 {
return worker.TranscribeResp{Text: "maven, что у меня сегодня", Confidence: 1.0}, nil
}
// vary phrase by first byte for visibility in logs/tests.
phrases := []string{
"maven, отметь что я выпил воды",
"maven, напомни через 4 часа размяться",
"maven, restart nginx",
"maven, что у меня сегодня по календарю",
"note: staggered cooldown by time of day",
"slept 6h",
}
idx := int(req.Audio.Bytes[0]) % len(phrases)
return worker.TranscribeResp{Text: phrases[idx], Confidence: 1.0}, nil
}
// defaultSocket returns XDG_RUNTIME_DIR/maven/<name> if set, falling back
// to a homedir-relative path (mirrors config.defaultRuntimeDir).
func defaultSocket(name string) string {
if x := os.Getenv("XDG_RUNTIME_DIR"); x != "" {
return x + "/maven/" + name
}
home, err := os.UserHomeDir()
if err != nil || home == "" {
return name
}
return home + "/.local/share/maven/" + name
}
+119
View File
@@ -0,0 +1,119 @@
package main
/*
#cgo CFLAGS: -I${SRCDIR}/../../deps/include -I${SRCDIR}/../../deps/whisper.cpp/ggml/include
#cgo LDFLAGS: -L${SRCDIR}/../../deps/lib -Wl,-rpath,${SRCDIR}/../../deps/lib -lwhisper -lggml -lggml-base -lggml-cpu -lggml-vulkan -lm -lstdc++ -fopenmp
#include <whisper.h>
#include <stdlib.h>
*/
import "C"
import (
"context"
"fmt"
"math"
"unsafe"
"github.com/kami/maven/internal/worker"
)
type whisperHandler struct {
ctx *C.struct_whisper_context
}
func newWhisperHandler(modelPath string) (*whisperHandler, error) {
cparams := C.whisper_context_default_params()
cPath := C.CString(modelPath)
defer C.free(unsafe.Pointer(cPath))
ctx := C.whisper_init_from_file_with_params(cPath, cparams)
if ctx == nil {
return nil, fmt.Errorf("whisper: failed to init from %s", modelPath)
}
return &whisperHandler{ctx: ctx}, nil
}
func (h *whisperHandler) Transcribe(ctx context.Context, req worker.TranscribeReq) (worker.TranscribeResp, error) {
a := req.Audio
if len(a.Bytes) == 0 {
return worker.TranscribeResp{}, fmt.Errorf("whisper: empty audio")
}
nSamples := len(a.Bytes) / 2
samples := make([]float32, nSamples)
for i := 0; i < nSamples; i++ {
s := int16(a.Bytes[i*2]) | int16(a.Bytes[i*2+1])<<8
samples[i] = float32(s) / 32768.0
}
params := C.whisper_full_default_params(C.WHISPER_SAMPLING_GREEDY)
params.print_progress = false
params.print_realtime = false
params.print_timestamps = false
params.print_special = false
params.n_threads = C.int(4)
params.single_segment = true
lang := C.CString(req.Lang)
defer C.free(unsafe.Pointer(lang))
params.language = lang
params.detect_language = false
cSamples := (*C.float)(unsafe.Pointer(&samples[0]))
res := C.whisper_full(h.ctx, params, cSamples, C.int(nSamples))
if res != 0 {
return worker.TranscribeResp{}, fmt.Errorf("whisper: full failed: %d", int(res))
}
nSegments := int(C.whisper_full_n_segments(h.ctx))
if nSegments == 0 {
return worker.TranscribeResp{Text: "", Confidence: 0}, nil
}
var text string
totalLogProb := float64(0)
totalTokens := 0
for i := 0; i < nSegments; i++ {
cSeg := C.whisper_full_get_segment_text(h.ctx, C.int(i))
if cSeg != nil {
text += C.GoString(cSeg)
}
nTokens := int(C.whisper_full_n_tokens(h.ctx, C.int(i)))
for j := 0; j < nTokens; j++ {
p := float64(C.whisper_full_get_token_p(h.ctx, C.int(i), C.int(j)))
if p > 0 {
totalLogProb += math.Log(p)
totalTokens++
}
}
}
confidence := 0.0
if totalTokens > 0 {
avgLogProb := totalLogProb / float64(totalTokens)
confidence = math.Exp(avgLogProb)
}
noSpeechProb := float64(C.whisper_full_get_segment_no_speech_prob(h.ctx, 0))
if noSpeechProb > 0.9 {
confidence = 0
}
if math.IsNaN(confidence) || math.IsInf(confidence, 0) {
confidence = 0
}
return worker.TranscribeResp{
Text: text,
Confidence: confidence,
}, nil
}
func (h *whisperHandler) Close() {
if h.ctx != nil {
C.whisper_free(h.ctx)
h.ctx = nil
}
}