cmd: handle sigint globally

This change also updates both client.do and client.stream to return
ctx.Err(). Previously this error is skipped so canceled contexts are
silently ignored

CMakeLists.txt

@@ -24,7 +24,7 @@ set(GGML_LLAMAFILE ON)
 set(GGML_CUDA_PEER_MAX_BATCH_SIZE 128)
 set(GGML_CUDA_GRAPHS ON)
 
-if((NOT CMAKE_OSX_ARCHITECTURES MATCHES "arm64")
+if((CMAKE_OSX_ARCHITECTURES AND NOT CMAKE_OSX_ARCHITECTURES MATCHES "arm64")
     OR (NOT CMAKE_OSX_ARCHITECTURES AND NOT CMAKE_SYSTEM_PROCESSOR MATCHES "arm|aarch64|ARM64|ARMv[0-9]+"))
     set(GGML_CPU_ALL_VARIANTS ON)
 endif()

api/client.go

@@ -126,7 +126,8 @@ func (c *Client) do(ctx context.Context, method, path string, reqData, respData
 			return err
 		}
 	}
-	return nil
+
+	return ctx.Err()
 }
 
 const maxBufferSize = 512 * format.KiloByte
@@ -189,7 +190,7 @@ func (c *Client) stream(ctx context.Context, method, path string, data any, fn f
 		}
 	}
 
-	return nil
+	return ctx.Err()
 }
 
 // GenerateResponseFunc is a function that [Client.Generate] invokes every time

cmd/cmd.go

@@ -15,13 +15,11 @@ import (
 	"net"
 	"net/http"
 	"os"
-	"os/signal"
 	"path/filepath"
 	"runtime"
 	"strconv"
 	"strings"
 	"sync/atomic"
-	"syscall"
 	"time"
 
 	"github.com/containerd/console"
@@ -330,6 +328,7 @@ func RunHandler(cmd *cobra.Command, args []string) error {
 			if err := PullHandler(cmd, []string{name}); err != nil {
 				return nil, err
 			}
+
 			return client.Show(cmd.Context(), &api.ShowRequest{Name: name})
 		}
 		return info, err
@@ -858,17 +857,6 @@ func chat(cmd *cobra.Command, opts runOptions) (*api.Message, error) {
 	spinner := progress.NewSpinner("")
 	p.Add("", spinner)
 
-	cancelCtx, cancel := context.WithCancel(cmd.Context())
-	defer cancel()
-
-	sigChan := make(chan os.Signal, 1)
-	signal.Notify(sigChan, syscall.SIGINT)
-
-	go func() {
-		<-sigChan
-		cancel()
-	}()
-
 	var state *displayResponseState = &displayResponseState{}
 	var latest api.ChatResponse
 	var fullResponse strings.Builder
@@ -903,10 +891,7 @@ func chat(cmd *cobra.Command, opts runOptions) (*api.Message, error) {
 		req.KeepAlive = opts.KeepAlive
 	}
 
-	if err := client.Chat(cancelCtx, req, fn); err != nil {
-		if errors.Is(err, context.Canceled) {
-			return nil, nil
-		}
+	if err := client.Chat(cmd.Context(), req, fn); err != nil {
 		return nil, err
 	}
 
@@ -946,17 +931,6 @@ func generate(cmd *cobra.Command, opts runOptions) error {
 		generateContext = []int{}
 	}
 
-	ctx, cancel := context.WithCancel(cmd.Context())
-	defer cancel()
-
-	sigChan := make(chan os.Signal, 1)
-	signal.Notify(sigChan, syscall.SIGINT)
-
-	go func() {
-		<-sigChan
-		cancel()
-	}()
-
 	var state *displayResponseState = &displayResponseState{}
 
 	fn := func(response api.GenerateResponse) error {
@@ -992,10 +966,7 @@ func generate(cmd *cobra.Command, opts runOptions) error {
 		KeepAlive: opts.KeepAlive,
 	}
 
-	if err := client.Generate(ctx, &request, fn); err != nil {
-		if errors.Is(err, context.Canceled) {
-			return nil
-		}
+	if err := client.Generate(cmd.Context(), &request, fn); err != nil {
 		return err
 	}
 
@@ -1017,8 +988,7 @@ func generate(cmd *cobra.Command, opts runOptions) error {
 		latest.Summary()
 	}
 
-	ctx = context.WithValue(cmd.Context(), generateContextKey("context"), latest.Context)
-	cmd.SetContext(ctx)
+	cmd.SetContext(context.WithValue(cmd.Context(), generateContextKey("context"), latest.Context))
 
 	return nil
 }

main.go

@@ -2,6 +2,8 @@ package main
 
 import (
 	"context"
+	"os"
+	"os/signal"
 
 	"github.com/spf13/cobra"
 
@@ -9,5 +11,15 @@ import (
 )
 
 func main() {
-	cobra.CheckErr(cmd.NewCLI().ExecuteContext(context.Background()))
+	ctx, cancel := context.WithCancel(context.Background())
+	defer cancel()
+
+	sigChan := make(chan os.Signal, 1)
+	signal.Notify(sigChan, os.Interrupt)
+	go func() {
+		<-sigChan
+		cancel()
+	}()
+
+	cobra.CheckErr(cmd.NewCLI().ExecuteContext(ctx))
 }

progress/progress.go

@@ -1,6 +1,7 @@
 package progress
 
 import (
+	"bufio"
 	"fmt"
 	"io"
 	"sync"
@@ -13,7 +14,8 @@ type State interface {
 
 type Progress struct {
 	mu sync.Mutex
-	w  io.Writer
+	// buffer output to minimize flickering on all terminals
+	w *bufio.Writer
 
 	pos int
 
@@ -22,7 +24,7 @@ type Progress struct {
 }
 
 func NewProgress(w io.Writer) *Progress {
-	p := &Progress{w: w}
+	p := &Progress{w: bufio.NewWriter(w)}
 	go p.start()
 	return p
 }
@@ -47,26 +49,29 @@ func (p *Progress) stop() bool {
 func (p *Progress) Stop() bool {
 	stopped := p.stop()
 	if stopped {
-		fmt.Fprint(p.w, "\n")
+		fmt.Fprintln(p.w)
 	}
+
+	// show cursor
+	fmt.Fprint(p.w, "\033[?25h")
+	p.w.Flush()
 	return stopped
 }
 
 func (p *Progress) StopAndClear() bool {
-	fmt.Fprint(p.w, "\033[?25l")
-	defer fmt.Fprint(p.w, "\033[?25h")
-
 	stopped := p.stop()
 	if stopped {
 		// clear all progress lines
-		for i := range p.pos {
-			if i > 0 {
-				fmt.Fprint(p.w, "\033[A")
-			}
-			fmt.Fprint(p.w, "\033[2K\033[1G")
+		for range p.pos - 1 {
+			fmt.Fprint(p.w, "\033[A")
 		}
+
+		fmt.Fprint(p.w, "\033[2K", "\033[1G")
 	}
 
+	// show cursor
+	fmt.Fprint(p.w, "\033[?25h")
+	p.w.Flush()
 	return stopped
 }
 
@@ -81,30 +86,31 @@ func (p *Progress) render() {
 	p.mu.Lock()
 	defer p.mu.Unlock()
 
-	fmt.Fprint(p.w, "\033[?25l")
-	defer fmt.Fprint(p.w, "\033[?25h")
+	fmt.Fprint(p.w, "\033[?2026h")
+	defer fmt.Fprint(p.w, "\033[?2026l")
 
-	// clear already rendered progress lines
-	for i := range p.pos {
-		if i > 0 {
-			fmt.Fprint(p.w, "\033[A")
-		}
-		fmt.Fprint(p.w, "\033[2K\033[1G")
+	for range p.pos - 1 {
+		fmt.Fprint(p.w, "\033[A")
 	}
 
+	fmt.Fprint(p.w, "\033[1G")
+
 	// render progress lines
 	for i, state := range p.states {
-		fmt.Fprint(p.w, state.String())
+		fmt.Fprint(p.w, state.String(), "\033[K")
 		if i < len(p.states)-1 {
 			fmt.Fprint(p.w, "\n")
 		}
 	}
 
 	p.pos = len(p.states)
+	p.w.Flush()
 }
 
 func (p *Progress) start() {
 	p.ticker = time.NewTicker(100 * time.Millisecond)
+	// hide cursor
+	fmt.Fprint(p.w, "\033[?25l")
 	for range p.ticker.C {
 		p.render()
 	}