fix: resolve back step, audio labels, progress parsing, file size pre-check, and stale job cleanup

- Fix handleBackStep to redirect to download prompt when step state is nil
  (previously silent no-op after container selection cleared the state)
- Improve formatLabel for audio-only entries without bitrate: use ID + extension
  instead of generic 'audio', fixing duplicate entries in format list
- Add yt-dlp --progress-template for reliable JSON progress parsing across all
  sites (fixes 0.0% progress on sites like xnxx.com where [download] lines
  are absent); keep regex parser as fallback for older yt-dlp versions
- Add file size pre-check in handleContainerSelection before download starts
  (previously checked only after download completed, wasting bandwidth)
- Clear stale active jobs and step state on /start for a clean slate

internal/dl/progress.go

@@ -12,6 +12,7 @@ import (
 )
 
 var progressRE = regexp.MustCompile(`\[\w+\]\s+([\d.]+)%\s+of\s+~?([\d.]+)(\w+)`)
+var progressTemplateRE = regexp.MustCompile(`^DLPROG\|([\d.]+)\|(.*?)\|(.*)$`)
 
 type progressUpdate struct {
 	pct   float64
@@ -19,11 +20,39 @@ type progressUpdate struct {
 	eta   string
 }
 
+// parseProgressTemplate parses a DLPROG-prefixed line from yt-dlp's --progress-template.
+func parseProgressTemplate(line string) *progressUpdate {
+	m := progressTemplateRE.FindStringSubmatch(line)
+	if m == nil {
+		return nil
+	}
+	pct, err := strconv.ParseFloat(m[1], 64)
+	if err != nil {
+		return nil
+	}
+	u := &progressUpdate{pct: pct}
+	speed := strings.TrimSpace(m[2])
+	eta := strings.TrimSpace(m[3])
+	if speed != "" && !strings.EqualFold(speed, "unknown") && speed != "N/A" {
+		u.speed = speed
+	}
+	if eta != "" && !strings.EqualFold(eta, "unknown") && eta != "N/A" {
+		u.eta = eta
+	}
+	return u
+}
+
 // parseProgressLine parses a single yt-dlp stderr progress line.
 // Returns nil if the line doesn't contain progress info.
 func parseProgressLine(line string) *progressUpdate {
 	line = strings.TrimSpace(line)
-	if line == "" || !strings.HasPrefix(line, "[download]") {
+	if line == "" {
+		return nil
+	}
+	if u := parseProgressTemplate(line); u != nil {
+		return u
+	}
+	if !strings.HasPrefix(line, "[download]") {
 		return nil
 	}
 	m := progressRE.FindStringSubmatch(line)

internal/dl/progress_test.go

@@ -100,6 +100,79 @@ func TestParseProgressLine(t *testing.T) {
 	}
 }
 
+func TestParseProgressTemplate(t *testing.T) {
+	tests := []struct {
+		line   string
+		pct    float64
+		speed  string
+		eta    string
+		hasRes bool
+	}{
+		{
+			line:   "DLPROG|45.2|2.34MiB/s|00:45",
+			pct:    45.2,
+			speed:  "2.34MiB/s",
+			eta:    "00:45",
+			hasRes: true,
+		},
+		{
+			line:   "DLPROG|100.0|5.00MiB/s|00:00",
+			pct:    100.0,
+			speed:  "5.00MiB/s",
+			eta:    "00:00",
+			hasRes: true,
+		},
+		{
+			line:   "DLPROG|0.0||",
+			pct:    0.0,
+			speed:  "",
+			eta:    "",
+			hasRes: true,
+		},
+		{
+			line:   "not a progress line",
+			hasRes: false,
+		},
+		{
+			line:   "",
+			hasRes: false,
+		},
+		{
+			line:   "[download] 45.2% of 123.45MiB at 2.34MiB/s ETA 00:45",
+			hasRes: false,
+		},
+		{
+			line:   "DLPROG|50.0|Unknown|Unknown",
+			pct:    50.0,
+			speed:  "",
+			eta:    "",
+			hasRes: true,
+		},
+	}
+	for _, tt := range tests {
+		got := parseProgressTemplate(tt.line)
+		if !tt.hasRes {
+			if got != nil {
+				t.Errorf("parseProgressTemplate(%q) = %+v, want nil", tt.line, got)
+			}
+			continue
+		}
+		if got == nil {
+			t.Errorf("parseProgressTemplate(%q) = nil, want result", tt.line)
+			continue
+		}
+		if math.Abs(got.pct-tt.pct) > 0.01 {
+			t.Errorf("parseProgressTemplate(%q).pct = %f, want %f", tt.line, got.pct, tt.pct)
+		}
+		if got.speed != tt.speed {
+			t.Errorf("parseProgressTemplate(%q).speed = %q, want %q", tt.line, got.speed, tt.speed)
+		}
+		if got.eta != tt.eta {
+			t.Errorf("parseProgressTemplate(%q).eta = %q, want %q", tt.line, got.eta, tt.eta)
+		}
+	}
+}
+
 func TestParseProgressLineEdgeCases(t *testing.T) {
 	tests := []struct {
 		line   string

internal/dl/ytdlp.go

@@ -246,8 +246,9 @@ func (c *Client) StartDownload(job *domain.DownloadJob, downloadDir string) (<-c
 	outputPath := filepath.Join(downloadDir, job.ID+"_%(id)s.%(ext)s")
 	args = append(args, "--output", outputPath)
 
-	// Progress reporting (--newline for line-based, no --no-progress so we get actual output)
+	// Progress reporting using progress-template for machine-readable output
 	args = append(args, "--newline")
+	args = append(args, "--progress-template", "stderr:DLPROG|%(progress.percent)s|%(progress._speed_str)s|%(progress._eta_str)s")
 
 	// Language selection if specified
 	if job.Language != "" {
@@ -435,7 +436,10 @@ func formatLabel(f domain.Format) string {
 		if f.Bitrate != "" {
 			return f.Bitrate
 		}
-		return "audio"
+		if f.Extension != "" {
+			return f.ID + " (" + f.Extension + ")"
+		}
+		return f.ID
 	}
 	if f.Resolution != "" {
 		return f.Resolution

internal/dl/ytdlp_test.go

@@ -89,7 +89,8 @@ func TestFormatLabel(t *testing.T) {
 		want string
 	}{
 		{domain.Format{IsAudioOnly: true, Bitrate: "128k"}, "128k"},
-		{domain.Format{IsAudioOnly: true}, "audio"},
+		{domain.Format{ID: "140", Extension: "m4a", IsAudioOnly: true}, "140 (m4a)"},
+		{domain.Format{ID: "251", IsAudioOnly: true}, "251"},
 		{domain.Format{Resolution: "1920x1080", Height: 1080}, "1920x1080"},
 		{domain.Format{Height: 720}, "720p"},
 		{domain.Format{ID: "137"}, "137"},

internal/handler/download.go

@@ -346,6 +346,14 @@ func (h *Handler) handleContainerSelection(ctx context.Context, chatID int64, ms
 		format = ss.MainFormat
 	}
 
+	fileSize := format.Filesize
+
+	if fileSize > 0 && fileSize > h.maxFileSize {
+		h.editText(ctx, chatID, msgID,
+			fmt.Sprintf("File too large (%s). Max: %s", formatBytes(fileSize), formatBytes(h.maxFileSize)), nil)
+		return
+	}
+
 	job := &domain.DownloadJob{
 		ID:             h.generateJobID(),
 		UserID:         userID,
@@ -359,9 +367,10 @@ func (h *Handler) handleContainerSelection(ctx context.Context, chatID int64, ms
 		Language:       ss.Language,
 		Status:         domain.StatusDownloading,
 		Title:          ss.MediaInfo.Title,
-		FileSize:       format.Filesize,
+		FileSize:       fileSize,
 	}
 
+	h.editText(ctx, chatID, msgID, "Processing...", nil)
 	h.setActiveJob(userID, job)
 	h.clearStepState(chatID)
 
@@ -526,6 +535,7 @@ func (h *Handler) handleBackStep(ctx context.Context, chatID int64, msgID int, c
 	defer h.answerCb(ctx, cbID)
 	ss := h.getUserStepState(chatID)
 	if ss == nil {
+		h.handleDownloadPrompt(ctx, chatID, msgID, cbID, userID)
 		return
 	}
 

internal/handler/handler.go

@@ -442,9 +442,11 @@ func (h *Handler) generateJobID() string {
 	return uuid.NewString()[:8]
 }
 
-// handleStart shows the main menu.
+// handleStart shows the main menu and cleans up any stale state.
 func (h *Handler) handleStart(ctx context.Context, msg *models.Message, userID int64) {
 	slog.Info("start", "user_id", userID, "chat_id", msg.Chat.ID)
+	h.clearActiveJob(userID)
+	h.clearStepState(msg.Chat.ID)
 	h.sendWithKB(ctx, msg.Chat.ID, "Main Menu:", mainKeyboard())
 }