move-data-nas/internal/syncengine/parser.go

package syncengine

import (
	"regexp"
	"strconv"
	"strings"
)

type RsyncStats struct {
	SentBytes     int64
	ReceivedBytes int64
	TotalSize     int64
	Speedup       float64
	FilesSent     int
	FilesTotal    int
}

type ProgressLine struct {
	Phase       string
	Percent     float64
	Files       int
	Total       int
	SentBytes   int64
	XferedBytes int64
}

type ProgressFields struct {
	FileBytes  int64 `json:"file_bytes"`
	Pct        int   `json:"pct"`
	SpeedBps   int64 `json:"speed_bps"`
	EtaSeconds int   `json:"eta_seconds"`
	XfrDone    int   `json:"xfr_done"`
	XfrTotal   int   `json:"xfr_total"`
}

var (
	progressRegex = regexp.MustCompile(`\s*([\d,]+)\s+([\d,]+)\s+([\d%]+)\s*`)
	sentRegex     = regexp.MustCompile(`sent\s+([\d,]+)\s+bytes`)
	recvRegex     = regexp.MustCompile(`received\s+([\d,]+)\s+bytes`)
	totalRegex    = regexp.MustCompile(`total size is\s+([\d,]+)`)
	filesRegex    = regexp.MustCompile(`Number of files: ([\d,]+)`)
)

var perFileProgressRegex = regexp.MustCompile(
	`^\s*(\d{1,3}(?:,\d{3})+)\s+(\d+)%\s+(\d+\.\d+)([kMG])B/s\s+(\d+:\d{2}:\d{2})(.*)`,
)

var xfrRegex = regexp.MustCompile(`xfr#(\d+).*to-chk=(\d+)/(\d+)`)

func parseXfrSuffix(suffix string) (done, total int) {
	m := xfrRegex.FindStringSubmatch(suffix)
	if m == nil {
		return 0, 0
	}
	done, _ = strconv.Atoi(m[1])
	t, _ := strconv.Atoi(m[2])
	_ = t
	total, _ = strconv.Atoi(m[3])
	return done, total
}

func ParseProgressLine(line string) *ProgressLine {
	if strings.Contains(line, "files to consider") || strings.Contains(line, "files...") {
		return &ProgressLine{Phase: "scanning"}
	}
	if strings.Contains(line, "building file list") {
		return &ProgressLine{Phase: "listing"}
	}
	if strings.Contains(line, "sent") && strings.Contains(line, "bytes") {
		return &ProgressLine{Phase: "stats"}
	}
	return nil
}

func ParseStatsLine(line string) (int64, bool) {
	m := sentRegex.FindStringSubmatch(line)
	if len(m) >= 2 {
		n, _ := strconv.ParseInt(strings.ReplaceAll(m[1], ",", ""), 10, 64)
		return n, true
	}
	return 0, false
}

func ParseFinalStats(output string) *RsyncStats {
	stats := &RsyncStats{}
	lines := strings.Split(output, "\n")
	for _, line := range lines {
		line = strings.TrimSpace(line)
		if m := sentRegex.FindStringSubmatch(line); len(m) >= 2 {
			stats.SentBytes, _ = strconv.ParseInt(strings.ReplaceAll(m[1], ",", ""), 10, 64)
		}
		if m := recvRegex.FindStringSubmatch(line); len(m) >= 2 {
			stats.ReceivedBytes, _ = strconv.ParseInt(strings.ReplaceAll(m[1], ",", ""), 10, 64)
		}
		if m := totalRegex.FindStringSubmatch(line); len(m) >= 2 {
			stats.TotalSize, _ = strconv.ParseInt(strings.ReplaceAll(m[1], ",", ""), 10, 64)
		}
		if m := filesRegex.FindStringSubmatch(line); len(m) >= 2 {
			stats.FilesTotal, _ = strconv.Atoi(strings.ReplaceAll(m[1], ",", ""))
		}
	}
	return stats
}

func isProgressOnlyLine(line string) bool {
	return perFileProgressRegex.MatchString(line)
}

func parseProgressFields(line string) *ProgressFields {
	m := perFileProgressRegex.FindStringSubmatch(line)
	if m == nil {
		return nil
	}
	bytes, _ := strconv.ParseInt(strings.ReplaceAll(m[1], ",", ""), 10, 64)
	pct, _ := strconv.Atoi(m[2])
	speed, _ := strconv.ParseFloat(m[3], 64)
	unit := m[4]
	eta := m[5]
	suffix := m[6]

	speedBps := int64(speed * 1e6)
	switch unit {
	case "k", "K":
		speedBps = int64(speed * 1e3)
	case "m", "M":
		speedBps = int64(speed * 1e6)
	case "g", "G":
		speedBps = int64(speed * 1e9)
	}

	etaSecs := 0
	parts := strings.Split(eta, ":")
	if len(parts) == 3 {
		h, _ := strconv.Atoi(parts[0])
		m, _ := strconv.Atoi(parts[1])
		s, _ := strconv.Atoi(parts[2])
		etaSecs = h*3600 + m*60 + s
	}

	pf := &ProgressFields{
		FileBytes:  bytes,
		Pct:        pct,
		SpeedBps:   speedBps,
		EtaSeconds: etaSecs,
	}

	if suffix != "" {
		done, total := parseXfrSuffix(suffix)
		pf.XfrDone = done
		pf.XfrTotal = total
	}

	return pf
}

func isFileNameLine(line string) bool {
	if line == "" || strings.TrimSpace(line) == "" {
		return false
	}
	if strings.Contains(line, "sending incremental file list") ||
		strings.Contains(line, "building file list") ||
		strings.Contains(line, "cannot open") ||
		strings.Contains(line, "skipping non-regular") ||
		strings.HasPrefix(line, "sent ") ||
		strings.HasPrefix(line, "total ") ||
		strings.HasPrefix(line, "Number of files:") ||
		strings.Contains(line, "bytes received") {
		return false
	}
	return !isProgressOnlyLine(line)
}