feature/pmem: 实现了pmem读取内存数据，根据大小端解析，并按fmt格式化展示

0_godbg/godbg/cmd/debug/pmem.go

@@ -4,7 +4,10 @@ import (
 	"errors"
 	"fmt"
 	"strconv"
+	"strings"
 	"syscall"
+	"text/tabwriter"
+	"unsafe"
 
 	"github.com/spf13/cobra"
 )
@@ -37,22 +40,22 @@ var pmemCmd = &cobra.Command{
 		if err != nil || n != int(bytes) {
 			return fmt.Errorf("read %d bytes, error: %v", n, err)
 		}
+		buf = buf[:n]
 
-		fmt.Printf("read %d bytes ok:", n)
-		for _, b := range buf[:n] {
-			fmt.Printf("%x", b)
-		}
-		fmt.Println()
+		fmt.Printf("read %d bytes ok:\n", n)
+		s := prettyPrintMem(uintptr(readAt), buf, isLittleEndian(), format[0], int(size))
+		fmt.Println(s)
 
 		return nil
 	},
 }
 
 func init() {
 	debugRootCmd.AddCommand(pmemCmd)
+
 	// 类似gdb的命令x/FMT，其中FMT=重复数字+格式化修饰符+size
 	pmemCmd.Flags().Uint("count", 16, "查看数值数量")
-	pmemCmd.Flags().String("fmt", "hex", "数值打印格式: b(binary), o(octal), x(hex), d(decimal), ud(unsigned decimal)")
+	pmemCmd.Flags().String("fmt", "hex", "数值打印格式: b(binary), o(octal), x(hex), d(decimal)") // TODO signed/unsigned
 	pmemCmd.Flags().Uint("size", 4, "数值占用字节")
 	pmemCmd.Flags().String("addr", "", "读取的内存地址")
 }
@@ -78,3 +81,100 @@ func checkPmemArgs(count uint, format string, size uint, addr string) error {
 	_, err := strconv.ParseUint(addr, 0, 64)
 	return err
 }
+
+// prettyPrintMem examine the memory and format data
+//
+// `format` specifies the data format (or data type), `size` specifies size of each data,
+// like 4byte integer, 1byte character, etc. `count` specifies the number of values.
+func prettyPrintMem(address uintptr, memArea []byte, littleEndian bool, format byte, size int) string {
+
+	var (
+		cols      int
+		colFormat string
+		colBytes  = size
+
+		addrLen int
+		addrFmt string
+	)
+
+	// Diffrent versions of golang output differently about '#'.
+	// See https://ci.appveyor.com/project/derekparker/delve-facy3/builds/30179356.
+	switch format {
+	case 'b':
+		cols = 4 // Avoid emitting rows that are too long when using binary format
+		colFormat = fmt.Sprintf("%%0%db", colBytes*8)
+	case 'o':
+		cols = 8
+		colFormat = fmt.Sprintf("0%%0%do", colBytes*3) // Always keep one leading zero for octal.
+	case 'd':
+		cols = 8
+		colFormat = fmt.Sprintf("%%0%dd", colBytes*3)
+	case 'x':
+		cols = 8
+		colFormat = fmt.Sprintf("0x%%0%dx", colBytes*2) // Always keep one leading '0x' for hex.
+	default:
+		return fmt.Sprintf("not supprted format %q\n", string(format))
+	}
+	colFormat += "\t"
+
+	l := len(memArea)
+	rows := l / (cols * colBytes)
+	if l%(cols*colBytes) != 0 {
+		rows++
+	}
+
+	// Avoid the lens of two adjacent address are different, so always use the last addr's len to format.
+	if l != 0 {
+		addrLen = len(fmt.Sprintf("%x", uint64(address)+uint64(l)))
+	}
+	addrFmt = "0x%0" + strconv.Itoa(addrLen) + "x:\t"
+
+	var b strings.Builder
+	w := tabwriter.NewWriter(&b, 0, 0, 3, ' ', 0)
+
+	for i := 0; i < rows; i++ {
+		fmt.Fprintf(w, addrFmt, address)
+
+		for j := 0; j < cols; j++ {
+			offset := i*(cols*colBytes) + j*colBytes
+			if offset+colBytes <= len(memArea) {
+				n := byteArrayToUInt64(memArea[offset:offset+colBytes], littleEndian)
+				fmt.Fprintf(w, colFormat, n)
+			}
+		}
+		fmt.Fprintln(w, "")
+		address += uintptr(cols)
+	}
+	w.Flush()
+	return b.String()
+}
+
+// 将byte slice转成uint64数值，注意字节序影响
+func byteArrayToUInt64(buf []byte, isLittleEndian bool) uint64 {
+	var n uint64
+	if isLittleEndian {
+		for i := len(buf) - 1; i >= 0; i-- {
+			n = n<<8 + uint64(buf[i])
+		}
+	} else {
+		for i := 0; i < len(buf); i++ {
+			n = n<<8 + uint64(buf[i])
+		}
+	}
+	return n
+}
+
+// 判断是否是小端字节序
+func isLittleEndian() bool {
+	buf := [2]byte{}
+	*(*uint16)(unsafe.Pointer(&buf[0])) = uint16(0xABCD)
+
+	switch buf {
+	case [2]byte{0xCD, 0xAB}:
+		return true
+	case [2]byte{0xAB, 0xCD}:
+		return false
+	default:
+		panic("Could not determine native endianness.")
+	}
+}