英文:
How does Go do string comparison?
问题
Go支持字符串比较而无需任何特殊函数。Go运行时是否在幕后进行工作以进行字符串字面量的比较?
英文:
http://golang.org/ref/spec#Comparison_operators
Go supports string comparison without any special functions. Does the Go runtime do work behind the scenes to make comparisons of string literals?
答案1
得分: 57
如您在以下的汇编转储中所见,字符串比较在短路检查(第11行)后,将委托给运行时的runtime.eqstring
函数(第17行),以查看两个操作数是否为相同的内存字符串(第11行):
$ cat foo.go
package main
func main() {
a := "hello"
b := "world"
_ = a == b
}
$ go tool 6g -S foo.go
--- 程序列表 "main" ---
0000 (foo.go:3) TEXT main+0(SB),$40-0
0001 (foo.go:3) LOCALS ,$0
0002 (foo.go:4) LEAQ go.string."hello"+0(SB),BX
0003 (foo.go:4) MOVQ (BX),SI
0004 (foo.go:4) MOVQ 8(BX),CX
0005 (foo.go:5) LEAQ go.string."world"+0(SB),BX
0006 (foo.go:5) MOVQ (BX),DX
0007 (foo.go:5) MOVQ 8(BX),AX
0008 (foo.go:6) JMP ,11
0009 (foo.go:6) MOVQ $1,AX
0010 (foo.go:6) JMP ,23
0011 (foo.go:6) CMPQ CX,AX
0012 (foo.go:6) JNE ,22
0013 (foo.go:6) MOVQ SI,(SP)
0014 (foo.go:6) MOVQ CX,8(SP)
0015 (foo.go:6) MOVQ DX,16(SP)
0016 (foo.go:6) MOVQ AX,24(SP)
0017 (foo.go:6) CALL ,runtime.eqstring+0(SB)
0018 (foo.go:6) MOVBQZX 32(SP),BX
0019 (foo.go:6) CMPB BX,$0
0020 (foo.go:6) JEQ ,22
0021 (foo.go:6) JMP ,9
0022 (foo.go:6) MOVQ $0,AX
0023 (foo.go:7) RET ,
--- 程序列表 "init" ---
0024 (foo.go:7) TEXT init+0(SB),$0-0
0025 (foo.go:7) MOVBQZX initdone·+0(SB),AX
0026 (foo.go:7) LOCALS ,$0
0027 (foo.go:7) CMPB AX,$0
0028 (foo.go:7) JEQ ,34
0029 (foo.go:7) CMPB AX,$2
0030 (foo.go:7) JNE ,32
0031 (foo.go:7) RET ,
0032 (foo.go:7) CALL ,runtime.throwinit+0(SB)
0033 (foo.go:7) UNDEF ,
0034 (foo.go:7) MOVB $2,initdone·+0(SB)
0035 (foo.go:7) RET ,
除非您正在编写编译器或运行时,否则这不应过多关注:只需按照规范定义使用运算符,并期望比较的时间复杂度为O(n),其中n为字符串的长度。
英文:
As you can see in the following assembly dump, string comparison is delegated to a runtime.eqstring
function from the runtime (line 17) after a short circuit check to see if the two operands are the same in-memory string (line 11):
$ cat foo.go
package main
func main() {
a := "hello"
b := "world"
_ = a == b
}
$ go tool 6g -S foo.go
--- prog list "main" ---
0000 (foo.go:3) TEXT main+0(SB),$40-0
0001 (foo.go:3) LOCALS ,$0
0002 (foo.go:4) LEAQ go.string."hello"+0(SB),BX
0003 (foo.go:4) MOVQ (BX),SI
0004 (foo.go:4) MOVQ 8(BX),CX
0005 (foo.go:5) LEAQ go.string."world"+0(SB),BX
0006 (foo.go:5) MOVQ (BX),DX
0007 (foo.go:5) MOVQ 8(BX),AX
0008 (foo.go:6) JMP ,11
0009 (foo.go:6) MOVQ $1,AX
0010 (foo.go:6) JMP ,23
0011 (foo.go:6) CMPQ CX,AX
0012 (foo.go:6) JNE ,22
0013 (foo.go:6) MOVQ SI,(SP)
0014 (foo.go:6) MOVQ CX,8(SP)
0015 (foo.go:6) MOVQ DX,16(SP)
0016 (foo.go:6) MOVQ AX,24(SP)
0017 (foo.go:6) CALL ,runtime.eqstring+0(SB)
0018 (foo.go:6) MOVBQZX 32(SP),BX
0019 (foo.go:6) CMPB BX,$0
0020 (foo.go:6) JEQ ,22
0021 (foo.go:6) JMP ,9
0022 (foo.go:6) MOVQ $0,AX
0023 (foo.go:7) RET ,
--- prog list "init" ---
0024 (foo.go:7) TEXT init+0(SB),$0-0
0025 (foo.go:7) MOVBQZX initdone·+0(SB),AX
0026 (foo.go:7) LOCALS ,$0
0027 (foo.go:7) CMPB AX,$0
0028 (foo.go:7) JEQ ,34
0029 (foo.go:7) CMPB AX,$2
0030 (foo.go:7) JNE ,32
0031 (foo.go:7) RET ,
0032 (foo.go:7) CALL ,runtime.throwinit+0(SB)
0033 (foo.go:7) UNDEF ,
0034 (foo.go:7) MOVB $2,initdone·+0(SB)
0035 (foo.go:7) RET ,
Unless you are working on the compiler or runtime, this shouldn't concern you too much: just use the operators as the spec defines, and expect that the comparison to be O(n) with the length of the string.
答案2
得分: 31
runtime/string.goc
(go1.3):
func eqstring(s1 String, s2 String) (v bool) {
if s1.len != s2.len {
v = false
return
}
if s1.str == s2.str {
v = true
return
}
v = runtime·memeq(s1.str, s2.str, s1.len)
}
int32
runtime·strcmp(byte *s1, byte *s2)
{
uintptr i
byte c1, c2
for i = 0;; i++ {
c1 = s1[i]
c2 = s2[i]
if c1 < c2 {
return -1
}
if c1 > c2 {
return +1
}
if c1 == 0 {
return 0
}
}
}
注意:runtime·
分隔符是一个 Unicode 中点,而不是句号。[1]: https://code.google.com/p/go/source/browse/src/pkg/runtime/string.goc
英文:
runtime/string.goc
(go1.3):
func eqstring(s1 String, s2 String) (v bool) {
if(s1.len != s2.len) {
v = false;
return;
}
if(s1.str == s2.str) {
v = true;
return;
}
v = runtime·memeq(s1.str, s2.str, s1.len);
}
int32
runtime·strcmp(byte *s1, byte *s2)
{
uintptr i;
byte c1, c2;
for(i=0;; i++) {
c1 = s1[i];
c2 = s2[i];
if(c1 < c2)
return -1;
if(c1 > c2)
return +1;
if(c1 == 0)
return 0;
}
}
Note: The runtime·
separator is a Unicode middle-dot, not a period.
[1]: https://code.google.com/p/go/source/browse/src/pkg/runtime/string.goc
通过集体智慧和协作来改善编程学习和解决问题的方式。致力于成为全球开发者共同参与的知识库,让每个人都能够通过互相帮助和分享经验来进步。
评论