我想衡量所花费的时间GoLang解析XML文件。所以,我决定写一个基准。
我确实有其生成包含XML文档的io.Reader
功能。
// PRIVATE: createSampleXMLReader creates an io.Reader instance that contains 10.000 '<Node />' elements which are
// suitable for running a benchmark test.
func createSampleXMLReader(
nodeElementCount int) io.Reader {
xmlContents := new(strings.Builder)
xmlContents.WriteString("<ROOT>\n")
for i := 0; i < nodeElementCount; i++ {
appendNodeXMLElement(xmlContents)
}
xmlContents.WriteString("</ROOT>")
return strings.NewReader(xmlContents.String())
}
// PRIVATE: appendNodeXMLElement appends a '<Node />' elements to an existing io.Reader instance.
func appendNodeXMLElement(
xmlDocument *strings.Builder) {
xmlDocument.WriteString("<Node id=\"0\" position=\"0\" depth=\"0\" parent=\"0\">\n")
xmlDocument.WriteString(" <Name>Name</Name>\n")
xmlDocument.WriteString(" <Description>Description</Description>\n")
xmlDocument.WriteString(" <OwnInformation>\n")
xmlDocument.WriteString(" <Title>Title</Title>\n")
xmlDocument.WriteString(" <Description>Description</Description>\n")
xmlDocument.WriteString(" </OwnInformation>\n")
xmlDocument.WriteString(" <Assets>\n")
xmlDocument.WriteString(" <Asset id=\"0\" position=\"0\" type=\"0\" category=\"0\">\n")
xmlDocument.WriteString(" <OriginalFile>OriginalFile</OriginalFile>\n")
xmlDocument.WriteString(" <Description>Description</Description>\n")
xmlDocument.WriteString(" <Uri>Uri</Uri>\n")
xmlDocument.WriteString(" </Asset>\n")
xmlDocument.WriteString(" <Asset id=\"1\" position=\"1\" type=\"1\" category=\"1\">\n")
xmlDocument.WriteString(" <OriginalFile>OriginalFile</OriginalFile>\n")
xmlDocument.WriteString(" <Description>Description</Description>\n")
xmlDocument.WriteString(" <Uri>Uri</Uri>\n")
xmlDocument.WriteString(" </Asset>\n")
xmlDocument.WriteString(" <Asset id=\"2\" position=\"2\" type=\"2\" category=\"2\">\n")
xmlDocument.WriteString(" <OriginalFile>OriginalFile</OriginalFile>\n")
xmlDocument.WriteString(" <Description>Description</Description>\n")
xmlDocument.WriteString(" <Uri>Uri</Uri>\n")
xmlDocument.WriteString(" </Asset>\n")
xmlDocument.WriteString(" <Asset id=\"3\" position=\"3\" type=\"3\" category=\"3\">\n")
xmlDocument.WriteString(" <OriginalFile>OriginalFile</OriginalFile>\n")
xmlDocument.WriteString(" <Description>Description</Description>\n")
xmlDocument.WriteString(" <Uri>Uri</Uri>\n")
xmlDocument.WriteString(" </Asset>\n")
xmlDocument.WriteString(" <Asset id=\"4\" position=\"4\" type=\"4\" category=\"4\">\n")
xmlDocument.WriteString(" <OriginalFile>OriginalFile</OriginalFile>\n")
xmlDocument.WriteString(" <Description>Description</Description>\n")
xmlDocument.WriteString(" <Uri>Uri</Uri>\n")
xmlDocument.WriteString(" </Asset>\n")
xmlDocument.WriteString(" </Assets>\n")
xmlDocument.WriteString(" <Synonyms>\n")
xmlDocument.WriteString(" <Synonym>Synonym 0</Synonym>\n")
xmlDocument.WriteString(" <Synonym>Synonym 1</Synonym>\n")
xmlDocument.WriteString(" <Synonym>Synonym 2</Synonym>\n")
xmlDocument.WriteString(" <Synonym>Synonym 3</Synonym>\n")
xmlDocument.WriteString(" <Synonym>Synonym 4</Synonym>\n")
xmlDocument.WriteString(" </Synonyms>\n")
xmlDocument.WriteString("</Node>\n")
}
然后,我实际上解析这个XML文档的功能。
// PRIVATE: parseXML parses an io.Reader instance into a 'Node' struct.
func parseXML(
xmlReader io.Reader) {
xmlDecoder := xml.NewDecoder(xmlReader)
for {
token, _ := xmlDecoder.Token()
if token == nil {
break
}
switch element := token.(type) {
case xml.StartElement:
if element.Name.Local == "Node" {
decodeNodeElement(xmlDecoder, &element)
}
}
}
}
// PRIVATE: decodeNodeElement decodes a '<Node />' element into a 'Node' struct.
func decodeNodeElement(
xmlDecoder *xml.Decoder,
element *xml.StartElement) {
node := new(model.Node)
xmlDecoder.DecodeElement(node, element)
}
然后,我有执行基准功能:
// PRIVATE: runBenchmarkParseXML performs a benchmark that parses an XML document that contains the given number of
// '<Node />' element.
func runBenchmarkParseXML(
nodeCount int,
benchmark *testing.B) {
// Arrange.
xmlReader := createSampleXMLReader(nodeCount)
// Act.
for i := 0; i < benchmark.N; i++ {
parseXML(xmlReader)
}
}
然后,我有5个功能执行的基准。这些是在XML文档中1,10,100,1000和10000元件。
func BenchmarkParseXML1(benchmark *testing.B) { runBenchmarkParseXML(1, benchmark) }
func BenchmarkParseXML10(benchmark *testing.B) { runBenchmarkParseXML(10, benchmark) }
func BenchmarkParseXML100(benchmark *testing.B) { runBenchmarkParseXML(100, benchmark) }
func BenchmarkParseXML1000(benchmark *testing.B) { runBenchmarkParseXML(1000, benchmark) }
func BenchmarkParseXML10000(benchmark *testing.B) { runBenchmarkParseXML(10000, benchmark) }
当我运行这个测试,我看到下面的输出:
BenchmarkParseXML1-4 5000000 226 ns/op
BenchmarkParseXML10-4 10000000 230 ns/op
BenchmarkParseXML100-4 5000000 226 ns/op
BenchmarkParseXML1000-4 5000000 254 ns/op
BenchmarkParseXML10000-4 1 1690998100 ns/op
这怎么可能,有在分析包含10.000元,其中基准定时其余都是稳定的一个XML文件中的基准这样的区别?
是我的标杆新的或它的ParseXML方法的实现不正确。
编辑:节点结构
// Node represents a '<Node />' element in the XML document.
type Node struct {
ID int `xml:"id,attr"`
Position int `xml:"position,attr"`
Depth int `xml:"depth,attr"`
Parent string `xml:"parent,attr"`
Name string `xml:"Name"`
Description string `xml:"Description"`
OwnInformation struct {
Title string `xml:"Title"`
Description string `xml:"Description"`
} `xml:"OwnInformation"`
Assets []struct {
ID string `xml:"id,attr"`
Position int `xml:"position,attr"`
Type string `xml:"type,attr"`
Category int `xml:"category,attr"`
OriginalFile string `xml:"OriginalFile"`
Description string `xml:"Description"`
URI string `xml:"Uri"`
} `xml:"Assets>Asset"`
Synonyms []string `xml:"Synonyms>Synonym"`
}
感谢您事先的任何指导。
您的基准是有缺陷的。你是10倍增加投入,但持续时间整个保持大体不变的事实应该让你很可疑这一基准的。
您是在同一读写器每一次重新使用。只有每个基准的第一次迭代实际上会做任何事情。进一步的调用会从阅读器已经在EOF阅读。
更改您的设置,以便它返回一个字节切片,构建一个新的阅读器每次迭代。这将产生预期的结果:
func createSampleXMLDoc(nodeElementCount int) []byte {
xmlContents := &bytes.Buffer{}
xmlContents.WriteString("<ROOT>\n")
for i := 0; i < nodeElementCount; i++ {
appendNodeXMLElement(xmlContents)
}
xmlContents.WriteString("</ROOT>")
return xmlContents.Bytes()
}
func runBenchmarkParseXML(nodeCount int, b *testing.B) {
doc := createSampleXMLDoc(nodeCount)
for i := 0; i < b.N; i++ {
xmlReader := bytes.NewReader(doc)
parseXML(xmlReader)
}
}
结果我的机器上(如预期,10倍增加投入了10倍,增加了时间):
$ go test -benchtime=5s -bench .
goos: linux
goarch: amd64
BenchmarkParseXML1-8 100000 115978 ns/op
BenchmarkParseXML10-8 10000 1147605 ns/op
BenchmarkParseXML100-8 1000 11586980 ns/op
BenchmarkParseXML1000-8 50 124199120 ns/op
BenchmarkParseXML10000-8 5 1003668966 ns/op
-benchtime=5s
增加了每个指标的默认时间从一秒到五秒。一秒钟是不够的,在后一种情况下不止一个迭代,产生不可靠的结果。这也是为什么看到你的原始基准的大量。第一次迭代是缓慢的,但每一个之后立即返回,这样的平均时间显着下降。在基准眼看只是一个迭代是另一个红旗,顺便说一句。即使是在五次迭代仍然是可靠的测量相当低,其实。
这个故事的寓意:检查你的错误!