我想出了以下代码:
private static Map<String, String> headerMap(String value) {
return Stream.of(value.split(";"))
.filter(entry -> entry.indexOf('=') != -1)
.map(entry -> {
int i = entry.indexOf('=');
return new String[] { entry.substring(0, i), entry.substring(i + 1) };
})
.collect(Collectors.toMap(array -> array[0], array -> array[1]));
}
它仅扫描两次,方法是将键和值存储在大小为 2 的数组中。我不确定它是否会像循环一样高性能,因为我们正在创建另一个对象来充当持有者。entry
for
另一个扫描唯一一次的解决方案是这个,尽管我不是很了解它:entry
private static Map<String, String> headerMap(String value) {
return Stream.of(value.split(";"))
.map(entry -> {
int i = entry.indexOf('=');
if (i == -1) {
return null;
}
return new String[] { entry.substring(0, i), entry.substring(i + 1) };
})
.filter(Objects::nonNull)
.collect(Collectors.toMap(array -> array[0], array -> array[1]));
}
我实现了一个JMH基准测试来测试这一点。以下是基准代码:
@Warmup(iterations = 5, time = 1000, timeUnit = TimeUnit.MILLISECONDS)
@Measurement(iterations = 10, time = 1000, timeUnit = TimeUnit.MILLISECONDS)
@BenchmarkMode(Mode.AverageTime)
@OutputTimeUnit(TimeUnit.MICROSECONDS)
@Fork(3)
@State(Scope.Benchmark)
public class StreamTest {
private static final String VALUE = "Accept=text/plain;"
+ "Accept-Charset=utf-8;"
+ "Accept-Encoding=gzip, deflate;"
+ "Accept-Language=en-US;"
+ "Accept-Datetime=Thu, 31 May 2007 20:35:00 GMT;"
+ "Cache-Control=no-cache;"
+ "Connection=keep-alive;"
+ "Content-Length=348;"
+ "Content-Type=application/x-www-form-urlencoded;"
+ "Date=Tue, 15 Nov 1994 08:12:31 GMT;"
+ "Expect=100-continue;"
+ "Max-Forwards=10;"
+ "Pragma=no-cache";
@Benchmark
public void loop() {
int eq;
Map<String, String> map = new HashMap<>();
for (String entry : VALUE.split(";")) {
if ((eq = entry.indexOf('=')) != -1) {
map.put(entry.substring(0, eq), entry.substring(eq + 1));
}
}
}
@Benchmark
public void stream1() {
Stream.of(VALUE.split(";"))
.filter(entry -> entry.indexOf('=') != -1)
.map(entry -> {
int i = entry.indexOf('=');
return new String[] { entry.substring(0, i), entry.substring(i + 1) };
})
.collect(Collectors.toMap(array -> array[0], array -> array[1]));
}
@Benchmark
public void stream2() {
Stream.of(VALUE.split(";"))
.map(entry -> {
int i = entry.indexOf('=');
if (i == -1) {
return null;
}
return new String[] { entry.substring(0, i), entry.substring(i + 1) };
})
.filter(Objects::nonNull)
.collect(Collectors.toMap(array -> array[0], array -> array[1]));
}
public static void main(String[] args) throws Exception {
Main.main(args);
}
}
这是结果(代码i5 3230M CPU @ 2.60 GHz,Windows 10,Oracle JDK 1.8.0_25):
Benchmark Mode Cnt Score Error Units
StreamTest.loop avgt 30 1,541 ± 0,038 us/op
StreamTest.stream1 avgt 30 1,633 ± 0,042 us/op
StreamTest.stream2 avgt 30 1,604 ± 0,058 us/op
这表明,流解决方案和 for 循环在性能方面实际上是等效的。