Skip to content

Java 速查表

企业级面向对象语言,一次编写,到处运行。

01

入门

Hello World

每个 Java 程序从 main() 开始。文件名必须与公共类名匹配(Main.java → Main.class)。javac 编译为字节码(.class),java 在 JVM 上运行它。System.out.println 打印到 stdout;printf 支持格式说明符(%s、%d、%f、%n 表示换行)。

java
// Main.java
public class Main {
    public static void main(String[] args) {
        System.out.println("Hello, World!");
        System.out.printf("Name: %s, Age: %d%n", "Alice", 30);
    }
}

// Compile: javac Main.java  -> Main.class
// Run:     java Main
// Package: java -cp . com.example.Main

// Every Java program needs:
// 1. A class (public class, file name must match)
// 2. A main method: public static void main(String[] args)

变量与基本类型

Java 有 8 种基本类型(int、long、double、float、boolean、char、byte、short)和引用类型(String、数组、对象)。使用 'final' 表示常量。'var'(Java 10+)在编译时推断类型——用于类型明显的局部变量。数字中的下划线(100_000)提高可读性。

java
// Primitive types (8 total)
int age = 30;              // 32-bit integer
long bigNum = 100_000L;   // 64-bit integer
double price = 19.99;     // 64-bit float (default for decimals)
float pi = 3.14f;         // 32-bit float
boolean active = true;    // true/false
char grade = 'A';         // 16-bit Unicode character
byte b = 127;             // 8-bit signed
short s = 32767;          // 16-bit signed

// Reference types
String name = "Alice";    // Object (not primitive)
int[] nums = {1, 2, 3};  // Array object

// Constants
final double PI = 3.14159; // can't be reassigned

// var (Java 10+, local type inference)
var count = 42;    // inferred as int
var list = new ArrayList<String>(); // inferred as ArrayList<String>

包装类与装箱

包装类(Integer、Double、Boolean 等)是基本类型的对象版本。自动装箱/拆箱自动转换。Integer 缓存 -128 到 127 的值,所以 == 对小数字有效但对大数字失败——始终使用 .equals()。集合需要包装类(不能持有基本类型)。

java
// Wrapper classes (Object versions of primitives)
Integer wrapped = Integer.valueOf(42);  // explicit
Integer auto = 42;                       // autoboxing
int unboxed = auto;                      // unboxing

// Useful methods
int max = Integer.MAX_VALUE;     // 2147483647
String bin = Integer.toBinaryString(42);
int parsed = Integer.parseInt("42");
String s = String.valueOf(42);

// Other wrappers: Double, Boolean, Character, Long, Float
Double d = 3.14;
Boolean b = Boolean.TRUE;
Character c = 'A';

// Be careful with == on wrappers
Integer a = 127, b2 = 127;  // a == b2: true (cached)
Integer x = 128, y = 128;   // x == y: false (not cached)
// Always use .equals() for Integer comparison

包与导入

包组织类并防止命名冲突。约定:反向域名(com.example.app)。导入特定类或使用通配符(*)。静态导入引入常量和方法(Math.PI、Math.sqrt)。完全限定名称无需导入即可工作但冗长。java.lang 包自动导入。

java
// Package declaration (must be first line)
package com.example.app;

// Import specific class
import java.util.List;
import java.util.ArrayList;

// Import all classes in a package
import java.util.*;

// Static import (for static members)
import static java.lang.Math.PI;
import static java.lang.Math.sqrt;

// Usage
double area = PI * 5 * 5;
double root = sqrt(16);

// Fully qualified name (no import needed)
java.time.LocalDate today = java.time.LocalDate.now();

// Package naming convention: reverse domain
// com.google.gson, org.apache.commons, io.netty.channel

输入与输出

System.out(stdout)、System.err(stderr)、System.in(stdin)。Scanner 是读取控制台输入的最简单方式——它解析 token(nextInt、nextDouble、nextLine)。始终关闭 Scanner 以释放资源。命令行参数在 args[] 中(args[0] 是第一个参数,不像 C 中的程序名)。

java
import java.util.Scanner;

// Console output
System.out.println("Hello");      // with newline
System.out.print("No newline");   // without newline
System.out.printf("Pi: %.2f%n", 3.14159); // formatted

// Console input with Scanner
Scanner scanner = new Scanner(System.in);
System.out.print("Enter name: ");
String name = scanner.nextLine();

System.out.print("Enter age: ");
int age = scanner.nextInt();

System.out.printf("Hi %s, age %d%n", name, age);
scanner.close(); // always close

// Command-line arguments
// java Main arg1 arg2
// args[0] = "arg1", args[1] = "arg2"
02

字符串与格式化

字符串方法

字符串是不可变的——方法返回新字符串。始终使用 .equals() 进行内容比较(== 比较引用)。compareTo() 返回负数/零/正数用于排序(适用于排序)。split() 返回 String[]。对于可变字符串,使用 StringBuilder。

java
String s = "Hello, World";

// Length and access
int len = s.length();        // 12
char c = s.charAt(0);        // 'H'

// Comparison
s.equals("Hello, World");    // true (content comparison)
s.equalsIgnoreCase("hello, world"); // true
s.compareTo("Apple");        // positive (s > "Apple")
"abc".compareTo("abd");      // negative

// Search
s.indexOf("World");          // 7 (-1 if not found)
s.lastIndexOf("l");          // 10
s.contains("World");         // true
s.startsWith("Hello");       // true
s.endsWith("World");         // true

// Extract
s.substring(7);              // "World"
s.substring(0, 5);           // "Hello"

// Transform
s.toUpperCase();             // "HELLO, WORLD"
s.toLowerCase();             // "hello, world"
s.replace("o", "0");         // "Hell0, W0rld"
s.trim();                    // remove whitespace
s.split(", ");               // ["Hello", "World"]

StringBuilder 与拼接

使用 + 的字符串拼接每次创建新 String(在循环中低效)。StringBuilder 是可变的,适用于增量构建字符串。StringBuffer 是线程安全版本(很少需要)。String.join() 用分隔符组合。Java 11+ 添加 repeat() 用于字符串乘法。

java
// String concatenation (creates new String each time)
String s = "Hello" + ", " + "World";
String formatted = String.format("%s is %d", "Alice", 30);

// StringBuilder (mutable, efficient for many concatenations)
StringBuilder sb = new StringBuilder();
sb.append("Hello");
sb.append(", ");
sb.append("World");
sb.insert(5, " there");
sb.delete(5, 11);
sb.reverse();
String result = sb.toString();

// StringBuffer (thread-safe, slower than StringBuilder)
StringBuffer sbf = new StringBuffer("thread-safe");

// Join strings
String joined = String.join(", ", "a", "b", "c"); // "a, b, c"

// Repeat (Java 11+)
String repeated = "ab".repeat(3); // "ababab"

字符串格式化

printf/format 使用 C 风格格式说明符:%d(int)、%f(float)、%s(string)、%c(char)、%b(boolean)、%x(hex)。宽度(%5d)、左对齐(%-5d)、零填充(%05d)、精度(%.2f)。%n 是平台换行符。文本块(Java 15+)用三引号启用多行字符串而无需转义。

java
// printf / format specifiers
System.out.printf("Int: %d%n", 42);
System.out.printf("Float: %.2f%n", 3.14159);  // 3.14
System.out.printf("String: %s%n", "hello");
System.out.printf("Char: %c%n", 'A');
System.out.printf("Bool: %b%n", true);
System.out.printf("Hex: %x%n", 255);           // ff
System.out.printf("Octal: %o%n", 8);           // 10

// Width and padding
System.out.printf("[%5d]%n", 42);      // [   42]
System.out.printf("[%-5d]%n", 42);     // [42   ]
System.out.printf("[%05d]%n", 42);     // [00042]
System.out.printf("[%8.2f]%n", 3.14);  // [    3.14]

// String.format returns a String
String s = String.format("Name: %s, Age: %d", "Alice", 30);

// Text blocks (Java 15+)
String json = """
    {
        "name": "Alice",
        "age": 30
    }
    """;

正则表达式

Java 正则使用 Pattern(编译)和 Matcher(应用于输入)。String 方法(matches、split、replaceAll)是方便的快捷方式。Java 字符串字面量中反斜杠必须加倍(\\d 表示 \d)。组用括号捕获,在替换中引用为 $1、$2。如果重复使用,始终编译模式一次。

java
import java.util.regex.*;

// String methods
"hello123".matches("[a-z]+\d+"); // true
"a,b,c".split(",");               // ["a", "b", "c"]
"hello".replaceAll("l", "L");     // "heLLo"

// Pattern and Matcher
Pattern p = Pattern.compile("\\d{4}-\\d{2}-\\d{2}");
Matcher m = p.matcher("Date: 2024-01-15");

if (m.find()) {
    System.out.println(m.group()); // "2024-01-15"
}

// Find all matches
while (m.find()) {
    System.out.println(m.group());
}

// Replace with regex
String result = "2024-01-15".replaceAll(
    "(\\d{4})-(\\d{2})-(\\d{2})",
    "$3/$2/$1"); // "15/01/2024"

// Common patterns
String email = "^[\\w.]+@[\\w.]+\\.\\w+$";
String phone = "^\\d{3}-\\d{4}$";

数字与数学

Math 类提供静态数学函数。Math.random() 返回 0.0-1.0。要更多控制,使用 java.util.Random(可设种子)或 java.security.SecureRandom(加密)。Integer/Double 有静态实用方法。注意浮点精度——财务计算使用 BigDecimal。

java
// Math class
double sqrt = Math.sqrt(16);     // 4.0
double pow = Math.pow(2, 10);    // 1024.0
int abs = Math.abs(-5);          // 5
int max = Math.max(3, 7);        // 7
int min = Math.min(3, 7);        // 3
double rounded = Math.round(3.7); // 4
double ceil = Math.ceil(3.1);    // 4.0
double floor = Math.floor(3.9);  // 3.0
double random = Math.random();   // 0.0 to 1.0

// Constants
double pi = Math.PI;             // 3.14159...
double e = Math.E;               // 2.71828...

// Integer/Long methods
int sum = Integer.sum(3, 4);     // 7
int max2 = Integer.max(3, 7);    // 7

// Rounding modes
double r = Math.round(3.5);      // 4 (round half up)
double r2 = Math.floor(3.5 + 0.5); // alternative

// Random (java.util.Random)
import java.util.Random;
Random rand = new Random();
int n = rand.nextInt(100);  // 0-99
double d = rand.nextDouble(); // 0.0-1.0
boolean b = rand.nextBoolean();
03

控制流

If / Else

Java if/else 工作方式类似 C/C++。条件必须是布尔值——没有像 JavaScript 那样的 truthy/falsy(0 和非空字符串不是 truthy)。三元运算符(cond ? a : b)是表达式,不是语句。即使单行主体也使用花括号(代码风格最佳实践)。

java
int score = 85;

if (score >= 90) {
    System.out.println("A");
} else if (score >= 80) {
    System.out.println("B");
} else if (score >= 70) {
    System.out.println("C");
} else {
    System.out.println("F");
}

// Ternary operator
String grade = score >= 60 ? "Pass" : "Fail";

// Nested if
if (score >= 60) {
    if (score >= 90) {
        System.out.println("Excellent");
    }
}

// Note: conditions must be boolean (no truthy/falsy)
// if (score) { } // Error: int is not boolean

Switch 与表达式

传统 switch 有贯穿(使用 break)。Java 14+ switch 表达式(->)不贯穿且可以返回值。使用逗号分隔多个 case 标签(case 1, 2, 3)。'yield' 从复杂块返回值。Switch 表达式是穷尽的——枚举需要 default 或所有 case。

java
// Traditional switch (fall-through)
int day = 3;
switch (day) {
    case 1:
        System.out.println("Mon");
        break;
    case 2:
    case 3:
    case 4:
        System.out.println("Midweek");
        break;
    case 6:
    case 7:
        System.out.println("Weekend");
        break;
    default:
        System.out.println("Invalid");
}

// Switch expression (Java 14+, no fall-through)
String type = switch (day) {
    case 1, 2, 3, 4, 5 -> "Weekday";
    case 6, 7 -> "Weekend";
    default -> "Invalid";
};

// Switch with yield (for complex blocks)
int result = switch (day) {
    case 1, 2, 3, 4, 5 -> {
        int hours = 8;
        yield hours * 5;
    }
    case 6, 7 -> 0;
    default -> -1;
};

循环

Java 有 for、while 和 do-while 循环。增强 for(for-each)适用于数组和任何 Iterable。break 退出循环;continue 跳到下一次迭代。对于集合,优先使用 for-each 或流而非索引循环。do-while 至少运行一次(很少使用)。

java
// For loop
for (int i = 0; i < 5; i++) {
    System.out.println(i);
}

// Enhanced for (for-each)
int[] nums = {1, 2, 3, 4, 5};
for (int n : nums) {
    System.out.println(n);
}

List<String> names = List.of("Alice", "Bob");
for (String name : names) {
    System.out.println(name);
}

// While loop
int count = 0;
while (count < 3) {
    System.out.println(count);
    count++;
}

// Do-while (runs at least once)
int i = 0;
do {
    System.out.println(i);
    i++;
} while (i < 3);

// Break and continue
for (int j = 0; j < 10; j++) {
    if (j == 5) break;      // exit loop
    if (j % 2 == 0) continue; // skip iteration
    System.out.println(j);
}

标签 Break 与 Continue

标签(outer:)允许从嵌套循环中 break/continue 外层循环。这很少需要——提取到带 return 的方法通常更干净。标签放在循环之前,后跟冒号。break label 退出标记的循环;continue label 跳到其下一次迭代。

java
// Labels for breaking out of nested loops
outer:
for (int i = 0; i < 3; i++) {
    for (int j = 0; j < 3; j++) {
        if (i == 1 && j == 1) {
            break outer; // exits both loops
        }
        System.out.println(i + "," + j);
    }
}

// Labeled continue
outer:
for (int i = 0; i < 3; i++) {
    for (int j = 0; j < 3; j++) {
        if (j == 1) {
            continue outer; // skip to next i
        }
        System.out.println(i + "," + j);
    }
}

// Alternative: extract to method and use return
void findPair(int[][] matrix, int target) {
    for (int i = 0; i < matrix.length; i++) {
        for (int j = 0; j < matrix[i].length; j++) {
            if (matrix[i][j] == target) return; // exit method
        }
    }
}

数组

数组有固定长度(动态使用 ArrayList)。Arrays.sort() 原地排序。Arrays.toString() 给出可读表示。Arrays.copyOf() 创建新长度副本。对于多维数组,每行可以有不同长度(锯齿数组)。使用 Arrays 进行数组实用方法。

java
// Declare and initialize
int[] nums = {1, 2, 3, 4, 5};
int[] empty = new int[5]; // [0, 0, 0, 0, 0]
String[] names = new String[3]; // [null, null, null]

// Access and modify
nums[0] = 10;
int first = nums[0]; // 10
int length = nums.length; // 5

// Multidimensional
int[][] matrix = {{1, 2}, {3, 4}};
int val = matrix[0][1]; // 2

// Arrays utility class
import java.util.Arrays;
int[] sorted = {3, 1, 2};
Arrays.sort(sorted); // [1, 2, 3]
int[] copy = Arrays.copyOf(nums, 3);
String str = Arrays.toString(nums); // "[10, 2, 3, 4, 5]"
boolean eq = Arrays.equals(nums, copy);

// Fill
int[] filled = new int[5];
Arrays.fill(filled, 42); // [42, 42, 42, 42, 42]

// Binary search (sorted array only)
int idx = Arrays.binarySearch(sorted, 2); // index of 2
04

方法与函数

方法定义

Java 方法始终在类内。'static' 意味着方法属于类(无需实例调用)。返回类型(int、String、void)在名称之前声明。参数有类型。Java 没有默认参数值——改用方法重载。

java
public class Calculator {
    // Method with return type
    public static int add(int a, int b) {
        return a + b;
    }

    // Void method (no return)
    public static void printResult(int result) {
        System.out.println("Result: " + result);
    }

    // Method with default (no overloading needed)
    public static String greet(String name, String greeting) {
        return greeting + ", " + name + "!";
    }

    public static void main(String[] args) {
        int sum = add(3, 4);
        printResult(sum);

        String msg = greet("Alice", "Hello");
        System.out.println(msg);
    }
}

方法重载

方法重载允许同名但不同参数列表(类型、数量或顺序)的多个方法。Java 在编译时根据参数类型解析重载。重载常用于构造函数和实用方法。它与重写不同(涉及继承和运行时分发)。

java
public class MathUtils {
    // Overloaded methods (same name, different params)
    public static int add(int a, int b) {
        return a + b;
    }

    public static double add(double a, double b) {
        return a + b;
    }

    public static int add(int a, int b, int c) {
        return a + b + c;
    }

    public static String add(String a, String b) {
        return a + b;
    }
}

// Java picks the most specific match
MathUtils.add(1, 2);        // int version -> 3
MathUtils.add(1.5, 2.5);    // double version -> 4.0
MathUtils.add(1, 2, 3);     // 3-param version -> 6
MathUtils.add("Hello", "!"); // String version -> "Hello!"

可变参数与按值传递

可变参数(Type... name)允许可变参数,作为数组接收。Java 始终按值传递:基本类型被复制,对象引用被复制(但指向相同对象)。所以在方法内修改参数不影响调用者的变量,但修改它指向的对象则会影响。

java
// Varargs: variable number of arguments
public static int sum(int... nums) {
    int total = 0;
    for (int n : nums) {
        total += n;
    }
    return total;
}

sum(1, 2, 3);           // 6
sum(1, 2, 3, 4, 5);     // 15
sum();                  // 0 (empty array)
int[] arr = {1, 2, 3};
sum(arr);               // 6 (pass array to varargs)

// Java is ALWAYS pass-by-value
public static void modify(int x) {
    x = 100; // doesn't affect the caller's variable
}

int n = 5;
modify(n);
System.out.println(n); // still 5

// For objects, the reference is passed by value
public static void addItem(List<String> list) {
    list.add("new"); // modifies the same list object
}

递归

递归是方法调用自身。始终有基本情况以停止。Java 不优化尾递归(不像某些语言),所以深递归可能导致 StackOverflowError。对于性能关键或深递归,转换为迭代。记忆化(缓存结果)可以加速递归解决方案如 Fibonacci。

java
// Factorial
public static int factorial(int n) {
    if (n <= 1) return 1;        // base case
    return n * factorial(n - 1); // recursive case
}
// factorial(5) = 5 * 4 * 3 * 2 * 1 = 120

// Fibonacci
public static int fib(int n) {
    if (n < 2) return n;
    return fib(n - 1) + fib(n - 2);
}

// Tail recursion (Java doesn't optimize this)
public static int factorialTail(int n, int acc) {
    if (n <= 1) return acc;
    return factorialTail(n - 1, n * acc);
}
// Call: factorialTail(5, 1)

// Be careful: deep recursion causes StackOverflowError
// For deep recursion, use iteration or a loop instead

Lambda 表达式

Lambda(Java 8+)是匿名函数。类型是函数式接口(一个抽象方法)。常见:Function<T,R>(输入→输出)、Predicate<T>(布尔测试)、Consumer<T>(消费,无返回)、Supplier<T>(生产,无输入)。方法引用(String::length)是调用单个方法的 lambda 简写。

java
import java.util.function.*;

// Lambda syntax: (params) -> expression
Function<Integer, Integer> square = x -> x * x;
Function<String, Integer> length = s -> s.length();
BiFunction<Integer, Integer, Integer> add = (a, b) -> a + b;

// With type annotations
BinaryOperator<Integer> multiply = (Integer a, Integer b) -> a * b;

// Multi-line lambda
Function<String, String> process = s -> {
    String upper = s.toUpperCase();
    return upper.substring(0, 3);
};

// Predicate (boolean test)
Predicate<Integer> isEven = n -> n % 2 == 0;
Predicate<String> isEmpty = String::isEmpty; // method reference

// Consumer (no return)
Consumer<String> printer = s -> System.out.println(s);
Consumer<String> printer2 = System.out::println; // method reference

// Supplier (no input, produces value)
Supplier<Double> random = () -> Math.random();

// Usage
int result = square.apply(5); // 25
boolean even = isEven.test(4); // true
printer.accept("Hello"); // prints "Hello"
05

类与 OOP

类与构造函数

类是对象的模板。字段持有状态,方法定义行为。构造函数初始化新对象(使用 'this' 区分字段和参数)。@Override 表示方法重写超类方法(toString 来自 Object)。封装:私有字段,公共 getter/setter。

java
public class Person {
    // Fields (instance variables)
    private String name;
    private int age;

    // Constructor
    public Person(String name, int age) {
        this.name = name;  // 'this' refers to the current instance
        this.age = age;
    }

    // Methods
    public String getName() { return name; }
    public int getAge() { return age; }

    public void setAge(int age) {
        if (age >= 0) this.age = age;
    }

    public String greet() {
        return "Hi, I'm " + name;
    }

    @Override
    public String toString() {
        return "Person{name='" + name + "', age=" + age + "}";
    }
}

// Usage
Person p = new Person("Alice", 30);
System.out.println(p.getName());  // "Alice"
System.out.println(p);            // uses toString()

访问修饰符与 Static

访问修饰符:public(到处)、private(仅类)、protected(类 + 子类 + 包)、default/package-private(同包)。静态成员属于类而非实例——在所有对象间共享。静态初始化器在类加载时运行一次。常量(static final)、实用方法和计数器使用 static。

java
public class BankAccount {
    // Access modifiers:
    public String owner;      // accessible everywhere
    private double balance;   // class only
    protected String type;    // class + subclasses + same package
    String id;                // package-private (default)

    // Static field (shared by all instances)
    private static int accountCount = 0;

    // Static constant
    public static final double MIN_BALANCE = 100.0;

    // Static method (call without instance)
    public static int getAccountCount() {
        return accountCount;
    }

    // Static initializer (runs once when class loads)
    static {
        System.out.println("BankAccount class loaded");
    }

    public BankAccount(String owner) {
        this.owner = owner;
        this.balance = MIN_BALANCE;
        accountCount++; // increment shared counter
    }
}

int count = BankAccount.getAccountCount(); // static method call

继承与 super

Java 使用 'extends' 进行类继承(仅单继承)。super() 调用父构造函数(必须是第一行)。@Override 表示方法重写(运行时多态)。Dog IS-A Animal。'is-a' 关系使用继承;代码重用使用组合(has-a)。Java 17+ 支持密封类以限制继承。

java
// Parent class
class Animal {
    protected String name;

    public Animal(String name) {
        this.name = name;
        System.out.println("Animal constructor");
    }

    public void eat() {
        System.out.println(name + " is eating");
    }
}

// Child class (extends)
class Dog extends Animal {
    private String breed;

    public Dog(String name, String breed) {
        super(name); // must be first line — call parent constructor
        this.breed = breed;
    }

    // Override parent method
    @Override
    public void eat() {
        super.eat(); // call parent's eat()
        System.out.println(name + " the " + breed + " eats dog food");
    }

    public void bark() {
        System.out.println("Woof!");
    }
}

Dog dog = new Dog("Rex", "Labrador");
dog.eat();   // calls Dog's eat()
dog.bark();  // Dog-specific method

抽象类与接口

抽象类不能被实例化,可以同时有抽象(无主体)和具体方法。接口定义契约——所有方法默认是 public abstract。Java 8+ 允许接口中的默认方法(有主体)和静态方法。一个类扩展一个抽象类但可以实现多个接口。共享代码使用抽象类,契约使用接口。

java
// Abstract class (can't be instantiated)
abstract class Shape {
    protected String color;

    public Shape(String color) {
        this.color = color;
    }

    // Abstract method (must be implemented by subclasses)
    public abstract double area();

    // Concrete method (inherited)
    public String describe() {
        return color + " " + this.getClass().getSimpleName();
    }
}

// Interface (pure contract, Java 8+ can have default methods)
interface Drawable {
    void draw(); // abstract by default

    // Default method (Java 8+)
    default void drawTwice() {
        draw();
        draw();
    }

    // Static method in interface
    static Drawable empty() {
        return () -> System.out.println("nothing");
    }
}

// A class can extend one class and implement multiple interfaces
class Circle extends Shape implements Drawable {
    private double radius;

    public Circle(String color, double radius) {
        super(color);
        this.radius = radius;
    }

    @Override
    public double area() {
        return Math.PI * radius * radius;
    }

    @Override
    public void draw() {
        System.out.println("Drawing " + describe());
    }
}

多态与转型

多态:父引用可以持有子对象。方法调用分发到实际对象的实现(运行时多态)。向下转型前使用 instanceof 以避免 ClassCastException。Java 16+ 模式匹配(instanceof Circle c)结合检查和转型。一起重写 equals() 和 hashCode() 以在集合中正确行为。

java
// Polymorphism: one interface, many forms
Shape s1 = new Circle("red", 5);
Shape s2 = new Square("blue", 3);

// Calls the overridden method (runtime dispatch)
System.out.println(s1.area()); // Circle's area
System.out.println(s2.area()); // Square's area

// instanceof check
if (s1 instanceof Circle) {
    Circle c = (Circle) s1; // downcast
    System.out.println("Radius: " + c.radius);
}

// Pattern matching (Java 16+)
if (s1 instanceof Circle c) {
    System.out.println("Radius: " + c.radius); // c is already cast
}

// Upcasting (automatic)
Circle circle = new Circle("green", 2);
Shape shape = circle; // upcast (no explicit cast needed)

// Object class methods (all classes inherit from Object)
circle.equals(circle);   // reference equality by default
circle.hashCode();       // hash code
circle.getClass();       // Class<Circle>
circle.toString();       // string representation

Record 与枚举

Record(Java 16+)是不可变数据类——编译器生成构造函数、getter、equals、hashCode 和 toString。用于 DTO 和值对象。枚举是类型安全的常量,可以有字段、方法和构造函数。枚举实现 Comparable 并有 values() 和 valueOf() 方法。两者都是现代 Java 的基础。

java
// Record (Java 16+): concise data class
public record Point(int x, int y) {}

// Equivalent to a class with:
// - final fields x, y
// - constructor
// - getters x(), y()
// - equals, hashCode, toString

Point p = new Point(3, 4);
System.out.println(p.x());      // 3
System.out.println(p.y());      // 4
System.out.println(p);          // Point[x=3, y=4]

// Compact constructor (validation)
public record Age(int value) {
    public Age {
        if (value < 0 || value > 150) {
            throw new IllegalArgumentException("Invalid age");
        }
    }
}

// Enum (named constants)
public enum Direction {
    UP, DOWN, LEFT, RIGHT;

    public Direction opposite() {
        return switch (this) {
            case UP -> DOWN;
            case DOWN -> UP;
            case LEFT -> RIGHT;
            case RIGHT -> LEFT;
        };
    }
}

Direction d = Direction.UP;
Direction opp = d.opposite(); // DOWN
06

集合与泛型

List(ArrayList 与 LinkedList)

ArrayList 由数组支持(快速 get/set,中间 insert/delete 慢)。LinkedList 由双向链表支持(两端快速 insert/delete,随机访问慢)。List.of() 创建不可变列表。大多数情况使用 ArrayList;仅频繁端操作使用 LinkedList。两者都实现 List 接口。

java
import java.util.*;

// ArrayList (fast random access, slow insert/delete in middle)
List<String> list = new ArrayList<>();
list.add("Alice");
list.add("Bob");
list.add(0, "Carol");     // insert at index
list.set(1, "Dave");      // replace at index
String name = list.get(0); // "Carol"
list.remove(0);            // remove by index
list.remove("Dave");       // remove by value
int size = list.size();    // 1
boolean has = list.contains("Bob");

// LinkedList (fast insert/delete at ends)
LinkedList<Integer> linked = new LinkedList<>();
linked.addFirst(1);
linked.addLast(2);
linked.removeFirst();
linked.peek(); // see first element

// Immutable list (Java 9+)
List<String> immutable = List.of("a", "b", "c");
// immutable.add("d"); // UnsupportedOperationException

// Iterate
for (String s : list) {
    System.out.println(s);
}
list.forEach(System.out::println); // method reference

Set(HashSet 与 TreeSet)

Set 存储唯一元素。HashSet 最快但无序。TreeSet 保持元素排序(自然顺序或 Comparator)。LinkedHashSet 维护插入顺序。集合操作:addAll(并集)、retainAll(交集)、removeAll(差集)。对于 HashSet 中的自定义对象,重写 equals() 和 hashCode()。

java
import java.util.*;

// HashSet (fast, unordered)
Set<String> set = new HashSet<>();
set.add("apple");
set.add("banana");
set.add("apple"); // duplicate ignored
System.out.println(set.size()); // 2
System.out.println(set.contains("apple")); // true
set.remove("banana");

// TreeSet (sorted, slower)
Set<Integer> sorted = new TreeSet<>();
sorted.add(3);
sorted.add(1);
sorted.add(2);
System.out.println(sorted); // [1, 2, 3]

// LinkedHashSet (maintains insertion order)
Set<String> ordered = new LinkedHashSet<>();
ordered.add("c");
ordered.add("a");
ordered.add("b");
System.out.println(ordered); // [c, a, b]

// Set operations
Set<Integer> a = new HashSet<>(Set.of(1, 2, 3));
Set<Integer> b = new HashSet<>(Set.of(2, 3, 4));
a.addAll(b);    // union: [1, 2, 3, 4]
a.retainAll(b); // intersection: [2, 3]
a.removeAll(b); // difference: [1]

// Immutable set
Set<String> immutable = Set.of("x", "y", "z");

Map(HashMap 与 TreeMap)

Map 存储键值对。HashMap 最快(无序)。TreeMap 按键排序。LinkedHashMap 维护插入顺序。getOrDefault 避免 null 检查。compute/merge 用于更新值很强大。对于自定义键,重写 equals() 和 hashCode()。Map.of() 创建不可变映射(Java 9+)。

java
import java.util.*;

// HashMap (fast, unordered)
Map<String, Integer> ages = new HashMap<>();
ages.put("Alice", 30);
ages.put("Bob", 25);
ages.put("Alice", 31); // overwrite

// Access
int age = ages.get("Alice"); // 31
int defaultAge = ages.getOrDefault("Eve", 0); // 0

// Check
boolean has = ages.containsKey("Alice");
boolean hasVal = ages.containsValue(25);

// Remove
ages.remove("Bob");

// Iterate
for (Map.Entry<String, Integer> entry : ages.entrySet()) {
    System.out.println(entry.getKey() + ": " + entry.getValue());
}

ages.forEach((key, val) -> System.out.println(key + "=" + val));

// Compute (Java 8+)
ages.compute("Alice", (k, v) -> v + 1); // increment
ages.putIfAbsent("Carol", 28);
ages.merge("Alice", 1, Integer::sum); // add 1

// TreeMap (sorted by keys)
Map<String, Integer> sorted = new TreeMap<>();
// LinkedHashMap (maintains insertion order)
Map<String, Integer> ordered = new LinkedHashMap<>();

Queue 与 Deque

Queue 是 FIFO(先进先出)。Deque 是双端的(可以从两端添加/删除)。PriorityQueue 按自然顺序或 Comparator 排序元素(默认最小堆)。对于栈,使用 ArrayDeque(push/pop)而不是遗留 Stack 类。ArrayDeque 在队列/deque 操作上比 LinkedList 快。

java
import java.util.*;

// Queue (FIFO)
Queue<String> queue = new LinkedList<>();
queue.add("first");   // throws if full (capacity-restricted)
queue.offer("second"); // returns false if full
String head = queue.peek(); // see head (null if empty)
String removed = queue.poll(); // remove and return head

// Deque (double-ended)
Deque<Integer> deque = new ArrayDeque<>();
deque.addFirst(1);
deque.addLast(2);
deque.peekFirst(); // 1
deque.peekLast();  // 2
deque.pollFirst(); // 1
deque.pollLast();  // 2

// PriorityQueue (min-heap by default)
PriorityQueue<Integer> pq = new PriorityQueue<>();
pq.add(3);
pq.add(1);
pq.add(2);
System.out.println(pq.poll()); // 1 (smallest first)

// Max-heap (reverse order)
PriorityQueue<Integer> maxHeap = new PriorityQueue<>(Collections.reverseOrder());
maxHeap.add(1);
maxHeap.add(3);
System.out.println(maxHeap.poll()); // 3 (largest first)

// Stack (legacy, prefer Deque)
Deque<Integer> stack = new ArrayDeque<>();
stack.push(1); // add to front
stack.push(2);
stack.pop();   // 2 (remove from front)

泛型

泛型启用类型安全的集合和类。<T> 是类型参数。有界类型(<T extends Comparable<T>>)限制为具有特定行为的类型。通配符:?(任何)、? extends T(协变,只读)、? super T(逆变,只写)。泛型使用类型擦除——类型在编译时检查,运行时擦除。

java
// Generic class
public class Box<T> {
    private T value;

    public void set(T value) { this.value = value; }
    public T get() { return value; }
}

Box<String> stringBox = new Box<>();
stringBox.set("hello");
String s = stringBox.get();

Box<Integer> intBox = new Box<>();
intBox.set(42);

// Generic method
public static <T> T firstOf(List<T> list) {
    return list.get(0);
}

String first = firstOf(List.of("a", "b"));

// Bounded type parameter
public static <T extends Comparable<T>> T max(List<T> list) {
    T result = list.get(0);
    for (T item : list) {
        if (item.compareTo(result) > 0) {
            result = item;
        }
    }
    return result;
}

// Wildcards
void process(List<?> list) { }          // any type
void processNums(List<? extends Number> list) { } // Number or subclass
void addNums(List<? super Integer> list) { }      // Integer or superclass

迭代器与 Comparable

Iterator 允许在迭代期间安全删除(it.remove())。ListIterator 添加双向遍历和 set/add。Comparable 定义自然顺序(compareTo)。Comparator 定义自定义顺序(comparing、comparingInt、reversed、thenComparing)。使用 Comparator.comparing() 进行流式排序。Collections.sort() 使用自然顺序。

java
import java.util.*;

// Iterator
List<String> list = List.of("a", "b", "c");
Iterator<String> it = list.iterator();
while (it.hasNext()) {
    String s = it.next();
    System.out.println(s);
    // it.remove(); // safe removal during iteration
}

// ListIterator (bidirectional)
ListIterator<String> lit = list.listIterator();
while (lit.hasNext()) {
    lit.set(lit.next().toUpperCase()); // replace
}

// Comparable (natural ordering)
public class Person implements Comparable<Person> {
    String name;
    int age;

    @Override
    public int compareTo(Person other) {
        return Integer.compare(this.age, other.age);
    }
}

// Comparator (custom ordering)
Comparator<Person> byName = Comparator.comparing(p -> p.name);
Comparator<Person> byAgeDesc = Comparator.comparingInt((Person p) -> p.age).reversed();

List<Person> people = new ArrayList<>();
people.sort(byName);
people.sort(byAgeDesc);
Collections.sort(people); // uses Comparable
07

流与函数式

Stream 基础

Stream(Java 8+)提供声明式数据处理。用 .stream()(集合)或 Stream.of() 创建。中间操作(filter、map、sorted)是惰性的——它们仅在调用终端操作(collect、reduce、count、forEach)时执行。toList()(Java 16+)是 collect(Collectors.toList()) 的简洁替代。

java
import java.util.*;
import java.util.stream.*;

List<Integer> nums = List.of(1, 2, 3, 4, 5, 6);

// Filter and collect
List<Integer> evens = nums.stream()
    .filter(n -> n % 2 == 0)
    .collect(Collectors.toList()); // [2, 4, 6]

// Map (transform)
List<String> doubled = nums.stream()
    .map(n -> "num" + n)
    .collect(Collectors.toList());

// Reduce
int sum = nums.stream().reduce(0, Integer::sum); // 21
int product = nums.stream().reduce(1, (a, b) -> a * b);

// Count
long count = nums.stream().filter(n -> n > 3).count(); // 3

// Find
Optional<Integer> first = nums.stream().filter(n -> n > 3).findFirst();
boolean anyMatch = nums.stream().anyMatch(n -> n > 5);
boolean allMatch = nums.stream().allMatch(n -> n > 0);

// ForEach
nums.stream().forEach(System.out::println);

// toList() shortcut (Java 16+)
List<Integer> result = nums.stream().filter(n -> n > 3).toList();

Stream 操作

sorted() 排序元素(自然或用 Comparator)。distinct() 移除重复。limit(n)/skip(n) 分页。flatMap 展平嵌套流——一对多转换的关键。peek() 用于调试(副作用)。groupingBy 创建按键分组元素的映射。Stream 是惰性的——操作链式高效。

java
List<String> names = List.of("Alice", "Bob", "Charlie", "David");

// Sorted
List<String> sorted = names.stream()
    .sorted()
    .toList();

// Sorted by length
List<String> byLength = names.stream()
    .sorted(Comparator.comparing(String::length))
    .toList();

// Distinct
List<Integer> distinct = List.of(1, 2, 2, 3, 3, 3).stream()
    .distinct()
    .toList(); // [1, 2, 3]

// Limit and Skip
List<Integer> limited = nums.stream()
    .skip(2)   // skip first 2
    .limit(3)  // take next 3
    .toList();

// FlatMap (flatten nested structures)
List<List<Integer>> nested = List.of(List.of(1, 2), List.of(3, 4));
List<Integer> flat = nested.stream()
    .flatMap(List::stream)
    .toList(); // [1, 2, 3, 4]

// Peek (debug, side-effect)
nums.stream()
    .peek(n -> System.out.println("before: " + n))
    .filter(n -> n > 2)
    .peek(n -> System.out.println("after: " + n))
    .toList();

// Grouping
Map<Integer, List<String>> byLength = names.stream()
    .collect(Collectors.groupingBy(String::length));

Collector 与归约

Collector 提供丰富的归约操作:joining(连接字符串)、groupingBy(按键分组)、partitioningBy(按布尔拆分)、toMap(创建映射)、summarizingInt(统计:count、sum、min、max、average)。Collector 可以组合(groupingBy 带下游 collector)。这些用声明式单行代码替换冗长的循环。

java
import java.util.stream.*;

List<Person> people = List.of(
    new Person("Alice", 30),
    new Person("Bob", 25),
    new Person("Charlie", 35)
);

// Join strings
String joined = people.stream()
    .map(Person::getName)
    .collect(Collectors.joining(", ")); // "Alice, Bob, Charlie"

// Group by
Map<Integer, List<Person>> byAge = people.stream()
    .collect(Collectors.groupingBy(Person::getAge));

// Partition (boolean)
Map<Boolean, List<Person>> partition = people.stream()
    .collect(Collectors.partitioningBy(p -> p.getAge() > 28));

// Count by group
Map<Integer, Long> countByAge = people.stream()
    .collect(Collectors.groupingBy(Person::getAge, Collectors.counting()));

// Summarizing
IntSummaryStatistics stats = people.stream()
    .collect(Collectors.summarizingInt(Person::getAge));
System.out.println(stats.getAverage()); // 30.0
System.out.println(stats.getMax());     // 35

// To map
Map<String, Integer> nameToAge = people.stream()
    .collect(Collectors.toMap(Person::getName, Person::getAge));

// Reducing
int totalAge = people.stream()
    .collect(Collectors.reducing(0, Person::getAge, Integer::sum));

Optional

Optional<T> 是一个可能包含也可能不包含值的容器。它强制显式处理缺失——不再有 NullPointerException。非 null 值使用 of(),可能 null 使用 ofNullable()。用 map/flatMap/filter 链式。永远不要在没有 isPresent() 的情况下使用 get()——优先使用 orElse/orElseThrow。Optional 设计用于返回类型,不是字段。

java
import java.util.Optional;

// Creating Optional
Optional<String> present = Optional.of("hello");
Optional<String> empty = Optional.empty();
Optional<String> nullable = Optional.ofNullable(null); // empty if null

// Checking
present.isPresent(); // true
empty.isEmpty();     // true (Java 11+)

// Getting values
String val = present.get(); // throws if empty (avoid!)
String safe = present.orElse("default");
String computed = present.orElseGet(() -> computeDefault());
String orThrow = present.orElseThrow(() -> new RuntimeException("missing"));

// Transform (map/flatMap)
Optional<Integer> length = present.map(String::length); // Optional[5]
Optional<String> upper = present.map(s -> s.toUpperCase());

// Filter
Optional<String> filtered = present.filter(s -> s.length() > 3);

// ifPresent
present.ifPresent(s -> System.out.println(s));
present.ifPresentOrElse(
    s -> System.out.println("Got: " + s),
    () -> System.out.println("Empty")
);

// Chaining (avoid null checks)
String result = getUser(1)
    .map(User::getProfile)
    .map(Profile::getEmail)
    .orElse("no email");

函数式接口

函数式接口恰好有一个抽象方法(可以有多个默认方法)。@FunctionalInterface 是可选的但记录意图。Java 在 java.util.function 中提供许多:Function、Predicate、Consumer、Supplier,加上 Bi- 和基本类型变体。尽可能使用这些而不是创建自定义接口。它们启用 lambda 表达式和方法引用。

java
import java.util.function.*;

// Built-in functional interfaces
Function<String, Integer> strToInt = Integer::parseInt;
BiFunction<String, String, String> concat = String::concat;

Predicate<String> isEmpty = String::isEmpty;
BiPredicate<String, String> contains = String::contains;

Consumer<String> printer = System.out::println;
BiConsumer<String, Integer> printPair = (s, i) -> System.out.println(s + ":" + i);

Supplier<List<String>> listFactory = ArrayList::new;

// Primitive specializations
IntFunction<String> intToStr = String::valueOf;
ToIntFunction<String> length = String::length;
IntPredicate isPositive = n -> n > 0;
IntConsumer intPrinter = System.out::println;
IntSupplier randomInt = () -> (int)(Math.random() * 100);

// Binary operators
BinaryOperator<Integer> max = Integer::max;
IntBinaryOperator sum = Integer::sum;

// Unary operators
UnaryOperator<String> trim = String::trim;
IntUnaryOperator negate = n -> -n;

// Custom functional interface
@FunctionalInterface
interface StringProcessor {
    String process(String input);

    // Can have default methods
    default StringProcessor andThen(StringProcessor after) {
        return input -> after.process(process(input));
    }
}
08

异常与 I/O

Try / Catch / Finally

try/catch/finally 处理异常。finally 始终运行(用于清理)。多捕获(catch A | B)一起处理多个异常。Try-with-resources 自动关闭任何 AutoCloseable(文件、连接、流)——优先于手动 finally 清理。资源按声明相反顺序关闭。

java
try {
    int result = 10 / 0;
} catch (ArithmeticException e) {
    System.out.println("Math error: " + e.getMessage());
} catch (Exception e) {
    System.out.println("General error: " + e);
} finally {
    // Always runs (even if return/throw in try/catch)
    System.out.println("Cleanup");
}

// Multi-catch (Java 7+)
try {
    riskyOperation();
} catch (IOException | SQLException e) {
    // Handle both exceptions the same way
    log.error(e);
}

// Try-with-resources (auto-close, Java 7+)
try (BufferedReader br = new BufferedReader(new FileReader("file.txt"));
     PrintWriter pw = new PrintWriter("output.txt")) {
    String line = br.readLine();
    pw.println(line);
} catch (IOException e) {
    e.printStackTrace();
}
// br and pw are auto-closed (in reverse order)

受检与非受检

受检异常(extends Exception)必须用 'throws' 声明或捕获——编译器强制执行。用于可恢复条件(文件未找到、网络错误)。非受检异常(extends RuntimeException)不需要声明——用于编程错误(null 指针、无效参数)。争论:受检异常强制处理但可能使代码混乱;许多框架优先使用非受检。

java
// Checked exceptions (must be declared or caught)
public void readFile(String path) throws IOException {
    BufferedReader br = new BufferedReader(new FileReader(path));
    // IOException is checked — compiler enforces handling
}

// Unchecked exceptions (RuntimeException, no need to declare)
public int divide(int a, int b) {
    if (b == 0) {
        throw new IllegalArgumentException("Divisor cannot be zero");
        // RuntimeException — no 'throws' needed
    }
    return a / b;
}

// Common checked exceptions:
// IOException, SQLException, ClassNotFoundException

// Common unchecked exceptions:
// NullPointerException, IllegalArgumentException,
// IndexOutOfBoundsException, ArithmeticException,
// ClassCastException, IllegalStateException

// Custom checked exception
class DataException extends Exception {
    public DataException(String msg) { super(msg); }
}

// Custom unchecked exception
class ValidationException extends RuntimeException {
    public ValidationException(String msg) { super(msg); }
}

文件 I/O(NIO.2)

NIO.2(java.nio.file)是现代文件 API。Files.readString/writeString(Java 11+)对文本很方便。Files.lines() 返回惰性 Stream——对大文件高效(必须用 try-with-resources 关闭)。Path.of() 替代旧 File 类。Files.createDirectories() 创建完整路径。始终处理 IOException。

java
import java.nio.file.*;
import java.io.*;

// Read entire file (small files)
List<String> lines = Files.readAllLines(Path.of("input.txt"));
String content = Files.readString(Path.of("config.json")); // Java 11+
byte[] bytes = Files.readAllBytes(Path.of("image.png"));

// Write file
Files.writeString(Path.of("output.txt"), "Hello, World!");
Files.write(Path.of("data.bin"), bytes);

// Append
Files.writeString(Path.of("log.txt"), "entry\n",
    StandardOpenOption.APPEND, StandardOpenOption.CREATE);

// Stream lines (large files, lazy)
try (Stream<String> lineStream = Files.lines(Path.of("large.txt"))) {
    lineStream.filter(l -> l.contains("ERROR"))
              .forEach(System.out::println);
}

// Copy, move, delete
Files.copy(Path.of("src.txt"), Path.of("dest.txt"));
Files.move(Path.of("old.txt"), Path.of("new.txt"));
Files.delete(Path.of("temp.txt"));

// Check existence
boolean exists = Files.exists(Path.of("file.txt"));

// Create directories
Files.createDirectories(Path.of("a/b/c"));

Reader 与 Writer(文本)

BufferedReader/Writer 对文本 I/O 高效(缓冲减少系统调用)。PrintWriter 提供 printf 风格格式化。Scanner 解析输入(nextInt、nextDouble、nextLine)。InputStreamReader 将字节流桥接到字符流(为非 UTF-8 指定字符集)。始终使用 try-with-resources 确保流关闭。

java
import java.io.*;
import java.nio.file.*;

// BufferedReader (efficient text reading)
try (BufferedReader br = Files.newBufferedReader(Path.of("input.txt"))) {
    String line;
    while ((line = br.readLine()) != null) {
        System.out.println(line);
    }
}

// BufferedWriter (efficient text writing)
try (BufferedWriter bw = Files.newBufferedWriter(Path.of("output.txt"))) {
    bw.write("First line");
    bw.newLine();
    bw.write("Second line");
}

// PrintWriter (convenient formatting)
try (PrintWriter pw = new PrintWriter("output.txt")) {
    pw.println("Hello");
    pw.printf("Name: %s, Age: %d%n", "Alice", 30);
}

// Scanner (parsing input)
try (Scanner sc = new Scanner(Path.of("data.txt"))) {
    while (sc.hasNextLine()) {
        String line = sc.nextLine();
        // Parse tokens
        Scanner lineSc = new Scanner(line);
        if (lineSc.hasNextInt()) {
            int n = lineSc.nextInt();
        }
    }
}

// InputStreamReader (bytes to chars, e.g., from InputStream)
Reader reader = new InputStreamReader(System.in);

日期与时间(java.time)

java.time(Java 8+)是现代日期/时间 API,替代旧 Date/Calendar。LocalDate(仅日期)、LocalTime(仅时间)、LocalDateTime(两者)、ZonedDateTime(带时区)。都是不可变且线程安全的。日期差异使用 Period,时间差异使用 Duration。DateTimeFormatter 用于解析/格式化。Instant 用于机器时间戳(UTC)。

java
import java.time.*;
import java.time.format.*;
import java.time.temporal.*;

// Current date/time
LocalDate today = LocalDate.now();      // 2024-01-15
LocalTime now = LocalTime.now();        // 14:30:45.123
LocalDateTime dt = LocalDateTime.now(); // both
ZonedDateTime zdt = ZonedDateTime.now(); // with timezone

// Create specific
LocalDate date = LocalDate.of(2024, 1, 15);
LocalTime time = LocalTime.of(14, 30, 0);
LocalDateTime specific = LocalDateTime.of(2024, 1, 15, 14, 30);

// Parsing and formatting
LocalDate parsed = LocalDate.parse("2024-01-15");
String formatted = date.format(DateTimeFormatter.ofPattern("dd/MM/yyyy"));
// "15/01/2024"

// Manipulation (immutable, returns new)
LocalDate tomorrow = today.plusDays(1);
LocalDate lastMonth = today.minusMonths(1);
LocalDate nextYear = today.plusYears(1);

// Period (date-based)
Period age = Period.between(LocalDate.of(1990, 1, 1), today);
System.out.println(age.getYears()); // 34

// Duration (time-based)
Duration dur = Duration.between(time, LocalTime.now());
System.out.println(dur.toMinutes());

// Instant (machine time, UTC)
Instant instant = Instant.now();
Instant epoch = Instant.ofEpochSecond(0);

并发基础

Java 并发:Thread(低级)、ExecutorService(线程池——优先)、CompletableFuture(异步组合,类似 Promise)。parallelStream() 使用 ForkJoinPool 进行并行处理。synchronized 块保护共享状态。原子变量(AtomicInteger 等)提供无锁线程安全操作。对于复杂并发,使用 java.util.concurrent 集合(ConcurrentHashMap、BlockingQueue)。

java
import java.util.concurrent.*;

// Create a thread
Thread thread = new Thread(() -> {
    System.out.println("Running in: " + Thread.currentThread().getName());
});
thread.start();
thread.join(); // wait for completion

// ExecutorService (thread pool)
ExecutorService executor = Executors.newFixedThreadPool(4);
Future<Integer> future = executor.submit(() -> {
    Thread.sleep(1000);
    return 42;
});
Integer result = future.get(); // blocks until done
executor.shutdown();

// CompletableFuture (async, Java 8+)
CompletableFuture<String> cf = CompletableFuture
    .supplyAsync(() -> "Hello")
    .thenApply(s -> s + " World")
    .thenApply(String::toUpperCase);
String asyncResult = cf.join(); // "HELLO WORLD"

// Parallel stream
List<Integer> nums = List.of(1, 2, 3, 4, 5);
int sum = nums.parallelStream().mapToInt(Integer::intValue).sum();

// Synchronized
synchronized (this) {
    // only one thread at a time
}

// Atomic variables
AtomicInteger counter = new AtomicInteger(0);
counter.incrementAndGet();
counter.compareAndSet(0, 1);
09

Lambda 表达式

Lambda 语法基础

Lambda(Java 8+)是函数式接口的简洁实现。语法:(params) -> expression 或 (params) -> { statements; }。编译器从目标类型推断参数类型。单参数 lambda 可以省略括号;零参数需要空括号。Lambda 启用函数式编程,是 Stream API 的支柱。

java
// Anonymous class (verbose, pre-Java 8)
Runnable r1 = new Runnable() {
    public void run() {
        System.out.println("Old way");
    }
};

// Lambda expression (Java 8+)
Runnable r2 = () -> System.out.println("Lambda");

// With parameters and body
Comparator<Integer> cmp = (a, b) -> {
    int diff = a - b;
    return diff;
};

// Type inference (omit types)
Comparator<Integer> cmp2 = (a, b) -> a - b;

// Single param, no parentheses needed
Consumer<String> printer = s -> System.out.println(s);

// Zero params need empty parens
Runnable noop = () -> {};

函数式接口

函数式接口恰好有一个抽象方法(SAM 类型)。@FunctionalInterface 注解使编译器强制执行此规则。Lambda 只能针对函数式接口。默认和静态方法允许,不破坏单方法规则。这是使 lambda 在 Java 类型系统中工作的基础。

java
// Functional interface: exactly one abstract method
@FunctionalInterface
interface MathOperation {
    int operate(int a, int b);
}

// Implement with lambda
MathOperation add = (a, b) -> a + b;
MathOperation mul = (a, b) -> a * b;
MathOperation max = (a, b) -> Math.max(a, b);

int result = add.operate(3, 4);  // 7
int m = mul.operate(3, 4);       // 12

// @FunctionalInterface is optional but recommended
// It prevents accidentally adding a second abstract method
// Default and static methods don't count toward the limit
interface StringProcessor {
    String process(String s);
    default StringProcessor andThen(StringProcessor next) {
        return s -> next.process(this.process(s));
    }
}

内置函数式接口

java.util.function 提供约 40 个现成的函数式接口,所以你很少自己编写。核心四个:Function(转换)、Predicate(测试)、Consumer(消费)、Supplier(生产)。Bi- 变体接受两个参数。基本类型变体(IntFunction、ToIntFunction 等)避免自动装箱开销。使用这些而不是创建自定义接口。

java
import java.util.function.*;

// Function<T,R>: input -> output
Function<String, Integer> len = String::length;
int n = len.apply("hello");  // 5

// Predicate<T>: input -> boolean (for filtering)
Predicate<String> isEmpty = String::isEmpty;
boolean e = isEmpty.test("");  // true

// Consumer<T>: input -> void (side effects)
Consumer<String> print = System.out::println;
print.accept("hi");

// Supplier<T>: no input -> output (factories, lazy)
Supplier<Double> random = Math::random;
double r = random.get();

// BiFunction<T,U,R>: two inputs -> output
BiFunction<String, Integer, String> repeat =
    (s, i) -> s.repeat(i);

// Primitives variants avoid boxing
IntFunction<String> f = i -> "n=" + i;
IntPredicate positive = i -> i > 0;
ToIntFunction<String> length = String::length;
IntBinaryOperator sum = (a, b) -> a + b;

方法引用

方法引用(::)是只调用单个方法的 lambda 简写。四种:静态(Class::static)、绑定实例(obj::method)、未绑定实例(Class::method——第一个参数成为接收者)和构造函数(Class::new)。当 lambda 只转发到一个方法时使用它们——更易读。否则坚持使用显式 lambda。

java
import java.util.*;

List<String> names = List.of("alice", "bob", "charlie");

// Lambda form
names.forEach(s -> System.out.println(s));
// Method reference (shorthand)
names.forEach(System.out::println);

// Four kinds of method references:

// 1. Static method: ClassName::staticMethod
names.stream().map(String::toUpperCase);

// 2. Instance method of particular object: instance::method
var printer = System.out;
names.forEach(printer::println);

// 3. Instance method of arbitrary object: ClassName::instanceMethod
List<String> upper = names.stream()
    .map(String::toUpperCase)
    .toList();

// 4. Constructor: ClassName::new
Supplier<ArrayList<String>> factory = ArrayList::new;
ArrayList<String> list = factory.get();

捕获变量(事实上 final)

Lambda 可以捕获局部变量,但它们必须是 final 或'事实上 final'(从未重新赋值)。这是因为 lambda 可能比栈帧活得更久。要解决此问题,使用单元素数组或 AtomicInteger/holder 对象。实例和静态字段没有此限制。lambda 内的 'this' 指向封闭类实例,不是 lambda 本身。

java
import java.util.function.*;

int x = 10;
// Capturing a local variable — must be final or effectively final
Supplier<Integer> getter = () -> x * 2;
System.out.println(getter.get());  // 20

// x = 20;  // ERROR: would break the lambda capture
// Local variables captured by lambdas must be final/effectively final

// Workaround: use an array or wrapper (mutable container)
int[] counter = {0};
Runnable inc = () -> counter[0]++;
inc.run();
inc.run();
System.out.println(counter[0]);  // 2

// Instance/static fields CAN be modified (no restriction)
class Holder {
    int value = 0;
    Runnable bump = () -> value++;  // OK, field access
}

// 'this' inside a lambda refers to the enclosing instance
class Outer {
    String name = "Outer";
    Runnable r = () -> System.out.println(this.name);  // "Outer"
}

构造函数引用

构造函数引用(ClassName::new)简洁地创建新实例。与 Collectors.toCollection() 一起选择结果类型,与数组创建(Type[]::new)一起,以及在工厂模式中。对于 record 和不可变对象,构造函数引用是构建副本的惯用方式。它们自然地与 Function/Supplier 目标配对。

java
import java.util.*;
import java.util.function.*;
import java.util.stream.*;

// Supplier constructor reference
Supplier<StringBuilder> sbFactory = StringBuilder::new;
StringBuilder sb = sbFactory.get();

// Function constructor reference (with one arg)
Function<String, StringBuilder> sbFromString = StringBuilder::new;
StringBuilder named = sbFromString.apply("Hello");

// In streams: collect into a specific collection
List<String> names = List.of("a", "b", "c");
ArrayList<String> copy = names.stream()
    .collect(Collectors.toCollection(ArrayList::new));

// Array constructor reference
IntFunction<String[]> arrayFactory = String[]::new;
String[] arr = arrayFactory.apply(5);  // new String[5]

// Copying via constructor
record Point(int x, int y) {}
Function<Point, Point> copyCtor = Point::new;
Point p = copyCtor.apply(new Point(1, 2));
10

Optional 与空安全

创建 Optional

Optional 是一个可能持有也可能不持有值的容器。无值使用 empty(),确定值非 null 时使用 of()(否则抛出 NPE),null 可能时使用 ofNullable()。Optional 强制调用者显式处理缺失情况。永远不要在期望 Optional 的地方返回 null——那违背了目的。

java
import java.util.Optional;

// empty() — no value
Optional<String> empty = Optional.empty();

// of() — value must be non-null (throws NPE if null)
Optional<String> present = Optional.of("hello");

// ofNullable() — accepts null safely
Optional<String> maybe = Optional.ofNullable(getName());

// From a stream that may produce 0 or 1 elements
Optional<Integer> first = List.of(1, 2, 3).stream().findFirst();

// Common helper pattern
public Optional<User> findUser(long id) {
    User u = db.lookup(id);
    return Optional.ofNullable(u);
}

String getName() { return Math.random() > 0.5 ? "Alice" : null; }
class User {}
class Db { User lookup(long id) { return null; } }
Db db = new Db();

安全消费值

优先使用 ifPresent/ifPresentOrElse 而非 isPresent+get。orElse 返回常量默认值;orElseGet 接受 Supplier,所以默认值延迟计算(当默认值昂贵时重要)。orElseThrow 将缺失转换为异常。目标是永远不要盲目调用 .get()——那重新引入了 Optional 旨在消除的 NPE 风险。

java
import java.util.Optional;

Optional<String> name = Optional.of("Alice");

// isPresent / isEmpty (Java 11+)
if (name.isPresent()) {
    System.out.println(name.get());  // "Alice"
}
// Avoid .get() without checking — throws NoSuchElementException

// ifPresent: run action only if value exists
name.ifPresent(System.out::println);

// ifPresentOrElse (Java 9+)
name.ifPresentOrElse(
    System.out::println,
    () -> System.out.println("No name")
);

// orElse: provide default
String s1 = name.orElse("Anonymous");

// orElseGet: lazy default (computed only if needed)
String s2 = name.orElseGet(() -> expensiveDefault());

// orElseThrow: throw if absent
String s3 = name.orElseThrow(() -> new IllegalStateException("missing"));

String expensiveDefault() { return "computed"; }

用 map 与 flatMap 转换

map 转换包含的值(Optional<T> -> Optional<R>)。当映射函数本身返回 Optional 时使用 flatMap,防止嵌套 Optional。filter 仅在谓词匹配时保留值。链式 map/filter/flatMap 让你构建在第一个空值时短路的管道——比嵌套 null 检查干净得多。

java
import java.util.Optional;

Optional<String> name = Optional.of("Alice");

// map: transform the value if present
Optional<Integer> length = name.map(String::length);  // Optional[5]
Optional<String> upper = name.map(String::toUpperCase);

// flatMap: when the transform itself returns Optional
// (avoids Optional<Optional<T>>)
public Optional<String> findEmail(long id) {
    return Optional.ofNullable(db.get(id));
}
Optional<String> email = Optional.of(1L)
    .flatMap(this::findEmail);  // Optional<email> not Optional<Optional<email>>

// filter: keep only if predicate matches
Optional<Integer> adultAge = Optional.of(25)
    .filter(a -> a >= 18);  // Optional[25]
Optional<Integer> kid = Optional.of(10)
    .filter(a -> a >= 18);  // Optional.empty

// Chain transformations
String label = Optional.of("alice")
    .map(String::strip)
    .filter(s -> !s.isEmpty())
    .map(s -> s.substring(0, 1).toUpperCase() + s.substring(1))
    .orElse("unknown");  // "Alice"

class Db { String get(long id) { return "[email protected]"; } }
Db db = new Db();
Optional<String> findEmail(long id) { return Optional.ofNullable(db.get(id)); }

要避免的反模式

Optional 设计用于返回类型,不是字段或参数。它不是 Serializable 的,作为字段会增加开销。不要在没有检查的情况下使用 .get(),不要使用 isPresent()+get()——那只是冗长的 null 检查。集合已经表达了空性,所以不要用 Optional 包装它们。使用 Optional 作为值可能缺失的返回类型信号。

java
import java.util.Optional;

// BAD: using Optional for fields (not serializable, wastes memory)
class Bad {
    private Optional<String> name;  // DON'T
}

// GOOD: use plain field, return Optional from accessor
class Good {
    private String name;
    public Optional<String> getName() { return Optional.ofNullable(name); }
}

// BAD: Optional as method parameter (clutters API)
public void process(Optional<String> input) {}  // DON'T

// GOOD: method overloading or nullable param
public void process(String input) {}
public void process() { process(null); }

// BAD: .get() without check
String x = findName().get();  // throws if empty

// BAD: .isPresent() + .get() — defeats the purpose
Optional<String> opt = findName();
if (opt.isPresent()) {
    use(opt.get());  // just use ifPresent or map instead
}

// BAD: returning Optional from collections
public Optional<Item> find(...) {
    // Collections already express emptiness — return empty List, not Optional<List>
    return Optional.ofNullable(items);
}

Optional 与 Stream

Optional.stream()(Java 9+)产生 0 或 1 个元素的 Stream,让你优雅地从流中 flatMap Optional。这是在流处理期间跳过缺失值的最干净方式。它避免了冗长的 filter(isPresent).map(get) 模式并保持管道声明式。

java
import java.util.*;
import java.util.stream.*;

// stream() on Optional: 0 or 1 element stream
Optional<String> opt = Optional.of("hi");
opt.stream().forEach(System.out::println);

// Useful: flatMap Optional out of a stream
class User {
    String email;  // may be null
    User(String e) { email = e; }
    Optional<String> getEmail() { return Optional.ofNullable(email); }
}

List<User> users = List.of(
    new User("[email protected]"),
    new User(null),
    new User("[email protected]")
);

// Extract emails, skipping nulls — clean with Optional::stream
List<String> emails = users.stream()
    .flatMap(u -> u.getEmail().stream())
    .toList();  // [[email protected], [email protected]]

// Without Optional::stream you'd need filter+map
List<String> emails2 = users.stream()
    .map(User::getEmail)
    .filter(Optional::isPresent)
    .map(Optional::get)
    .toList();
11

Stream API 深入

创建 Stream

Stream 可以从集合、数组或静态工厂创建。iterate() 和 generate() 产生无限流——始终后跟 limit()。带谓词的 iterate(Java 9+)比裸 iterate 更安全。IntStream/LongStream/DoubleStream 避免数字工作的装箱。Stream 是一次性的:一旦终端操作运行,流就被消费。

java
import java.util.*;
import java.util.stream.*;

// From collections
Stream<String> s1 = List.of("a", "b").stream();
Stream<String> s2 = Set.of("x").stream();

// From arrays
int[] nums = {1, 2, 3};
IntStream s3 = Arrays.stream(nums);
Stream<String> s4 = Arrays.stream(new String[]{"a", "b"});

// Static factory methods
Stream<Integer> s5 = Stream.of(1, 2, 3);
Stream<Integer> s6 = Stream.empty();
Stream<Integer> s7 = Stream.iterate(1, n -> n * 2);  // infinite
Stream<Integer> s8 = Stream.iterate(1, n -> n < 100, n -> n + 1);  // bounded (Java 9+)
Stream<Double> s9 = Stream.generate(Math::random);  // infinite
Stream<String> s10 = Stream.ofNullable(null);  // 0 or 1 element (Java 9+)

// From functions (infinite, must limit)
List<Integer> powers = Stream.iterate(1, n -> n * 2)
    .limit(10)
    .toList();

// Numeric ranges
IntStream range = IntStream.range(0, 5);       // 0,1,2,3,4
IntStream closed = IntStream.rangeClosed(1, 5); // 1,2,3,4,5

中间操作

中间操作是惰性的——它们在调用终端操作之前不运行。filter 保留元素,map 1:1 转换,flatMap 1:多 转换。distinct/sorted/limit/skip 是有状态的。takeWhile/dropWhile(Java 9+)在第一个不匹配元素处停止(不像 filter 扫描所有)。peek 用于调试,不用于生产中的副作用。

java
import java.util.*;
import java.util.stream.*;

List<Integer> nums = List.of(3, 1, 4, 1, 5, 9, 2, 6, 5);

// filter: keep matching
List<Integer> evens = nums.stream()
    .filter(n -> n % 2 == 0).toList();

// map: transform
List<String> labels = nums.stream()
    .map(n -> "n=" + n).toList();

// flatMap: one-to-many
List<Integer> expanded = List.of(List.of(1, 2), List.of(3))
    .stream().flatMap(List::stream).toList();  // [1,2,3]

// distinct: remove duplicates
List<Integer> uniq = nums.stream().distinct().toList();

// sorted
List<Integer> asc = nums.stream().sorted().toList();
List<Integer> desc = nums.stream().sorted(Comparator.reverseOrder()).toList();

// peek: inspect (mainly for debugging)
nums.stream().peek(n -> System.out.println("seen " + n)).count();

// limit / skip
List<Integer> first3 = nums.stream().limit(3).toList();
List<Integer> after2 = nums.stream().skip(2).toList();

// takeWhile / dropWhile (Java 9+)
List<Integer> lt5 = nums.stream().takeWhile(n -> n < 5).toList();

Collector:分组与分区

Collectors.groupingBy 是 Java 流的 SQL GROUP BY。分类器函数定义键;可选的下游 collector 处理每个组(counting、summing、mapping 等)。partitioningBy 是带布尔谓词的特殊情况(恰好两个桶)。传递 TreeMap supplier 获得排序键。这些强大地组合——你可以构建多级分组。

java
import java.util.*;
import java.util.stream.*;

record Person(String name, String city, int age) {}

List<Person> people = List.of(
    new Person("Alice", "NYC", 30),
    new Person("Bob", "LA", 25),
    new Person("Carol", "NYC", 35),
    new Person("Dave", "LA", 40)
);

// groupingBy: Map<key, List<item>>
Map<String, List<Person>> byCity = people.stream()
    .collect(Collectors.groupingBy(Person::city));
// {NYC=[Alice,Carol], LA=[Bob,Dave]}

// groupingBy with downstream collector
Map<String, Long> countByCity = people.stream()
    .collect(Collectors.groupingBy(Person::city, Collectors.counting()));

Map<String, Integer> sumAgeByCity = people.stream()
    .collect(Collectors.groupingBy(Person::city,
        Collectors.summingInt(Person::age)));

Map<String, List<String>> namesByCity = people.stream()
    .collect(Collectors.groupingBy(Person::city,
        Collectors.mapping(Person::name, Collectors.toList())));

// partitioningBy: Map<Boolean, List> (2 buckets)
Map<Boolean, List<Person>> byAge = people.stream()
    .collect(Collectors.partitioningBy(p -> p.age() >= 30));

// groupingBy with TreeMap for sorted keys
Map<String, List<Person>> sorted = people.stream()
    .collect(Collectors.groupingBy(Person::city, TreeMap::new, Collectors.toList()));

归约与统计

reduce 将所有元素合并为单个值——为空流安全提供 identity。summaryStatistics 一次通过给出 count/sum/min/max/avg。Collectors.joining 用于构建分隔字符串很方便。teeing(Java 12+)并行运行两个 collector 并合并结果——当你需要一次通过中的两个聚合(如 min 和 max)时有用。

java
import java.util.*;
import java.util.stream.*;

List<Integer> nums = List.of(1, 2, 3, 4, 5);

// reduce: combine all into one
int sum = nums.stream().reduce(0, Integer::sum);          // 15
Optional<Integer> product = nums.stream().reduce((a, b) -> a * b);  // Optional[120]

// Built-in summary collectors
IntSummaryStatistics stats = nums.stream()
    .mapToInt(Integer::intValue)
    .summaryStatistics();
// stats.getCount(), getSum(), getMin(), getMax(), getAverage()

// Common terminal collectors
long count = nums.stream().collect(Collectors.counting());
double avg = nums.stream().collect(Collectors.averagingInt(Integer::intValue));
int max = nums.stream().collect(Collectors.maxBy(Comparator.naturalOrder())).orElse(0);

// join strings
String joined = List.of("a", "b", "c").stream()
    .collect(Collectors.joining(", ", "[", "]"));  // "[a, b, c]"

// teeing (Java 12+): two collectors, merge results
record Range(int min, int max) {}
Range range = nums.stream().collect(Collectors.teeing(
    Collectors.minBy(Comparator.naturalOrder()),
    Collectors.maxBy(Comparator.naturalOrder()),
    (mn, mx) -> new Range(mn.orElse(0), mx.orElse(0))
));

数字流

IntStream/LongStream/DoubleStream 是避免自动装箱开销的基本特化——数字工作使用它们。mapToInt/mapToLong/mapToDouble 将对象流转换为基本流;boxed() 回去。基本流有专门的终端操作(sum、average、max)返回 OptionalInt/Double 以处理空流。适用于性能敏感的数字管道。

java
import java.util.stream.*;
import java.util.*;

// IntStream / LongStream / DoubleStream avoid boxing
IntStream range = IntStream.rangeClosed(1, 100);
int sum = range.sum();                       // 5050
double avg = IntStream.of(1, 2, 3).average().orElse(0);

// mapToInt / mapToLong / mapToDouble from object stream
int totalAge = people().stream().mapToInt(Person::age).sum();

// boxed: convert primitive stream back to object stream
List<Integer> list = IntStream.range(0, 5).boxed().toList();

// mapToObj: primitive -> objects
List<String> labels = IntStream.range(1, 4)
    .mapToObj(i -> "item-" + i).toList();

// asLongStream / asDoubleStream
LongStream longs = IntStream.range(0, 5).asLongStream();

// Common numeric operations
int max = IntStream.of(3, 1, 4, 1, 5).max().orElse(Integer.MIN_VALUE);
boolean anyEven = IntStream.of(1, 3, 5).anyMatch(n -> n % 2 == 0);

// Iterate to build numeric sequences
List<Integer> fib = Stream.iterate(new int[]{0, 1}, a -> new int[]{a[1], a[0] + a[1]})
    .limit(10).mapToInt(a -> a[0]).boxed().toList();

List<Person> people() { return List.of(new Person("a", "c", 30)); }
record Person(String name, String city, int age) {}

并行流

parallelStream 跨公共 ForkJoinPool(大小为 CPU 核心数)拆分工作。仅对大数据集的 CPU 密集型、无状态、顺序无关操作使用——小数据时开销超过收益。避免共享可变状态(导致竞争)。并行流中的 I/O 阻塞共享池——为阻塞工作使用自定义 ForkJoinPool。在假设并行更快之前测量。

java
import java.util.*;
import java.util.stream.*;

// parallelStream: uses common ForkJoinPool
long sum = List.of(1, 2, 3, 4, 5).parallelStream()
    .mapToInt(Integer::intValue).sum();

// Convert sequential to parallel
long count = IntStream.range(0, 1_000_000).parallel()
    .filter(n -> n % 2 == 0).count();

// Order may differ — use forEachOrdered if order matters
List.of(1, 2, 3, 4).parallelStream()
    .forEachOrdered(System.out::println);

// Collecting preserves encounter order (but work is parallel)
List<Integer> doubled = IntStream.range(0, 1000).parallel()
    .map(n -> n * 2).boxed().toList();

// Custom thread pool (avoid blocking the common pool)
import java.util.concurrent.ForkJoinPool;
ForkJoinPool pool = new ForkJoinPool(8);
int result = pool.submit(() ->
    IntStream.range(0, 1000).parallel().sum()
).get();

// WHEN to use parallel: large dataset, CPU-heavy per element,
// order-independent, stateless operations
// WHEN NOT: small data, I/O-bound, shared mutable state, ordered ops
12

泛型深入

泛型类与方法

泛型启用类型安全的可重用代码。类声明类型参数(<T>);方法也可以(返回类型前的 <T>)。菱形运算符 <> 在构造时推断类型。泛型在编译时检查——它们通过早期捕获类型错误而不是运行时通过 ClassCastException 使集合和 API 更安全。

java
// Generic class
public class Box<T> {
    private T value;
    public void set(T v) { value = v; }
    public T get() { return value; }
}

Box<String> strBox = new Box<>();
strBox.set("hello");
String s = strBox.get();  // no cast needed

// Multiple type parameters
public class Pair<K, V> {
    private final K key;
    private final V value;
    public Pair(K k, V v) { key = k; value = v; }
    public K key() { return key; }
    public V value() { return value; }
}

Pair<String, Integer> p = new Pair<>("age", 30);

// Generic method (independent of class type params)
public static <T> T first(List<T> list) {
    return list.get(0);
}

// Generic method with multiple type params
public static <K, V> Map<K, V> zip(List<K> keys, List<V> values) {
    Map<K, V> m = new HashMap<>();
    for (int i = 0; i < keys.size(); i++) m.put(keys.get(i), values.get(i));
    return m;
}

有界类型参数

有界类型参数(<T extends Bound>)限制可以使用的类型并让你调用 bound 的方法。<T extends Number> 意味着 T 必须是 Number 或子类型。多个 bound 使用 &——最多一个类(必须在前),其余接口。Bound 对于编写需要特定能力(可比性、数字操作)的算法很重要。

java
// Upper bound: T must be a subtype of Number
public static <T extends Number> double sum(List<T> nums) {
    double total = 0;
    for (Number n : nums) total += n.doubleValue();
    return total;
}

sum(List.of(1, 2, 3));        // Integer is a Number
sum(List.of(1.0, 2.5));       // Double is a Number
// sum(List.of("a"));          // compile error

// Multiple bounds: T must extend all (first is class, rest interfaces)
interface Comparable<T> { int compareTo(T o); }
interface Serializable {}

public static <T extends Number & Comparable<T> & Serializable>
    T max(List<T> list) {
    T best = list.get(0);
    for (T t : list) if (t.compareTo(best) > 0) best = t;
    return best;
}

// Bound lets you call methods of the bound
public static <T extends CharSequence> int totalLength(List<T> items) {
    int len = 0;
    for (CharSequence c : items) len += c.length();  // can call .length()
    return len;
}

通配符:?、extends、super

通配符使泛型类型灵活。? extends T(协变)让你读 T 但不能写——用于生产者。? super T(逆变)让你写 T 但只能读 Object——用于消费者。PECS 规则(Producer Extends, Consumer Super)指导使用哪个。copy(dest, src) 是经典示例:dest 是消费者(super),src 是生产者(extends)。

java
import java.util.*;

// ? (unbounded wildcard) — any type
void printAll(List<?> list) {
    for (Object o : list) System.out.println(o);
}

// ? extends T (upper-bounded / covariant) — producer (PECS: Producer Extends)
double sum(List<? extends Number> nums) {
    double total = 0;
    for (Number n : nums) total += n.doubleValue();
    return total;
}
sum(List.of(1, 2, 3));     // List<Integer> OK
sum(List.of(1.0, 2.0));    // List<Double> OK
// nums.add(5);  // ERROR: can't add (don't know exact type)

// ? super T (lower-bounded / contravariant) — consumer (PECS: Consumer Super)
void addNumbers(List<? super Integer> list) {
    list.add(1); list.add(2); list.add(3);  // OK to add Integer
}
addNumbers(new ArrayList<Number>());  // OK
addNumbers(new ArrayList<Object>());  // OK
// Number n = list.get(0);  // only safe to read as Object

// PECS rule: Producer Extends, Consumer Super
// If you read from a collection, use ? extends T
// If you write to a collection, use ? super T
public static <T> void copy(List<? super T> dest, List<? extends T> src) {
    for (T t : src) dest.add(t);
}

类型擦除

Java 泛型使用类型擦除——泛型类型仅在编译时存在;运行时,List<String> 和 List<Integer> 都只是 List。这启用向后兼容但有局限:你不能 new T()、创建泛型数组、使用 instanceof 与泛型或有相同擦除签名的重载方法。当未检查转换将错误类型放入泛型时发生堆污染,将错误延迟到运行时。

java
import java.util.*;

// At runtime, generic types are erased to their bounds (or Object)
// List<String>, List<Integer>, List<?> all become List at runtime

List<String> strings = new ArrayList<>();
List<Integer> ints = new ArrayList<>();
// Runtime: both are just ArrayList

// You CANNOT do these due to erasure:
// new T()              — can't instantiate type param
// new T[]              — can't create generic array
// instanceof List<String>  — only instanceof List (raw)
// class MyException<T> extends Exception  — can't extend Throwable generically
// static T field       — no static generic fields

// Erasure means overloads clash:
// void process(List<String> list) {}
// void process(List<Integer> list) {}  // ERROR: same erasure

// Checking types at runtime requires Class<T>
public static <T> List<T> filter(List<?> items, Class<T> type) {
    List<T> result = new ArrayList<>();
    for (Object o : items) {
        if (type.isInstance(o)) result.add(type.cast(o));
    }
    return result;
}

// Heap pollution: when unchecked warnings lead to runtime ClassCastException
List<String> polluted = (List<String>)(List) List.of(1, 2);  // unchecked
// String s = polluted.get(0);  // ClassCastException at runtime

泛型方法与推断

类型推断让编译器从上下文(参数和目标类型)确定类型参数,所以你很少显式编写它们。菱形运算符 <> 是构造函数的推断。目标类型使用变量的期望类型。仅当推断无法解决歧义时使用显式类型见证(Class.<T>method())。推断使泛型代码读起来和非泛型代码一样干净。

java
import java.util.*;

// Type inference: compiler figures out T from arguments
public static <T> T pick(T a, T b) { return Math.random() > 0.5 ? a : b; }
String s = pick("hello", "world");      // T inferred as String
Number n = pick(1, 2.0);                // T inferred as Number (common supertype)

// Target typing: inference uses the expected type
List<String> list = Collections.emptyList();  // T inferred from target

// Inference with method chains
List<Integer> nums = List.of(1, 2, 3);
String joined = nums.stream()
    .map(Object::toString)              // Stream<String>
    .collect(Collectors.joining(","));  // inferred

// Explicit type witness (rarely needed)
Collections.<String>emptyList();

// Generic constructor
class Holder<T> {
    private T value;
    <U extends T> Holder(U init) { value = init; }  // constructor type param
    T get() { return value; }
}
Holder<Number> h = new Holder<>(42);  // U=Integer, T=Number

import java.util.stream.Collectors;
class Math { static double random() { return 0.5; } }

泛型接口与模式

泛型接口(如 Repository<T,ID>)定义可重用契约。自引用 bound 模式(class X implements Comparable<X>)确保 compareTo 只接受相同类型。类型令牌模式(使用 Class<T> 作为键)绕过擦除以在异构容器中提供运行时类型安全。这些模式是 Spring Data 等框架的支柱。

java
// Generic interface
interface Repository<T, ID> {
    Optional<T> findById(ID id);
    List<T> findAll();
    void save(T entity);
}

// Implement with concrete types
class UserRepository implements Repository<User, Long> {
    public Optional<User> findById(Long id) { /* ... */ return Optional.empty(); }
    public List<User> findAll() { return List.of(); }
    public void save(User entity) {}
}

// Generic interface with self-referencing bound (Comparable pattern)
interface Comparable<T> {
    int compareTo(T other);
}
class Temperature implements Comparable<Temperature> {
    private final double celsius;
    Temperature(double c) { celsius = c; }
    public int compareTo(Temperature other) {
        return Double.compare(celsius, other.celsius);
    }
}

// Generic builder pattern
class Builder<T> {
    private T value;
    public Builder<T> set(T v) { value = v; return this; }
    public T build() { return value; }
}

// Type token pattern for runtime type safety
class TypeSafeMap {
    private final Map<Class<?>, Object> map = new HashMap<>();
    public <T> void put(Class<T> type, T value) { map.put(type, value); }
    public <T> T get(Class<T> type) { return type.cast(map.get(type)); }
}

record User(String name) {}
import java.util.Optional;
13

注解

内置注解

Java 的内置注解:@Override(捕获重写中的拼写错误——始终使用它)、@Deprecated(信号 API 不应使用,带 since/forRemoval 元数据)、@SuppressWarnings(静默特定警告——窄范围使用)、@FunctionalInterface(强制 SAM 规则)。这些是提高编译时安全性和文档的日常注解。

java
import java.util.*;

// @Override: declares intent to override (compiler checks)
class Animal {
    public void sound() { System.out.println("..."); }
}
class Dog extends Animal {
    @Override
    public void sound() { System.out.println("Woof"); }
}

// @Deprecated: marks API as outdated
class OldApi {
    @Deprecated(since = "1.5", forRemoval = true)
    public void legacy() {}
}

// @SuppressWarnings: silence compiler warnings
@SuppressWarnings("unchecked")
List<String> list = (List<String>) new ArrayList();

// @FunctionalInterface: enforces single abstract method
@FunctionalInterface
interface Op { int apply(int a, int b); }

// Common warning keys: unchecked, deprecation, rawtypes, null

// Java 17+ sealed-related
@Deprecated
class ToRemove {}

自定义注解

自定义注解用 @interface 声明。成员看起来像方法但它们是注解属性——可以有默认值。使用 @Target 限制其应用位置(TYPE、METHOD、FIELD 等)和 @Retention 控制可用性。标记注解(无成员)只是标记元素。注解本身不携带行为——处理器(反射、注解工具)读取它们并行动。

java
import java.lang.annotation.*;

// Define an annotation
@Retention(RetentionPolicy.RUNTIME)
@Target(ElementType.METHOD)
public @interface Test {
    String value() default "";
    long timeout() default 0L;
}

// Use it
class MyTests {
    @Test
    public void quickCheck() {}

    @Test(timeout = 5000)
    public void slowCheck() {}

    @Test("custom-name")
    public void named() {}
}

// Annotation with default values
@Retention(RetentionPolicy.RUNTIME)
@Target({ElementType.TYPE, ElementType.FIELD})
public @interface Entity {
    String table() default "";
    String[] columns() default {};
}

@Entity(table = "users", columns = {"id", "name"})
class User {}

// Marker annotation (no members)
@Retention(RetentionPolicy.RUNTIME)
@Target(ElementType.TYPE)
public @interface Service {}

保留与目标

@Retention 控制注解存活多久:SOURCE(仅编译时,如 @Override)、CLASS(在字节码中但运行时不可见——默认)、RUNTIME(通过反射访问)。@Target 限制注解可以出现的位置。Java 8+ 添加了 TYPE_USE 和 TYPE_PARAMETER,让你注解泛型和转换(List<@NonNull String>)。仅在需要反射访问时选择 RUNTIME。

java
import java.lang.annotation.*;

// RetentionPolicy.SOURCE: discarded by compiler (e.g., @Override)
@Retention(RetentionPolicy.SOURCE)
@interface CompileOnly {}

// RetentionPolicy.CLASS: kept in .class but not loaded (default)
@Retention(RetentionPolicy.CLASS)
@interface BytecodeOnly {}

// RetentionPolicy.RUNTIME: available via reflection at runtime
@Retention(RetentionPolicy.RUNTIME)
@interface RuntimeVisible {}

// ElementType targets
@Target(ElementType.TYPE)         // classes, interfaces, enums
@interface ForType {}

@Target(ElementType.METHOD)
@interface ForMethod {}

@Target(ElementType.FIELD)
@interface ForField {}

@Target({ElementType.TYPE_USE, ElementType.TYPE_PARAMETER})
@interface ForTypeUse {}

// Java 8+ type-use annotations (annotate any type occurrence)
@RuntimeVisible String[] names;  // example usage
List<@RuntimeVisible String> typed;

通过反射读取注解

具有 RUNTIME 保留的注解可以通过反射读取:isAnnotationPresent() 检查存在,getAnnotation() 检索它。这就是框架(Spring、JUnit、JAX-RS)声明式连接行为的方式——你标记方法/类,框架扫描并分发。编译时注解处理(注解处理器)是无运行时反射成本的代码生成替代方案。

java
import java.lang.annotation.*;
import java.lang.reflect.*;
import java.util.*;

@Retention(RetentionPolicy.RUNTIME)
@Target(ElementType.METHOD)
@interface Route {
    String path();
    String method() default "GET";
}

class Api {
    @Route(path = "/users", method = "GET")
    public void listUsers() {}

    @Route(path = "/users", method = "POST")
    public void createUser() {}
}

// Scan methods for @Route at runtime
for (Method m : Api.class.getDeclaredMethods()) {
    if (m.isAnnotationPresent(Route.class)) {
        Route r = m.getAnnotation(Route.class);
        System.out.println(r.method() + " " + r.path() + " -> " + m.getName());
    }
}
// Output:
// GET /users -> listUsers
// POST /users -> createUser

// Reading annotations on a class
@Retention(RetentionPolicy.RUNTIME)
@Target(ElementType.TYPE)
@interface Table { String name(); }

@Table(name = "orders")
class Order {}
Table t = Order.class.getAnnotation(Table.class);
System.out.println(t.name());  // "orders"

可重复与元注解

@Repeatable(Java 8+)让你通过定义容器注解多次应用相同注解。@Inherited 使注解传播到子类(仅适用于类级注解)。@Documented 在 Javadoc 中包含注解。带 ANNOTATION_TYPE 的 @Target 创建元注解(注解其他注解的注解)——这就是 Spring 构建可组合注解原型如 @RestController = @Controller + @ResponseBody 的方式。

java
import java.lang.annotation.*;

// Repeatable: allow same annotation multiple times
@Retention(RetentionPolicy.RUNTIME)
@Target(ElementType.METHOD)
@interface Schedule {
    String cron();
}

@Retention(RetentionPolicy.RUNTIME)
@Target(ElementType.METHOD)
@interface Schedules {
    Schedule[] value();  // container annotation
}

// Make Schedule repeatable
@Repeatable(Schedules.class)
@interface Schedule2 {
    String cron();
}

// Now you can repeat it (Java 8+)
class Job {
    @Schedule2(cron = "0 0 * * *")
    @Schedule2(cron = "0 30 * * *")
    public void run() {}
}

// Meta-annotations: annotations on annotations
@Retention(RetentionPolicy.RUNTIME)
@Target(ElementType.ANNOTATION_TYPE)  // can only annotate other annotations
@interface TestCategory {}

@TestCategory
@Retention(RetentionPolicy.RUNTIME)
@interface UnitTest {}

// Inherited: subclass inherits the annotation
@Inherited
@Retention(RetentionPolicy.RUNTIME)
@Target(ElementType.TYPE)
@interface Persistent {}
@Persistent class Base {}
class Child extends Base {}  // Child also has @Persistent

// Documented: appears in Javadoc
@Documented
@interface PublicApi {}
14

反射

Class 对象与获取类

每个加载的类型都有唯一的 Class 对象——反射的入口点。通过类字面量(Type.class)、instance.getClass() 或 Class.forName()(动态加载,抛出 ClassNotFoundException)获取。Class 对象暴露名称、修饰符、超类、接口和类型检查(isInterface、isArray、isEnum、isRecord)。isAssignableFrom 检查多态关系。

java
import java.lang.reflect.*;

// Three ways to get a Class object
Class<String> c1 = String.class;                  // class literal
Class<?> c2 = "hello".getClass();                 // from instance
Class<?> c3 = Class.forName("java.lang.String");  // by name (throws checked)

// Basic introspection
Class<?> c = String.class;
System.out.println(c.getName());          // "java.lang.String"
System.out.println(c.getSimpleName());    // "String"
System.out.println(c.getPackage());       // "package java.lang"
System.out.println(c.getSuperclass());    // "class java.lang.Object"
System.out.println(Modifier.toString(c.getModifiers()));  // "public final"

// Check type relationships
System.out.println(c.isInterface());      // false
System.out.println(c.isArray());          // false
System.out.println(c.isEnum());           // false
System.out.println(c.isRecord());         // false (Java 16+)
System.out.println(CharSequence.class.isAssignableFrom(c));  // true

// Primitive class objects
Class<?> intClass = int.class;
Class<?> intArrayClass = int[].class;
System.out.println(intClass.isPrimitive());  // true

检查字段、方法、构造函数

getDeclaredFields/Methods/Constructors 返回此类中声明的所有成员(包括私有)。getFields/getMethods 仅返回公共成员但包括继承的。要查找特定成员,使用 getDeclaredField(name) 或 getDeclaredMethod(name, paramTypes...)——需要参数类型来消除重载歧义。反射绕过访问控制,除非你调用 setAccessible(true)。

java
import java.lang.reflect.*;
import java.util.*;

class Sample {
    public String name;
    private int count;
    public Sample() {}
    public Sample(String n) { name = n; }
    private void secret() {}
    public int compute(int x) { return x * 2; }
}

Class<?> c = Sample.class;

// Fields: getDeclaredFields includes private; getFields only public
for (Field f : c.getDeclaredFields()) {
    System.out.println(f.getName() + " : " + f.getType().getSimpleName()
        + " (" + Modifier.toString(f.getModifiers()) + ")");
}

// Methods
for (Method m : c.getDeclaredMethods()) {
    System.out.println(m.getName()
        + " params=" + Arrays.toString(m.getParameterTypes())
        + " returns=" + m.getReturnType().getSimpleName());
}

// Constructors
for (Constructor<?> ctor : c.getConstructors()) {
    System.out.println("ctor params=" + Arrays.toString(ctor.getParameterTypes()));
}

// Lookup specific member
Field nameField = c.getDeclaredField("name");
Method compute = c.getDeclaredMethod("compute", int.class);
Constructor<?> ctor = c.getDeclaredConstructor(String.class);

调用方法与创建实例

Method.invoke(obj, args...) 反射式调用方法——始终返回 Object,所以转换结果。setAccessible(true) 绕过 Java 访问检查(私有成员变得可达;可能需要模块上的 --add-opens)。Constructor.newInstance() 创建对象——new 的反射等价物。Array.newInstance 创建运行时已知组件类型的数组。反射比直接调用慢且绕过编译时安全。

java
import java.lang.reflect.*;

class Greeter {
    public String greet(String name) { return "Hello, " + name; }
    private String secret() { return "hidden"; }
}

Object obj = new Greeter();
Class<?> c = obj.getClass();

// Invoke public method
Method greet = c.getMethod("greet", String.class);
String result = (String) greet.invoke(obj, "Alice");  // "Hello, Alice"

// Invoke private method
Method sec = c.getDeclaredMethod("secret");
sec.setAccessible(true);  // bypass access check
String s = (String) sec.invoke(obj);  // "hidden"

// Create instances via constructor
Constructor<?> noArg = c.getConstructor();
Object o1 = noArg.newInstance();

// With args
class Person {
    String name;
    public Person(String n) { name = n; }
    public String toString() { return "Person(" + name + ")"; }
}
Constructor<?> ctor = Person.class.getConstructor(String.class);
Object p = ctor.newInstance("Bob");
System.out.println(p);  // "Person(Bob)"

// Array creation via reflection
Object strArray = Array.newInstance(String.class, 5);
Array.set(strArray, 0, "first");
String v = (String) Array.get(strArray, 0);

读取与修改字段

Field.get(instance) 读取字段值;Field.set(instance, value) 写入。对于基本类型,使用类型特定的访问器(getInt/setInt)以避免装箱。静态字段接受 null 作为实例参数。私有字段需要 setAccessible(true)。反射字段访问是序列化库(Jackson、Gson)和 ORM 框架通用读/写对象状态的方式。

java
import java.lang.reflect.*;

class Config {
    public String env = "dev";
    private int retries = 3;
    public static String VERSION = "1.0";
}

Config cfg = new Config();
Class<?> c = cfg.getClass();

// Read public field
Field env = c.getField("env");
String e = (String) env.get(cfg);  // "dev"

// Read private field
Field retries = c.getDeclaredField("retries");
retries.setAccessible(true);
int r = retries.getInt(cfg);  // 3
// For objects: Object val = field.get(instance);

// Modify fields
env.set(cfg, "prod");
retries.setInt(cfg, 5);
System.out.println(cfg.env);     // "prod"

// Static fields: pass null as the instance
Field version = c.getField("VERSION");
String v = (String) version.get(null);  // "1.0"
version.set(null, "2.0");

// Type-specific getters/setters avoid boxing
// getInt/setInt, getLong/setLong, getBoolean/setBoolean, etc.
// For reference types, use get()/set()

动态代理与用例

java.lang.reflect.Proxy 创建在运行时实现接口的动态代理——InvocationHandler 拦截每个调用。这就是 Spring AOP、Hibernate 懒加载和 Mockito mock 的工作方式。反射驱动大多数 Java 框架(DI、ORM、序列化、测试)但有成本:比直接调用慢、类型安全较弱、模块系统限制。当你需要运行时灵活性时使用它,不是用于普通代码。

java
import java.lang.reflect.*;
import java.util.*;

// JDK dynamic proxy: implements interfaces at runtime
interface UserService {
    String getUser(long id);
    void deleteUser(long id);
}

// InvocationHandler intercepts every method call
class LoggingHandler implements InvocationHandler {
    private final Object target;
    LoggingHandler(Object t) { target = t; }

    public Object invoke(Object proxy, Method method, Object[] args) throws Throwable {
        System.out.println("Calling " + method.getName() + "(" + Arrays.toString(args) + ")");
        long start = System.nanoTime();
        Object result = method.invoke(target, args);
        System.out.println("  -> " + result + " (" + (System.nanoTime() - start) + "ns)");
        return result;
    }
}

UserService real = id -> "user-" + id;
UserService proxied = (UserService) Proxy.newProxyInstance(
    UserService.class.getClassLoader(),
    new Class<?>[]{UserService.class},
    new LoggingHandler(real)
);
proxied.getUser(42);  // logs the call and result

// Common reflection use cases:
// - Frameworks: Spring DI, JPA entities, Jackson serialization
// - AOP/proxies: transactions, logging, security
// - Annotation processing at runtime
// - Test frameworks: JUnit discovers @Test methods
// - Code generation tools

// Caution: reflection is slower, bypasses compile-time checks,
// and can break under modules (--add-opens). Prefer alternatives when possible.
15

JDBC 与数据库访问

连接与 DriverManager

DriverManager.getConnection() 打开数据库连接——始终用 try-with-resources 包装以避免泄漏。从 JDBC 4 起,驱动通过 ServiceLoader 自动注册,所以很少需要 Class.forName()。URL 格式因供应商而异。使用 Properties 设置连接选项(SSL、超时)。在生产中,优先使用连接池(HikariCP)而非直接 DriverManager 调用。

java
import java.sql.*;

// Basic connection (try-with-resources auto-closes)
String url = "jdbc:postgresql://localhost:5432/mydb";
try (Connection conn = DriverManager.getConnection(url, "user", "pass")) {
    System.out.println("Connected: " + conn.getSchema());
    // ... use connection
}

// Modern: no need for Class.forName() with JDBC 4+ (auto-discovery)
// Legacy: Class.forName("org.postgresql.Driver");

// Common URL patterns:
// jdbc:postgresql://host:5432/db
// jdbc:mysql://host:3306/db
// jdbc:oracle:thin:@host:1521:db
// jdbc:sqlite:/path/to/db.sqlite
// jdbc:h2:mem:test  (in-memory H2)

// Connection properties
import java.util.Properties;
Properties props = new Properties();
props.setProperty("user", "user");
props.setProperty("password", "pass");
props.setProperty("ssl", "true");
try (Connection c = DriverManager.getConnection(url, props)) {
    // ...
}

Statement 与 PreparedStatement

对于任何带参数的查询,始终使用 PreparedStatement 而非 Statement——它通过将 SQL 结构与数据分离来防止 SQL 注入。参数通过 1 基索引和类型特定的 setter 设置。PreparedStatement 可以重用(重新设置参数并再次执行)并支持批处理(addBatch/executeBatch)用于批量操作。Statement 仅适用于静态、可信 SQL 如 DDL。

java
import java.sql.*;

// Statement: plain SQL (vulnerable to injection — avoid for user input)
try (Connection c = DriverManager.getConnection("jdbc:h2:mem:", "sa", "");
     Statement st = c.createStatement()) {

    st.execute("CREATE TABLE users (id INT PRIMARY KEY, name VARCHAR(100))");
    st.execute("INSERT INTO users VALUES (1, 'Alice')");

    // NEVER do this with user input — SQL injection!
    // st.execute("SELECT * FROM users WHERE name = '" + userInput + "'");
}

// PreparedStatement: parameterized, safe from injection, can be reused
String sql = "INSERT INTO users (id, name) VALUES (?, ?)";
try (Connection c = DriverManager.getConnection("jdbc:h2:mem:", "sa", "");
     PreparedStatement ps = c.prepareStatement(sql)) {

    // Set parameters by index (1-based!)
    ps.setInt(1, 1);
    ps.setString(2, "Alice");
    ps.executeUpdate();

    // Reuse with different params
    ps.setInt(1, 2);
    ps.setString(2, "Bob");
    ps.executeUpdate();

    // Batch inserts
    for (int i = 3; i <= 100; i++) {
        ps.setInt(1, i);
        ps.setString(2, "user" + i);
        ps.addBatch();
    }
    ps.executeBatch();
}

ResultSet 与查询

ResultSet 是查询行的游标——调用 next() 前进(结束时返回 false)。按名称(可读)或 1 基索引读取列。wasNull() 区分 SQL NULL 和基本默认值(例如,getInt 对 NULL 返回 0)。默认 ResultSet 是只向前的;TYPE_SCROLL_INSENSITIVE + CONCUR_UPDATABLE 启用随机访问和原地更新,尽管这在现代应用中很少使用。

java
import java.sql.*;
import java.util.*;

record User(int id, String name, String email) {}

// Execute query and map rows
String sql = "SELECT id, name, email FROM users WHERE active = ?";
List<User> users = new ArrayList<>();
try (Connection c = getConnection();
     PreparedStatement ps = c.prepareStatement(sql)) {
    ps.setBoolean(1, true);
    try (ResultSet rs = ps.executeQuery()) {
        while (rs.next()) {  // advance to next row
            users.add(new User(
                rs.getInt("id"),       // by column name
                rs.getString("name"),
                rs.getString("email")  // by column name (preferred)
            ));
        }
    }
}

// Column access by index (1-based) or name (more readable)
// rs.getInt(1), rs.getString(2), rs.getBoolean("active")

// Handle NULLs
String nick = rs.getString("nickname");
if (rs.wasNull()) nick = "anonymous";  // distinguish NULL from real null

// Scrollable/updatable ResultSet (needs specific flags)
Statement st = c.createStatement(
    ResultSet.TYPE_SCROLL_INSENSITIVE,
    ResultSet.CONCUR_UPDATABLE
);
ResultSet rs = st.executeQuery("SELECT * FROM users");
rs.absolute(5);  // jump to row 5
rs.updateString("name", "newname");
rs.updateRow();  // persist change

Connection getConnection() throws SQLException {
    return DriverManager.getConnection("jdbc:h2:mem:", "sa", "");
}

事务与批处理

JDBC 默认自动提交每条语句——设置 autoCommit(false) 将语句分组为事务。成功提交,失败回滚。Savepoint 允许事务内部分回滚。隔离级别控制并发更改的可见性(READ_COMMITTED 是常见默认值;SERIALIZABLE 最安全但最慢)。始终恢复 autoCommit 或关闭连接以避免泄漏事务状态。

java
import java.sql.*;

// Transactions: disabled by default (auto-commit = true)
try (Connection c = getConnection()) {
    c.setAutoCommit(false);  // start transaction
    try (PreparedStatement ps = c.prepareStatement(
            "UPDATE accounts SET balance = balance - ? WHERE id = ?")) {
        ps.setInt(1, 100); ps.setInt(2, 1); ps.executeUpdate();  // debit
        ps.setInt(1, -100); ps.setInt(2, 2); ps.executeUpdate();  // credit
    }
    c.commit();  // commit both

    // If any step fails, rollback
} catch (SQLException e) {
    // connection auto-closed; rollback happens implicitly on close if not committed
}

// Explicit rollback pattern
try (Connection c = getConnection()) {
    c.setAutoCommit(false);
    try {
        // ... multiple statements
        c.commit();
    } catch (SQLException e) {
        c.rollback();  // undo all changes in this transaction
        throw e;
    }
}

// Savepoints: partial rollback
Statement st = c.createStatement();
st.execute("INSERT INTO log VALUES (1)");
Savepoint sp = c.setSavepoint("before-risky");
st.execute("INSERT INTO log VALUES (2)");
c.rollback(sp);  // undo only after savepoint
c.commit();

// Transaction isolation levels
c.setTransactionIsolation(Connection.TRANSACTION_READ_COMMITTED);
// Levels: NONE, READ_UNCOMMITTED, READ_COMMITTED, REPEATABLE_READ, SERIALIZABLE

Connection getConnection() throws SQLException {
    return DriverManager.getConnection("jdbc:h2:mem:", "sa", "");
}

连接池(HikariCP)

连接池(HikariCP 是事实标准)保持连接温暖并重用它们,避免每个请求打开新 TCP+auth 连接的 10-100ms 成本。配置最大池大小(受 DB 容量限制)、超时和生命周期。用 getConnection() 借用,通过关闭返回(它回到池中,不是关闭)。始终在关闭时关闭 DataSource。在 Spring Boot 中,HikariCP 自动配置。

java
import com.zaxxer.hikari.*;
import java.sql.*;
import java.util.*;

// HikariCP: high-performance JDBC connection pool
HikariConfig config = new HikariConfig();
config.setJdbcUrl("jdbc:postgresql://localhost:5432/mydb");
config.setUsername("user");
config.setPassword("pass");
config.setMaximumPoolSize(10);
config.setMinimumIdle(2);
config.setConnectionTimeout(30_000);   // ms to wait for a connection
config.setIdleTimeout(600_000);        // ms before idle connections close
config.setMaxLifetime(1_800_000);      // ms max connection lifetime
config.setPoolName("app-pool");

HikariDataSource ds = new HikariDataSource(config);

// Borrow a connection, use it, return it (auto via try-with-resources)
try (Connection c = ds.getConnection();
     PreparedStatement ps = c.prepareStatement("SELECT * FROM users")) {
    try (ResultSet rs = ps.executeQuery()) {
        while (rs.next()) System.out.println(rs.getString("name"));
    }
}  // connection returns to pool here

// Shutdown pool when app stops
ds.close();

// Why pool? Opening a TCP+auth connection is expensive (~10-100ms).
// A pool keeps connections warm and reuses them across requests,
// dramatically reducing latency under load.

// Alternative pools: Apache DBCP, c3p0, Tomcat JDBC, Agroal
16

I/O 与 NIO 深入

InputStream 与 OutputStream(字节)

InputStream/OutputStream 处理原始字节。始终包装在 Buffered* 变体中——未缓冲 I/O 每字节一次系统调用,对性能是灾难性的。read() 在流结束时返回 -1。transferTo()(Java 9+)进行高效批量复制。readAllBytes() 方便但将整个流加载到内存——仅适用于小文件。始终关闭流(try-with-resources)。

java
import java.io.*;

// Read bytes from a file
try (InputStream in = new FileInputStream("input.bin")) {
    int b;
    while ((b = in.read()) != -1) {  // -1 = end of stream
        // process byte
    }
}

// Buffered for performance (8KB default buffer)
try (InputStream in = new BufferedInputStream(new FileInputStream("big.bin"))) {
    byte[] buffer = new byte[8192];
    int read;
    while ((read = in.read(buffer)) != -1) {
        // process buffer[0..read]
    }
}

// readAllBytes (small files only — loads everything into memory)
byte[] all = new FileInputStream("small.bin").readAllBytes();

// Write bytes
try (OutputStream out = new BufferedOutputStream(new FileOutputStream("out.bin"))) {
    out.write(65);  // single byte
    out.write(new byte[]{66, 67, 68});
    out.flush();  // force buffered data to disk
}

// Copy streams (Java 9+)
try (InputStream in = new FileInputStream("src.bin");
     OutputStream out = new FileOutputStream("dst.bin")) {
    in.transferTo(out);  // efficient bulk copy
}

// Standard streams
System.in.read();   // stdin (InputStream)
System.out.write(65);  // stdout (PrintStream)

通道与缓冲区(NIO)

NIO Channel + ByteBuffer 是流的高性能替代。缓冲区有 position/limit/capacity;flip() 从写切换到读模式,clear() 重置为写入,compact() 保留未读数据。直接缓冲区(allocateDirect)存在于 JVM 堆外,避免大 I/O 的复制步骤。通道之间的 transferTo 可以在支持的 OS 上使用零拷贝。当流性能重要时使用 NIO。

java
import java.nio.*;
import java.nio.channels.*;
import java.nio.file.*;
import java.io.*;

// Channel: high-performance, block-oriented I/O
try (FileChannel ch = FileChannel.open(Path.of("data.bin"),
        StandardOpenOption.READ, StandardOpenOption.WRITE)) {

    // ByteBuffer: fixed-capacity block of bytes
    ByteBuffer buf = ByteBuffer.allocate(1024);
    int read = ch.read(buf);  // fill buffer from channel
    buf.flip();  // switch from write mode to read mode

    while (buf.hasRemaining()) {
        byte b = buf.get();
    }
    buf.clear();  // reset for next read (or compact() to preserve unread)
}

// Direct buffer: outside JVM heap, faster for large I/O (no copy)
ByteBuffer direct = ByteBuffer.allocateDirect(64 * 1024);

// Scatter/gather: read into multiple buffers / write from multiple
ByteBuffer header = ByteBuffer.allocate(128);
ByteBuffer body = ByteBuffer.allocate(1024);
ch.read(new ByteBuffer[]{header, body});  // scatter

// Transfer between channels (zero-copy on some OSes)
try (FileChannel src = FileChannel.open(Path.of("a.bin"));
     FileChannel dst = FileChannel.open(Path.of("b.bin"), StandardOpenOption.WRITE)) {
    src.transferTo(0, src.size(), dst);
}

// ByteOrder
buf.order(ByteOrder.LITTLE_ENDIAN);
int value = buf.getInt();

Path 操作(NIO.2)

Path(NIO.2)替代旧 File 类。Path 操作(getFileName、getParent、resolve、normalize、relativize)是纯字符串数学——它们不触碰磁盘。Files.* 方法与文件系统交互:大小、时间戳、权限、符号链接。normalize() 清理 . 和 .. 段。resolveSibling 用于重命名很方便(相同目录,不同名称)。现代代码中优先使用 Path 而非 File。

java
import java.nio.file.*;
import java.nio.file.attribute.*;
import java.io.IOException;

// Creating Path objects
Path p1 = Path.of("a/b/c.txt");          // relative
Path p2 = Path.of("/usr/local/bin");     // absolute
Path p3 = Paths.get("config.json");      // older API, same thing

// Path manipulation (doesn't touch filesystem)
System.out.println(p1.getFileName());    // "c.txt"
System.out.println(p1.getParent());      // "a/b"
System.out.println(p1.getRoot());        // "" (relative)
System.out.println(p2.getRoot());        // "/"
System.out.println(p1.toAbsolutePath()); // "/cwd/a/b/c.txt"
System.out.println(p1.normalize());      // removes . and ..
System.out.println(p1.resolve("d.txt")); // "a/b/c.txt/d.txt"
System.out.println(p1.resolveSibling("x.txt")); // "a/b/x.txt"
System.out.println(p1.relativize(Path.of("a/b")));  // "../.."

// File metadata
Path file = Path.of("notes.txt");
System.out.println(Files.size(file));            // bytes
System.out.println(Files.getLastModifiedTime(file));
System.out.println(Files.isReadable(file));
System.out.println(Files.isWritable(file));

// PosixFilePermissions (Linux/macOS)
String perms = "rwxr-xr--";
Set<PosixFilePermission> set = PosixFilePermissions.fromString(perms);
Files.setPosixFilePermissions(file, set);

// Symbolic links
Path link = Path.of("link.txt");
Files.createSymbolicLink(link, Path.of("target.txt"));
Path target = Files.readSymbolicLink(link);

目录遍历与文件树

Files.list() 列出一层;Files.walk() 递归遍历树(返回 Stream,必须关闭)。Files.find() 在遍历期间按路径和属性过滤。要完全控制,walkFileTree 与 FileVisitor 让你跳过子树、处理错误并在访问目录之前/之后行动。所有都返回持有文件句柄的 Stream——始终使用 try-with-resources。使用最大深度限制昂贵的遍历。

java
import java.nio.file.*;
import java.io.IOException;
import java.util.stream.*;

// List directory entries (one level)
try (Stream<Path> entries = Files.list(Path.of("."))) {
    entries.filter(Files::isRegularFile)
           .map(Path::getFileName)
           .forEach(System.out::println);
}

// Walk file tree recursively (depth-first)
try (Stream<Path> walk = Files.walk(Path.of("/project"))) {
    walk.filter(Files::isRegularFile)
        .filter(p -> p.toString().endsWith(".java"))
        .forEach(System.out::println);
}

// Walk with depth limit
try (Stream<Path> walk = Files.walk(Path.of("/project"), 3)) {
    walk.forEach(System.out::println);
}

// Find with BiPredicate (path + attributes)
try (Stream<Path> found = Files.find(Path.of("/logs"), 10,
        (path, attrs) -> attrs.isRegularFile()
            && attrs.size() > 1_000_000
            && path.toString().endsWith(".log"))) {
    found.forEach(p -> System.out.println("Large log: " + p));
}

// FileVisitor for full control (pre/post visit, skip subtrees)
Files.walkFileTree(Path.of("/project"), new SimpleFileVisitor<>() {
    public FileVisitResult visitFile(Path file, BasicFileAttributes attrs) {
        System.out.println("Visiting " + file);
        return FileVisitResult.CONTINUE;
    }
    public FileVisitResult preVisitDirectory(Path dir, BasicFileAttributes attrs) {
        if (dir.getFileName().toString().equals("target")) {
            return FileVisitResult.SKIP_SUBTREE;  // skip target dirs
        }
        return FileVisitResult.CONTINUE;
    }
});

import java.nio.file.attribute.BasicFileAttributes;
import java.nio.file.attribute.BasicFileAttributes;

WatchService 与文件事件

WatchService 接收已注册目录的文件系统事件(创建、修改、删除)。take() 阻塞直到事件到达;pollEvents() 排空它们。监视父目录并通过 context() 过滤以跟踪特定文件。事件可能被合并或丢失(OVERFLOW)。递归监视需要注册每个子目录。WatchService 是 OS 原生的(Linux 上 inotify,macOS 上 FSEvents)但其 API 是低级的——对于复杂需求考虑库。

java
import java.nio.file.*;
import static java.nio.file.StandardWatchEventKinds.*;
import java.io.IOException;

// Watch a directory for changes
try (WatchService watcher = FileSystems.getDefault().newWatchService()) {
    Path dir = Path.of("/var/log");
    dir.register(watcher,
        ENTRY_CREATE,
        ENTRY_DELETE,
        ENTRY_MODIFY);

    System.out.println("Watching " + dir + "...");
    while (true) {
        WatchKey key = watcher.take();  // blocks until event
        for (WatchEvent<?> event : key.pollEvents()) {
            Path changed = dir.resolve((Path) event.context());
            System.out.println(event.kind() + " -> " + changed);

            if (event.kind() == OVERFLOW) continue;  // events lost

            // React to change
            if (event.kind() == ENTRY_CREATE) {
                System.out.println("New file: " + changed);
            }
        }
        if (!key.reset()) break;  // key no longer valid (dir deleted)
    }
} catch (InterruptedException e) {
    Thread.currentThread().interrupt();
}

// Note: WatchService watches directories, not individual files.
// To watch a file, watch its parent dir and filter by context().
// Events may be coalesced (rapid modifications may yield one event).
// For recursive watching, register each subdirectory manually
// or use a library like Apache Commons IO's FileAlterationMonitor.
17

Record 与模式匹配

Record 基础

Record(Java 16+)是透明的不可变数据载体。头部 `record Name(Type1 f1, Type2 f2)` 生成构造函数、访问器(f1()、f2()——不是 getF1())、equals、hashCode 和 toString。它们是 DTO、值对象和函数结果的理想选择。紧凑构造函数(只是 `{ ... }`)验证或规范化而不重新赋值字段。Record 可以实现接口但不能扩展类。

java
// Record: concise immutable data carrier (Java 16+)
public record Point(int x, int y) {}

// Equivalent to writing:
// - final class with private final int x, y
// - constructor Point(int x, int y)
// - accessors x(), y()
// - equals, hashCode, toString (all auto-generated)

Point p = new Point(3, 4);
System.out.println(p.x());          // 3 (accessor, not getX())
System.out.println(p.y());          // 4
System.out.println(p);              // "Point[x=3, y=4]"
System.out.println(p.equals(new Point(3, 4)));  // true

// Immutable: no setters
// p.setX(5);  // no such method

// Custom record with validation
public record Age(int value) {
    public Age {
        if (value < 0 || value > 150) {
            throw new IllegalArgumentException("Invalid age: " + value);
        }
    }
}

// Record with multiple components
public record Employee(String name, int id, double salary, String dept) {}

// Records implement interfaces
public interface Named { String name(); }
public record Customer(String name, long id) implements Named {}

紧凑构造函数与自定义方法

紧凑构造函数(`public Name { ... }`)在字段赋值之前运行——赋值给参数以规范化,编译器将其赋值给字段。Record 可以有额外方法和静态工厂但头部之外没有实例字段。非规范构造函数必须通过 this(...) 委托给规范构造函数。使用静态工厂进行更清晰的构造(Point.origin())和缓存常见实例。

java
// Compact constructor: validation without reassigning
public record Email(String address) {
    public Email {
        if (!address.contains("@")) {
            throw new IllegalArgumentException("Bad email: " + address);
        }
        address = address.toLowerCase().strip();  // normalize
    }
}

// Add custom methods (but not mutable fields)
public record Money(double amount, String currency) {
    public Money plus(Money other) {
        if (!currency.equals(other.currency)) {
            throw new IllegalArgumentException("Currency mismatch");
        }
        return new Money(amount + other.amount, currency);
    }
    public Money times(int factor) {
        return new Money(amount * factor, currency);
    }
    public static Money usd(double amt) { return new Money(amt, "USD"); }
}

// Canonical + non-canonical constructors
public record Range(int start, int end) {
    // Compact canonical (validation)
    public Range {
        if (start > end) throw new IllegalArgumentException();
    }
    // Non-canonical convenience constructor
    public Range(int end) {
        this(0, end);
    }
}

// Static factories are common on records
public record Point(int x, int y) {
    public static Point origin() { return new Point(0, 0); }
    public static Point of(int x, int y) { return new Point(x, y); }
}

密封类

密封类/接口(Java 17+)通过 `permits` 子句限制哪些类型可以扩展它们。每个允许的子类型必须是 final、sealed 或 non-sealed。与 record 结合,它们形成代数数据类型(封闭继承 + 不可变数据)。关键好处:编译器知道所有子类型,所以 switch 表达式可以无需 default 分支就穷尽——如果你添加新子类型,编译器标记每个需要更新的 switch。

java
// Sealed class: restricts which classes can extend it (Java 17+)
public sealed interface Shape
    permits Circle, Square, Triangle {}

public record Circle(double radius) implements Shape {}
public record Square(double side) implements Shape {}
public record Triangle(double base, double height) implements Shape {}

// Every permitted subtype must be final, sealed, or non-sealed
public non-sealed class WeirdShape implements Shape {}  // open again

// Why sealed? Enables exhaustive pattern matching
public double area(Shape s) {
    return switch (s) {
        case Circle c -> Math.PI * c.radius() * c.radius();
        case Square sq -> sq.side() * sq.side();
        case Triangle t -> 0.5 * t.base() * t.height();
        // No default needed — compiler knows all subtypes
    };
}

// Sealed class hierarchy with records = algebraic data types
// Combines: closed inheritance (sealed) + immutable data (record)

// Sealed classes also work with classes (not just interfaces)
public sealed abstract class Result permits Success, Failure {}
public final class Success<T>(T value) extends Result {}
public final class Failure(String error) extends Result {}

instanceof 模式匹配

instanceof 模式匹配(Java 16+)声明一个仅在测试成功时绑定的变量,消除显式转换。变量的作用域从模式的真值流出——可在 && 延续和提前返回后使用。Java 21 添加了 switch 中的模式匹配(case Type var when guard),启用带守卫的基于类型的分发。这使类型检查代码更简洁且不易出错。

java
// Old way: cast after instanceof
Object obj = "hello";
if (obj instanceof String) {
    String s = (String) obj;  // redundant cast
    System.out.println(s.length());
}

// Java 16+: pattern variable
if (obj instanceof String s) {
    System.out.println(s.length());  // s is in scope, no cast
}

// The variable is scoped by the pattern's truth
if (obj instanceof String s && s.length() > 3) {
    System.out.println("Long string: " + s);
}
// s is NOT in scope here if the && short-circuited

// Negation scoping
if (!(obj instanceof String s)) {
    // s NOT in scope here
    return;
}
// s IS in scope here (the instanceof must have been true)

// Combining with other patterns
Object data = 42;
String desc = switch (data) {
    case Integer i when i > 0 -> "positive int: " + i;
    case Integer i -> "non-positive int: " + i;
    case String s -> "string: " + s;
    case null -> "null";
    default -> "other";
};

Switch 模式匹配(Java 21)

Switch 模式匹配(Java 21,最终)让你按类型切换、解构 record 并添加守卫(when)。与密封类型结合,编译器验证穷尽性——如果覆盖所有子类型则无需 default。Record 模式(case Point(int x, int y))一步解构。null case 是显式的(无 NPE)。这使 Java 接近 ML/Scala 的模式匹配,用于声明式建模领域。

java
// Java 21: switch pattern matching with type patterns and guards
sealed interface Event permits Login, Logout, Message, Error {}
record Login(String user) implements Event {}
record Logout(String user) implements Event {}
record Message(String from, String text) implements Event {}
record Error(int code, String detail) implements Event {}

String describe(Event e) {
    return switch (e) {
        case Login l -> l.user() + " logged in";
        case Logout l -> l.user() + " logged out";
        case Message m when m.from().equals("system") -> "[system] " + m.text();
        case Message m -> m.from() + ": " + m.text();
        case Error(int code, String detail) when code >= 500 -> "SERVER ERROR " + code;
        case Error(int code, String detail) -> "error " + code + ": " + detail;
        case null -> "no event";  // explicit null handling
    };
}

// Record patterns: destructure records in one go
record Point(int x, int y) {}
String classify(Object o) {
    return switch (o) {
        case Point(int x, int y) when x == y -> "diagonal";
        case Point(int x, int y) -> "point at (" + x + "," + y + ")";
        default -> "not a point";
    };
}

// Nested patterns
record Box(Point p) {}
String describe2(Object o) {
    return switch (o) {
        case Box(Point(int x, int y)) -> "box at " + x + "," + y;
        default -> "unknown";
    };
}
18

模块(JPMS)

module-info.java 基础

module-info.java 声明模块(Java 9+ JPMS)。requires 添加依赖;exports 使包可访问;opens 允许反射访问(序列化/DI 框架需要);uses/provides 连接 ServiceLoader。没有此文件,代码作为'未命名模块'存在于类路径上,具有遗留行为。模块提供强封装(仅导出的包是公共的)和可靠配置(显式依赖)。

java
// File: src/com.example.app/module-info.java
module com.example.app {
    requires java.sql;           // depends on java.sql module
    requires transitive com.example.lib;  // re-export (consumers see it too)
    requires static java.annotation;  // compile-time only (optional at runtime)

    exports com.example.app.api;       // public API visible to all
    exports com.example.app.internal to com.example.test;  // qualified export

    opens com.example.app.model to com.fasterxml.jackson.databind;  // reflection only
    opens com.example.app.dynamic;  // open to all for reflection

    uses com.example.app.spi.Plugin;  // service consumer
    provides com.example.app.spi.Plugin with com.example.app.plugins.DefaultPlugin;  // service provider
}

// Key directives:
// requires: depends on another module
// exports: makes packages public to other modules
// opens: allows deep reflection (for frameworks like Jackson, Hibernate)
// uses/provides: service loader integration

// A module without module-info.java is an "unnamed module" (classpath behavior)

requires、exports、opens

requires 声明依赖;requires transitive 传播它(当你的公共 API 暴露该模块的类型时使用)。exports 使包公共;exports to 限制为命名模块(限定导出)。opens 授予反射访问(深度反射)——对于 setAccessible(true) 的框架很重要。exports(公共 API)和 opens(反射)之间的区别是关键:强封装是默认的,你按包选择加入。

java
// Module A: com.example.library
module com.example.library {
    // Public API anyone can use
    exports com.example.library.api;

    // Internal package: only visible to specific modules
    exports com.example.library.internal to com.example.app;

    // Allow reflection for frameworks (Jackson, JPA)
    opens com.example.library.model;

    // Only specific module can reflect
    opens com.example.library.config to com.example.app;

    // Dependencies
    requires java.logging;
    requires transitive java.sql;  // consumers of A also get java.sql
}

// Module B: com.example.app (consumer)
module com.example.app {
    requires com.example.library;  // use A's exported packages
    // Note: transitive means java.sql is also available here

    requires com.fasterxml.jackson.databind;
}

// 'requires transitive X' means: any module requiring this one
// also reads X. Use when your exported API exposes X's types.

// 'opens' vs 'exports':
// exports: compile-time + runtime access to public members
// opens: runtime reflective access to ALL members (including private)

// Reflective access without 'opens' fails with InaccessibleObjectException
// in Java 16+ (strong encapsulation enforced by default).

ServiceLoader 与服务

ServiceLoader 实现 SPI 模式:一个模块中的接口,运行时发现的实现。API 模块导出接口;提供者模块声明 `provides X with Y`;消费者模块声明 `uses X`。ServiceLoader.load(X.class) 找到模块路径上的所有提供者。这将接口与实现解耦——JDBC 驱动、日志后端(SLF4J)和 Charset 提供者都这样工作。对实现没有编译时依赖。

java
// SPI (Service Provider Interface) pattern with modules

// 1. Define the service interface in an API module
module com.example.spi {
    exports com.example.spi;
}
package com.example.spi;
public interface Plugin {
    String name();
    void run();
}

// 2. Implement in a provider module
module com.example.plugin.impl {
    requires com.example.spi;
    provides com.example.spi.Plugin with com.example.plugin.impl.DefaultPlugin;
}
package com.example.plugin.impl;
import com.example.spi.Plugin;
public class DefaultPlugin implements Plugin {
    public String name() { return "default"; }
    public void run() { System.out.println("running"); }
}

// 3. Consume in an app module
module com.example.app {
    requires com.example.spi;
    uses com.example.spi.Plugin;  // declares intent to load services
}

// Loading services at runtime
import java.util.ServiceLoader;
ServiceLoader<Plugin> loader = ServiceLoader.load(Plugin.class);
for (Plugin p : loader) {
    System.out.println("Found: " + p.name());
    p.run();
}

// ServiceLoader is how JDBC drivers, SLF4J backends, and many
// plugin systems are discovered without compile-time dependencies.

模块路径与类路径

模块路径(--module-path)持有具有强封装的模块化 JAR;类路径(-cp)持有作为未命名模块的遗留 JAR,无封装。模块化 JAR 在两者上都工作。自动模块是没有 module-info 放在模块路径上的 JAR——其名称来自文件名或 Automatic-Module-Name 清单属性。--add-opens 是强封装下破坏的遗留反射的逃生舱。

java
// Compile a module
//   javac -d out/com.example.app \
//         --module-source-path src \
//         --module com.example.app

// Run a modular app
//   java --module-path out --module com.example.app/com.example.app.Main

// Package as a modular JAR (includes module-info.class)
//   jar --create --file app.jar --main-class com.example.app.Main -C out/com.example.app .

// Module path vs classpath:
// --module-path: modules with module-info, strong encapsulation enforced
// --class-path (or -cp): legacy "unnamed module", everything public, no encapsulation

// Mixing: modular JARs can be used on the classpath too (automatic module)
//   java -cp lib/app.jar:lib/dep.jar com.example.app.Main
// An automatic module: a JAR without module-info on the module path.
// Its name is derived from the JAR filename (Automatic-Module-Name in MANIFEST.MF
// gives an explicit name).

// Inspect a module JAR
//   jar --describe-module --file app.jar

// List observable modules
//   java --list-modules
//   java --describe-module java.sql

// Add opens at runtime for legacy reflection (escape hatch)
//   java --add-opens com.example.app/com.example.app.internal=ALL-UNNAMED

jdeps 与 jlink(自定义运行时)

jdeps 分析字节码以列出模块依赖——适用于迁移到模块和查找未使用的依赖。jlink 创建仅包含你的应用所需模块的自定义 JRE,产生自包含、更小、启动更快的运行时。这适用于 Docker 镜像和安装程序:发布应用加上 30-50MB JRE,而不是需要 300MB JDK 安装。jdeps + jlink 一起启用精简、自包含的 Java 部署。

java
# jdeps: analyze dependencies (find unused, list required modules)

# List dependencies of a JAR
jdeps --module-path lib app.jar

# Generate module-info.java for an existing JAR
jdeps --generate-module-info ./out app.jar

# Show which JDK modules a JAR uses
jdeps --print-module-deps --ignore-missing-deps app.jar
# Output: java.base,java.logging,java.sql

# jlink: create a custom stripped-down JRE containing only needed modules
jlink \
  --module-path "$JAVA_HOME/jmods:./out" \
  --add-modules com.example.app \
  --output ./custom-jre \
  --strip-debug \
  --compress=zip-6 \
  --no-header-files \
  --no-man-pages \
  --launcher app=com.example.app/com.example.app.Main

# The custom JRE is self-contained:
#   ./custom-jre/bin/app   # launches the app
#   ./custom-jre/bin/java  # the stripped JVM

# Benefits of jlink:
# - Smaller distribution (only needed modules)
# - Faster startup (less to load)
# - No need to install Java on target machine
# - Can cross-target (different OS/arch) with matching jmods

# Common workflow:
# 1. jdeps to find required modules
# 2. jlink to build a custom runtime
# 3. Package app + custom JRE together (Docker image, installer)
19

并发深入

锁:ReentrantLock 与 ReadWriteLock

ReentrantLock 比 synchronized 提供更多控制:tryLock(非阻塞/定时)、公平性、可中断性和锁状态检查。始终在 finally 中 unlock。ReadWriteLock 允许多个并发读取者但独占写入者——适用于读密集型缓存。StampedLock(Java 8+)添加乐观读取以获得更好的读取吞吐量。简单情况优先使用 synchronized;需要其高级功能时使用 Lock。

java
import java.util.concurrent.locks.*;

// ReentrantLock: more flexible than synchronized
class Counter {
    private final ReentrantLock lock = new ReentrantLock();
    private int count = 0;

    public void increment() {
        lock.lock();
        try {
            count++;
        } finally {
            lock.unlock();  // MUST be in finally
        }
    }

    public int get() {
        lock.lock();
        try { return count; }
        finally { lock.unlock(); }
    }
}

// tryLock with timeout (avoids deadlock-induced hangs)
if (lock.tryLock(1, java.util.concurrent.TimeUnit.SECONDS)) {
    try { /* work */ } finally { lock.unlock(); }
}

// Fair lock (FIFO ordering, slower)
ReentrantLock fair = new ReentrantLock(true);

// ReadWriteLock: many readers OR one writer
class Cache {
    private final ReentrantReadWriteLock rw = new ReentrantReadWriteLock();
    private final Lock read = rw.readLock();
    private final Lock write = rw.writeLock();
    private java.util.Map<String, String> data = new java.util.HashMap<>();

    public String get(String key) {
        read.lock();
        try { return data.get(key); }
        finally { read.unlock(); }
    }
    public void put(String key, String val) {
        write.lock();
        try { data.put(key, val); }
        finally { write.unlock(); }
    }
}

并发集合

ConcurrentHashMap 是主力线程安全映射——使用 compute/merge 进行原子更新而非 check-then-act。CopyOnWriteArrayList 最适合读密集、写稀少的列表(事件监听器)——写入复制数组。BlockingQueue 是生产者-消费者管道的支柱(满时 put 阻塞,空时 take 阻塞)。ConcurrentLinkedQueue 是无界非阻塞的。这些替代 synchronized 包装器(Collections.synchronizedX),后者粗粒度且较慢。

java
import java.util.concurrent.*;
import java.util.*;

// ConcurrentHashMap: thread-safe HashMap (no null keys/values)
ConcurrentHashMap<String, Integer> map = new ConcurrentHashMap<>();
map.put("a", 1);
map.putIfAbsent("a", 2);  // only if absent
map.compute("a", (k, v) -> v == null ? 1 : v + 1);  // atomic update
map.merge("a", 1, Integer::sum);  // add 1 atomically
Integer val = map.getOrDefault("a", 0);

// CopyOnWriteArrayList: snapshot semantics, fast reads, slow writes
CopyOnWriteArrayList<String> listeners = new CopyOnWriteArrayList<>();
listeners.add("x");  // copies entire array
for (String l : listeners) { /* safe iteration, no ConcurrentModificationException */ }

// BlockingQueue: producer-consumer pattern
BlockingQueue<String> queue = new ArrayBlockingQueue<>(100);
// Producer
queue.put("task");        // blocks if full
boolean added = queue.offer("task", 1, TimeUnit.SECONDS);  // timed
// Consumer
String task = queue.take();  // blocks if empty
String polled = queue.poll(1, TimeUnit.SECONDS);

// ConcurrentLinkedQueue: non-blocking, unbounded
Queue<String> q = new ConcurrentLinkedQueue<>();
q.offer("a");

// SkipListMap / SkipListSet: concurrent sorted collections
ConcurrentNavigableMap<Integer, String> sorted = new ConcurrentSkipListMap<>();

import java.util.concurrent.TimeUnit;

CountDownLatch 与 CyclicBarrier

CountDownLatch 是一次性门——N 个线程倒计时,其他等待;不可重置。CyclicBarrier 可重用——线程在汇合点互相等待,所有到达时可选执行操作。Phaser 最灵活:可变参与方、多阶段、树结构。启动协调使用 latch,并行多阶段算法使用 barrier,动态参与者计数使用 phaser。

java
import java.util.concurrent.*;
import java.util.*;

// CountDownLatch: one-shot gate, N threads must arrive before proceeding
CountDownLatch ready = new CountDownLatch(3);
List<String> results = Collections.synchronizedList(new ArrayList<>());

for (int i = 0; i < 3; i++) {
    final int id = i;
    new Thread(() -> {
        try { Thread.sleep(id * 100); } catch (InterruptedException e) {}
        results.add("worker-" + id);
        ready.countDown();  // signal done
    }).start();
}
ready.await();  // main thread blocks until count reaches 0
System.out.println("All done: " + results);

// CyclicBarrier: reusable barrier, threads wait for each other
CyclicBarrier barrier = new CyclicBarrier(3, () ->
    System.out.println("--- phase complete ---"));

Runnable worker = () -> {
    try {
        System.out.println(Thread.currentThread().getName() + " phase 1");
        barrier.await();  // wait for all 3
        System.out.println(Thread.currentThread().getName() + " phase 2");
        barrier.await();  // reusable for next phase
    } catch (Exception e) {}
};
for (int i = 0; i < 3; i++) new Thread(worker).start();

// Phaser: more flexible (variable parties, multiple phases)
Phaser phaser = new Phaser(3);
phaser.register();  // dynamically add a party
phaser.arriveAndAwaitAdvance();

Semaphore 与 Exchanger

Semaphore 控制对 N 个许可的访问——acquire 阻塞直到有可用,release 返回它。用于速率限制、连接池或任何有界资源场景。tryAcquire 提供非阻塞和定时变体。Exchanger 让两个线程在汇合点交换值——小众但优雅,适用于两个线程交换缓冲区的管道设计。两者都在 java.util.concurrent 中,对于某些协调模式比锁更低级。

java
import java.util.concurrent.*;

// Semaphore: limit concurrent access to N permits
Semaphore pool = new Semaphore(5);  // 5 concurrent allowed

pool.acquire();  // blocks until a permit is available
try {
    // critical section (at most 5 threads here at once)
    System.out.println("Working, permits left: " + pool.availablePermits());
} finally {
    pool.release();  // return the permit
}

// Try-acquire (non-blocking)
if (pool.tryAcquire()) {
    try { /* work */ } finally { pool.release(); }
} else {
    System.out.println("Too busy, try later");
}

// Timed acquire
if (pool.tryAcquire(1, TimeUnit.SECONDS)) {
    try { /* work */ } finally { pool.release(); }
}

// Use case: rate limiting, connection pools, parking lots

// Exchanger: two threads swap values
Exchanger<String> ex = new Exchanger<>();
new Thread(() -> {
    try {
        String got = ex.exchange("from-A");  // gives "from-A", receives "from-B"
        System.out.println("A got: " + got);
    } catch (InterruptedException e) {}
}).start();
String got = ex.exchange("from-B");  // gives "from-B", receives "from-A"
System.out.println("B got: " + got);

import java.util.concurrent.TimeUnit;

CompletableFuture 高级

CompletableFuture 是 Java 的 Promise——用 thenApply(map)、thenCompose(flatMap)、thenCombine(zip 两个)组合异步工作。allOf 等待所有,anyOf 等待第一个。exceptionally 从错误恢复;handle 覆盖两者。orTimeout(Java 9+)如果太长则取消。始终传递显式 executor——默认 commonPool 在阻塞工作下可能饥饿。这是 Java 中响应式异步代码的基础。

java
import java.util.concurrent.*;
import java.util.*;

ExecutorService pool = Executors.newFixedThreadPool(4);

// Async composition (like JS Promises)
CompletableFuture<String> future = CompletableFuture
    .supplyAsync(() -> "hello", pool)            // async supplier
    .thenApply(String::toUpperCase)              // transform
    .thenCompose(s -> CompletableFuture.supplyAsync(  // flatMap
        () -> s + " world"))
    .thenApply(s -> s + "!");

System.out.println(future.join());  // "HELLO world!"

// Combine two independent futures
CompletableFuture<Integer> a = CompletableFuture.supplyAsync(() -> 10);
CompletableFuture<Integer> b = CompletableFuture.supplyAsync(() -> 20);
CompletableFuture<Integer> sum = a.thenCombine(b, Integer::sum);
System.out.println(sum.join());  // 30

// Run multiple and wait for all (or any)
List<CompletableFuture<Integer>> futures = List.of(
    CompletableFuture.supplyAsync(() -> 1, pool),
    CompletableFuture.supplyAsync(() -> 2, pool),
    CompletableFuture.supplyAsync(() -> 3, pool)
);
CompletableFuture<Void> all = CompletableFuture.allOf(
    futures.toArray(new CompletableFuture[0]));
all.join();
List<Integer> results = futures.stream().map(CompletableFuture::join).toList();

// Error handling
CompletableFuture<String> safe = CompletableFuture
    .supplyAsync(() -> { throw new RuntimeException("boom"); })
    .exceptionally(ex -> "fallback: " + ex.getMessage());
System.out.println(safe.join());  // "fallback: boom"

// handle: both success and failure
CompletableFuture<String> handled = CompletableFuture
    .supplyAsync(() -> "ok")
    .handle((val, ex) -> ex == null ? val : "error");

// Timeout (Java 9+)
CompletableFuture<String> timed = CompletableFuture
    .supplyAsync(() -> { Thread.sleep(5000); return "slow"; })
    .orTimeout(1, TimeUnit.SECONDS)
    .exceptionally(ex -> "timed out");

虚拟线程(Java 21)

虚拟线程(Java 21)是由 JVM 在少量载体(平台)线程池上调度的轻量级线程。当虚拟线程在 I/O 上阻塞时,它被挂起,载体运行另一个——所以你可以有数百万并发阻塞操作。这让你编写简单的阻塞代码而不是复杂的响应式/异步链。I/O 密集型工作负载(HTTP 处理程序、DB 调用)使用它们;CPU 密集型工作,平台线程或 parallelStream 仍然合适。

java
import java.util.concurrent.*;
import java.util.*;

// Virtual threads: lightweight, cheap, millions possible
// Java 21 LTS feature — "Project Loom"

// Start a virtual thread
Thread vt = Thread.ofVirtual().start(() -> {
    System.out.println("Running on: " + Thread);
});

// Builder pattern
Thread vt2 = Thread.ofVirtual().name("worker-1").start(() -> {
    // blocking I/O here is fine — virtual thread yields, not the OS thread
});

// Per-thread factory
ThreadFactory factory = Thread.ofVirtual().factory();

// ExecutorService for virtual threads (Java 21+)
try (ExecutorService es = Executors.newVirtualThreadPerTaskExecutor()) {
    // Submit a million tasks — each gets its own virtual thread
    List<Future<String>> futures = new ArrayList<>();
    for (int i = 0; i < 1_000_000; i++) {
        final int id = i;
        futures.add(es.submit(() -> {
            Thread.sleep(100);  // blocking call is cheap on virtual threads
            return "done-" + id;
        }));
    }
    // virtual threads yield when blocking, so 1M concurrent is feasible
}

// Why virtual threads?
// - Platform threads (OS threads) are heavy (~1MB stack, kernel scheduling)
// - Virtual threads are user-mode, ~KB, scheduled by JVM on a small carrier pool
// - Blocking I/O on a virtual thread doesn't block a platform thread
// - Lets you write straightforward blocking code at scale (no reactive complexity)

// Best practice: use virtual threads for I/O-bound work,
// NOT for CPU-bound work (use platform threads / parallelStream).
20

集合框架深入

Comparator 与排序

Comparator.comparing(keyExtractor) 从键函数构建比较器——比编写原始 compare 逻辑干净得多。thenComparing 链接次要排序键。nullsFirst/nullsLast 安全处理 null。使用 comparingInt/comparingLong/comparingDouble 避免自动装箱。List.sort() 原地排序(仅可变列表);Stream.sorted() 返回新的排序流。Comparator 驱动排序、TreeSet/TreeMap 排序和流操作。

java
import java.util.*;
import java.util.stream.*;

record Person(String name, int age) {}

List<Person> people = List.of(
    new Person("Alice", 30),
    new Person("Bob", 25),
    new Person("Carol", 30),
    new Person("Dave", 25)
);

// Comparator.comparing (key extractor)
people.stream().sorted(Comparator.comparing(Person::name)).toList();
people.stream().sorted(Comparator.comparingInt(Person::age)).toList();

// Reversed
people.stream().sorted(Comparator.comparing(Person::age).reversed()).toList();

// Chained (thenComparing): sort by age, then by name
List<Person> sorted = people.stream().sorted(
    Comparator.comparingInt(Person::age)
              .thenComparing(Person::name)
).toList();
// [Bob(25), Dave(25), Alice(30), Carol(30)]

// Nulls handling
Comparator<String> cmp = Comparator.nullsFirst(Comparator.naturalOrder());
List.of("b", null, "a").stream().sorted(cmp).toList();  // [null, a, b]

// Custom comparator
Comparator<Person> byNameLen = (a, b) -> a.name().length() - b.name().length();

// Mutable list sort
List<String> names = new ArrayList<>(List.of("charlie", "alice", "bob"));
names.sort(Comparator.naturalOrder());
// names = [alice, bob, charlie]

// Comparing with primitive specializations avoids boxing
Comparator<Person> byAge = Comparator.comparingInt(Person::age);

不可修改与不可变集合

List.of/Set.of/Map.of(Java 9+)创建真正不可变的集合——无 null、无修改。Collections.unmodifiableX 创建只读视图,仍反映对支持集合的更改。List.copyOf(Java 10+)制作独立的不可变副本。Arrays.asList 是数组的固定大小视图(set 工作,add/remove 不工作)。按需选择:常量使用不可变工厂,安全暴露内部使用不可修改视图,防御性副本使用 copyOf。

java
import java.util.*;

// Java 9+ immutable factories (List.of, Set.of, Map.of)
List<String> immutable = List.of("a", "b", "c");
Set<Integer> set = Set.of(1, 2, 3);
Map<String, Integer> map = Map.of("a", 1, "b", 2);
Map<String, Integer> bigMap = Map.ofEntries(
    Map.entry("x", 1), Map.entry("y", 2), Map.entry("z", 3)
);
// immutable.add("d");  // throws UnsupportedOperationException
// immutable.set(0, "z");  // throws
// nulls not allowed in these immutable collections

// Unmodifiable view (wraps an existing collection)
List<String> mutable = new ArrayList<>(List.of("a", "b"));
List<String> view = Collections.unmodifiableList(mutable);
// view.add("c");  // throws
mutable.add("c");  // but changes to backing list ARE visible in view
System.out.println(view);  // [a, b, c]

// CopyOf (Java 10+): creates immutable copy
List<String> copy = List.copyOf(mutable);  // independent immutable copy

// Arrays.asList: fixed-size view of an array
String[] arr = {"a", "b"};
List<String> fixed = Arrays.asList(arr);
fixed.set(0, "x");  // OK (writes through to array)
// fixed.add("c");  // throws (size fixed)

// To make a truly mutable copy:
List<String> mut = new ArrayList<>(Arrays.asList(arr));

Queue 与 Deque 实现

ArrayDeque 是首选的栈和队列实现——比遗留 Stack(它是同步的)和 LinkedList 快。PriorityQueue 按 Comparator 排序元素(默认最小堆)——用于调度、top-K 问题。Deque 支持两端;栈语义使用 addFirst/removeFirst,队列语义使用 addLast/removeFirst。对于并发队列,使用 java.util.concurrent 实现(LinkedBlockingQueue 等)。

java
import java.util.*;

// Queue: FIFO (offer/poll/peek)
Queue<String> queue = new LinkedList<>();
queue.offer("a"); queue.offer("b");
System.out.println(queue.peek());  // "a" (head)
System.out.println(queue.poll());  // "a" (remove head)

// Deque: double-ended (add/remove at both ends)
Deque<String> deque = new ArrayDeque<>();
deque.addFirst("a"); deque.addLast("b");
System.out.println(deque.peekFirst());  // "a"
System.out.println(deque.peekLast());   // "b"
deque.removeFirst(); deque.removeLast();

// ArrayDeque as a stack (push/pop/peek)
Deque<String> stack = new ArrayDeque<>();
stack.push("first");  // addFirst
stack.push("second");
System.out.println(stack.pop());  // "second" (LIFO)
System.out.println(stack.peek()); // "first"

// PriorityQueue: orders by Comparator (not insertion order)
PriorityQueue<Integer> pq = new PriorityQueue<>();
pq.offer(5); pq.offer(1); pq.offer(3);
System.out.println(pq.poll());  // 1 (smallest first)
System.out.println(pq.poll());  // 3

// With custom comparator
PriorityQueue<String> byLen = new PriorityQueue<>(Comparator.comparingInt(String::length));
byLen.offer("aaa"); byLen.offer("a"); byLen.offer("aa");
System.out.println(byLen.poll());  // "a"

// BlockingQueue implementations (concurrency): see concurrency-deep section
// ArrayDeque is faster than Stack/LinkedList for stack/queue use

Map merge、compute、getOrDefault

这些 Map 方法使常见模式原子且简洁。merge 是字计数惯用法——组合现有和新值,如果函数返回 null 则删除条目。computeIfAbsent 是惰性缓存模式(记忆化)。getOrDefault 避免 null 检查。replaceAll 转换所有值。这些比 check-then-act 的 get/put 舞蹈干净得多,它们是 ConcurrentHashMap 在并发代码中原子更新的构建块。

java
import java.util.*;

Map<String, Integer> counts = new HashMap<>();

// getOrDefault: safe read with default
int n = counts.getOrDefault("missing", 0);  // 0, no null

// putIfAbsent: only set if not present
counts.putIfAbsent("a", 1);  // sets to 1
counts.putIfAbsent("a", 2);  // no change (already present)

// compute: recompute value for a key
counts.compute("a", (k, v) -> v == null ? 1 : v + 1);  // increment

// computeIfAbsent: lazy initialization (cache pattern)
Map<String, List<String>> groups = new HashMap<>();
groups.computeIfAbsent("key", k -> new ArrayList<>()).add("item");
// creates list if absent, then adds — atomic, no race in single thread

// computeIfPresent: update only if present
counts.computeIfPresent("a", (k, v) -> v > 0 ? v - 1 : null);  // decrement, remove at 0

// merge: combine existing and new value (great for counting)
counts.merge("word", 1, Integer::sum);  // word count pattern
counts.merge("word", 1, Integer::sum);  // now 2
counts.merge("word", 1, Integer::sum);  // now 3

// merge with removal: function returns null -> entry removed
counts.merge("word", 1, (old, v) -> old > 1 ? old - 1 : null);

// replaceAll: transform all values
Map<String, Integer> doubled = new HashMap<>(counts);
doubled.replaceAll((k, v) -> v * 2);

// Word frequency counter (idiomatic)
String text = "the cat the dog the bird";
Map<String, Integer> freq = new HashMap<>();
for (String w : text.split(" ")) {
    freq.merge(w, 1, Integer::sum);
}

集合实用方法

Collections.* 提供经典实用:sort、binarySearch(需要排序输入)、shuffle、reverse、frequency、min/max。singleton/empty 工厂返回不可变的单元素或空集合——优先使用 emptyList() 而非返回 null。nCopies 内存高效(一个元素共享)。同步包装器存在用于遗留代码但优先使用 java.util.concurrent 集合。检查包装器在运行时捕获泛型类型违规,适用于与原始类型互操作。

java
import java.util.*;
import java.util.stream.*;

List<Integer> nums = new ArrayList<>(List.of(3, 1, 4, 1, 5, 9, 2, 6));

// Sorting
Collections.sort(nums);                    // in-place, natural order
Collections.sort(nums, Comparator.reverseOrder());
nums.sort(Comparator.naturalOrder());      // List.sort (preferred)

// Searching (list must be sorted first)
int idx = Collections.binarySearch(nums, 4);  // index, or negative if absent

// Shuffling & reversing
Collections.shuffle(nums);
Collections.reverse(nums);

// Frequency & disjoint
int freq = Collections.frequency(nums, 1);  // count occurrences
boolean dis = Collections.disjoint(List.of(1, 2), List.of(3, 4));  // true

// Min/max
int min = Collections.min(nums);
int max = Collections.max(nums, Comparator.reverseOrder());

// Singleton collections (immutable, single element)
Set<String> one = Collections.singleton("only");
List<Integer> oneList = Collections.singletonList(42);
Map<String, Integer> oneMap = Collections.singletonMap("k", 1);

// Empty collections (prefer over returning null)
List<Object> empty = Collections.emptyList();
Set<Object> emptySet = Collections.emptySet();

// nCopies (immutable list of n copies)
List<String> padding = Collections.nCopies(5, "x");  // [x,x,x,x,x]

// Synchronized wrappers (legacy — prefer concurrent collections)
List<String> sync = Collections.synchronizedList(new ArrayList<>());

// Checked wrappers (catch heap pollution at runtime)
List<String> checked = Collections.checkedList(new ArrayList<>(), String.class);
// checked.add(123);  // throws ClassCastException at the add site
21

测试(JUnit 5 与 Mockito)

JUnit 5 基础

JUnit 5(Jupiter)注解:@Test 标记测试;@BeforeEach/@AfterEach 在每个测试前后运行;@BeforeAll/@AfterAll 为类运行一次(必须是 static)。@DisplayName 自定义测试名称。@Disabled 跳过测试。assertThrows 验证异常。测试应该独立——使用 @BeforeEach 重置状态,不是静态字段。JUnit 5 在 org.junit.jupiter.api 中(与 JUnit 4 的 org.junit 不同)。

java
import org.junit.jupiter.api.*;
import static org.junit.jupiter.api.Assertions.*;

class CalculatorTest {
    private Calculator calc;

    @BeforeAll
    static void setUpOnce() {
        // runs once before all tests (must be static)
        System.out.println("Starting CalculatorTest");
    }

    @BeforeEach
    void setUp() {
        // runs before each test
        calc = new Calculator();
    }

    @Test
    @DisplayName("2 + 2 should equal 4")
    void addsTwoNumbers() {
        assertEquals(4, calc.add(2, 2));
    }

    @Test
    void dividesByZeroThrows() {
        ArithmeticException ex = assertThrows(
            ArithmeticException.class,
            () -> calc.divide(1, 0)
        );
        assertEquals("Division by zero", ex.getMessage());
    }

    @AfterEach
    void tearDown() {
        // runs after each test (cleanup)
        calc = null;
    }

    @AfterAll
    static void tearDownOnce() {
        // runs once after all tests
    }

    @Disabled("until bug #42 is fixed")
    @Test
    void skippedTest() {}
}

class Calculator {
    int add(int a, int b) { return a + b; }
    int divide(int a, int b) {
        if (b == 0) throw new ArithmeticException("Division by zero");
        return a / b;
    }
}

断言

JUnit 5 断言:assertEquals/assertNotEquals、assertTrue/False、assertNull/NotNull、assertSame(身份)。assertAll 分组检查,所以即使某些失败也全部运行。assertTimeout 使慢测试失败。消息可以是字符串或 Supplier(惰性——仅在失败时构建,避免测试通过时的字符串拼接)。assertIterableEquals 比较有序集合。这些来自 org.junit.jupiter.api.Assertions。

java
import org.junit.jupiter.api.*;
import static org.junit.jupiter.api.Assertions.*;

class AssertionsTest {

    @Test
    void equality() {
        assertEquals(4, 2 + 2);
        assertEquals(4, 2 + 2, "math is broken");
        assertNotEquals(5, 2 + 2);
    }

    @Test
    void booleans() {
        assertTrue(5 > 3);
        assertFalse(5 < 3, "5 should not be less than 3");
    }

    @Test
    void nullness() {
        assertNull(null);
        assertNotNull(new Object());
    }

    @Test
    void sameInstance() {
        String a = "x";
        assertSame(a, a);          // ==
        assertNotSame(a, new String("x"));
    }

    @Test
    void collections() {
        assertIterableEquals(List.of(1, 2, 3), List.of(1, 2, 3));
        assertLinesMatch(List.of("a.*", "b"), List.of("abc", "b"));
    }

    @Test
    void groupedAssertions() {
        // All run even if one fails — reports all failures
        assertAll("person",
            () -> assertEquals("Alice", "Alice"),
            () -> assertEquals(30, 30),
            () -> assertNotNull("x")
        );
    }

    @Test
    void timeout() {
        // Fails if it takes longer than 100ms
        assertTimeout(Duration.ofMillis(100), () -> {
            Thread.sleep(10);
        });
        // assertTimeoutPreemptively: stops the task early (in another thread)
    }

    @Test
    void customMessage() {
        int result = 5;
        assertEquals(4, result, () -> "expected 4 but got " + result);
        // Supplier<String> — message built lazily only on failure
    }

    import java.time.Duration;
    import java.util.List;
}

参数化测试

参数化测试用多个输入运行相同测试逻辑。@ValueSource 提供单参数数组。@CsvSource 将 CSV 行映射到多个参数。@MethodSource(最灵活)使用静态 Stream<Arguments>。@EnumSource 迭代枚举值。@NullAndEmptySource 添加 null/空案例。这消除了复制粘贴测试方法并使数据驱动测试干净。显示名称自动包含参数以便诊断。

java
import org.junit.jupiter.params.*;
import org.junit.jupiter.params.provider.*;
import static org.junit.jupiter.api.Assertions.*;

class ParameterizedTests {

    @ParameterizedTest
    @ValueSource(ints = {1, 2, 3, 4, 5})
    void positiveNumbersArePositive(int n) {
        assertTrue(n > 0);
    }

    @ParameterizedTest
    @ValueSource(strings = {"", "  ", "\t"})
    void blankStrings(String s) {
        assertTrue(s.isBlank());
    }

    @ParameterizedTest
    @NullAndEmptySource
    @ValueSource(strings = {"  ", "\t"})
    void blankOrNull(String s) {
        assertTrue(s == null || s.isBlank());
    }

    @ParameterizedTest
    @CsvSource({
        "1, 1, 2",
        "2, 3, 5",
        "10, -5, 5"
    })
    void addition(int a, int b, int expected) {
        assertEquals(expected, a + b);
    }

    @ParameterizedTest
    @CsvFileSource(resources = "/testdata.csv", numLinesToSkip = 1)
    void fromCsv(int a, int b, int expected) {
        assertEquals(expected, a + b);
    }

    @ParameterizedTest
    @MethodSource("additionProvider")
    void fromMethod(int a, int b, int expected) {
        assertEquals(expected, a + b);
    }
    static java.util.stream.Stream<Arguments> additionProvider() {
        return java.util.stream.Stream.of(
            Arguments.of(1, 1, 2),
            Arguments.of(2, 3, 5)
        );
    }

    @ParameterizedTest
    @EnumSource(TimeUnit.class)
    void allEnums(TimeUnit unit) {
        assertNotNull(unit);
    }
}

生命周期、嵌套与条件

@Nested 创建共享生命周期的内部测试类——适用于 BDD 风格的'when X then Y'结构,其中外部设置适用于内部测试。条件注解(@EnabledOnOs、@EnabledIfSystemProperty、@EnabledIfEnvironmentVariable)基于环境跳过测试。@Tag 分组测试以选择性执行(例如,快 vs 慢,单元 vs 集成)。嵌套类不能有 @BeforeAll(它们是非静态的)。这些功能使测试组织富有表现力。

java
import org.junit.jupiter.api.*;
import static org.junit.jupiter.api.Assertions.*;
import org.junit.jupiter.api.condition.*;

class LifecycleTest {

    @Test
    void topLevel() {}

    @Nested
    @DisplayName("when stack is empty")
    class WhenEmpty {
        @BeforeEach
        void setUp() { /* stack is empty here */ }

        @Test
        void throwsOnPop() {
            assertThrows(Exception.class, () -> {});
        }

        @Nested
        @DisplayName("after pushing one element")
        class AfterPush {
            @BeforeEach
            void push() { /* push one */ }

            @Test
            void popReturnsElement() {
                // ...
            }
        }
    }

    // Conditional execution
    @Test
    @EnabledOnOs(OS.LINUX)
    void onlyOnLinux() {}

    @Test
    @EnabledIfSystemProperty(named = "env", matches = "ci")
    void onlyInCi() {}

    @Test
    @EnabledIfEnvironmentVariable(named = "DATABASE", matches = "postgres")
    void onlyWithPostgres() {}

    @Test
    @DisabledIf("customCondition")
    void conditional() {}
    static boolean customCondition() { return java.time.LocalTime.now().getHour() < 9; }

    // Tagging for selective runs
    @Test
    @Tag("slow")
    void slowIntegrationTest() {}

    @Test
    @Tag("fast")
    void fastUnitTest() {}

    // Run only fast: mvn test -Dgroups=fast
    import org.junit.jupiter.api.condition.OS;
    import java.util.concurrent.TimeUnit;

Mockito 模拟

Mockito 为依赖创建测试替身。@Mock 创建模拟;@InjectMocks 构建注入模拟的真实对象。when(...).thenReturn(...) 存根返回值;verify(...) 检查交互。参数匹配器(any()、eq()、argThat())灵活匹配调用。Spy 包装真实对象(部分模拟)。Arrange-Act-Assert 模式保持测试可读。模拟将被测单元与其依赖(数据库、网络、时间)隔离。

java
import org.junit.jupiter.api.*;
import org.mockito.*;
import static org.mockito.Mockito.*;
import static org.junit.jupiter.api.Assertions.*;

class UserServiceTest {

    @Mock
    UserRepository repo;  // mock dependency

    @InjectMocks
    UserService service;  // real service with mocks injected

    @BeforeEach
    void setUp() {
        MockitoAnnotations.openMocks(this);
    }

    @Test
    void returnsUserWhenFound() {
        // Arrange: stub the mock
        User alice = new User(1, "Alice");
        when(repo.findById(1)).thenReturn(alice);

        // Act
        String name = service.getUserName(1);

        // Assert
        assertEquals("Alice", name);
        verify(repo).findById(1);       // called once
        verify(repo, never()).findById(2);
    }

    @Test
    void throwsWhenNotFound() {
        when(repo.findById(99)).thenReturn(null);
        // or: when(repo.findById(99)).thenThrow(new RuntimeException());

        assertThrows(RuntimeException.class, () -> service.getUserName(99));
    }

    @Test
    void argumentMatchers() {
        when(repo.findById(anyInt())).thenReturn(new User(0, "default"));
        // matchers: eq(), any(), anyInt(), contains(), argThat()

        service.getUserName(42);
        verify(repo).findById(intThat(n -> n > 0));
    }

    @Test
    void verifyInteractionDetails() {
        service.getUserName(1);
        verify(repo, times(1)).findById(1);
        verify(repo, atLeastOnce()).findById(anyInt());
        verifyNoMoreInteractions(repo);
    }

    @Test
    void spy_partialMock() {
        List<String> spy = spy(new ArrayList<>());
        spy.add("real");
        when(spy.size()).thenReturn(100);  // stub one method
        assertEquals(100, spy.size());     // stubbed
        assertEquals(1, spy.size());       // wait, this would be 100 too
    }
}

interface UserRepository { User findById(int id); }
record User(int id, String name) {}
class UserService {
    UserRepository repo;
    UserService(UserRepository r) { repo = r; }
    String getUserName(int id) {
        User u = repo.findById(id);
        if (u == null) throw new RuntimeException("not found");
        return u.name();
    }
}
22

JUnit 测试

基本测试

JUnit 5 使用 org.junit.jupiter.api 中的 @Test。assertEquals 验证期望等于实际。其他断言:assertTrue、assertThrows、assertAll。

java
import org.junit.jupiter.api.Test;
import static org.junit.jupiter.api.Assertions.*;
class CalcTest {
    @Test
    void testAdd() {
        assertEquals(5, calc.add(2, 3));
    }
}

参数化测试

参数化测试用不同输入运行相同测试。@ValueSource 提供单参数。@CsvSource 提供多参数。减少测试重复。

java
@ParameterizedTest
@ValueSource(ints = {1, 2, 3, 4})
void testPositive(int n) { assertTrue(n > 0); }

@ParameterizedTest
@CsvSource({"1,2,3", "4,5,9"})
void testAdd(int a, int b, int expected) {
    assertEquals(expected, calc.add(a, b));
}

生命周期方法

@BeforeAll/@AfterAll 每类运行一次(必须是 static)。@BeforeEach/@AfterEach 在每个测试前后运行。用于数据库连接和模拟设置。

java
class DbTest {
    @BeforeAll static void setupAll() { /* once before all */ }
    @AfterAll static void tearDownAll() { /* once after all */ }
    @BeforeEach void setup() { /* before each test */ }
    @AfterEach void tearDown() { /* after each test */ }
}

断言

assertAll 即使某些失败也运行所有断言。assertThrows 验证代码抛出特定异常。时间限制测试使用 assertTimeout。

java
@Test
void testAll() {
    assertAll("person",
        () -> assertEquals("Alice", p.getName()),
        () -> assertEquals(30, p.getAge())
    );
}
@Test
void testException() {
    assertThrows(IllegalArgumentException.class, () -> service.process(-1));
}

Mockito

@Mock 创建模拟对象,@InjectMocks 注入它们。when().thenReturn() 存根调用。verify() 检查方法是否被调用。Mockito 是标准模拟框架。

java
@ExtendWith(MockitoExtension.class)
class UserServiceTest {
    @Mock UserRepository repo;
    @InjectMocks UserService service;
    @Test
    void testFind() {
        when(repo.findById(1)).thenReturn(Optional.of(new User("Alice")));
        assertEquals("Alice", service.findUser(1).getName());
        verify(repo).findById(1);
    }
}
23

Maven/Gradle

Maven POM

Maven 使用 pom.xml。groupId/artifactId/version 标识项目。依赖有 scope(compile、test、provided)。Maven 强制标准目录结构。

java
<project>
  <modelVersion>4.0.0</modelVersion>
  <groupId>com.example</groupId>
  <artifactId>myapp</artifactId>
  <version>1.0.0</version>
  <dependencies>
    <dependency>
      <groupId>org.junit.jupiter</groupId>
      <artifactId>junit-jupiter</artifactId>
      <version>5.10.0</version>
      <scope>test</scope>
    </dependency>
  </dependencies>
</project>

Maven 命令

Maven 生命周期:clean、compile、test、package、install、deploy。每个阶段运行前面的阶段。使用 -DskipTests 跳过测试。

java
mvn clean          # Remove target/
mvn compile        # Compile sources
mvn test           # Run tests
mvn package        # Build JAR
mvn install        # Install to local repo
mvn dependency:tree  # Show dependency tree

Gradle 构建

Gradle 使用 build.gradle(Groovy)或 build.gradle.kts(Kotlin)。implementation 用于编译依赖,testImplementation 用于测试。Gradle 比 Maven 快。

java
plugins { id 'java'; id 'application' }
repositories { mavenCentral() }
dependencies {
    implementation 'com.google.guava:guava:32.1.3-jre'
    testImplementation 'org.junit.jupiter:junit-jupiter:5.10.0'
}
application { mainClass = 'com.example.App' }

Gradle 命令

Gradle 任务:build、test、run、clean。包装器(gradlew)确保版本一致。使用 --parallel 进行并行模块构建。

java
gradle clean       # Clean build
gradle build       # Build + test
gradle test        # Run tests
gradle run         # Run application
gradle bootRun     # Run Spring Boot
gradle dependencies  # Show dependency tree

多模块项目

多模块项目拆分大型应用。settings.gradle 列出模块。project(:core) 创建模块间依赖。每个模块有自己的 build.gradle。

java
// settings.gradle
include 'core', 'web', 'api'
// build.gradle (root)
subprojects {
    apply plugin: 'java'
    repositories { mavenCentral() }
}
// In web/build.gradle
dependencies { implementation project(':core') }
24

Spring 基础

Spring Boot 应用

@SpringBootApplication 启用自动配置、组件扫描和配置。SpringApplication.run 启动嵌入式服务器。消除 XML 配置。

java
@SpringBootApplication
public class App {
    public static void main(String[] args) {
        SpringApplication.run(App.class, args);
    }
}

REST 控制器

@RestController 结合 @Controller 和 @ResponseBody。@GetMapping、@PostMapping 是快捷方式。@PathVariable 提取 URL 参数,@RequestBody 绑定 JSON。

java
@RestController
@RequestMapping("/api/users")
public class UserController {
    @GetMapping("/{id}")
    public User getUser(@PathVariable Long id) {
        return userService.findById(id);
    }
    @PostMapping
    public User create(@RequestBody User user) {
        return userService.save(user);
    }
}

依赖注入

@Autowired 注入依赖。推荐构造函数注入(可测试、不可变)。@Service、@Repository、@Component 是注入的原型。

java
@Service
public class UserService {
    private final UserRepository repo;
    @Autowired  // Constructor injection (recommended)
    public UserService(UserRepository repo) {
        this.repo = repo;
    }
}

配置

@Configuration 标记配置类。@Bean 声明由 Spring 管理的 bean。@Primary 使 bean 优先。用于第三方类。

java
@Configuration
public class AppConfig {
    @Bean
    public RestTemplate restTemplate() {
        return new RestTemplate();
    }
    @Bean @Primary
    public DataSource primaryDataSource() {
        return DataSourceBuilder.create().build();
    }
}

应用属性

application.properties 配置 Spring Boot。Profile 启用环境特定配置。用 spring.profiles.active=dev 激活。使用 @Value 或 @ConfigurationProperties。

java
# application.properties
server.port=8080
spring.datasource.url=jdbc:postgresql://localhost/mydb
spring.jpa.hibernate.ddl-auto=update
# Profile-specific
# application-dev.properties
server.port=9090
25

JDBC 深入

连接与 Statement

DriverManager.getConnection 建立连接。Statement 执行静态 SQL。ResultSet 迭代结果。始终关闭资源或使用 try-with-resources。

java
Connection conn = DriverManager.getConnection(url, user, pass);
Statement stmt = conn.createStatement();
ResultSet rs = stmt.executeQuery("SELECT * FROM users");
while (rs.next()) {
    System.out.println(rs.getString("name"));
}
rs.close(); stmt.close(); conn.close();

PreparedStatement

PreparedStatement 通过参数化查询防止 SQL 注入。按索引(1 基)设置值。try-with-resources 自动关闭。通过预编译提高性能。

java
String sql = "INSERT INTO users (name, email) VALUES (?, ?)";
try (PreparedStatement ps = conn.prepareStatement(sql)) {
    ps.setString(1, "Alice");
    ps.setString(2, "[email protected]");
    ps.executeUpdate();
}

事务管理

setAutoCommit(false) 启动事务。commit 持久化,rollback 撤销。如果任何语句失败,rollback 以维护完整性。Spring 中 @Transactional 处理此问题。

java
conn.setAutoCommit(false);
try {
    stmt.executeUpdate("UPDATE accounts SET bal = bal - 100 WHERE id = 1");
    stmt.executeUpdate("UPDATE accounts SET bal = bal + 100 WHERE id = 2");
    conn.commit();
} catch (SQLException e) {
    conn.rollback();
}

连接池

连接池重用连接。HikariCP 是最快的池。maximumPoolSize 限制并发连接。始终关闭(返回到池)。Spring Boot 自动配置 HikariCP。

java
HikariConfig config = new HikariConfig();
config.setJdbcUrl(url);
config.setMaximumPoolSize(10);
HikariDataSource ds = new HikariDataSource(config);
Connection conn = ds.getConnection();

ResultSet 元数据

ResultSetMetaData 描述结果结构:列名、类型、属性。适用于通用数据访问。列索引是 1 基的。

java
ResultSet rs = stmt.executeQuery("SELECT * FROM users");
ResultSetMetaData meta = rs.getMetaData();
int cols = meta.getColumnCount();
for (int i = 1; i <= cols; i++)
    System.out.println(meta.getColumnName(i) + ": " + meta.getColumnTypeName(i));
26

并发工具

ExecutorService

ExecutorService 管理线程池。submit 返回 Future 用于异步结果。get 阻塞直到完成(带超时)。始终关闭 executor。

java
ExecutorService executor = Executors.newFixedThreadPool(4);
Future<String> future = executor.submit(() -> {
    Thread.sleep(1000);
    return "Result";
});
String result = future.get(5, TimeUnit.SECONDS);
executor.shutdown();

CompletableFuture

CompletableFuture 启用函数式异步编程。supplyAsync 在 ForkJoinPool 中运行。thenApply 转换,thenAccept 消费,exceptionally 处理错误。

java
CompletableFuture.supplyAsync(() -> fetchData())
    .thenApply(data -> process(data))
    .thenAccept(result -> System.out.println(result))
    .exceptionally(ex -> { ex.printStackTrace(); return null; });

并发集合

ConcurrentHashMap 无需完全锁定即线程安全。CopyOnWriteArrayList 写入时复制(读密集)。BlockingQueue 支持生产者-消费者模式。

java
ConcurrentHashMap<String, Integer> map = new ConcurrentHashMap<>();
map.computeIfAbsent("b", k -> k.length());
CopyOnWriteArrayList<String> list = new CopyOnWriteArrayList<>();
BlockingQueue<Task> queue = new LinkedBlockingQueue<>(100);

CountDownLatch 与 CyclicBarrier

CountDownLatch 等待 N 个线程(一次性)。CyclicBarrier 等待 N 个线程然后重置(可重用)。启动使用 latch,分阶段计算使用 barrier。

java
CountDownLatch latch = new CountDownLatch(3);
for (int i = 0; i < 3; i++)
    new Thread(() -> { work(); latch.countDown(); }).start();
latch.await();  // Wait for all

CyclicBarrier barrier = new CyclicBarrier(3, () -> System.out.println("All ready"));

原子变量

原子变量提供无锁线程安全操作。compareAndSet 启用乐观锁定。高竞争计数器 LongAdder 比 AtomicLong 快。

java
AtomicInteger counter = new AtomicInteger(0);
counter.incrementAndGet();
counter.compareAndSet(0, 1);
LongAdder adder = new LongAdder();
adder.increment();
27

JVM 内部

内存区域

JVM 内存:堆(对象、GC 管理)、栈(方法调用、每线程)、元空间(类元数据)。年轻代使用复制 GC,老年代使用标记-清除-压缩。

java
// Heap: objects and arrays (shared)
// - Young Gen: Eden, S0, S1
// - Old Gen: long-lived objects
// Stack: method frames (per thread)
// Metaspace: class metadata (Java 8+)
// JVM flags: -Xms512m -Xmx2g

类加载

类加载是惰性的。Bootstrap 加载核心 Java,Extension 加载扩展,Application 加载类路径。静态初始化器运行一次。自定义类加载器启用热重载。

java
// Bootstrap -> Extension -> Application classloaders
class MyClass {
    static { System.out.println("Static init"); }
}
// Class.forName("MyClass") triggers loading
// -verbose:class shows class loading

字节码

Java 编译为字节码(基于栈)。javap -c 反汇编类文件。每条指令压入/弹出操作数栈。Java 代理可以在加载时修改字节码。

java
// javap -c MyClass.class
// Method int add(int, int):
//   iload_1      // Load local var 1
//   iload_2      // Load local var 2
//   iadd         // Add
//   ireturn      // Return int

JIT 编译

JIT 将频繁执行的字节码编译为本地代码。分层编译平衡启动和峰值性能。热点方法被内联和优化。

java
// JIT compiles hot methods to native code
// -XX:+PrintCompilation  // Show JIT activity
// -XX:CompileThreshold=10000  // Method call count
// Tiered: Interpreter -> C1 -> C2

线程转储

线程转储显示所有线程状态和堆栈跟踪。调试死锁和挂起的关键。jstack 是命令行工具。查找 BLOCKED 和 WAITING 线程。

java
// Get thread dump
jstack <pid>
// Or: kill -3 <pid>
// Deadlock detection
jstack -l <pid> | grep -A 20 "Found deadlock"
28

垃圾回收

GC 算法

Serial GC 用于小应用。Parallel GC 最大化吞吐量。G1 GC 平衡吞吐量和延迟(默认)。ZGC 为大堆提供亚毫秒级暂停。

java
# Serial GC (single-threaded)
-XX:+UseSerialGC
# Parallel GC (throughput)
-XX:+UseParallelGC
# G1 GC (balanced, default in Java 9+)
-XX:+UseG1GC
# ZGC (low-latency)
-XX:+UseZGC

G1 GC 调优

G1 将堆划分为区域。MaxGCPauseMillis 设置软暂停目标。G1 优先处理垃圾最多的区域。使用 GCViewer 或 GCEasy 分析日志。

java
# Set heap
-Xms4g -Xmx4g
# Max GC pause target
-XX:MaxGCPauseMillis=200
# Region size (1-32MB)
-XX:G1HeapRegionSize=16m
# Enable GC logging
-Xlog:gc*:file=gc.log:time,uptime

内存泄漏

内存泄漏由意外对象保留引起。静态集合、未关闭资源和监听器注册是常见的。jmap 显示对象计数。用 MAT 分析 hprof。

java
// Common leak: static collections
static Map<String, Object> cache = new HashMap<>();
// Objects never removed -> leak
// Detect with:
jmap -histo <pid> | head -20
// Heap dump:
jmap -dump:format=b,file=heap.hprof <pid>

弱引用

WeakReference 在没有强引用时允许 GC。SoftReference 在内存压力下存活。WeakHashMap 键是弱的。用于不应阻止 GC 的缓存。

java
WeakReference<Object> weakRef = new WeakReference<>(new Object());
SoftReference<byte[]> softRef = new SoftReference<>(new byte[1024]);
WeakHashMap<Object, String> map = new WeakHashMap<>();
map.put(key, "value");  // Entry removed when key is GC'd

终结

finalize() 已弃用(不可预测、慢)。Cleaner API(Java 9+)提供更好的清理。try-with-resources 优先用于确定性清理。

java
// Cleaner API (Java 9+)
class Resource implements AutoCloseable {
    private final Cleaner.Cleanable cleanable;
    Resource() {
        cleanable = Cleaner.create().register(this, () -> cleanup());
    }
    public void close() { cleanable.clean(); }
}
29

Stream Collector

分组

groupingBy 按分类器分区元素。第二个参数是用于聚合的下游 collector。counting、averaging、summing 是常见的下游 collector。

java
Map<String, List<Person>> byCity =
    people.stream().collect(Collectors.groupingBy(Person::getCity));
Map<String, Long> countByCity =
    people.stream().collect(Collectors.groupingBy(
        Person::getCity, Collectors.counting()));

分区

partitioningBy 拆分为两组(true/false)。比布尔键的 groupingBy 更高效。结果始终有两个键。下游 collector 聚合每个分区。

java
Map<Boolean, List<Person>> partition =
    people.stream().collect(Collectors.partitioningBy(p -> p.getAge() >= 18));
// {false=[minors], true=[adults]}
Map<Boolean, Long> count =
    people.stream().collect(Collectors.partitioningBy(
        p -> p.getAge() >= 18, Collectors.counting()));

连接

joining 用可选分隔符、前缀和后缀连接字符串。元素必须是字符串;先使用 map。内部使用 StringBuilder。

java
String names = people.stream()
    .map(Person::getName)
    .collect(Collectors.joining(", "));
// "Alice, Bob, Charlie"
String csv = people.stream()
    .map(p -> p.getName() + "=" + p.getAge())
    .collect(Collectors.joining("\n", "[", "]"));

归约

reducing 执行折叠操作。三参数版本接受 identity、mapper、reducer。两参数版本返回 Optional。标准 collector 不足时使用。

java
int totalAge = people.stream()
    .collect(Collectors.reducing(0, Person::getAge, Integer::sum));
Optional<Person> oldest = people.stream()
    .collect(Collectors.reducing((p1, p2) ->
        p1.getAge() > p2.getAge() ? p1 : p2));

自定义 Collector

Collector.of 创建自定义 collector:supplier、accumulator、combiner、finisher。combiner 为并行流合并部分结果。适用于专门输出格式。

java
Collector<Person, ?, String> toJson = Collector.of(
    StringBuilder::new,
    (sb, p) -> sb.append(`{"name":"${p.getName()}"}`),
    StringBuilder::append,
    StringBuilder::toString
);
30

常见陷阱

Integer 缓存

Java 缓存 -128 到 127 的 Integer 值。== 比较引用,不是值。对于缓存范围外的 Integer,== 返回 false。Integer 始终使用 .equals()。

java
Integer a = 127; Integer b = 127;
System.out.println(a == b);  // true (cached)
Integer c = 128; Integer d = 128;
System.out.println(c == d);  // false (not cached)
System.out.println(c.equals(d));  // true

字符串不可变性

字符串是不可变的:concat 等方法返回新字符串。忘记赋值结果是常见 bug。重复拼接使用 StringBuilder。

java
String s = "Hello";
s.concat(" World");  // Returns new string, s unchanged
System.out.println(s);  // "Hello"
// Use StringBuilder for mutation
StringBuilder sb = new StringBuilder("Hello");
sb.append(" World");  // Mutates in place

Equals 与 HashCode

相等的对象必须有相等的哈希码。HashMap 和 HashSet 使用 hashCode 进行分桶,使用 equals 进行比较。如果重写 equals,必须重写 hashCode。

java
class Person {
    String name;
    public boolean equals(Object o) {
        if (!(o instanceof Person)) return false;
        return name.equals(((Person)o).name);
    }
    public int hashCode() { return name.hashCode(); }
}

受检与非受检

受检异常必须声明或捕获(IOException、SQLException)。非受检(RuntimeException)不需要处理。避免捕获宽泛的 Exception。

java
// Checked: must catch or declare
try { Thread.sleep(1000); }
catch (InterruptedException e) { /* handle */ }
// Unchecked: RuntimeException
throw new IllegalArgumentException("bad input");

资源泄漏

资源必须关闭。try-with-resources(Java 7+)自动关闭 AutoCloseable。没有它,异常导致泄漏。永远不要让资源未关闭。

java
// BAD: resource leak
FileInputStream fis = new FileInputStream("file.txt");
// If exception here, fis never closed
// GOOD: try-with-resources
try (FileInputStream fis = new FileInputStream("file.txt")) {
    // Use resource
}  // Auto-closed even on exception
31

设计模式

单例

带 volatile 的双重检查锁定确保线程安全的惰性初始化。volatile 防止指令重排序。枚举单例更简单:public enum Singleton { INSTANCE; }。

java
public class Singleton {
    private static volatile Singleton instance;
    private Singleton() {}
    public static Singleton getInstance() {
        if (instance == null) {
            synchronized (Singleton.class) {
                if (instance == null) instance = new Singleton();
            }
        }
        return instance;
    }
}

构建器

构建器模式处理具有许多可选参数的对象。避免伸缩构造函数。内部 Builder 流式收集参数。build() 创建不可变对象。

java
public class Pizza {
    private final String size;
    private final boolean cheese;
    private Pizza(Builder b) { size = b.size; cheese = b.cheese; }
    public static class Builder {
        private String size; private boolean cheese;
        public Builder size(String s) { size = s; return this; }
        public Builder cheese(boolean c) { cheese = c; return this; }
        public Pizza build() { return new Pizza(this); }
    }
}

策略

策略模式封装可互换的算法。上下文委托给策略接口。避免大型 if-else 链。遵循开闭原则。

java
interface PaymentStrategy { void pay(double amount); }
class CreditCard implements PaymentStrategy {
    public void pay(double amount) { System.out.println("Card: " + amount); }
}
class Cart { private PaymentStrategy strategy; void checkout() { strategy.pay(100); } }

观察者

观察者模式定义一对多依赖。当主题更改时,所有观察者被通知。用于 MVC、事件系统和响应式编程。

java
interface Observer { void update(String event); }
class Subject {
    private List<Observer> observers = new ArrayList<>();
    void subscribe(Observer o) { observers.add(o); }
    void notify(String event) { observers.forEach(o -> o.update(event)); }
}

工厂方法

工厂方法定义创建对象的接口但让子类决定实例化哪个类。将客户端代码与具体类解耦。

java
abstract class Document { abstract void open(); }
class PDF extends Document { void open() { /* ... */ } }
abstract class DocFactory { abstract Document create(); }
class PDFFactory extends DocFactory { Document create() { return new PDF(); } }

这篇内容对您有帮助吗?

学习路径

从零开始学习

通过结构化课程从头学习这个语言。