java 正則表達(dá)式

正則表達(dá)式是一個(gè)非常強(qiáng)大的字符串處理工具，通過一種特殊的語法來描述一種模式，再通過模式可以完成字符串的匹配，萃取，替換等操作

簡(jiǎn)例

要判斷一個(gè)字符串是否是一個(gè)郵箱，可能需要很多的判斷邏輯，使用則表達(dá)式，只需要下面代碼即可

Pattern pattern = Pattern.compile("^([a-z0-9._%+-]+)@([a-z0-9.-]+)\\.[a-z]{2,4}$");
Matcher matcher = pattern.matcher("hatlonely@foxmail.com");
assertTrue(matcher.matches());

主要有 Pattern 對(duì)象和 Matcher 對(duì)象

• Pattern: 模式對(duì)象
• Matcher: 匹配的結(jié)果

部分匹配與完全匹配

Matcher 提供 matches 方法用于完全匹配，find 方法用于部分匹配，還提供 asPredicate 方法返回一個(gè)部分匹配的謂詞

Pattern 提供一個(gè)靜態(tài)方法 matches 用于完全匹配，這個(gè)方法會(huì)臨時(shí)構(gòu)造 Pattern 對(duì)象，如果 Pattern 會(huì)被多次重復(fù)使用，盡量別直接使用這種方法

assertTrue(Pattern.compile("hello").matcher("hello").matches());         // 完全匹配
assertTrue(Pattern.compile("hello").matcher("hello world").find());      // 部分匹配
assertTrue(Pattern.compile("hello").asPredicate().test("hello world"));  // 部分匹配
assertTrue(Pattern.matches("hello", "hello"));                           // 完全匹配

通配符

assertTrue(Pattern.matches("[0-9]*", ""));
assertTrue(Pattern.matches("[0-9]?", ""));
assertTrue(Pattern.matches("[0-9]?", "1"));
assertTrue(Pattern.matches("[0-9]+", "123"));
assertTrue(Pattern.matches("[0-9]{3}", "123"));
assertTrue(Pattern.matches("[0-9]{3,4}", "123"));
assertTrue(Pattern.matches("[0-9]{3,4}", "1234"));
assertTrue(Pattern.matches("[0-9]{3,}", "1234"));
assertTrue(Pattern.matches("\\d+", "123"));
assertTrue(Pattern.matches("\\s+", " \t"));
assertTrue(Pattern.matches("\\w+", "abc"));
assertTrue(Pattern.matches("(f|z)oo", "foo"));
assertTrue(Pattern.matches("(f|z)oo", "zoo"));
assertTrue(Pattern.matches(".*", "any string"));

捕獲分組

支持用小括號(hào) () 將模式分組，Matcher 提供 group 方法獲取分組的內(nèi)容

使用 groupCount 獲取捕獲分組的數(shù)量，由于下面例子中模式串中沒有小括號(hào)，所以沒有捕獲的分組

Pattern pattern = Pattern.compile("^[a-z0-9]+@[a-z0-9.]+[.][a-z]{2,4}$");
Matcher matcher = pattern.matcher("hatlonely@foxmail.com");
assertTrue(matcher.matches());
assertEquals(matcher.groupCount(), 0);

使用 group 方法獲取分組的內(nèi)容，分組的編號(hào)以左括號(hào)為準(zhǔn)，gruop(i) 返回第 i 個(gè)分組，group(0) 代表整個(gè)匹配串

Pattern pattern = Pattern.compile("^([a-z0-9]+)@(([a-z0-9.]+)[.]([a-z]{2,4}))$");
Matcher matcher = pattern.matcher("hatlonely@foxmail.com");
assertTrue(matcher.matches());
assertEquals(matcher.groupCount(), 4);
assertEquals(matcher.group(), "hatlonely@foxmail.com");
assertEquals(matcher.group(1), "hatlonely");
assertEquals(matcher.group(2), "foxmail.com");
assertEquals(matcher.group(3), "foxmail");
assertEquals(matcher.group(4), "com");

可以用 ?: 阻止捕獲，如下面代碼所示，foxmail.com 就沒有被捕獲

Pattern pattern = Pattern.compile("^([a-z0-9]+)@(?:([a-z0-9.]+)[.]([a-z]{2,4}))$");
Matcher matcher = pattern.matcher("hatlonely@foxmail.com");
assertTrue(matcher.matches());
assertEquals(matcher.groupCount(), 3);
assertEquals(matcher.group(), "hatlonely@foxmail.com");
assertEquals(matcher.group(0), "hatlonely@foxmail.com");
assertEquals(matcher.group(1), "hatlonely");
assertEquals(matcher.group(2), "foxmail");
assertEquals(matcher.group(3), "com");

預(yù)測(cè)

正向預(yù)測(cè)是指匹配開始或者結(jié)束的字符串需要匹配(或者反向預(yù)測(cè)不能匹配)分組，分組內(nèi)的模式僅用于預(yù)測(cè)，不會(huì)出現(xiàn)在最終的匹配串中，這種匹配只能用于部分匹配，?=，?! 匹配開始，?<=，?<! 匹配結(jié)束

{
    Pattern pattern = Pattern.compile("Windows (?=95|98|NT|2000)");
    Matcher matcher = pattern.matcher("Windows 2000");
    assertTrue(matcher.find());
    assertEquals(matcher.group(), "Windows ");
    assertEquals(matcher.groupCount(), 0);
}
{
    Pattern pattern = Pattern.compile("Windows (?!95|98|NT|2000)");
    Matcher matcher = pattern.matcher("Windows vista");
    assertTrue(matcher.find());
    assertEquals(matcher.group(), "Windows ");
    assertEquals(matcher.groupCount(), 0);
}
{
    Pattern pattern = Pattern.compile("(?<=95|98|NT|2000) Windows");
    Matcher matcher = pattern.matcher("2000 Windows");
    assertTrue(matcher.find());
    assertEquals(matcher.group(), " Windows");
    assertEquals(matcher.groupCount(), 0);
}
{
    Pattern pattern = Pattern.compile("(?<!95|98|NT|2000) Windows");
    Matcher matcher = pattern.matcher("vista Windows");
    assertTrue(matcher.find());
    assertEquals(matcher.group(), " Windows");
    assertEquals(matcher.groupCount(), 0);
}

反向引用

前面通過捕獲獲得的分組可以再后面引用，比如 (\w+) \1 表示兩個(gè)重復(fù)的單詞

assertTrue(Pattern.matches("(\\w+) \\1", "ab ab"));
assertTrue(Pattern.matches("(\\w+) \\1", "abc abc"));
assertFalse(Pattern.matches("(\\w+) \\1", "abc def"));

正則替換

String 和 Pattern 都提供了 replaceAll 和 replaceFirst 方法來作正則替換，替換串中可以 $i 來獲取正則匹配捕獲的分組

Pattern pattern = Pattern.compile("^([a-z0-9]+)@(?:([a-z0-9.]+)[.]([a-z]{2,4}))$");
assertEquals(pattern.matcher("hatlonely@foxmail.com").replaceAll(
        "$0 $1 $2 $3"
), "hatlonely@foxmail.com hatlonely foxmail com");

assertEquals("hatlonely@foxmail.com".replaceAll(
        "^([a-z0-9]+)@(?:([a-z0-9.]+)[.]([a-z]{2,4}))$", "$0 $1 $2 $3"
), "hatlonely@foxmail.com hatlonely foxmail com");

搜索所有匹配

每次 find 調(diào)用都會(huì)匹配目標(biāo)串中的一個(gè)匹配，通過多次 find 調(diào)用，可以找出目標(biāo)串中所有的匹配

String str = "abab x acac y aeae";
Pattern pattern = Pattern.compile("(\\w+)\\1");
Matcher matcher = pattern.matcher(str);

List<String> li = new ArrayList<>();
while (matcher.find()) {
    li.add(matcher.group());
    assertThat(str.substring(matcher.start(), matcher.end()), equalTo(matcher.group()));
}
assertThat(li, equalTo(List.of("abab", "acac", "aeae")));

鏈接

• 測(cè)試代碼: https://github.com/hatlonely/hellojava/blob/master/src/test/java/util/RegexTest.java