Merge pull request #4134 from Lettuceleaves/perf/optimize-str-finder

Perf/optimize str finder

hutool-core/src/main/java/cn/hutool/v7/core/text/CharSequenceUtil.java

@@ -2416,8 +2416,32 @@ public class CharSequenceUtil extends StrValidator {
 		if (null == str1 || null == str2) {
 			return false;
 		}
+		if (str1 instanceof String && str2 instanceof String) {
+			return ((String) str1).regionMatches(ignoreCase, offset1, (String) str2, offset2, length);
+		}
+		if (offset1 < 0 || offset2 < 0 || length < 0) {
+			return false;
+		}
+		if (str1.length() - offset1 < length || str2.length() - offset2 < length) {
+			return false;
+		}
+		for (int i = 0; i < length; i++) {
+			final char c1 = str1.charAt(offset1 + i);
+			final char c2 = str2.charAt(offset2 + i);
+			if (c1 == c2) {
+				continue;
+			}
+			if (ignoreCase) {
+				final char u1 = Character.toLowerCase(c1);
+				final char u2 = Character.toLowerCase(c2);
+				if (u1 == u2) {
+					continue;
+				}
+			}
+			return false;
+		}
 
-		return str1.toString().regionMatches(ignoreCase, offset1, str2.toString(), offset2, length);
+		return true;
 	}
 	// endregion
 

hutool-core/src/main/java/cn/hutool/v7/core/text/finder/StrFinder.java

@@ -20,6 +20,10 @@ import cn.hutool.v7.core.lang.Assert;
 import cn.hutool.v7.core.text.CharSequenceUtil;
 
 import java.io.Serial;
+import java.util.HashMap;
+import java.util.Map;
+
+import static cn.hutool.v7.core.text.CharSequenceUtil.isSubEquals;
 
 /**
  * 字符串查找器
@@ -44,6 +48,8 @@ public class StrFinder extends TextFinder {
 
 	private final CharSequence strToFind;
 	private final boolean caseInsensitive;
+	private Map<Character, Integer> forwardOffsetMap;
+	private Map<Character, Integer> reverseOffsetMap;
 
 	/**
 	 * 构造
@@ -61,23 +67,54 @@ public class StrFinder extends TextFinder {
 	public int start(int from) {
 		Assert.notNull(this.text, "Text to find must be not null!");
 		final int subLen = strToFind.length();
+		final int textLen = text.length();
 
-		if (from < 0) {
-			from = 0;
-		}
-		int endLimit = getValidEndIndex();
+		// 基于Sunday算法实现高效子串查询
 		if (negative) {
-			for (int i = from; i > endLimit; i--) {
-				if (CharSequenceUtil.isSubEquals(text, i, strToFind, 0, subLen, caseInsensitive)) {
+			if (this.reverseOffsetMap == null) {
+				this.reverseOffsetMap = buildReverseOffsetMap(strToFind, caseInsensitive);
+			}
+			int maxIndex = textLen - subLen;
+			if (from > maxIndex) {
+				from = maxIndex;
+			}
+			int i = from;
+			while (i >= 0) {
+				if (isSubEquals(text, i, strToFind, 0, subLen, caseInsensitive)) {
 					return i;
 				}
+				if (i - 1 < 0) {
+					break;
+				}
+				char preChar = text.charAt(i - 1);
+				int jump = reverseOffsetMap.getOrDefault(
+					caseInsensitive ? Character.toLowerCase(preChar) : preChar,
+					subLen + 1
+				);
+				i -= jump;
 			}
 		} else {
-			endLimit = endLimit - subLen + 1;
-			for (int i = from; i < endLimit; i++) {
-				if (CharSequenceUtil.isSubEquals(text, i, strToFind, 0, subLen, caseInsensitive)) {
+			if (this.forwardOffsetMap == null) {
+				this.forwardOffsetMap = buildForwardOffsetMap(strToFind, caseInsensitive);
+			}
+			if (from < 0) {
+				from = 0;
+			}
+			int endLimit = textLen - subLen;
+			int i = from;
+			while (i <= endLimit) {
+				if (isSubEquals(text, i, strToFind, 0, subLen, caseInsensitive)) {
 					return i;
 				}
+				if (i + subLen >= textLen) {
+					break;
+				}
+				char nextChar = text.charAt(i + subLen);
+				int jump = forwardOffsetMap.getOrDefault(
+					caseInsensitive ? Character.toLowerCase(nextChar) : nextChar,
+					subLen + 1
+				);
+				i += jump;
 			}
 		}
 
@@ -91,4 +128,44 @@ public class StrFinder extends TextFinder {
 		}
 		return start + strToFind.length();
 	}
+
+	/**
+	 * 构建正向偏移表
+	 */
+	private static Map<Character, Integer> buildForwardOffsetMap(CharSequence pattern, boolean caseInsensitive) {
+		int m = pattern.length();
+		Map<Character, Integer> map = new HashMap<>(Math.min(m, 128));
+
+		for (int i = 0; i < m; i++) {
+			char c = pattern.charAt(i);
+			int jump = m - i;
+
+			if (caseInsensitive) {
+				map.put(Character.toLowerCase(c), jump);
+			} else {
+				map.put(c, jump);
+			}
+		}
+		return map;
+	}
+
+	/**
+	 * 构建反向偏移表
+	 */
+	private static Map<Character, Integer> buildReverseOffsetMap(CharSequence pattern, boolean caseInsensitive) {
+		int m = pattern.length();
+		Map<Character, Integer> map = new HashMap<>(Math.min(m, 128));
+
+		for (int i = m - 1; i >= 0; i--) {
+			char c = pattern.charAt(i);
+			int jump = i + 1;
+
+			if (caseInsensitive) {
+				map.put(Character.toLowerCase(c), jump);
+			} else {
+				map.put(c, jump);
+			}
+		}
+		return map;
+	}
 }

hutool-core/src/test/java/cn/hutool/v7/core/text/finder/StrFinderTest.java

@@ -0,0 +1,146 @@
+
+package cn.hutool.v7.core.text.finder;
+
+import org.junit.jupiter.api.Assertions;
+import org.junit.jupiter.api.Test;
+
+/**
+ * @author Lettuceleaves
+ */
+public class StrFinderTest {
+
+	@Test
+	public void testForward() {
+		// 正向查找
+		String text = "Hello Hutool World";
+		StrFinder finder = StrFinder.of("Hutool", false);
+		finder.setText(text);
+
+		Assertions.assertEquals(6, finder.start(0));
+		Assertions.assertEquals(-1, finder.start(7));
+	}
+
+	@Test
+	public void testForwardIgnoreCase() {
+		// 正向查找，忽略大小写
+		String text = "Hello HUTOOL World";
+		StrFinder finder = StrFinder.of("hutool", true);
+		finder.setText(text);
+
+		Assertions.assertEquals(6, finder.start(0));
+	}
+
+	@Test
+	public void testReverseBasic() {
+		// 反向查找
+		String text = "abc abc abc";
+		StrFinder finder = StrFinder.of("abc", false);
+		finder.setNegative(true);
+		finder.setText(text);
+
+		Assertions.assertEquals(8, finder.start(text.length() - 1));
+		Assertions.assertEquals(4, finder.start(7));
+		Assertions.assertEquals(0, finder.start(2));
+	}
+
+	@Test
+	public void testReverseIgnoreCase() {
+		// 反向查找，忽略大小写
+		String text = "ABC abc Abc";
+		StrFinder finder = StrFinder.of("abc", true);
+		finder.setNegative(true);
+		finder.setText(text);
+
+		Assertions.assertEquals(8, finder.start(text.length() - 1));
+	}
+
+	@Test
+	public void testAlgorithmEdgeCase() {
+
+		String text = "ababa";
+
+		StrFinder forward = StrFinder.of("aba", false);
+		forward.setText(text);
+		Assertions.assertEquals(0, forward.start(0));
+		Assertions.assertEquals(2, forward.start(1));
+
+		StrFinder reverse = StrFinder.of("aba", false);
+		reverse.setNegative(true);
+		reverse.setText(text);
+		Assertions.assertEquals(2, reverse.start(4));
+		Assertions.assertEquals(0, reverse.start(1));
+	}
+
+	@Test
+	public void testZeroCopy() {
+		// 验证toString()优化
+		StringBuilder bigText = new StringBuilder();
+		bigText.append("ignore-".repeat(1000));
+		bigText.append("TARGET");
+		bigText.append("-ignore");
+
+		StrFinder finder = StrFinder.of("TARGET", false);
+		finder.setText(bigText);
+
+		Assertions.assertEquals(7000, finder.start(0));
+	}
+
+	@Test
+	public void testChinese() {
+		// 中文测试
+		String text = "希望pr能够通过";
+		StrFinder finder = StrFinder.of("通过", false);
+		finder.setText(text);
+
+		Assertions.assertEquals(6, finder.start(0));
+	}
+
+	@Test
+	public void testNotFound() {
+		// 不包含字符串
+		StrFinder finder = StrFinder.of("NotExists", false);
+		finder.setText("Hello World");
+		Assertions.assertEquals(-1, finder.start(0));
+	}
+
+	@Test
+	public void benchmark() {
+		System.out.println("正在生成测试数据...");
+		StringBuilder sb = new StringBuilder();
+		String base = "abcdefghijklmnopqrstuvwxyz0123456789-";
+		sb.append(base.repeat(500));
+		String target = "HUTOOL_TARGET";
+		sb.append(target);
+
+		String textStr = sb.toString();
+
+		int loop = 100;
+		for (int i = 0; i < 100; i++) {
+			StrFinder.of(target, false).setText(textStr).start(0);
+		}
+
+		long start = System.currentTimeMillis();
+		for (int i = 0; i < loop; i++) {
+			int index = textStr.indexOf(target);
+			if (index == -1) throw new RuntimeException("Bug!");
+		}
+		long end = System.currentTimeMillis();
+		System.out.println("1. JDK String.indexOf耗时: " + (end - start) + "ms");
+
+		start = System.currentTimeMillis();
+		for (int i = 0; i < loop; i++) {
+			int index = StrFinder.of(target, false).setText(textStr).start(0);
+			if (index == -1) throw new RuntimeException("Bug!");
+		}
+		end = System.currentTimeMillis();
+		System.out.println("2. StrFinder (String) 耗时: " + (end - start) + "ms");
+
+		start = System.currentTimeMillis();
+		for (int i = 0; i < loop; i++) {
+			int index = StrFinder.of(target, false).setText(sb).start(0);
+			if (index == -1) throw new RuntimeException("Bug!");
+		}
+		end = System.currentTimeMillis();
+		System.out.println("3. StrFinder (Builder)耗时: " + (end - start) + "ms");
+	}
+}