admin/hutool
1
0
Fork 0
mirror of https://gitee.com/chinabugotech/hutool.git synced 2025-12-06 17:18:54 +08:00
Code Issues Packages Projects Releases Wiki Activity

优化EscapeUtil，兼容不规范的转义（pr#4150@Github）

Browse Source
This commit is contained in:
Looly 2025-11-25 17:07:17 +08:00
parent 654db66ecb
commit 3c067f1871
4 changed files with 183 additions and 24 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

4
CHANGELOG.md
View File

@ -1,7 +1,7 @@
# 🚀Changelog
-------------------------------------------------------------------------------------------------------------
# 5.8.42(2025-11-24)
# 5.8.42(2025-11-25)
### 🐣新特性
* 【core 】 `ListUtil`增加`zip`方法（pr#4052@Github）
@ -9,6 +9,7 @@
* 【ai 】 增加代理支持（pr#4107@Github）
* 【core 】 `CharSequenceUtil`增加`builder`方法重载（pr#4107@Github）
* 【core 】 `Combination`和`Arrangement `重构避免数组频繁拷贝，并避免溢出（pr#4144@Github）
* 【core 】 优化`EscapeUtil`，兼容不规范的转义（pr#4150@Github）
### 🐞Bug修复
* 【jwt 】 修复verify方法在定义alg为`none`时验证失效问题（issue#4105@Github）
@ -27,6 +28,7 @@
* 【core 】 修复`CreditCodeUtil.randomCreditCode`部分字母未使用问题（pr#4149@Github）
* 【core 】 修复`CacheableAnnotationAttribute`可能并发问题（pr#4149@Github）
* 【core 】 修复`URLUtil.url`未断开连接问题（pr#4149@Github）
* 【core 】 修复`Bimap.put`重复put问题（pr#4150@Github）
-------------------------------------------------------------------------------------------------------------
# 5.8.41(2025-10-12)

10
hutool-core/src/main/java/cn/hutool/core/map/BiMap.java
View File

@ -39,7 +39,7 @@ public class BiMap<K, V> extends MapWrapper<K, V> {
}
this.inverse.put(value, key);
}
return super.put(key, value);
return oldValue;
}
@Override
@ -94,10 +94,12 @@ public class BiMap<K, V> extends MapWrapper<K, V> {
@Override
public V putIfAbsent(K key, V value) {
if (null != this.inverse) {
this.inverse.putIfAbsent(value, key);
final V oldValue = super.putIfAbsent(key, value);
// 只有当oldValue为null时(即key之前不存在),才更新反向Map
if (null == oldValue && null != this.inverse) {
this.inverse.put(value, key);
}
return super.putIfAbsent(key, value);
return oldValue;
}
@Override

51
hutool-core/src/main/java/cn/hutool/core/util/EscapeUtil.java
View File

@ -10,7 +10,7 @@ import cn.hutool.core.text.escape.XmlUnescape;
* 转义和反转义工具类Escape / Unescape<br>
* escape采用ISO Latin字符集对指定的字符串进行编码。<br>
* 所有的空格符、标点符号、特殊字符以及其他非ASCII字符都将被转化成%xx格式的字符编码(xx等于该字符在字符集表里面的编码的16进制数字)。
* TODO 6.x迁移到core.text.escape包下
* TODO 7.x迁移到core.text.escape包下
*
* @author xiaoleilu
*/
@ -20,7 +20,7 @@ public class EscapeUtil {
* 不转义的符号编码
*/
private static final String NOT_ESCAPE_CHARS = "*@-_+./";
private static final Filter<Character> JS_ESCAPE_FILTER = c -> false == (
private static final Filter<Character> JS_ESCAPE_FILTER = c -> !(
Character.isDigit(c)
|| Character.isLowerCase(c)
|| Character.isUpperCase(c)
@ -122,7 +122,7 @@ public class EscapeUtil {
char c;
for (int i = 0; i < content.length(); i++) {
c = content.charAt(i);
if (false == filter.accept(c)) {
if (!filter.accept(c)) {
tmp.append(c);
} else if (c < 256) {
tmp.append("%");
@ -143,36 +143,69 @@ public class EscapeUtil {
}
/**
* Escape解码
* Escape解码支持两种转义格式的解码：
* <ul>
* <li>%XX - 两位十六进制数字，用于表示ASCII字符（0-255）</li>
* <li>%uXXXX - 四位十六进制数字，用于表示Unicode字符</li>
* </ul>
* <p>
* 对于不完整的转义序列，本方法会将其原样保留而不抛出异常：
* <ul>
* <li>字符串末尾的单独"%"字符会被原样保留</li>
* <li>"%u"后面不足4位十六进制数字时，整个不完整序列会被原样保留</li>
* <li>"%"后面不足2位十六进制数字时（非%u格式），整个不完整序列会被原样保留</li>
* </ul>
* 例如：
* <pre>
* unescape("test%") = "test%" // 末尾的%被保留
* unescape("test%u12") = "test%u12" // 不足4位，原样保留
* unescape("test%2") = "test%2" // 不足2位，原样保留
* unescape("test%20") = "test " // 正常解码空格
* unescape("test%u4E2D") = "test中" // 正常解码中文字符
* </pre>
*
* @param content 被转义的内容
* @return 解码后的字符串
*/
public static String unescape(String content) {
public static String unescape(final String content) {
if (StrUtil.isBlank(content)) {
return content;
}
StringBuilder tmp = new StringBuilder(content.length());
final int len = content.length();
final StringBuilder tmp = new StringBuilder(len);
int lastPos = 0;
int pos;
char ch;
while (lastPos < content.length()) {
while (lastPos < len) {
pos = content.indexOf("%", lastPos);
if (pos == lastPos) {
if (content.charAt(pos + 1) == 'u') {
if (pos + 1 < len && content.charAt(pos + 1) == 'u') {
if (pos + 6 <= len) {
ch = (char) Integer.parseInt(content.substring(pos + 2, pos + 6), 16);
tmp.append(ch);
lastPos = pos + 6;
} else {
// Not enough characters, append as-is
tmp.append(content.substring(pos));
lastPos = len;
}
} else {
// Check if there's enough characters for hex escape (%XX)
if (pos + 3 <= len) {
ch = (char) Integer.parseInt(content.substring(pos + 1, pos + 3), 16);
tmp.append(ch);
lastPos = pos + 3;
} else {
// Not enough characters, append as-is
tmp.append(content.substring(pos));
lastPos = len;
}
}
} else {
if (pos == -1) {
tmp.append(content.substring(lastPos));
lastPos = content.length();
lastPos = len;
} else {
tmp.append(content, lastPos, pos);
lastPos = pos;

122
hutool-core/src/test/java/cn/hutool/core/util/EscapeUtilTest.java
View File

@ -66,4 +66,126 @@ public class EscapeUtilTest {
final String s = EscapeUtil.unescapeHtml4(str);
assertEquals("'some text with single quotes'", s);
}
@Test
public void escapeXmlTest(){
final String a = "<>";
final String escape = EscapeUtil.escapeXml(a);
assertEquals("&lt;&gt;", escape);
assertEquals("中文“双引号”", EscapeUtil.escapeXml("中文“双引号”"));
}
@Test
void testUnescapeNull() {
assertNull(EscapeUtil.unescape(null));
}
@Test
void testUnescapeEmpty() {
assertEquals("", EscapeUtil.unescape(""));
}
@Test
void testUnescapeBlank() {
assertEquals(" ", EscapeUtil.unescape(" "));
}
@Test
void testUnescapeAsciiCharacters() {
// 测试ASCII字符转义
assertEquals("hello", EscapeUtil.unescape("hello"));
assertEquals("test space", EscapeUtil.unescape("test%20space"));
assertEquals("A", EscapeUtil.unescape("%41"));
assertEquals("a", EscapeUtil.unescape("%61"));
assertEquals("0", EscapeUtil.unescape("%30"));
assertEquals("!", EscapeUtil.unescape("%21"));
assertEquals("@", EscapeUtil.unescape("%40"));
assertEquals("#", EscapeUtil.unescape("%23"));
}
@Test
void testUnescapeUnicodeCharacters() {
// 测试Unicode字符转义
assertEquals("中", EscapeUtil.unescape("%u4E2D"));
assertEquals("文", EscapeUtil.unescape("%u6587"));
assertEquals("测", EscapeUtil.unescape("%u6D4B"));
assertEquals("试", EscapeUtil.unescape("%u8BD5"));
assertEquals("😊", EscapeUtil.unescape("%uD83D%uDE0A")); // 笑脸表情
}
@Test
void testUnescapeMixedContent() {
// 测试混合内容
assertEquals("Hello 世界!", EscapeUtil.unescape("Hello%20%u4E16%u754C%21"));
assertEquals("测试: 100%", EscapeUtil.unescape("%u6D4B%u8BD5%3A%20100%25"));
assertEquals("a+b=c", EscapeUtil.unescape("a%2Bb%3Dc"));
}
@Test
void testUnescapeIncompleteEscapeSequences() {
// 测试不完整的转义序列
assertEquals("test%", EscapeUtil.unescape("test%"));
assertEquals("test%u", EscapeUtil.unescape("test%u"));
assertEquals("test%u1", EscapeUtil.unescape("test%u1"));
assertEquals("test%u12", EscapeUtil.unescape("test%u12"));
assertEquals("test%u123", EscapeUtil.unescape("test%u123"));
assertEquals("test%1", EscapeUtil.unescape("test%1"));
assertEquals("test%2", EscapeUtil.unescape("test%2"));
}
@Test
void testUnescapeEdgeCases() {
// 测试边界情况
assertEquals("%", EscapeUtil.unescape("%"));
assertEquals("%u", EscapeUtil.unescape("%u"));
assertEquals("%%", EscapeUtil.unescape("%%"));
assertEquals("%u%", EscapeUtil.unescape("%u%"));
assertEquals("100% complete", EscapeUtil.unescape("100%25%20complete"));
}
@Test
void testUnescapeMultipleEscapeSequences() {
// 测试多个连续的转义序列
assertEquals("ABC", EscapeUtil.unescape("%41%42%43"));
assertEquals("中文测试", EscapeUtil.unescape("%u4E2D%u6587%u6D4B%u8BD5"));
assertEquals("A 中 B", EscapeUtil.unescape("%41%20%u4E2D%20%42"));
}
@Test
void testUnescapeSpecialCharacters() {
// 测试特殊字符
assertEquals("\n", EscapeUtil.unescape("%0A"));
assertEquals("\r", EscapeUtil.unescape("%0D"));
assertEquals("\t", EscapeUtil.unescape("%09"));
assertEquals(" ", EscapeUtil.unescape("%20"));
assertEquals("<", EscapeUtil.unescape("%3C"));
assertEquals(">", EscapeUtil.unescape("%3E"));
assertEquals("&", EscapeUtil.unescape("%26"));
}
@Test
void testUnescapeComplexScenario() {
// 测试复杂场景
final String original = "Hello 世界! 这是测试。Email: test@example.com";
final String escaped = "Hello%20%u4E16%u754C%21%20%u8FD9%u662F%u6D4B%u8BD5%u3002Email%3A%20test%40example.com";
assertEquals(original, EscapeUtil.unescape(escaped));
}
@Test
void testUnescapeWithIncompleteAtEnd() {
// 测试末尾有不完整转义序列
assertEquals("normal%", EscapeUtil.unescape("normal%"));
assertEquals("normal%u", EscapeUtil.unescape("normal%u"));
assertEquals("normal%u1", EscapeUtil.unescape("normal%u1"));
assertEquals("normal%1", EscapeUtil.unescape("normal%1"));
}
@Test
void testUnescapeUppercaseHex() {
// 测试大写十六进制
assertEquals("A", EscapeUtil.unescape("%41"));
assertEquals("A", EscapeUtil.unescape("%41"));
assertEquals("中", EscapeUtil.unescape("%u4E2D"));
assertEquals("中", EscapeUtil.unescape("%u4E2D"));
}
}