View Javadoc
1   /*
2    * Licensed to the Apache Software Foundation (ASF) under one or more
3    * contributor license agreements.  See the NOTICE file distributed with
4    * this work for additional information regarding copyright ownership.
5    * The ASF licenses this file to You under the Apache License, Version 2.0
6    * (the "License"); you may not use this file except in compliance with
7    * the License.  You may obtain a copy of the License at
8    *
9    *      http://www.apache.org/licenses/LICENSE-2.0
10   *
11   * Unless required by applicable law or agreed to in writing, software
12   * distributed under the License is distributed on an "AS IS" BASIS,
13   * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
14   * See the License for the specific language governing permissions and
15   * limitations under the License.
16   */
17  package org.apache.commons.text;
18  
19  import static org.apache.commons.text.StringEscapeUtils.escapeXSI;
20  import static org.apache.commons.text.StringEscapeUtils.unescapeXSI;
21  import static org.junit.jupiter.api.Assertions.assertEquals;
22  import static org.junit.jupiter.api.Assertions.assertFalse;
23  import static org.junit.jupiter.api.Assertions.assertNotNull;
24  import static org.junit.jupiter.api.Assertions.assertNull;
25  import static org.junit.jupiter.api.Assertions.assertThrows;
26  import static org.junit.jupiter.api.Assertions.assertTrue;
27  import static org.junit.jupiter.api.Assertions.fail;
28  
29  import java.io.IOException;
30  import java.io.StringWriter;
31  import java.lang.reflect.Constructor;
32  import java.lang.reflect.Modifier;
33  import java.nio.charset.StandardCharsets;
34  import java.nio.file.Files;
35  import java.nio.file.Paths;
36  
37  import org.junit.jupiter.api.Test;
38  
39  /**
40   * Tests {@link StringEscapeUtils}.
41   *
42   * <p>
43   * This code has been adapted from Apache Commons Lang 3.5.
44   * </p>
45   */
46  public class StringEscapeUtilsTest {
47      private static final String FOO = "foo";
48  
49      private static final String[][] HTML_ESCAPES = {
50              {"no escaping", "plain text", "plain text"},
51              {"no escaping", "plain text", "plain text"},
52              {"empty string", "", ""},
53              {"null", null, null},
54              {"ampersand", "bread &amp; butter", "bread & butter"},
55              {"quotes", "&quot;bread&quot; &amp; butter", "\"bread\" & butter"},
56              {"final character only", "greater than &gt;", "greater than >"},
57              {"first character only", "&lt; less than", "< less than"},
58              {"apostrophe", "Huntington's chorea", "Huntington's chorea"},
59              {"languages", "English,Fran&ccedil;ais,\u65E5\u672C\u8A9E (nihongo)",
60                  "English,Fran\u00E7ais,\u65E5\u672C\u8A9E (nihongo)"},
61              {"8-bit ascii shouldn't number-escape", "\u0080\u009F", "\u0080\u009F"},
62      };
63  
64      private void assertEscapeJava(final String escaped, final String original) throws IOException {
65          assertEscapeJava(escaped, original, null);
66      }
67  
68      private void assertEscapeJava(final String expected, final String original, String message) throws IOException {
69          final String converted = StringEscapeUtils.escapeJava(original);
70          message = "escapeJava(String) failed" + (message == null ? "" : ": " + message);
71          assertEquals(expected, converted, message);
72  
73          final StringWriter writer = new StringWriter();
74          StringEscapeUtils.ESCAPE_JAVA.translate(original, writer);
75          assertEquals(expected, writer.toString());
76      }
77  
78      private void assertUnescapeJava(final String unescaped, final String original) throws IOException {
79          assertUnescapeJava(unescaped, original, null);
80      }
81  
82      private void assertUnescapeJava(final String unescaped, final String original, final String message)
83              throws IOException {
84          final String actual = StringEscapeUtils.unescapeJava(original);
85  
86          assertEquals(unescaped, actual, "unescape(String) failed"
87                          + (message == null ? "" : ": " + message)
88                          + ": expected '" + StringEscapeUtils.escapeJava(unescaped)
89                          // we escape this so we can see it in the error message
90                          + "' actual '" + StringEscapeUtils.escapeJava(actual) + "'");
91  
92          final StringWriter writer = new StringWriter();
93          StringEscapeUtils.UNESCAPE_JAVA.translate(original, writer);
94          assertEquals(unescaped, writer.toString());
95      }
96  
97      private void checkCsvEscapeWriter(final String expected, final String value) throws IOException {
98          final StringWriter writer = new StringWriter();
99          StringEscapeUtils.ESCAPE_CSV.translate(value, writer);
100         assertEquals(expected, writer.toString());
101     }
102 
103     private void checkCsvUnescapeWriter(final String expected, final String value) throws IOException {
104         final StringWriter writer = new StringWriter();
105         StringEscapeUtils.UNESCAPE_CSV.translate(value, writer);
106         assertEquals(expected, writer.toString());
107     }
108 
109     @Test
110     public void testBuilder() {
111         final String result =
112                 StringEscapeUtils.builder(StringEscapeUtils.ESCAPE_XML10).escape("<").append(">").toString();
113         assertEquals("&lt;>", result);
114     }
115 
116     @Test
117     public void testConstructor() {
118         assertNotNull(new StringEscapeUtils());
119         final Constructor<?>[] cons = StringEscapeUtils.class.getDeclaredConstructors();
120         assertEquals(1, cons.length);
121         assertTrue(Modifier.isPublic(cons[0].getModifiers()));
122         assertTrue(Modifier.isPublic(StringEscapeUtils.class.getModifiers()));
123         assertFalse(Modifier.isFinal(StringEscapeUtils.class.getModifiers()));
124     }
125 
126     // HTML and XML
127     @Test
128     public void testDeleteCharacter() {
129       final String deleteString = "Delete: \u007F";
130       assertEquals("Delete: \\u007F", StringEscapeUtils.escapeJson(deleteString));
131     }
132 
133     @Test
134     public void testEscapeCsvString() {
135         assertEquals("foo.bar", StringEscapeUtils.escapeCsv("foo.bar"));
136         assertEquals("\"foo,bar\"", StringEscapeUtils.escapeCsv("foo,bar"));
137         assertEquals("\"foo\nbar\"", StringEscapeUtils.escapeCsv("foo\nbar"));
138         assertEquals("\"foo\rbar\"", StringEscapeUtils.escapeCsv("foo\rbar"));
139         assertEquals("\"foo\"\"bar\"", StringEscapeUtils.escapeCsv("foo\"bar"));
140         assertEquals("foo\uD84C\uDFB4bar", StringEscapeUtils.escapeCsv("foo\uD84C\uDFB4bar"));
141         assertEquals("", StringEscapeUtils.escapeCsv(""));
142         assertNull(StringEscapeUtils.escapeCsv(null));
143     }
144 
145     @Test
146     public void testEscapeCsvWriter() throws IOException {
147         checkCsvEscapeWriter("foo.bar", "foo.bar");
148         checkCsvEscapeWriter("\"foo,bar\"", "foo,bar");
149         checkCsvEscapeWriter("\"foo\nbar\"", "foo\nbar");
150         checkCsvEscapeWriter("\"foo\rbar\"", "foo\rbar");
151         checkCsvEscapeWriter("\"foo\"\"bar\"", "foo\"bar");
152         checkCsvEscapeWriter("foo\uD84C\uDFB4bar", "foo\uD84C\uDFB4bar");
153         checkCsvEscapeWriter("", null);
154         checkCsvEscapeWriter("", "");
155     }
156 
157 @Test
158 public void testEscapeEcmaScript() {
159     assertNull(StringEscapeUtils.escapeEcmaScript(null));
160     try {
161         StringEscapeUtils.ESCAPE_ECMASCRIPT.translate(null, null);
162         fail("Exception expected!");
163     } catch (final IOException ex) {
164         fail("Exception expected!");
165     } catch (final IllegalArgumentException ex) {
166         // expected
167     }
168     try {
169         StringEscapeUtils.ESCAPE_ECMASCRIPT.translate("", null);
170         fail("Exception expected!");
171     } catch (final IOException ex) {
172         fail("Exception expected!");
173     } catch (final IllegalArgumentException ex) {
174         // expected
175     }
176 
177     assertEquals("He didn\\'t say, \\\"stop!\\\"", StringEscapeUtils.escapeEcmaScript("He didn't say, \"stop!\""));
178     assertEquals("document.getElementById(\\\"test\\\").value = \\'<script>alert(\\'aaa\\');<\\/script>\\';",
179             StringEscapeUtils.escapeEcmaScript(
180                     "document.getElementById(\"test\").value = '<script>alert('aaa');</script>';"));
181 }
182 
183     /**
184      * Tests https://issues.apache.org/jira/browse/LANG-339
185      */
186     @Test
187     public void testEscapeHiragana() {
188         // Some random Japanese Unicode characters
189         final String original = "\u304B\u304C\u3068";
190         final String escaped = StringEscapeUtils.escapeHtml4(original);
191         assertEquals(original, escaped,
192                 "Hiragana character Unicode behavior should not be being escaped by escapeHtml4");
193 
194         final String unescaped = StringEscapeUtils.unescapeHtml4(escaped);
195 
196         assertEquals(escaped, unescaped, "Hiragana character Unicode behavior has changed - expected no unescaping");
197     }
198 
199     @Test
200     public void testEscapeHtml3() {
201         for (final String[] element : HTML_ESCAPES) {
202             final String message = element[0];
203             final String expected = element[1];
204             final String original = element[2];
205             assertEquals(expected, StringEscapeUtils.escapeHtml4(original), message);
206             final StringWriter sw = new StringWriter();
207             try {
208                 StringEscapeUtils.ESCAPE_HTML3.translate(original, sw);
209             } catch (final IOException e) {
210                 // expected
211             }
212             final String actual = original == null ? null : sw.toString();
213             assertEquals(expected, actual, message);
214         }
215     }
216 
217     @Test
218         public void testEscapeHtml4() {
219             for (final String[] element : HTML_ESCAPES) {
220                 final String message = element[0];
221                 final String expected = element[1];
222                 final String original = element[2];
223                 assertEquals(expected, StringEscapeUtils.escapeHtml4(original), message);
224                 final StringWriter sw = new StringWriter();
225                 try {
226                     StringEscapeUtils.ESCAPE_HTML4.translate(original, sw);
227                 } catch (final IOException e) {
228                     // expected
229                 }
230                 final String actual = original == null ? null : sw.toString();
231                 assertEquals(expected, actual, message);
232             }
233         }
234 
235     /**
236      * Tests // https://issues.apache.org/jira/browse/LANG-480
237      */
238     @Test
239     public void testEscapeHtmlHighUnicode() {
240         // this is the utf8 representation of the character:
241         // COUNTING ROD UNIT DIGIT THREE
242         // in Unicode
243         // code point: U+1D362
244         final byte[] data = {(byte) 0xF0, (byte) 0x9D, (byte) 0x8D, (byte) 0xA2};
245 
246         final String original = new String(data, StandardCharsets.UTF_8);
247 
248         final String escaped = StringEscapeUtils.escapeHtml4(original);
249         assertEquals(original, escaped, "High Unicode should not have been escaped");
250 
251         final String unescaped = StringEscapeUtils.unescapeHtml4(escaped);
252         assertEquals(original, unescaped, "High Unicode should have been unchanged");
253 
254         // TODO: I think this should hold, needs further investigation
255         //        String unescapedFromEntity = StringEscapeUtils.unescapeHtml4("&#119650;");
256         //        assertEquals("High Unicode should have been unescaped", original, unescapedFromEntity);
257     }
258 
259     @Test
260     public void testEscapeHtmlThree() {
261         assertNull(StringEscapeUtils.escapeHtml3(null));
262         assertEquals("a", StringEscapeUtils.escapeHtml3("a"));
263         assertEquals("&lt;b&gt;a", StringEscapeUtils.escapeHtml3("<b>a"));
264     }
265 
266     @Test
267     public void testEscapeHtmlVersions() {
268         assertEquals("&Beta;", StringEscapeUtils.escapeHtml4("\u0392"));
269         assertEquals("\u0392", StringEscapeUtils.unescapeHtml4("&Beta;"));
270 
271         // TODO: refine API for escaping/unescaping specific HTML versions
272     }
273 
274     @Test
275     public void testEscapeJava() throws IOException {
276         assertNull(StringEscapeUtils.escapeJava(null));
277         try {
278             StringEscapeUtils.ESCAPE_JAVA.translate(null, null);
279             fail("Exception expected!");
280         } catch (final IOException ex) {
281             fail("Exception expected!");
282         } catch (final IllegalArgumentException ex) {
283             // expected
284         }
285         try {
286             StringEscapeUtils.ESCAPE_JAVA.translate("", null);
287             fail("Exception expected!");
288         } catch (final IOException ex) {
289             fail("Exception expected!");
290         } catch (final IllegalArgumentException ex) {
291             // expected
292         }
293 
294         assertEscapeJava("", "", "empty string");
295         assertEscapeJava(FOO, FOO);
296         assertEscapeJava("\\t", "\t", "tab");
297         assertEscapeJava("\\\\", "\\", "backslash");
298         assertEscapeJava("'", "'", "single quote should not be escaped");
299         assertEscapeJava("\\\\\\b\\t\\r", "\\\b\t\r");
300         assertEscapeJava("\\u1234", "\u1234");
301         assertEscapeJava("\\u0234", "\u0234");
302         assertEscapeJava("\\u00EF", "\u00ef");
303         assertEscapeJava("\\u0001", "\u0001");
304         assertEscapeJava("\\uABCD", "\uabcd", "Should use capitalized Unicode hex");
305 
306         assertEscapeJava("He didn't say, \\\"stop!\\\"",
307                 "He didn't say, \"stop!\"");
308         assertEscapeJava("This space is non-breaking:" + "\\u00A0", "This space is non-breaking:\u00a0",
309                 "non-breaking space");
310         assertEscapeJava("\\uABCD\\u1234\\u012C",
311                 "\uABCD\u1234\u012C");
312     }
313 
314     /**
315      * Tests https://issues.apache.org/jira/browse/LANG-421
316      */
317     @Test
318     public void testEscapeJavaWithSlash() {
319         final String input = "String with a slash (/) in it";
320 
321         final String actual = StringEscapeUtils.escapeJava(input);
322 
323         /*
324          * In 2.4 StringEscapeUtils.escapeJava(String) escapes '/' characters, which are not a valid character
325          * to escape in a Java string.
326          */
327         assertEquals(input, actual);
328     }
329 
330     @Test
331     public void testEscapeJson() {
332         assertNull(StringEscapeUtils.escapeJson(null));
333         try {
334             StringEscapeUtils.ESCAPE_JSON.translate(null, null);
335             fail("Exception expected!");
336         } catch (final IOException ex) {
337             fail("Exception expected!");
338         } catch (final IllegalArgumentException ex) {
339             // expected
340         }
341         try {
342             StringEscapeUtils.ESCAPE_JSON.translate("", null);
343             fail("Exception expected!");
344         } catch (final IOException ex) {
345             fail("Exception expected!");
346         } catch (final IllegalArgumentException ex) {
347             // expected
348         }
349 
350         assertEquals("He didn't say, \\\"stop!\\\"", StringEscapeUtils.escapeJson("He didn't say, \"stop!\""));
351 
352         final String expected = "\\\"foo\\\" isn't \\\"bar\\\". specials: \\b\\r\\n\\f\\t\\\\\\/";
353         final String input = "\"foo\" isn't \"bar\". specials: \b\r\n\f\t\\/";
354 
355         assertEquals(expected, StringEscapeUtils.escapeJson(input));
356     }
357 
358     @Test
359     public void testEscapeXml10() {
360         assertEquals("a&lt;b&gt;c&quot;d&apos;e&amp;f", StringEscapeUtils.escapeXml10("a<b>c\"d'e&f"));
361         assertEquals("a\tb\rc\nd", StringEscapeUtils.escapeXml10("a\tb\rc\nd"),
362                 "XML 1.0 should not escape \t \n \r");
363         assertEquals("ab", StringEscapeUtils.escapeXml10("a\u0000\u0001\u0008\u000b\u000c\u000e\u001fb"),
364                 "XML 1.0 should omit most #x0-x8 | #xb | #xc | #xe-#x19");
365         assertEquals("a\ud7ff  \ue000b", StringEscapeUtils.escapeXml10("a\ud7ff\ud800 \udfff \ue000b"),
366                 "XML 1.0 should omit #xd800-#xdfff");
367         assertEquals("a\ufffdb", StringEscapeUtils.escapeXml10("a\ufffd\ufffe\uffffb"),
368                 "XML 1.0 should omit #xfffe | #xffff");
369         assertEquals("a\u007e&#127;&#132;\u0085&#134;&#159;\u00a0b",
370                 StringEscapeUtils.escapeXml10("a\u007e\u007f\u0084\u0085\u0086\u009f\u00a0b"),
371                 "XML 1.0 should escape #x7f-#x84 | #x86 - #x9f, for XML 1.1 compatibility");
372     }
373 
374     @Test
375     public void testEscapeXml11() {
376         assertEquals("a&lt;b&gt;c&quot;d&apos;e&amp;f", StringEscapeUtils.escapeXml11("a<b>c\"d'e&f"));
377         assertEquals("a\tb\rc\nd", StringEscapeUtils.escapeXml11("a\tb\rc\nd"),
378                 "XML 1.1 should not escape \t \n \r");
379         assertEquals("ab", StringEscapeUtils.escapeXml11("a\u0000b"),
380                 "XML 1.1 should omit #x0");
381         assertEquals("a&#1;&#8;&#11;&#12;&#14;&#31;b",
382                 StringEscapeUtils.escapeXml11("a\u0001\u0008\u000b\u000c\u000e\u001fb"),
383                 "XML 1.1 should escape #x1-x8 | #xb | #xc | #xe-#x19");
384         assertEquals("a\u007e&#127;&#132;\u0085&#134;&#159;\u00a0b",
385                 StringEscapeUtils.escapeXml11("a\u007e\u007f\u0084\u0085\u0086\u009f\u00a0b"),
386                 "XML 1.1 should escape #x7F-#x84 | #x86-#x9F");
387         assertEquals("a\ud7ff  \ue000b", StringEscapeUtils.escapeXml11("a\ud7ff\ud800 \udfff \ue000b"),
388                 "XML 1.1 should omit #xd800-#xdfff");
389         assertEquals("a\ufffdb", StringEscapeUtils.escapeXml11("a\ufffd\ufffe\uffffb"),
390                 "XML 1.1 should omit #xfffe | #xffff");
391     }
392 
393     @Test
394     public void testEscapeXSI() {
395         assertNull(null, escapeXSI(null));
396         assertEquals("He\\ didn\\'t\\ say,\\ \\\"Stop!\\\"", escapeXSI("He didn't say, \"Stop!\""));
397         assertEquals("\\\\", escapeXSI("\\"));
398         assertEquals("", escapeXSI("\n"));
399     }
400 
401     @Test
402     public void testLang313() {
403         assertEquals("& &", StringEscapeUtils.unescapeHtml4("& &amp;"));
404     }
405 
406     /**
407      * Tests https://issues.apache.org/jira/browse/LANG-708
408      *
409      * @throws IOException
410      *             if an I/O error occurs
411      */
412     @Test
413     public void testLang708() throws IOException {
414         final byte[] inputBytes = Files.readAllBytes(Paths.get("src/test/resources/org/apache/commons/text/stringEscapeUtilsTestData.txt"));
415         final String input = new String(inputBytes, StandardCharsets.UTF_8);
416         final String escaped = StringEscapeUtils.escapeEcmaScript(input);
417         // just the end:
418         assertTrue(escaped.endsWith("}]"), escaped);
419         // a little more:
420         assertTrue(escaped.endsWith("\"valueCode\\\":\\\"\\\"}]"), escaped);
421     }
422 
423     /**
424      * Tests https://issues.apache.org/jira/browse/LANG-911
425      */
426     @Test
427     public void testLang911() {
428         final String bellsTest = "\ud83d\udc80\ud83d\udd14";
429         final String value = StringEscapeUtils.escapeJava(bellsTest);
430         final String valueTest = StringEscapeUtils.unescapeJava(value);
431         assertEquals(bellsTest, valueTest);
432     }
433 
434     // Tests issue #38569
435     // https://issues.apache.org/bugzilla/show_bug.cgi?id=38569
436     @Test
437     public void testStandaloneAmphersand() {
438         assertEquals("<P&O>", StringEscapeUtils.unescapeHtml4("&lt;P&O&gt;"));
439         assertEquals("test & <", StringEscapeUtils.unescapeHtml4("test & &lt;"));
440         assertEquals("<P&O>", StringEscapeUtils.unescapeXml("&lt;P&O&gt;"));
441         assertEquals("test & <", StringEscapeUtils.unescapeXml("test & &lt;"));
442     }
443 
444     @Test
445     public void testUnescapeCsvString() {
446         assertEquals("foo.bar", StringEscapeUtils.unescapeCsv("foo.bar"));
447         assertEquals("foo,bar", StringEscapeUtils.unescapeCsv("\"foo,bar\""));
448         assertEquals("foo\nbar", StringEscapeUtils.unescapeCsv("\"foo\nbar\""));
449         assertEquals("foo\rbar", StringEscapeUtils.unescapeCsv("\"foo\rbar\""));
450         assertEquals("foo\"bar", StringEscapeUtils.unescapeCsv("\"foo\"\"bar\""));
451         assertEquals("foo\uD84C\uDFB4bar", StringEscapeUtils.unescapeCsv("foo\uD84C\uDFB4bar"));
452         assertEquals("", StringEscapeUtils.unescapeCsv(""));
453         assertNull(StringEscapeUtils.unescapeCsv(null));
454 
455         assertEquals("foo.bar", StringEscapeUtils.unescapeCsv("\"foo.bar\""));
456     }
457 
458     @Test
459     public void testUnescapeCsvWriter() throws IOException {
460         checkCsvUnescapeWriter("foo.bar", "foo.bar");
461         checkCsvUnescapeWriter("foo,bar", "\"foo,bar\"");
462         checkCsvUnescapeWriter("foo\nbar", "\"foo\nbar\"");
463         checkCsvUnescapeWriter("foo\rbar", "\"foo\rbar\"");
464         checkCsvUnescapeWriter("foo\"bar", "\"foo\"\"bar\"");
465         checkCsvUnescapeWriter("foo\uD84C\uDFB4bar", "foo\uD84C\uDFB4bar");
466         checkCsvUnescapeWriter("", null);
467         checkCsvUnescapeWriter("", "");
468 
469         checkCsvUnescapeWriter("foo.bar", "\"foo.bar\"");
470     }
471 
472     @Test
473     public void testUnescapeEcmaScript() {
474         assertNull(StringEscapeUtils.unescapeEcmaScript(null));
475         assertEquals("8lvc1u+6B#-I", StringEscapeUtils.unescapeEcmaScript("8lvc1u+6B#-I"));
476         assertEquals("<script src=\"build/main.bundle.js\"></script>",
477                 StringEscapeUtils.unescapeEcmaScript("<script src=\"build/main.bundle.js\"></script>"));
478         assertEquals("<script src=\"build/main.bundle.js\"></script>>",
479                 StringEscapeUtils.unescapeEcmaScript("<script src=\"build/main.bundle.js\"></script>>"));
480     }
481 
482     @Test
483     public void testUnescapeHexCharsHtml() {
484         // Simple easy to grok test
485         assertEquals("\u0080\u009F", StringEscapeUtils.unescapeHtml4("&#x80;&#x9F;"), "hex number unescape");
486         assertEquals("\u0080\u009F", StringEscapeUtils.unescapeHtml4("&#X80;&#X9F;"), "hex number unescape");
487         // Test all Character values:
488         for (char i = Character.MIN_VALUE; i < Character.MAX_VALUE; i++) {
489             final char c2 = (char) (i + 1);
490             final String expected = Character.toString(i) + Character.toString(c2);
491             final String escapedC1 = "&#x" + Integer.toHexString(i) + ";";
492             final String escapedC2 = "&#x" + Integer.toHexString(c2) + ";";
493             assertEquals(expected, StringEscapeUtils.unescapeHtml4(escapedC1 + escapedC2),
494                     "hex number unescape index " + i);
495         }
496     }
497 
498     @Test
499     public void testUnescapeHtml3() {
500         for (final String[] element : HTML_ESCAPES) {
501             final String message = element[0];
502             final String expected = element[2];
503             final String original = element[1];
504             assertEquals(expected, StringEscapeUtils.unescapeHtml3(original), message);
505 
506             final StringWriter sw = new StringWriter();
507             try {
508                 StringEscapeUtils.UNESCAPE_HTML3.translate(original, sw);
509             } catch (final IOException e) {
510                 // expected
511             }
512             final String actual = original == null ? null : sw.toString();
513             assertEquals(expected, actual, message);
514         }
515         // \u00E7 is a cedilla (c with wiggle under)
516         // note that the test string must be 7-bit-clean (Unicode escaped) or else it will compile incorrectly
517         // on some locales
518         assertEquals("Fran\u00E7ais", StringEscapeUtils.unescapeHtml3("Fran\u00E7ais"), "funny chars pass through OK");
519 
520         assertEquals("Hello&;World", StringEscapeUtils.unescapeHtml3("Hello&;World"));
521         assertEquals("Hello&#;World", StringEscapeUtils.unescapeHtml3("Hello&#;World"));
522         assertEquals("Hello&# ;World", StringEscapeUtils.unescapeHtml3("Hello&# ;World"));
523         assertEquals("Hello&##;World", StringEscapeUtils.unescapeHtml3("Hello&##;World"));
524     }
525 
526     @Test
527     public void testUnescapeHtml4() {
528         for (final String[] element : HTML_ESCAPES) {
529             final String message = element[0];
530             final String expected = element[2];
531             final String original = element[1];
532             assertEquals(expected, StringEscapeUtils.unescapeHtml4(original), message);
533 
534             final StringWriter sw = new StringWriter();
535             try {
536                 StringEscapeUtils.UNESCAPE_HTML4.translate(original, sw);
537             } catch (final IOException e) {
538                 // expected
539             }
540             final String actual = original == null ? null : sw.toString();
541             assertEquals(expected, actual, message);
542         }
543         // \u00E7 is a cedilla (c with wiggle under)
544         // note that the test string must be 7-bit-clean (Unicode escaped) or else it will compile incorrectly
545         // on some locales
546         assertEquals("Fran\u00E7ais", StringEscapeUtils.unescapeHtml4("Fran\u00E7ais"), "funny chars pass through OK");
547 
548         assertEquals("Hello&;World", StringEscapeUtils.unescapeHtml4("Hello&;World"));
549         assertEquals("Hello&#;World", StringEscapeUtils.unescapeHtml4("Hello&#;World"));
550         assertEquals("Hello&# ;World", StringEscapeUtils.unescapeHtml4("Hello&# ;World"));
551         assertEquals("Hello&##;World", StringEscapeUtils.unescapeHtml4("Hello&##;World"));
552     }
553 
554     @Test
555     public void testUnescapeJava() throws IOException {
556         assertNull(StringEscapeUtils.unescapeJava(null));
557         try {
558             StringEscapeUtils.UNESCAPE_JAVA.translate(null, null);
559             fail("Exception expected!");
560         } catch (final IOException ex) {
561             fail("Exception expected!");
562         } catch (final IllegalArgumentException ex) {
563             // expected
564         }
565         try {
566             StringEscapeUtils.UNESCAPE_JAVA.translate("", null);
567             fail("Exception expected!");
568         } catch (final IOException ex) {
569             fail("Exception expected!");
570         } catch (final IllegalArgumentException ex) {
571             // expected
572         }
573         assertThrows(RuntimeException.class, () -> StringEscapeUtils.unescapeJava("\\u02-3"));
574 
575         assertUnescapeJava("", "");
576         assertUnescapeJava("test", "test");
577         assertUnescapeJava("\ntest\b", "\\ntest\\b");
578         assertUnescapeJava("\u123425foo\ntest\b", "\\u123425foo\\ntest\\b");
579         assertUnescapeJava("'\foo\teste\r", "\\'\\foo\\teste\\r");
580         assertUnescapeJava("", "\\");
581         //foo
582         assertUnescapeJava("\uABCDx", "\\uabcdx", "lowercase Unicode");
583         assertUnescapeJava("\uABCDx", "\\uABCDx", "uppercase Unicode");
584         assertUnescapeJava("\uABCD", "\\uabcd", "Unicode as final character");
585     }
586 
587     @Test
588     public void testUnescapeJson() {
589         final String jsonString =
590                 "{\"age\":100,\"name\":\"kyong.com\n\",\"messages\":[\"msg 1\",\"msg 2\",\"msg 3\"]}";
591 
592         assertEquals("", StringEscapeUtils.unescapeJson(""));
593         assertEquals(" ", StringEscapeUtils.unescapeJson(" "));
594         assertEquals("a:b", StringEscapeUtils.unescapeJson("a:b"));
595         assertEquals(jsonString, StringEscapeUtils.unescapeJson(jsonString));
596     }
597 
598     @Test // TEXT-120
599     public void testUnescapeJsonDoubleQuoteAndForwardSlash() {
600       final String escapedJsonString = "double quote: \\\" and a forward slash: \\/";
601       final String jsonString = "double quote: \" and a forward slash: /";
602 
603       assertEquals(jsonString, StringEscapeUtils.unescapeJson(escapedJsonString));
604     }
605 
606     @Test
607     public void testUnescapeUnknownEntity() {
608         assertEquals("&zzzz;", StringEscapeUtils.unescapeHtml4("&zzzz;"));
609     }
610 
611     /**
612      * Reverse of the above.
613      *
614      * @see <a href="https://issues.apache.org/jira/browse/LANG-729">LANG-729</a>
615      */
616     @Test
617     public void testUnescapeXmlSupplementaryCharacters() {
618         assertEquals("\uD84C\uDFB4", StringEscapeUtils.unescapeXml("&#144308;"),
619                 "Supplementary character must be represented using a single escape");
620 
621         assertEquals("a b c \uD84C\uDFB4", StringEscapeUtils.unescapeXml("a b c &#144308;"),
622                 "Supplementary characters mixed with basic characters should be decoded correctly");
623     }
624 
625     @Test
626     public void testUnscapeXSI() {
627         assertNull(null, unescapeXSI(null));
628         assertEquals("\"", unescapeXSI("\\\""));
629         assertEquals("He didn't say, \"Stop!\"", unescapeXSI("He\\ didn\\'t\\ say,\\ \\\"Stop!\\\""));
630         assertEquals("\\", unescapeXSI("\\\\"));
631         assertEquals("", unescapeXSI("\\"));
632     }
633 }