feat(objectionary#801): fix the bug with unicode characters in strings

volodya-lombrozo · Oct 26, 2024 · a9eceed · a9eceed
1 parent edf791d
commit a9eceed
Show file tree

Hide file tree

Showing 4 changed files with 30 additions and 9 deletions.
diff --git a/src/it/phi-unphi/verify.groovy b/src/it/phi-unphi/verify.groovy
@@ -26,7 +26,7 @@ import java.nio.file.Files
 String log = new File(basedir, 'build.log').text;
 assert log.contains("BUILD SUCCESS"): assertionMessage("BUILD FAILED")
 assert log.contains("sin(42.000000) = -0.916522"): assertionMessage("sin(42.000000) = -0.916522 not found")
-assert log.contains("We have the field with the unicaode character 'Φ'"): assertionMessage("We can't find the field with the unicode character 'Φ'")
+assert log.contains("We have the field with the unicode character 'Φ'"): assertionMessage("We can't find the field with the unicode character 'Φ'")
 
 private String assertionMessage(String message) {
     generateGitHubIssue()

diff --git a/src/main/java/org/eolang/jeo/representation/bytecode/DataType.java b/src/main/java/org/eolang/jeo/representation/bytecode/DataType.java
@@ -123,7 +123,9 @@ enum DataType {
      * String.
      */
     STRING("string", String.class,
-        value -> Optional.ofNullable(value).map(String::valueOf).map(String::getBytes).orElse(null),
+        value -> Optional.ofNullable(value).map(String::valueOf)
+            .map(unicode -> unicode.getBytes(StandardCharsets.UTF_8))
+            .orElse(null),
         bytes -> new String(bytes, StandardCharsets.UTF_8)
     ),
 

diff --git a/src/main/java/org/eolang/jeo/representation/xmir/XmlValue.java b/src/main/java/org/eolang/jeo/representation/xmir/XmlValue.java
@@ -23,6 +23,7 @@
  */
 package org.eolang.jeo.representation.xmir;
 
+import java.nio.charset.StandardCharsets;
 import java.util.Arrays;
 import java.util.regex.Pattern;
 import java.util.stream.Collectors;
@@ -70,10 +71,17 @@ public String string() {
             if (hex.isEmpty()) {
                 result = "";
             } else {
-                result = Arrays.stream(hex.split("(?<=\\G.{2})"))
-                    .map(ch -> (char) Integer.parseInt(ch, XmlValue.RADIX))
-                    .map(String::valueOf)
-                    .collect(Collectors.joining());
+                final String[] split = hex.split("(?<=\\G.{2})");
+                byte[] bytes = new byte[split.length];
+                for (int i = 0; i < split.length; i++) {
+                    bytes[i] = (byte) Integer.parseInt(split[i], XmlValue.RADIX);
+                }
+                result = new String(bytes, StandardCharsets.UTF_8);
+
+//                result = Arrays.stream(hex.split("(?<=\\G.{2})"))
+//                    .map(ch -> (char) Integer.parseInt(ch, XmlValue.RADIX))
+//                    .map(String::valueOf)
+//                    .collect(Collectors.joining());
             }
             return result;
         } catch (final NumberFormatException exception) {

diff --git a/src/test/java/org/eolang/jeo/representation/xmir/XmlValueTest.java b/src/test/java/org/eolang/jeo/representation/xmir/XmlValueTest.java
@@ -23,6 +23,7 @@
  */
 package org.eolang.jeo.representation.xmir;
 
+import java.nio.charset.StandardCharsets;
 import java.util.stream.Stream;
 import org.eolang.jeo.representation.directives.DirectivesValue;
 import org.hamcrest.MatcherAssert;
@@ -75,19 +76,29 @@ void decodesEncodesCorrectly(final Object origin) {
         "Φ", "Ψ", "Ω", "Ϊ", "Ϋ", "ά", "έ", "ή", "ί", "ΰ", "α", "β", "γ", "δ", "ε", "ζ", "η", "θ", "ι", "κ", "λ", "μ",
     })
     void decodesUnicodeCharacters(final String unicode) throws ImpossibleModificationException {
+        final String xml = new Xembler(
+            new DirectivesValue(unicode)
+        ).xml();
+        System.out.println(xml);
         MatcherAssert.assertThat(
             "Can't decode unicode characters",
             new XmlValue(
                 new XmlNode(
-                    new Xembler(
-                        new DirectivesValue(unicode)
-                    ).xml()
+                    xml
                 )
             ).string(),
             Matchers.equalTo(unicode)
         );
     }
 
+    @Test
+    void test() {
+        System.out.println(
+            new String(new byte[]{(byte) 0xCE, (byte) 0xA6}, StandardCharsets.UTF_8));
+//        CEA6
+//        new String({ce, a6});
+    }
+
     /**
      * Arguments for {@link XmlValue#decodesEncodesCorrectly(Object, String)}.
      * @return Stream of arguments.