Phase 4

2026-08-28 13:26:40 -04:00
parent a44ea00d3a
commit f310d39fc1
38 changed files with 2765 additions and 96 deletions
@@ -0,0 +1,263 @@
+package haus.nightmare.lib3270j.charset;
+
+import org.junit.jupiter.api.Test;
+import static org.junit.jupiter.api.Assertions.*;
+
+/**
+ * Unit tests for Single-Byte Character Set (SBCS) and Euro EBCDIC codepages.
+ */
+public class CodePageTest {
+
+    @Test
+    public void testCp037USCanada() {
+        CodePage cp = CodePageRegistry.getCodePage("037");
+        assertEquals("037", cp.getCodePageId());
+        assertEquals(37, cp.getCpgid());
+        assertEquals(697, cp.getCgcsgid());
+        assertFalse(cp.isDBCS());
+
+        // Basic ASCII roundtrip
+        assertEquals('A', cp.ebcdicToUnicode(0xC1));
+        assertEquals(0xC1, cp.unicodeToEbcdic('A'));
+        assertEquals('Z', cp.ebcdicToUnicode(0xE9));
+        assertEquals('0', cp.ebcdicToUnicode(0xF0));
+        assertEquals(' ', cp.ebcdicToUnicode(0x40));
+
+        // CP037 special punctuation
+        assertEquals('[', cp.ebcdicToUnicode(0xBA));
+        assertEquals(']', cp.ebcdicToUnicode(0xBB));
+        assertEquals('{', cp.ebcdicToUnicode(0xC0));
+        assertEquals('}', cp.ebcdicToUnicode(0xD0));
+        assertEquals('\\', cp.ebcdicToUnicode(0xE0));
+        assertEquals('~', cp.ebcdicToUnicode(0xA1));
+        assertEquals('^', cp.ebcdicToUnicode(0xB0));
+    }
+
+    @Test
+    public void testCp1047OpenSystems() {
+        CodePage cp = CodePageRegistry.getCodePage("1047");
+        assertEquals("1047", cp.getCodePageId());
+        assertEquals(1047, cp.getCpgid());
+        assertEquals(103, cp.getCgcsgid());
+
+        // CP1047 bracket differences
+        assertEquals('[', cp.ebcdicToUnicode(0xAD));
+        assertEquals(']', cp.ebcdicToUnicode(0xBD));
+        assertEquals(0xAD, cp.unicodeToEbcdic('['));
+        assertEquals(0xBD, cp.unicodeToEbcdic(']'));
+        assertEquals('^', cp.ebcdicToUnicode(0x5F));
+        assertEquals(0x5F, cp.unicodeToEbcdic('^'));
+        assertEquals('¬', cp.ebcdicToUnicode(0xB0));
+    }
+
+    @Test
+    public void testCp500International() {
+        CodePage cp = CodePageRegistry.getCodePage("500");
+        assertEquals("500", cp.getCodePageId());
+        assertEquals(500, cp.getCpgid());
+        assertEquals(697, cp.getCgcsgid());
+
+        // CP500 bracket & exclamation mappings
+        assertEquals('[', cp.ebcdicToUnicode(0x4A));
+        assertEquals(']', cp.ebcdicToUnicode(0x5A));
+        assertEquals(0x4A, cp.unicodeToEbcdic('['));
+        assertEquals(0x5A, cp.unicodeToEbcdic(']'));
+        assertEquals('!', cp.ebcdicToUnicode(0x4F));
+        assertEquals('$', cp.ebcdicToUnicode(0x5B));
+    }
+
+    @Test
+    public void testCp273GermanyAustria() {
+        CodePage cp = CodePageRegistry.getCodePage("273");
+        assertEquals("273", cp.getCodePageId());
+        assertEquals(273, cp.getCpgid());
+
+        // German umlauts and eszett roundtrip
+        int ebcAe = cp.unicodeToEbcdic('ä');
+        assertTrue(ebcAe >= 0);
+        assertEquals('ä', cp.ebcdicToUnicode(ebcAe));
+
+        int ebcOe = cp.unicodeToEbcdic('ö');
+        assertTrue(ebcOe >= 0);
+        assertEquals('ö', cp.ebcdicToUnicode(ebcOe));
+
+        int ebcUe = cp.unicodeToEbcdic('ü');
+        assertTrue(ebcUe >= 0);
+        assertEquals('ü', cp.ebcdicToUnicode(ebcUe));
+
+        int ebcSs = cp.unicodeToEbcdic('ß');
+        assertTrue(ebcSs >= 0);
+        assertEquals('ß', cp.ebcdicToUnicode(ebcSs));
+    }
+
+    @Test
+    public void testCp277DenmarkNorway() {
+        CodePage cp = CodePageRegistry.getCodePage("277");
+        assertEquals("277", cp.getCodePageId());
+
+        int ebcAe = cp.unicodeToEbcdic('æ');
+        assertTrue(ebcAe >= 0);
+        assertEquals('æ', cp.ebcdicToUnicode(ebcAe));
+
+        int ebcOe = cp.unicodeToEbcdic('ø');
+        assertTrue(ebcOe >= 0);
+        assertEquals('ø', cp.ebcdicToUnicode(ebcOe));
+
+        int ebcAa = cp.unicodeToEbcdic('å');
+        assertTrue(ebcAa >= 0);
+        assertEquals('å', cp.ebcdicToUnicode(ebcAa));
+    }
+
+    @Test
+    public void testCp278SwedenFinland() {
+        CodePage cp = CodePageRegistry.getCodePage("278");
+        assertEquals("278", cp.getCodePageId());
+
+        int ebcAe = cp.unicodeToEbcdic('ä');
+        assertTrue(ebcAe >= 0);
+        assertEquals('ä', cp.ebcdicToUnicode(ebcAe));
+
+        int ebcOe = cp.unicodeToEbcdic('ö');
+        assertTrue(ebcOe >= 0);
+        assertEquals('ö', cp.ebcdicToUnicode(ebcOe));
+
+        int ebcAa = cp.unicodeToEbcdic('å');
+        assertTrue(ebcAa >= 0);
+        assertEquals('å', cp.ebcdicToUnicode(ebcAa));
+    }
+
+    @Test
+    public void testCp280Italy() {
+        CodePage cp = CodePageRegistry.getCodePage("280");
+        assertEquals("280", cp.getCodePageId());
+
+        int ebcA = cp.unicodeToEbcdic('à');
+        assertTrue(ebcA >= 0);
+        assertEquals('à', cp.ebcdicToUnicode(ebcA));
+
+        int ebcE = cp.unicodeToEbcdic('é');
+        assertTrue(ebcE >= 0);
+        assertEquals('é', cp.ebcdicToUnicode(ebcE));
+    }
+
+    @Test
+    public void testCp284SpainLatinAmerica() {
+        CodePage cp = CodePageRegistry.getCodePage("284");
+        assertEquals("284", cp.getCodePageId());
+
+        int ebcN = cp.unicodeToEbcdic('ñ');
+        assertTrue(ebcN >= 0);
+        assertEquals('ñ', cp.ebcdicToUnicode(ebcN));
+
+        int ebcNu = cp.unicodeToEbcdic('Ñ');
+        assertTrue(ebcNu >= 0);
+        assertEquals('Ñ', cp.ebcdicToUnicode(ebcNu));
+    }
+
+    @Test
+    public void testCp285UK() {
+        CodePage cp = CodePageRegistry.getCodePage("285");
+        assertEquals("285", cp.getCodePageId());
+
+        int ebcPound = cp.unicodeToEbcdic('£');
+        assertTrue(ebcPound >= 0);
+        assertEquals('£', cp.ebcdicToUnicode(ebcPound));
+    }
+
+    @Test
+    public void testCp297France() {
+        CodePage cp = CodePageRegistry.getCodePage("297");
+        assertEquals("297", cp.getCodePageId());
+
+        int ebcA = cp.unicodeToEbcdic('à');
+        assertTrue(ebcA >= 0);
+        assertEquals('à', cp.ebcdicToUnicode(ebcA));
+
+        int ebcE = cp.unicodeToEbcdic('é');
+        assertTrue(ebcE >= 0);
+        assertEquals('é', cp.ebcdicToUnicode(ebcE));
+    }
+
+    @Test
+    public void testCp870Latin2() {
+        CodePage cp = CodePageRegistry.getCodePage("870");
+        assertEquals("870", cp.getCodePageId());
+        assertEquals(870, cp.getCpgid());
+        assertEquals(959, cp.getCgcsgid());
+
+        int ebcS = cp.unicodeToEbcdic('š');
+        assertTrue(ebcS >= 0);
+        assertEquals('š', cp.ebcdicToUnicode(ebcS));
+    }
+
+    @Test
+    public void testCp871Iceland() {
+        CodePage cp = CodePageRegistry.getCodePage("871");
+        assertEquals("871", cp.getCodePageId());
+
+        int ebcThorn = cp.unicodeToEbcdic('þ');
+        assertTrue(ebcThorn >= 0);
+        assertEquals('þ', cp.ebcdicToUnicode(ebcThorn));
+
+        int ebcEth = cp.unicodeToEbcdic('ð');
+        assertTrue(ebcEth >= 0);
+        assertEquals('ð', cp.ebcdicToUnicode(ebcEth));
+    }
+
+    @Test
+    public void testCp875Greece() {
+        CodePage cp = CodePageRegistry.getCodePage("875");
+        assertEquals("875", cp.getCodePageId());
+
+        int ebcAlpha = cp.unicodeToEbcdic('α');
+        assertTrue(ebcAlpha >= 0);
+        assertEquals('α', cp.ebcdicToUnicode(ebcAlpha));
+
+        int ebcOmega = cp.unicodeToEbcdic('Ω');
+        assertTrue(ebcOmega >= 0);
+        assertEquals('Ω', cp.ebcdicToUnicode(ebcOmega));
+    }
+
+    @Test
+    public void testCp1026Turkey() {
+        CodePage cp = CodePageRegistry.getCodePage("1026");
+        assertEquals("1026", cp.getCodePageId());
+
+        int ebcS = cp.unicodeToEbcdic('ş');
+        assertTrue(ebcS >= 0);
+        assertEquals('ş', cp.ebcdicToUnicode(ebcS));
+    }
+
+    @Test
+    public void testEuroVariantsCp1140To1149() {
+        CodePage cp1140 = CodePageRegistry.getCodePage("1140");
+        assertEquals(1140, cp1140.getCpgid());
+        assertEquals(695, cp1140.getCgcsgid());
+        assertEquals(0x9F, cp1140.unicodeToEbcdic('€'));
+        assertEquals('€', cp1140.ebcdicToUnicode(0x9F));
+
+        CodePage cp1141 = CodePageRegistry.getCodePage("1141");
+        assertEquals(0x9F, cp1141.unicodeToEbcdic('€'));
+        assertEquals('€', cp1141.ebcdicToUnicode(0x9F));
+
+        CodePage cp1148 = CodePageRegistry.getCodePage("1148");
+        assertEquals(1148, cp1148.getCpgid());
+        assertEquals(0x9F, cp1148.unicodeToEbcdic('€'));
+        assertEquals('€', cp1148.ebcdicToUnicode(0x9F));
+    }
+
+    @Test
+    public void testStringRoundTrips() {
+        CodePage cp273 = CodePageRegistry.getCodePage("273");
+        String german = "Grüße aus München: Äpfel & Öfen!";
+        byte[] ebcdic = cp273.stringToEbcdic(german);
+        String restored = cp273.ebcdicToString(ebcdic, 0, ebcdic.length);
+        assertEquals(german, restored);
+
+        CodePage cp284 = CodePageRegistry.getCodePage("284");
+        String spanish = "España y México: ¡Hola año nuevo!";
+        byte[] espEbcdic = cp284.stringToEbcdic(spanish);
+        String espRestored = cp284.ebcdicToString(espEbcdic, 0, espEbcdic.length);
+        assertEquals(spanish, espRestored);
+    }
+}
@@ -0,0 +1,98 @@
+package haus.nightmare.lib3270j.charset;
+
+import haus.nightmare.lib3270j.TerminalModel;
+import haus.nightmare.lib3270j.input.InputProcessor;
+import haus.nightmare.lib3270j.screen.ScreenBuffer;
+import org.junit.jupiter.api.Test;
+import static org.junit.jupiter.api.Assertions.*;
+
+/**
+ * Unit tests for Mixed Double-Byte Character Set (DBCS) codepages and SO/SI state processing.
+ */
+public class DBCSTranslationTest {
+
+    @Test
+    public void testDBCSProperties() {
+        CodePage cp930 = CodePageRegistry.getCodePage("930");
+        assertTrue(cp930.isDBCS());
+        assertEquals("930", cp930.getCodePageId());
+        assertEquals(930, cp930.getCpgid());
+        assertEquals(1172, cp930.getCgcsgid());
+
+        CodePage cp935 = CodePageRegistry.getCodePage("935");
+        assertTrue(cp935.isDBCS());
+        assertEquals(935, cp935.getCpgid());
+
+        CodePage cp937 = CodePageRegistry.getCodePage("937");
+        assertTrue(cp937.isDBCS());
+        assertEquals(937, cp937.getCpgid());
+
+        CodePage cp933 = CodePageRegistry.getCodePage("933");
+        assertTrue(cp933.isDBCS());
+        assertEquals(933, cp933.getCpgid());
+    }
+
+    @Test
+    public void testIdeographicSpace() {
+        CodePage cp930 = CodePageRegistry.getCodePage("930");
+        assertEquals('\u3000', cp930.dbcsToUnicode(0x40, 0x40));
+        assertEquals(0x4040, cp930.unicodeToDbcs('\u3000'));
+    }
+
+    @Test
+    public void testDBCSStreamSOSIProcessing() {
+        AbstractDBCSCodePage cp = (AbstractDBCSCodePage) CodePageRegistry.getCodePage("930");
+        cp.registerDbcsPair(0x4341, '\u6771'); // '東'
+        cp.registerDbcsPair(0x4342, '\u4EAC'); // '京'
+
+        // Mixed byte sequence: "IBM" + SO + 東 + 京 + SI + "123"
+        // 'I'=0xC9, 'B'=0xC2, 'M'=0xD4, SO=0x0E, 0x43, 0x41, 0x43, 0x42, SI=0x0F, '1'=0xF1, '2'=0xF2, '3'=0xF3
+        byte[] mixed = new byte[] {
+            (byte) 0xC9, (byte) 0xC2, (byte) 0xD4,
+            0x0E, (byte) 0x43, (byte) 0x41, (byte) 0x43, (byte) 0x42, 0x0F,
+            (byte) 0xF1, (byte) 0xF2, (byte) 0xF3
+        };
+
+        String translated = cp.ebcdicToString(mixed, 0, mixed.length);
+        assertEquals("IBM東京123", translated);
+
+        // Reverse conversion: stringToEbcdic
+        byte[] reEncoded = cp.stringToEbcdic("IBM東京123");
+        assertArrayEquals(mixed, reEncoded);
+    }
+
+    @Test
+    public void testInputProcessorDBCSInput() {
+        EbcdicTranslator translator = new EbcdicTranslator("930");
+        AbstractDBCSCodePage cp = (AbstractDBCSCodePage) translator.getCodePage();
+        cp.registerDbcsPair(0x4545, '\u6771');
+
+        ScreenBuffer screen = new ScreenBuffer(TerminalModel.IBM_3279_2, translator);
+        InputProcessor input = new InputProcessor(screen, translator, null);
+
+        // Enter DBCS character at pos 0
+        boolean ok = input.DBCSinputChar('\u6771', 0);
+        assertTrue(ok);
+        assertEquals(0x45, screen.getCell(0).ec & 0xFF);
+        assertEquals(0x45, screen.getCell(1).ec & 0xFF);
+        assertEquals('\u6771', (char) screen.getCell(0).ucs4);
+        assertEquals(2, screen.getCursorAddress());
+    }
+
+    @Test
+    public void testClearUnusedSISO() {
+        EbcdicTranslator translator = new EbcdicTranslator("930");
+        ScreenBuffer screen = new ScreenBuffer(TerminalModel.IBM_3279_2, translator);
+        InputProcessor input = new InputProcessor(screen, translator, null);
+
+        // Place orphaned SO (0x0E) followed immediately by SI (0x0F)
+        screen.setCell(5, 0x0E);
+        screen.setCell(6, 0x0F);
+        assertEquals(0x0E, screen.getCell(5).ec & 0xFF);
+        assertEquals(0x0F, screen.getCell(6).ec & 0xFF);
+
+        input.clearUnusedSISO(0);
+        assertEquals(0, screen.getCell(5).ec & 0xFF);
+        assertEquals(0, screen.getCell(6).ec & 0xFF);
+    }
+}
@@ -0,0 +1,133 @@
+package haus.nightmare.lib3270j.charset;
+
+import haus.nightmare.lib3270j.ConnectionConfig;
+import haus.nightmare.lib3270j.Telnet3270Client;
+import haus.nightmare.lib3270j.TerminalModel;
+import haus.nightmare.lib3270j.datastream.QueryReplyBuilder;
+import haus.nightmare.lib3270j.screen.ScreenBuffer;
+import org.junit.jupiter.api.Test;
+import static org.junit.jupiter.api.Assertions.*;
+
+/**
+ * High-level integration tests for Phase 4 Character Translation & Multi-Codepage subsystem.
+ */
+public class EbcdicTranslatorPhase4Test {
+
+    @Test
+    public void testRegistryAliases() {
+        assertEquals("037", CodePageRegistry.getCodePage("cp037").getCodePageId());
+        assertEquals("037", CodePageRegistry.getCodePage("IBM-037").getCodePageId());
+        assertEquals("037", CodePageRegistry.getCodePage("ebcdic-cp-us").getCodePageId());
+
+        assertEquals("1047", CodePageRegistry.getCodePage("posix").getCodePageId());
+        assertEquals("1047", CodePageRegistry.getCodePage("open-systems").getCodePageId());
+        assertEquals("1047", CodePageRegistry.getCodePage("IBM1047").getCodePageId());
+
+        assertEquals("273", CodePageRegistry.getCodePage("de").getCodePageId());
+        assertEquals("273", CodePageRegistry.getCodePage("germany").getCodePageId());
+
+        assertEquals("285", CodePageRegistry.getCodePage("uk").getCodePageId());
+        assertEquals("285", CodePageRegistry.getCodePage("gb").getCodePageId());
+
+        assertEquals("297", CodePageRegistry.getCodePage("fr").getCodePageId());
+        assertEquals("297", CodePageRegistry.getCodePage("france").getCodePageId());
+
+        assertEquals("930", CodePageRegistry.getCodePage("ja").getCodePageId());
+        assertEquals("935", CodePageRegistry.getCodePage("chinese-simplified").getCodePageId());
+    }
+
+    @Test
+    public void testDynamicCodePageSwitching() {
+        EbcdicTranslator translator = new EbcdicTranslator("037");
+        assertEquals("037", translator.getCodePageId());
+        assertEquals(37, translator.getCpgid());
+        assertEquals('[', translator.ebcdicToUnicode(0xBA));
+
+        // Switch to CP1047
+        translator.setCodePage("1047");
+        assertEquals("1047", translator.getCodePageId());
+        assertEquals(1047, translator.getCpgid());
+        assertEquals('[', translator.ebcdicToUnicode(0xAD));
+
+        // Switch to German CP273
+        translator.setCodePage("273");
+        assertEquals("273", translator.getCodePageId());
+        assertEquals(273, translator.getCpgid());
+        int ebcAe = translator.unicodeToEbcdic('ä');
+        assertTrue(ebcAe >= 0);
+        assertEquals('ä', translator.ebcdicToUnicode(ebcAe));
+
+        // Switch to Euro CP1140
+        translator.setCodePage("1140");
+        assertEquals("1140", translator.getCodePageId());
+        assertEquals(1140, translator.getCpgid());
+        assertEquals('€', translator.ebcdicToUnicode(0x9F));
+    }
+
+    @Test
+    public void testQueryReplyBuilderCharsetsSynchronization() {
+        ScreenBuffer screen = new ScreenBuffer(TerminalModel.IBM_3279_2, new EbcdicTranslator("273"));
+        QueryReplyBuilder builder = new QueryReplyBuilder(screen);
+
+        byte[] qrs = builder.buildAllQueryReplies(80, 24, 1920);
+        assertNotNull(qrs);
+        assertTrue(qrs.length > 0);
+
+        // Scan for QR_CHARSETS (0x85)
+        int idx = 1; // skip AID_SF
+        boolean foundCharsets = false;
+        while (idx < qrs.length) {
+            int len = ((qrs[idx] & 0xFF) << 8) | (qrs[idx + 1] & 0xFF);
+            int sfid = qrs[idx + 2] & 0xFF;
+            int qrCode = qrs[idx + 3] & 0xFF;
+
+            if (qrCode == 0x85) { // QR_CHARSETS
+                foundCharsets = true;
+                // Check Set 0 Descriptor (CPGID 273 = 0x0111)
+                // Descriptor 1 starts after 9-byte header
+                int descOffset = idx + 4 + 9;
+                int cpgidHigh = qrs[descOffset + 5] & 0xFF;
+                int cpgidLow = qrs[descOffset + 6] & 0xFF;
+                int cpgid = (cpgidHigh << 8) | cpgidLow;
+                assertEquals(273, cpgid, "Query Reply Charsets should advertise active CPGID 273");
+                break;
+            }
+            idx += len;
+        }
+        assertTrue(foundCharsets, "QR_CHARSETS must be present in Query Replies");
+    }
+
+    @Test
+    public void testScreenBufferTranslateWithActiveCodePage() {
+        EbcdicTranslator translator = new EbcdicTranslator("273");
+        ScreenBuffer screen = new ScreenBuffer(TerminalModel.IBM_3279_2, translator);
+
+        // Put German characters in buffer
+        screen.setCell(0, translator.unicodeToEbcdic('ä'));
+        screen.setCell(1, translator.unicodeToEbcdic('ö'));
+        screen.setCell(2, translator.unicodeToEbcdic('ü'));
+        screen.setCell(3, translator.unicodeToEbcdic('ß'));
+
+        screen.translateToUnicode();
+
+        assertEquals('ä', (char) screen.getCell(0).ucs4);
+        assertEquals('ö', (char) screen.getCell(1).ucs4);
+        assertEquals('ü', (char) screen.getCell(2).ucs4);
+        assertEquals('ß', (char) screen.getCell(3).ucs4);
+    }
+
+    @Test
+    public void testTelnet3270ClientWithConfiguredCodePage() {
+        ConnectionConfig config = new ConnectionConfig("mainframe.example.com", 23);
+        config.setCodePage("500");
+
+        Telnet3270Client client = new Telnet3270Client(config);
+        assertEquals("500", client.getCodePage());
+        assertEquals(500, client.getTranslator().getCpgid());
+
+        // Test changing at runtime
+        client.setCodePage("1047");
+        assertEquals("1047", client.getCodePage());
+        assertEquals(1047, client.getTranslator().getCpgid());
+    }
+}