Fix bugs related to UTF-8 (#604)

Author: yutaro-sakamoto

cobj/codegen.c

@@ -2452,6 +2452,7 @@ static void joutput_initialize_one(struct cb_initialize *p, cb_tree x) {
           joutput_data(x);
           joutput(".fillBytes(%d, %d);\n", buffchar, f->size);
         } else {
+#if !I18N_UTF8
           if (f->size >= 8) {
             buffchar = *(buff + f->size - 1);
             int n = 0;
@@ -2471,6 +2472,7 @@ static void joutput_initialize_one(struct cb_initialize *p, cb_tree x) {
               return;
             }
           }
+#endif
           joutput_data(x);
 #if I18N_UTF8
           joutput(".setByByteArrayAndPaddingSpaces (");

libcobj/app/src/main/java/jp/osscons/opensourcecobol/libcobj/termio/CobolTerminal.java

@@ -75,9 +75,7 @@ private static void displayAlnum(AbstractCobolField f, PrintStream stream) {
         CobolDataStorage storage = f.getDataStorage();
         if (CobolUtil.terminalEncoding == CobolEncoding.UTF8) {
             byte[] utf8Bytes =
-                    new String(
-                                    storage.getByteArrayRef(0, f.getSize()),
-                                    AbstractCobolField.charSetSJIS)
+                    new String(storage.getByteArray(0, f.getSize()), AbstractCobolField.charSetSJIS)
                             .getBytes(StandardCharsets.UTF_8);
             stream.write(utf8Bytes, 0, utf8Bytes.length);
         } else {

tests/cobol_utf8.src/pic-x.at

@@ -561,6 +561,59 @@ AT_CHECK([diff out1.txt out2.txt])
 
 AT_CLEANUP
 
+AT_SETUP([Group item COB_TERMINAL_ENCODING])
+
+AT_DATA([prog.cbl], [
+       IDENTIFICATION DIVISION.
+       PROGRAM-ID. prog.
+       DATA DIVISION.
+       WORKING-STORAGE SECTION.
+       01  WK-AREA.
+         03  WK-CD        PIC X(04).
+         03  WK-NAME      PIC X(20).
+         03  WK-DPT-CD    PIC X(02).
+         03  WK-ENT-DATE  PIC 9(08).
+         03  WK-RETURN    PIC 9(01).
+       PROCEDURE DIVISION.
+       MAIN.
+           MOVE "0012千葉  二郎          02199904010" TO WK-AREA.
+           DISPLAY "従業員番号: " WK-CD.
+           DISPLAY "名前　　　: " WK-NAME.
+           DISPLAY "部署コード: " WK-DPT-CD.
+           DISPLAY "入社日　　: " WK-ENT-DATE.
+           STOP RUN.
+])
+
+AT_CHECK([cobj prog.cbl])
+AT_CHECK([COB_TERMINAL_ENCODING=UTF-8 java prog], [0], 
+[従業員番号: 0012
+名前　　　: 千葉  二郎          @&t@
+部署コード: 02
+入社日　　: 19990401
+])
+
+AT_CLEANUP
+
+AT_SETUP([Initializing Group items])
+
+AT_DATA([prog.cbl], [
+       IDENTIFICATION   DIVISION.
+       PROGRAM-ID.      prog.
+       DATA             DIVISION.
+       WORKING-STORAGE  SECTION.
+       01 F0 PIC X(17) VALUE "千葉  二郎".
+       PROCEDURE        DIVISION.
+       DISPLAY F0.
+       STOP RUN.
+])
+
+AT_CHECK([cobj prog.cbl])
+AT_CHECK([COB_TERMINAL_ENCODING=UTF-8 java prog], [0], 
+[千葉  二郎       @&t@
+])
+
+AT_CLEANUP
+
 #AT_SETUP([Readable string literals])
 #export LC_ALL=''
 ## Older compilers converts string literals "日本語" in COBOL source code