Handle Japanese characters in file identifiers correctly (#540)

Author: yutaro-sakamoto

cobj/tree.c

@@ -124,8 +124,13 @@ static char *to_cname(const char *s) {
   unsigned char *p;
 
   copy = strdup(s);
-  for (p = (unsigned char *)copy; *p; p++) {
-    *p = (*p == '-') ? '_' : (unsigned char)toupper(*p);
+  for (p = (unsigned char *)copy; *p;) {
+    if ((0x81 <= *p && *p <= 0x9F) || (0xE0 <= *p && *p <= 0xFC)) {
+      p += 2;
+    } else {
+      *p = (*p == '-') ? '_' : (unsigned char)toupper(*p);
+      p++;
+    }
   }
   return copy;
 }

tests/Makefile.am

@@ -219,7 +219,8 @@ misc_DEPENDENCIES = \
 	misc.src/read_prev_after_start.at \
 	misc.src/japanese-char-section-var.at \
 	misc.src/evaluate-switch.at \
-	misc.src/fserial-variable.at
+	misc.src/fserial-variable.at \
+	misc.src/file-handler-japanese.at
 
 EXTRA_DIST = $(srcdir)/package.m4 \
 	$(TESTS) \

tests/Makefile.in

@@ -758,7 +758,8 @@ misc_DEPENDENCIES = \
 	misc.src/read_prev_after_start.at \
 	misc.src/japanese-char-section-var.at \
 	misc.src/evaluate-switch.at \
-	misc.src/fserial-variable.at
+	misc.src/fserial-variable.at \
+	misc.src/file-handler-japanese.at
 
 EXTRA_DIST = $(srcdir)/package.m4 \
 	$(TESTS) \

tests/misc.at

@@ -49,3 +49,4 @@ m4_include([read_prev_after_start.at])
 m4_include([japanese-char-section-var.at])
 m4_include([evaluate-switch.at])
 m4_include([fserial-variable.at])
+m4_include([file-handler-japanese.at])

tests/misc.src/file-handler-japanese.at

@@ -0,0 +1,25 @@
+AT_SETUP([file-handler-japanese])
+
+# This test verifies that Issue 539 is fixed.
+
+AT_DATA([prog.cbl], [
+       IDENTIFICATION   DIVISION.
+       PROGRAM-ID.      prog.
+       ENVIRONMENT      DIVISION.
+       INPUT-OUTPUT     SECTION.
+       FILE-CONTROL.
+       SELECT �e�X�g�t�@�C��
+           ORGANIZATION IS SEQUENTIAL.
+       DATA             DIVISION.
+       FILE             SECTION.
+       FD �e�X�g�t�@�C��.
+       01 REC      PIC X(10).
+       PROCEDURE        DIVISION.
+           STOP RUN.
+])
+
+AT_CHECK([${COBJ} prog.cbl])
+AT_CHECK([grep 'h_�e�X�g�t�@�C��' prog.java > /dev/null])
+AT_CHECK([grep 'h_�E�X�G�T�@�C��' prog.java], [1])
+
+AT_CLEANUP