Added UTF8 multibyte to wide char printing in some apps (#106)

* Added UTF8 multibyte to wide char printing in some apps

* Fix test compile

* Reduce diff churn

Co-authored-by: Steve Karg <skarg@users.sourceforge.net>

apps/epics/main.c

@@ -31,6 +31,9 @@
 #include <stdio.h>
 #include <stdlib.h>
 #include <time.h> /* for time */
+#ifdef __STDC_ISO_10646__
+#include <locale.h>
+#endif
 #include <errno.h>
 #include <assert.h>
 #include "bacnet/config.h"
@@ -1438,6 +1441,17 @@ int main(int argc, char *argv[])
     address_init();
     Init_Service_Handlers();
     dlenv_init();
+#ifdef __STDC_ISO_10646__
+    /* Internationalized programs must call setlocale()
+     * to initiate a specific language operation.
+     * This can be done by calling setlocale() as follows.
+     * If your native locale doesn't use UTF-8 encoding
+     * you need to replace the empty string with a
+     * locale like "en_US.utf8" which is the same as the string
+     * used in the enviromental variable "LANG=en_US.UTF-8".
+     */
+    setlocale(LC_ALL, "");
+#endif
     atexit(datalink_cleanup);
 
     /* configure the timeout values */

apps/readprop/main.c

@@ -29,6 +29,9 @@
 #include <stdlib.h>
 #include <errno.h>
 #include <time.h> /* for time */
+#ifdef __STDC_ISO_10646__
+#include <locale.h>
+#endif
 
 #define PRINT_ENABLED 1
 
@@ -363,6 +366,17 @@ int main(int argc, char *argv[])
     Device_Set_Object_Instance_Number(BACNET_MAX_INSTANCE);
     Init_Service_Handlers();
     dlenv_init();
+#ifdef __STDC_ISO_10646__
+    /* Internationalized programs must call setlocale()
+     * to initiate a specific language operation.
+     * This can be done by calling setlocale() as follows.
+     * If your native locale doesn't use UTF-8 encoding
+     * you need to replace the empty string with a
+     * locale like "en_US.utf8" which is the same as the string
+     * used in the enviromental variable "LANG=en_US.UTF-8".
+     */
+    setlocale(LC_ALL, "");
+#endif
     atexit(datalink_cleanup);
     /* configure the timeout values */
     last_seconds = time(NULL);

apps/readpropm/main.c

@@ -29,6 +29,9 @@
 #include <stdlib.h>
 #include <errno.h>
 #include <time.h> /* for time */
+#ifdef __STDC_ISO_10646__
+#include <locale.h>
+#endif
 
 #define PRINT_ENABLED 1
 
@@ -423,6 +426,17 @@ int main(int argc, char *argv[])
     address_init();
     Init_Service_Handlers();
     dlenv_init();
+#ifdef __STDC_ISO_10646__
+    /* Internationalized programs must call setlocale()
+     * to initiate a specific language operation.
+     * This can be done by calling setlocale() as follows.
+     * If your native locale doesn't use UTF-8 encoding
+     * you need to replace the empty string with a
+     * locale like "en_US.utf8" which is the same as the string
+     * used in the enviromental variable "LANG=en_US.UTF-8".
+     */
+    setlocale(LC_ALL, "");
+#endif
     atexit(datalink_cleanup);
     /* configure the timeout values */
     last_seconds = time(NULL);

apps/readrange/main.c

@@ -31,6 +31,9 @@
 #include <stdlib.h>
 #include <errno.h>
 #include <time.h> /* for time */
+#ifdef __STDC_ISO_10646__
+#include <locale.h>
+#endif
 #include "bacnet/bacdef.h"
 #include "bacnet/config.h"
 #include "bacnet/bactext.h"
@@ -320,6 +323,17 @@ int main(int argc, char *argv[])
     address_init();
     Init_Service_Handlers();
     dlenv_init();
+#ifdef __STDC_ISO_10646__
+    /* Internationalized programs must call setlocale()
+     * to initiate a specific language operation.
+     * This can be done by calling setlocale() as follows.
+     * If your native locale doesn't use UTF-8 encoding
+     * you need to replace the empty string with a
+     * locale like "en_US.utf8" which is the same as the string
+     * used in the enviromental variable "LANG=en_US.UTF-8".
+     */
+    setlocale(LC_ALL, "");
+#endif
     atexit(datalink_cleanup);
     /* configure the timeout values */
     last_seconds = time(NULL);

src/bacnet/bacapp.c

@@ -35,8 +35,12 @@
 #include <stdbool.h>
 #include <string.h>
 #include <stdio.h>
-#include <stdlib.h>
-#include <ctype.h>
+#include <stdlib.h> /* for strtol */
+#include <ctype.h> /* for isalnum */
+#ifdef __STDC_ISO_10646__
+#include <wchar.h>
+#include <wctype.h>
+#endif
 #include "bacnet/bacenum.h"
 #include "bacnet/bacdcode.h"
 #include "bacnet/bacint.h"
@@ -1038,6 +1042,12 @@ int bacapp_snprintf_value(
     char *p_str = str;
     size_t rem_str_len = str_len;
     char temp_str[32];
+#ifdef __STDC_ISO_10646__
+    /* Wide character (decoded from multi-byte character). */
+    wchar_t wc;
+    /* Wide character length in bytes. */
+    int wclen;
+#endif
 
     if (object_value && object_value->value) {
         value = object_value->value;
@@ -1102,17 +1112,50 @@ int bacapp_snprintf_value(
                 char_str = characterstring_value(&value->type.Character_String);
                 if (!append_str(&p_str, &rem_str_len, "\"")) {
                     break;
-}
-                for (i = 0; i < len; i++) {
-                    if (isprint(*((unsigned char *)char_str))) {
-                        snprintf(temp_str, sizeof(temp_str), "%c", *char_str);
-                    } else {
-                        snprintf(temp_str, sizeof(temp_str), "%c", '.');
+                }
+                #ifdef __STDC_ISO_10646__
+                if (characterstring_encoding(&value->type.Character_String) ==
+                    CHARACTER_UTF8) {
+                    while (len > 0) {
+                        wclen = mbtowc(&wc, char_str, MB_CUR_MAX);
+                        if (wclen == -1) {
+                            /* Encoding error, reset state: */
+                            mbtowc(NULL, NULL, MB_CUR_MAX);
+                            /* After handling an invalid byte,
+                               retry with the next one. */
+                            wclen = 1;
+                            wc = L'?';
+                        } else {
+                            if (!iswprint(wc)) {
+                                wc = L'.';
+                            }
+                        }
+                        snprintf(temp_str, sizeof(temp_str), "%lc", wc);
+                        if (!append_str(&p_str, &rem_str_len, temp_str)) {
+                            break;
+                        }
+                        if (len > wclen) {
+                            len -= wclen;
+                            char_str += wclen;
+                        } else {
+                            len = 0;
+                        }
+                    }
+                } else
+                #endif
+                {
+                    for (i = 0; i < len; i++) {
+                        if (isprint(*((unsigned char *)char_str))) {
+                            snprintf(temp_str, sizeof(temp_str), "%c",
+                            *char_str);
+                        } else {
+                            snprintf(temp_str, sizeof(temp_str), "%c", '.');
+                        }
+                        if (!append_str(&p_str, &rem_str_len, temp_str)) {
+                            break;
+                        }
+                        char_str++;
                     }
-                    if (!append_str(&p_str, &rem_str_len, temp_str)) {
-                        break;
-}
-                    char_str++;
                 }
                 if ((i == len) && append_str(&p_str, &rem_str_len, "\"")) {
                     /* Everything is fine. Indicate how many bytes were */