Use UTF-8 data read from config file #72

- shapes get additional MBCS values
- metadata is already converted for real to use bxstr_t*

src/boxes.c

@@ -84,23 +84,29 @@ static int build_design(design_t **adesigns, const char *cld)
 
     dp->name = "<Command Line Definition>";
     dp->aliases = (char **) calloc(1, sizeof(char *));
-    dp->created = "now";
+    dp->created = bxs_from_ascii("now");
     dp->revision = "1.0";
-    dp->sample = "n/a";
+    dp->sample = bxs_from_ascii("n/a");
     dp->indentmode = DEF_INDENTMODE;
     dp->padding[BLEF] = 1;
-    dp->defined_in = "(command line)";
+    dp->defined_in = bxs_from_ascii("(command line)");
 
     dp->tags = (char **) calloc(2, sizeof(char *));
     dp->tags[0] = "transient";
 
+    uint32_t *cld_u32 = u32_strconv_from_arg(cld, "UTF-8");  /* CHECK wrong on Windows (UTF-16) or different IME */
+    bxstr_t *cldW = bxs_from_unicode(cld_u32);
+    BFREE(cld_u32);
+
     dp->shape[W].height = 1;
-    dp->shape[W].width = strlen(cld);
+    dp->shape[W].width = cldW->num_columns;
     dp->shape[W].elastic = 1;
-    rc = genshape(dp->shape[W].width, dp->shape[W].height, &(dp->shape[W].chars));
+    rc = genshape(dp->shape[W].width, dp->shape[W].height, &(dp->shape[W].chars), &(dp->shape[W].mbcs));
     if (rc) {
         return rc;
     }
+    bxs_free(dp->shape[W].mbcs[0]);
+    dp->shape[W].mbcs[0] = cldW;
     strcpy(dp->shape[W].chars[0], cld);
 
     for (i = 0; i < NUM_SHAPES; ++i) {
@@ -139,7 +145,7 @@ static int build_design(design_t **adesigns, const char *cld)
                 return 1;                /* never happens ;-) */
         }
 
-        rc = genshape(c->width, c->height, &(c->chars));
+        rc = genshape(c->width, c->height, &(c->chars), &(c->mbcs));
         if (rc) {
             return rc;
         }
@@ -185,7 +191,7 @@ static void handle_command_line(int argc, char *argv[])
  */
 static void handle_config_parsing()
 {
-    char *config_file = discover_config_file(0);
+    bxstr_t *config_file = discover_config_file(0);
     if (config_file == NULL) {
         exit(EXIT_FAILURE);
     }
@@ -397,10 +403,13 @@ int main(int argc, char *argv[])
         fprintf (stderr, "BOXES STARTING ...\n");
     #endif
 
+    /* Temporarily set the system encoding, for proper output of --help text etc. */
+    setlocale(LC_ALL, "");    /* switch from default "C" encoding to system encoding */
+    encoding = locale_charset();
+
     handle_command_line(argc, argv);
 
     /* Store system character encoding */
-    setlocale(LC_ALL, "");    /* switch from default "C" encoding to system encoding */
     encoding = check_encoding(opt.encoding, locale_charset());
     #ifdef DEBUG
         fprintf (stderr, "Character Encoding = %s\n", encoding);

src/boxes.in.h

@@ -26,9 +26,10 @@
 /* #define LEXER_DEBUG 1 */
 /* #define DISCOVERY_DEBUG 1 */
 
-
 #include <stdio.h>
 #include <unitypes.h>
+
+#include "bxstring.h"
 #include "regulex.h"
 #include "shape.h"
 
@@ -56,6 +57,9 @@
 #define LINE_MAX_BYTES        16382
 #endif
 
+/* Macro to declare a function parameter as intentionally unused in order to avoid compiler warnings */
+#define UNUSED(variable) ((void)(variable))
+
 
 #define BTOP 0                       /* for use with sides */
 #define BRIG 1
@@ -64,8 +68,8 @@
 
 
 typedef struct {
-    char       *search;
-    char       *repstr;
+    bxstr_t    *search;
+    bxstr_t    *repstr;
     pcre2_code *prog;                /* compiled search pattern */
     int         line;                /* line of definition in config file */
     char        mode;                /* 'g' or 'o' */
@@ -75,20 +79,20 @@ typedef struct {
 typedef struct {
     char      *name;                 /* primary name of the box design */
     char     **aliases;              /* zero-terminated array of alias names of the design */
-    char      *author;               /* creator of the configuration file entry */
-    char      *designer;             /* creator of the original ASCII artwork */
-    char      *created;              /* date created, free format */
+    bxstr_t   *author;               /* creator of the configuration file entry */
+    bxstr_t   *designer;             /* creator of the original ASCII artwork */
+    bxstr_t   *created;              /* date created, free format */
     char      *revision;             /* revision number of design */
-    char      *revdate;              /* date of current revision */
-    char      *sample;               /* the complete sample block in one string */
+    bxstr_t   *revdate;              /* date of current revision */
+    bxstr_t   *sample;               /* the complete sample block in one string */
     char       indentmode;           /* 'b', 't', or 'n' */
     sentry_t   shape[NUM_SHAPES];
     size_t     maxshapeheight;       /* height of highest shape in design */
     size_t     minwidth;
     size_t     minheight;
     int        padding[NUM_SIDES];
     char     **tags;
-    char      *defined_in;           /* path to config file where this was defined */
+    bxstr_t   *defined_in;           /* path to config file where this was defined */
 
     reprule_t *current_rule;
     reprule_t *reprules;             /* applied when drawing a box */

src/cmdline.c

@@ -76,7 +76,7 @@ static void usage_short(FILE *st)
  */
 void usage_long(FILE *st)
 {
-    char *config_file = discover_config_file(0);
+    bxstr_t *config_file = discover_config_file(0);
 
     fprintf(st, "%s - draws any kind of box around your text (or removes it)\n", PROJECT);
     fprintf(st, "        Website: https://boxes.thomasjensen.com/\n");
@@ -86,7 +86,8 @@ void usage_long(FILE *st)
     fprintf(st, "        -d name  box design [default: first one in file]\n");
     fprintf(st, "        -e eol   Override line break type (experimental) [default: %s]\n",
                                   strcmp(EOL_DEFAULT, "\r\n") == 0 ? "CRLF" : "LF");
-    fprintf(st, "        -f file  configuration file [default: %s]\n", config_file != NULL ? config_file : "none");
+    fprintf(st, "        -f file  configuration file [default: %s]\n",
+                                  config_file != NULL ? bxs_to_output(config_file) : "none");
     fprintf(st, "        -h       print usage information\n");
     fprintf(st, "        -i mode  indentation mode [default: box]\n");
     fprintf(st, "        -k bool  leading/trailing blank line retention on removal\n");
@@ -100,7 +101,7 @@ void usage_long(FILE *st)
     fprintf(st, "        -t str   tab stop distance and expansion [default: %de]\n", DEF_TABSTOP);
     fprintf(st, "        -v       print version information\n");
 
-    BFREE(config_file);
+    bxs_free(config_file);
 }