Adapt Windows-specific code to recent changes

This included renaming the `WORD` token in our lexer to `BXWORD`,
in order to avoid a conflict with a symbol in windows.h.

Author: tsjensen

src/boxes.c

@@ -24,6 +24,9 @@
 #include <string.h>
 #include <uniconv.h>
 #include <unistd.h>
+#ifdef _WIN32
+#include <windows.h>
+#endif
 
 #include "boxes.h"
 #include "bxstring.h"
@@ -103,7 +106,8 @@ static int build_design(design_t **adesigns, const char *cld)
     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 */
+    /* We always use UTF-8, which is correct for Linux and MacOS, and for modern Windows configured for UTF-8. */
+    uint32_t *cld_u32 = u32_strconv_from_arg(cld, "UTF-8");
     bxstr_t *cldW = bxs_from_unicode(cld_u32);
     BFREE(cld_u32);
 
@@ -453,6 +457,22 @@ static int check_color_support(int opt_color)
 
 
 
+/**
+ * Switch from default "C" encoding to system encoding.
+ */
+static void activateSystemEncoding()
+{
+    #ifdef _WIN32
+        SetConsoleOutputCP(CP_ACP);
+        SetConsoleCP(CP_ACP);  
+        /* If it should one day turn out that this doesn't have the desired effect, try setlocale(LC_ALL, ".UTF8"). */
+    #else
+        setlocale(LC_ALL, "");
+    #endif
+}
+
+
+
 /*       _\|/_
          (o o)
  +----oOO-{_}-OOo------------------------------------------------------------+
@@ -470,7 +490,7 @@ int main(int argc, char *argv[])
     #endif
 
     /* Temporarily set the system encoding, for proper output of --help text etc. */
-    setlocale(LC_ALL, "");    /* switch from default "C" encoding to system encoding */
+    activateSystemEncoding();
     encoding = locale_charset();
 
     handle_command_line(argc, argv);

src/lexer.l

@@ -137,7 +137,7 @@ PFILENAME [^\r\n]+
 
 <DELIMSPEC>[^ \t\r\n]+ {
     /*
-     * String delimiter spec - like WORD, but allow any character
+     * String delimiter spec - like BXWORD, but allow any character
      */
     yylval->s = bxs_from_ascii("IGNORED");
     char *str = (char *) strdup(yytext);
@@ -415,10 +415,10 @@ PFILENAME [^\r\n]+
         exit (EXIT_FAILURE);
     }
     #ifdef LEXER_DEBUG
-        fprintf (stderr, "   WORD: %s\n", u32_strconv_to_output(utf8));
+        fprintf (stderr, " BXWORD: %s\n", u32_strconv_to_output(utf8));
     #endif
     BFREE(utf8);
-    return WORD;
+    return BXWORD;
 }
 
 

src/parser.y

@@ -127,7 +127,7 @@ typedef struct {
 %token YPARENT YSHAPES YELASTIC YPADDING YSAMPLE YENDSAMPLE YBOX YEND YUNREC
 %token YREPLACE YREVERSE YTO YWITH YCHGDEL YTAGS
 %token <ascii> KEYWORD
-%token <s> WORD
+%token <s> BXWORD
 %token <ascii> ASCII_ID
 %token <s> STRING
 %token <s> FILENAME
@@ -224,7 +224,7 @@ alias_list: alias | alias_list ',' alias;
 
 design_id: ASCII_ID | ASCII_ID ',' alias_list
 
-| WORD
+| BXWORD
     {
         yyerror(bison_args, "box design name must consist of printable standard ASCII characters.");
         YYERROR;
@@ -282,7 +282,7 @@ entry: KEYWORD STRING
 
 | YTAGS '(' tag_list ')' | YTAGS tag_entry
 
-| WORD STRING | ASCII_ID STRING
+| BXWORD STRING | ASCII_ID STRING
     {
         #ifdef PARSER_DEBUG
             fprintf (stderr, " Parser: Discarding entry [%s = %s].\n", $1, bxs_to_output($2));

src/tools.c

@@ -741,9 +741,10 @@ FILE *bx_fopen(char *pathname, char *mode)
     /*
      * On Linux/UNIX and OS X (Mac), one can access files with non-ASCII file names by passing them to fopen() as UTF-8.
      * On Windows, a different function must be called. (Info: https://stackoverflow.com/a/35065142/1005481)
+     * On newer Windows, we're good:
+     * https://learn.microsoft.com/en-us/cpp/c-runtime-library/reference/setlocale-wsetlocale#utf-8-support
      */
     FILE *f = fopen(pathname, mode);
-    // TODO Windows
     return f;
 }
 

utest/unicode_test.c

@@ -196,6 +196,7 @@ void test_u32_insert_space_at(void **state)
     u32_insert_space_at(&s, 1, 1);
     u32_insert_space_at(&s, 10000, 2);
 
+    assert_non_null(s);
     assert_int_equal(0, u32_strcmp(expected, s));
 
     BFREE(s);