Implement sprintf("%c") for UTF-8.

* sprintf("%c") is changed to accept a string for which String#size returns
  1, even if it is longer than one byte, and to convert a Fixnum via
  Fixnum#chr (possibly returning more than one byte).  Thus, if the UTF-8
  gem is in use, a character will be understood as a single UTF-8 character.

* The change to sprintf depends on the implementation of Fixnum#chr added
  to mrbgems/mruby-string-utf8/src/string.c.

This should work with any other gem that implements a multibyte encoding, as
long as it implements String#size and Fixnum#chr as appropriate.

2 files changed, 47 insertions, 11 deletions

diff --git a/mrbgems/mruby-sprintf/src/sprintf.c b/mrbgems/mruby-sprintf/src/sprintf.c
index b20cbe1df..5b255d28e 100644
--- a/mrbgems/mruby-sprintf/src/sprintf.c
+++ b/mrbgems/mruby-sprintf/src/sprintf.c
@@ -666,38 +666,37 @@ retry:
       case 'c': {
         mrb_value val = GETARG();
         mrb_value tmp;
-        unsigned int c;
+        char *c;
 
         tmp = mrb_check_string_type(mrb, val);
         if (!mrb_nil_p(tmp)) {
-          if (RSTRING_LEN(tmp) != 1 ) {
+          if (mrb_fixnum(mrb_funcall(mrb, tmp, "size", 0)) != 1 ) {
             mrb_raise(mrb, E_ARGUMENT_ERROR, "%c requires a character");
           }
-          c = RSTRING_PTR(tmp)[0];
-          n = 1;
         }
-        else {
-          c = mrb_fixnum(val);
-          n = 1;
+        else if (mrb_fixnum_p(val)) {
+          tmp = mrb_funcall(mrb, val, "chr", 0);
         }
-        if (n <= 0) {
+        else {
           mrb_raise(mrb, E_ARGUMENT_ERROR, "invalid character");
         }
+        c = RSTRING_PTR(tmp);
+        n = RSTRING_LEN(tmp);
         if (!(flags & FWIDTH)) {
           CHECK(n);
-          buf[blen] = c;
+          memcpy(buf+blen, c, n);
           blen += n;
         }
         else if ((flags & FMINUS)) {
           CHECK(n);
-          buf[blen] = c;
+          memcpy(buf+blen, c, n);
           blen += n;
           FILL(' ', width-1);
         }
         else {
           FILL(' ', width-1);
           CHECK(n);
-          buf[blen] = c;
+          memcpy(buf+blen, c, n);
           blen += n;
         }
       }
diff --git a/mrbgems/mruby-string-utf8/src/string.c b/mrbgems/mruby-string-utf8/src/string.c
index 2dd848c86..4f3833944 100644
--- a/mrbgems/mruby-string-utf8/src/string.c
+++ b/mrbgems/mruby-string-utf8/src/string.c
@@ -279,6 +279,41 @@ mrb_str_reverse(mrb_state *mrb, mrb_value str)
   return mrb_str_reverse_bang(mrb, mrb_str_dup(mrb, str));
 }
 
+static mrb_value
+mrb_fixnum_chr(mrb_state *mrb, mrb_value num)
+{
+  mrb_int cp = mrb_fixnum(num);
+  char utf8[4];
+  int len;
+
+  if (cp < 0 || 0x10FFFF < cp) {
+    mrb_raisef(mrb, E_RANGE_ERROR, "%S out of char range", num);
+  }
+  if (cp < 0x80) {
+    utf8[0] = (char)cp;
+    len = 1;
+  }
+  else if (cp < 0x800) {
+    utf8[0] = (char)(0xC0 | (cp >> 6));
+    utf8[1] = (char)(0x80 | (cp & 0x3F));
+    len = 2;
+  }
+  else if (cp < 0x10000) {
+    utf8[0] = (char)(0xE0 |  (cp >> 12));
+    utf8[1] = (char)(0x80 | ((cp >>  6) & 0x3F));
+    utf8[2] = (char)(0x80 | ( cp        & 0x3F));
+    len = 3;
+  }
+  else {
+    utf8[0] = (char)(0xF0 |  (cp >> 18));
+    utf8[1] = (char)(0x80 | ((cp >> 12) & 0x3F));
+    utf8[2] = (char)(0x80 | ((cp >>  6) & 0x3F));
+    utf8[3] = (char)(0x80 | ( cp        & 0x3F));
+    len = 4;
+  }
+  return mrb_str_new(mrb, utf8, len);
+}
+
 void
 mrb_mruby_string_utf8_gem_init(mrb_state* mrb)
 {
@@ -290,6 +325,8 @@ mrb_mruby_string_utf8_gem_init(mrb_state* mrb)
   mrb_define_method(mrb, s, "slice", mrb_str_aref_m, MRB_ARGS_ANY());
   mrb_define_method(mrb, s, "reverse",  mrb_str_reverse,      MRB_ARGS_NONE());
   mrb_define_method(mrb, s, "reverse!", mrb_str_reverse_bang, MRB_ARGS_NONE());
+
+  mrb_define_method(mrb, mrb->fixnum_class, "chr", mrb_fixnum_chr, MRB_ARGS_NONE());
 }
 
 void