add toupper/tolower functions (for JuliaLang/julia#11471)

Author: stevengj

.gitignore

@@ -21,3 +21,4 @@ graphemetest
 utf8proc_data.c.new
 printproperty
 charwidth
+case

Makefile

@@ -105,8 +105,12 @@ test/printproperty: test/printproperty.c utf8proc.o utf8proc.h test/tests.h
 test/charwidth: test/charwidth.c utf8proc.o utf8proc.h test/tests.h
 	$(cc) test/charwidth.c utf8proc.o -o $@
 
-check: test/normtest data/NormalizationTest.txt test/graphemetest data/GraphemeBreakTest.txt test/printproperty test/charwidth bench/bench.c bench/util.c bench/util.h utf8proc.o
+test/case: test/case.c utf8proc.o utf8proc.h test/tests.h
+	$(cc) test/case.c utf8proc.o -o $@
+
+check: test/normtest data/NormalizationTest.txt test/graphemetest data/GraphemeBreakTest.txt test/printproperty test/charwidth test/case bench/bench.c bench/util.c bench/util.h utf8proc.o
 	$(MAKE) -C bench
 	test/normtest data/NormalizationTest.txt
 	test/graphemetest data/GraphemeBreakTest.txt
 	test/charwidth
+	test/case

test/case.c

@@ -0,0 +1,50 @@
+#include "tests.h"
+#include <wctype.h>
+
+int main(int argc, char **argv)
+{
+     int error = 0, better = 0;
+     utf8proc_int32_t c;
+
+     (void) argc; /* unused */
+     (void) argv; /* unused */
+
+     /* some simple sanity tests of the character widths */
+     for (c = 0; c <= 0x110000; ++c) {
+          utf8proc_int32_t l = utf8proc_tolower(c);
+          utf8proc_int32_t u = utf8proc_toupper(c);
+
+          check(l == c || utf8proc_codepoint_valid(l), "invalid tolower");
+          check(u == c || utf8proc_codepoint_valid(u), "invalid toupper");
+
+          if (sizeof(wint_t) > 2 || c < (1<<16)) {
+               wint_t l0 = towlower(c), u0 = towupper(c);
+               
+               /* OS unicode tables may be out of date.  But if they
+                  do have a lower/uppercase mapping, hopefully it
+                  is correct? */
+               if (l0 != c && l0 != l) {
+                    fprintf(stderr, "MISMATCH %x != towlower(%x) == %x\n",
+                            l, c, l0);
+                    ++error;
+               }
+               else if (l0 != l) { /* often true for out-of-date OS unicode */
+                    ++better;
+                    /* printf("%x != towlower(%x) == %x\n", l, c, l0); */
+               }
+               if (u0 != c && u0 != u) {
+                    fprintf(stderr, "MISMATCH %x != towupper(%x) == %x\n",
+                            u, c, u0);
+                    ++error;
+               }
+               else if (u0 != u) { /* often true for out-of-date OS unicode */
+                    ++better;
+                    /* printf("%x != towupper(%x) == %x\n", u, c, u0); */
+               }
+          }
+     }
+     check(!error, "utf8proc case conversion FAILED %d tests.", error);
+     printf("More up-to-date than OS unicode tables for %d tests.\n", better);
+     printf("utf8proc case conversion tests SUCCEEDED.\n");
+     return 0;
+}

utf8proc.c

@@ -231,6 +231,18 @@ UTF8PROC_DLLEXPORT utf8proc_bool utf8proc_grapheme_break(utf8proc_int32_t c1, ut
                         utf8proc_get_property(c2)->boundclass);
 }
 
+UTF8PROC_DLLEXPORT utf8proc_int32_t utf8proc_tolower(utf8proc_int32_t c)
+{
+  utf8proc_int32_t cl = utf8proc_get_property(c)->lowercase_mapping;
+  return cl >= 0 ? cl : c;
+}
+
+UTF8PROC_DLLEXPORT utf8proc_int32_t utf8proc_toupper(utf8proc_int32_t c)
+{
+  utf8proc_int32_t cu = utf8proc_get_property(c)->uppercase_mapping;
+  return cu >= 0 ? cu : c;
+}
+
 /* return a character width analogous to wcwidth (except portable and
    hopefully less buggy than most system wcwidth functions). */
 UTF8PROC_DLLEXPORT int utf8proc_charwidth(utf8proc_int32_t c) {

utf8proc.h

@@ -506,6 +506,21 @@ UTF8PROC_DLLEXPORT utf8proc_ssize_t utf8proc_reencode(utf8proc_int32_t *buffer,
  */
 UTF8PROC_DLLEXPORT utf8proc_bool utf8proc_grapheme_break(utf8proc_int32_t codepoint1, utf8proc_int32_t codepoint2);
 
+
+/**
+ * Given a codepoint `c`, return the codepoint of the corresponding
+ * lower-case character, if any; otherwise (if there is no lower-case
+ * variant, or if `c` is not a valid codepoint) return `c`.
+ */
+UTF8PROC_DLLEXPORT utf8proc_int32_t utf8proc_tolower(utf8proc_int32_t c);
+
+/**
+ * Given a codepoint `c`, return the codepoint of the corresponding
+ * upper-case character, if any; otherwise (if there is no upper-case
+ * variant, or if `c` is not a valid codepoint) return `c`.
+ */
+UTF8PROC_DLLEXPORT utf8proc_int32_t utf8proc_toupper(utf8proc_int32_t c);
+
 /**
  * Given a codepoint, return a character width analogous to `wcwidth(codepoint)`,
  * except that a width of 0 is returned for non-printable codepoints