#include <errno.h>
#include <ctype.h>
#include <assert.h>
+#include <sys/types.h>
+#include <sys/stat.h>
#include "utf8.h"
#include "mem.h"
#include "hex.h"
#include "words.h"
#include "heap.h"
+#include "unicode.h"
+#include "inputline.h"
+#include "wstat.h"
static int tests, errors;
return d.vec;
}
+static const char *format_utf32(const uint32_t *s) {
+ struct dynstr d;
+ uint32_t c;
+ char buf[64];
+
+ dynstr_init(&d);
+ while((c = *s++)) {
+ if(c >= 32 && c <= 127)
+ dynstr_append(&d, c);
+ else {
+ sprintf(buf, "\\x%04lX", (unsigned long)c);
+ dynstr_append_string(&d, buf);
+ }
+ }
+ dynstr_terminate(&d);
+ return d.vec;
+}
+
#define check_string(GOT, WANT) do { \
const char *g = GOT; \
const char *w = WANT; \
}
static void test_casefold(void) {
- uint32_t c, l, u[2];
- const char *s, *ls;
+ uint32_t c, l;
+ const char *input, *canon_folded, *compat_folded, *canon_expected, *compat_expected;
fprintf(stderr, "test_casefold\n");
+ /* This isn't a very exhaustive test. Unlike for normalization, there don't
+ * seem to be any public test vectors for these algorithms. */
+
for(c = 1; c < 256; ++c) {
- u[0] = c;
- u[1] = 0;
- s = ucs42utf8(u);
- ls = casefold(s);
+ input = utf32_to_utf8(&c, 1, 0);
+ canon_folded = utf8_casefold_canon(input, strlen(input), 0);
+ compat_folded = utf8_casefold_compat(input, strlen(input), 0);
switch(c) {
default:
if((c >= 'A' && c <= 'Z')
l = 0x3BC; /* GREEK SMALL LETTER MU */
break;
case 0xDF: /* LATIN SMALL LETTER SHARP S */
- insist(!strcmp(ls, "ss"));
+ insist(!strcmp(canon_folded, "ss"));
+ insist(!strcmp(compat_folded, "ss"));
l = 0;
break;
}
if(l) {
- u[0] = l;
- u[1] = 0;
- s = ucs42utf8(u);
- if(strcmp(s, ls)) {
- fprintf(stderr, "%s:%d: casefolding %#lx got '%s', expected '%s'\n",
+ /* Case-folded data is now normalized */
+ canon_expected = ucs42utf8(utf32_decompose_canon(&l, 1, 0));
+ if(strcmp(canon_folded, canon_expected)) {
+ fprintf(stderr, "%s:%d: canon-casefolding %#lx got '%s', expected '%s'\n",
__FILE__, __LINE__, (unsigned long)c,
- format(ls), format(s));
+ format(canon_folded), format(canon_expected));
+ ++errors;
+ }
+ ++tests;
+ compat_expected = ucs42utf8(utf32_decompose_compat(&l, 1, 0));
+ if(strcmp(compat_folded, compat_expected)) {
+ fprintf(stderr, "%s:%d: compat-casefolding %#lx got '%s', expected '%s'\n",
+ __FILE__, __LINE__, (unsigned long)c,
+ format(compat_folded), format(compat_expected));
++errors;
}
++tests;
/** @struct iheap
* @brief A heap with @c int elements */
HEAP_TYPE(iheap, int, int_lt);
+HEAP_DEFINE(iheap, int, int_lt);
/** @brief Tests for @ref heap.h */
static void test_heap(void) {
putchar('\n');
}
+/** @brief Open a Unicode test file */
+static FILE *open_unicode_test(const char *path) {
+ const char *base;
+ FILE *fp;
+ char buffer[1024];
+ int w;
+
+ if((base = strrchr(path, '/')))
+ ++base;
+ else
+ base = path;
+ if(!(fp = fopen(base, "r"))) {
+ snprintf(buffer, sizeof buffer,
+ "wget http://www.unicode.org/Public/5.0.0/ucd/%s", path);
+ if((w = system(buffer)))
+ fatal(0, "%s: %s", buffer, wstat(w));
+ if(chmod(base, 0444) < 0)
+ fatal(errno, "chmod %s", base);
+ if(!(fp = fopen(base, "r")))
+ fatal(errno, "%s", base);
+ }
+ return fp;
+}
+
+/** @brief Tests for @ref lib/unicode.h */
+static void test_unicode(void) {
+ FILE *fp;
+ int lineno = 0;
+ char *l, *lp;
+ uint32_t buffer[1024];
+ uint32_t *c[6], *NFD_c[6], *NFKD_c[6]; /* 1-indexed */
+ int cn, bn;
+ char break_allowed[1024];
+
+ fprintf(stderr, "test_unicode\n");
+ fp = open_unicode_test("NormalizationTest.txt");
+ while(!inputline("NormalizationTest.txt", fp, &l, '\n')) {
+ ++lineno;
+ if(*l == '#' || *l == '@')
+ continue;
+ bn = 0;
+ cn = 1;
+ lp = l;
+ c[cn++] = &buffer[bn];
+ while(*lp && *lp != '#') {
+ if(*lp == ' ') {
+ ++lp;
+ continue;
+ }
+ if(*lp == ';') {
+ buffer[bn++] = 0;
+ if(cn == 6)
+ break;
+ c[cn++] = &buffer[bn];
+ ++lp;
+ continue;
+ }
+ buffer[bn++] = strtoul(lp, &lp, 16);
+ }
+ buffer[bn] = 0;
+ assert(cn == 6);
+ for(cn = 1; cn <= 5; ++cn) {
+ NFD_c[cn] = utf32_decompose_canon(c[cn], utf32_len(c[cn]), 0);
+ NFKD_c[cn] = utf32_decompose_compat(c[cn], utf32_len(c[cn]), 0);
+ }
+#define unt_check(T, A, B) do { \
+ ++tests; \
+ if(utf32_cmp(c[A], T##_c[B])) { \
+ fprintf(stderr, \
+ "NormalizationTest.txt:%d: c%d != "#T"(c%d)\n", \
+ lineno, A, B); \
+ fprintf(stderr, " c%d: %s\n", \
+ A, format_utf32(c[A])); \
+ fprintf(stderr, "%4s(c%d): %s\n", \
+ #T, B, format_utf32(T##_c[B])); \
+ ++errors; \
+ } \
+ } while(0)
+ unt_check(NFD, 3, 1);
+ unt_check(NFD, 3, 2);
+ unt_check(NFD, 3, 3);
+ unt_check(NFD, 5, 4);
+ unt_check(NFD, 5, 5);
+ unt_check(NFKD, 5, 1);
+ unt_check(NFKD, 5, 2);
+ unt_check(NFKD, 5, 3);
+ unt_check(NFKD, 5, 4);
+ unt_check(NFKD, 5, 5);
+ for(cn = 1; cn <= 5; ++cn) {
+ xfree(NFD_c[cn]);
+ xfree(NFKD_c[cn]);
+ }
+ xfree(l);
+ }
+ fclose(fp);
+ fp = open_unicode_test("auxiliary/GraphemeBreakTest.txt");
+ lineno = 0;
+ while(!inputline("GraphemeBreakTest.txt.txt", fp, &l, '\n')) {
+ ++lineno;
+ if(l[0] == '#') continue;
+ bn = 0;
+ lp = l;
+ while(*lp) {
+ if(*lp == ' ' || *lp == '\t') {
+ ++lp;
+ continue;
+ }
+ if(*lp == '#')
+ break;
+ if((unsigned char)*lp == 0xC3 && (unsigned char)lp[1] == 0xB7) {
+ /* 00F7 DIVISION SIGN */
+ break_allowed[bn] = 1;
+ lp += 2;
+ continue;
+ }
+ if((unsigned char)*lp == 0xC3 && (unsigned char)lp[1] == 0x97) {
+ /* 00D7 MULTIPLICATION SIGN */
+ break_allowed[bn] = 0;
+ lp += 2;
+ continue;
+ }
+ if(isxdigit((unsigned char)*lp)) {
+ buffer[bn++] = strtoul(lp, &lp, 16);
+ continue;
+ }
+ fatal(0, "GraphemeBreakTest.txt:%d: evil line: %s", lineno, l);
+ }
+ for(cn = 0; cn <= bn; ++cn) {
+ if(utf32_is_gcb(buffer, bn, cn) != break_allowed[cn]) {
+ fprintf(stderr,
+ "GraphemeBreakTest.txt:%d: offset %d: utf32_is_gcb wrong\n",
+ lineno, cn);
+ ++errors;
+ }
+ ++tests;
+ }
+ xfree(l);
+ }
+}
+
int main(void) {
insist('\n' == 0x0A);
insist('\r' == 0x0D);
/* split.c */
/* syscalls.c */
/* table.c */
+ /* unicode.c */
+ test_unicode();
/* utf8.c */
test_utf8();
/* vector.c */
Local Variables:
c-basic-offset:2
comment-column:40
+fill-column:79
+indent-tabs-mode:nil
End:
*/
-
~mdw / disorder / blobdiff