+ format(compat_folded), format(compat_expected));
+ count_error();
+ }
+ ++tests;
+ }
+ }
+ check_string(utf8_casefold_canon("", 0, 0), "");
+}
+
+struct {
+ const char *in;
+ const char *expect[10];
+} wtest[] = {
+ /* Empty string */
+ { "", { 0 } },
+ /* Only whitespace and punctuation */
+ { " ", { 0 } },
+ { " ' ", { 0 } },
+ { " ! ", { 0 } },
+ { " \"\" ", { 0 } },
+ { " @ ", { 0 } },
+ /* Basics */
+ { "wibble", { "wibble", 0 } },
+ { " wibble", { "wibble", 0 } },
+ { " wibble ", { "wibble", 0 } },
+ { "wibble ", { "wibble", 0 } },
+ { "wibble spong", { "wibble", "spong", 0 } },
+ { " wibble spong", { "wibble", "spong", 0 } },
+ { " wibble spong ", { "wibble", "spong", 0 } },
+ { "wibble spong ", { "wibble", "spong", 0 } },
+ { "wibble spong splat foo zot ", { "wibble", "spong", "splat", "foo", "zot", 0 } },
+ /* Apostrophes */
+ { "wibble 'spong", { "wibble", "spong", 0 } },
+ { " wibble's", { "wibble's", 0 } },
+ { " wibblespong' ", { "wibblespong", 0 } },
+ { "wibble sp''ong ", { "wibble", "sp", "ong", 0 } },
+};
+#define NWTEST (sizeof wtest / sizeof *wtest)
+
+static void test_words(void) {
+ size_t t, nexpect, ngot, i;
+ int right;
+
+ fprintf(stderr, "test_words\n");
+ for(t = 0; t < NWTEST; ++t) {
+ char **got = utf8_word_split(wtest[t].in, strlen(wtest[t].in), &ngot, 0);
+
+ for(nexpect = 0; wtest[t].expect[nexpect]; ++nexpect)
+ ;
+ if(nexpect == ngot) {
+ for(i = 0; i < ngot; ++i)
+ if(strcmp(wtest[t].expect[i], got[i]))
+ break;
+ right = i == ngot;
+ } else
+ right = 0;
+ if(!right) {
+ fprintf(stderr, "word split %zu failed\n", t);
+ fprintf(stderr, "input: %s\n", wtest[t].in);
+ fprintf(stderr, " | %-30s | %-30s\n",
+ "expected", "got");
+ for(i = 0; i < nexpect || i < ngot; ++i) {
+ const char *e = i < nexpect ? wtest[t].expect[i] : "<none>";
+ const char *g = i < ngot ? got[i] : "<none>";
+ fprintf(stderr, " %2zu | %-30s | %-30s\n", i, e, g);
+ }
+ count_error();
+ }
+ ++tests;
+ }
+}
+
+/** @brief Less-than comparison function for integer heap */
+static inline int int_lt(int a, int b) { return a < b; }
+
+/** @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) {
+ struct iheap h[1];
+ int n;
+ int last = -1;
+
+ fprintf(stderr, "test_heap\n");
+
+ iheap_init(h);
+ for(n = 0; n < 1000; ++n)
+ iheap_insert(h, random() % 100);
+ for(n = 0; n < 1000; ++n) {
+ const int latest = iheap_remove(h);
+ if(last > latest)
+ fprintf(stderr, "should have %d <= %d\n", last, latest);
+ insist(last <= latest);
+ last = latest;
+ }
+ 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 Run breaking tests for utf32_grapheme_boundary() etc */
+static void breaktest(const char *path,
+ int (*breakfn)(const uint32_t *, size_t, size_t)) {
+ FILE *fp = open_unicode_test(path);
+ int lineno = 0;
+ char *l, *lp;
+ size_t bn, n;
+ char break_allowed[1024];
+ uint32_t buffer[1024];
+
+ while(!inputline(path, 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, "%s:%d: evil line: %s", path, lineno, l);
+ }
+ for(n = 0; n <= bn; ++n) {
+ if(breakfn(buffer, bn, n) != break_allowed[n]) {
+ fprintf(stderr,
+ "%s:%d: offset %zu: mismatch\n"
+ "%s\n"
+ "\n",
+ path, lineno, n, l);
+ count_error();