start on ut32__unidata() which will provide a uniform interface

[disorder] / lib / test.c

diff --git a/lib/test.c b/lib/test.c
index 50300adb67cd9eb0207931127d6a777595d08ca8..62494449ccbe5256ace2ee42e0d8f5c2f5d912ca 100644 (file)
--- a/lib/test.c
+++ b/lib/test.c
@@ -42,13 +42,21 @@
 #include "heap.h"
 #include "unicode.h"
 #include "inputline.h"
+#include "wstat.h"
 
 static int tests, errors;
+static int fail_first;
+
+static void count_error() {
+  ++errors;
+  if(fail_first)
+    abort();
+}
 
 /** @brief Checks that @p expr is nonzero */
 #define insist(expr) do {                              \
   if(!(expr)) {                                                \
-    ++errors;                                          \
+    count_error();                                             \
     fprintf(stderr, "%s:%d: error checking %s\n",      \
             __FILE__, __LINE__, #expr);                        \
   }                                                    \
@@ -98,11 +106,11 @@ static const char *format_utf32(const uint32_t *s) {
   if(w == 0) {                                                 \
     fprintf(stderr, "%s:%d: %s returned 0\n",                  \
             __FILE__, __LINE__, #GOT);                         \
-    ++errors;                                                  \
+    count_error();                                                     \
   } else if(strcmp(w, g)) {                                    \
     fprintf(stderr, "%s:%d: %s returned:\n%s\nexpected:\n%s\n",        \
            __FILE__, __LINE__, #GOT, format(g), format(w));    \
-    ++errors;                                                  \
+    count_error();                                                     \
   }                                                            \
   ++tests;                                                     \
  } while(0)
@@ -390,7 +398,7 @@ static void test_casefold(void) {
        fprintf(stderr, "%s:%d: canon-casefolding %#lx got '%s', expected '%s'\n",
                __FILE__, __LINE__, (unsigned long)c,
                format(canon_folded), format(canon_expected));
-       ++errors;
+       count_error();
       }
       ++tests;
       compat_expected = ucs42utf8(utf32_decompose_compat(&l, 1, 0));
@@ -398,7 +406,7 @@ static void test_casefold(void) {
        fprintf(stderr, "%s:%d: compat-casefolding %#lx got '%s', expected '%s'\n",
                __FILE__, __LINE__, (unsigned long)c,
                format(compat_folded), format(compat_expected));
-       ++errors;
+       count_error();
       }
       ++tests;
     }
@@ -435,6 +443,84 @@ 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 Run breaking tests for utf32_is_gcb() 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",
+                path, lineno, n);
+        count_error();
+      }
+      ++tests;
+    }
+    xfree(l);
+  }
+  fclose(fp);
+}
+
 /** @brief Tests for @ref lib/unicode.h */
 static void test_unicode(void) {
   FILE *fp;
@@ -445,16 +531,7 @@ static void test_unicode(void) {
   int cn, bn;
 
   fprintf(stderr, "test_unicode\n");
-  if(!(fp = fopen("NormalizationTest.txt", "r"))) {
-    system("wget http://www.unicode.org/Public/5.0.0/ucd/NormalizationTest.txt");
-    chmod("NormalizationTest.txt", 0444);
-    if(!(fp = fopen("NormalizationTest.txt", "r"))) {
-      perror("NormalizationTest.txt");
-      ++tests;                         /* don't know how many... */
-      ++errors;
-      return;
-    }
-  }
+  fp = open_unicode_test("NormalizationTest.txt");
   while(!inputline("NormalizationTest.txt", fp, &l, '\n')) {
     ++lineno;
     if(*l == '#' || *l == '@')
@@ -487,12 +564,14 @@ static void test_unicode(void) {
 #define unt_check(T, A, B) do {                                        \
     ++tests;                                                   \
     if(utf32_cmp(c[A], T##_c[B])) {                            \
-      fprintf(stderr, "L%d: c%d != "#T"(c%d)\n", lineno, A, B);        \
-      fprintf(stderr, "    c%d:      %s\n",                    \
+      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;                                                        \
+      count_error();                                           \
     }                                                          \
   } while(0)
     unt_check(NFD, 3, 1);
@@ -511,9 +590,13 @@ static void test_unicode(void) {
     }
     xfree(l);
   }
+  fclose(fp);
+  breaktest("auxiliary/GraphemeBreakTest.txt", utf32_is_gcb);
+  breaktest("auxiliary/WordBreakTest.txt", utf32_is_word_boundary);
 }
 
 int main(void) {
+  fail_first = !!getenv("FAIL_FIRST");
   insist('\n' == 0x0A);
   insist('\r' == 0x0D);
   insist(' ' == 0x20);