Add the framework for character classes.

Now we have the potential to recognize character classes.  But the
most important task for us now is to experiment with ((B)RN)GLR
algorithms, so we leave the character classes at the present state for
a moment.

1 files changed, 108 insertions, 0 deletions

diff --git a/src/test/check_dfa.c b/src/test/check_dfa.c
new file mode 100644
index 0000000..a728fb8
--- /dev/null
+++ b/src/test/check_dfa.c
@@ -0,0 +1,108 @@
+#include "../utf8.h"
+#include "../str.h"
+#include <stdio.h>
+#include "../util.h"
+#include "../dfa.h"
+
+#define ADD_RANGE(X, Y) do {                    \
+    for (int i = 0; i + (X) <= (Y); i++)        \
+      *(v+vlen++) = (NUM) i + (X);              \
+  } while (0)
+
+int
+main(int UNUSED argc, char ** UNUSED argv)
+{
+  UNUM vlen = 0;
+  NUM *v = MYALLOC(NUM, 1<<21);
+
+  ADD_RANGE(0x0020, 0x007f);
+  ADD_RANGE(0x2580, 0x259f);
+  ADD_RANGE(0x00a0, 0x00ff);
+  ADD_RANGE(0x25a0, 0x25ff);
+  ADD_RANGE(0x0100, 0x017f);
+  ADD_RANGE(0x2600, 0x26ff);
+  ADD_RANGE(0x0180, 0x024f);
+  ADD_RANGE(0x2700, 0x27bf);
+  ADD_RANGE(0x0250, 0x02af);
+  ADD_RANGE(0x27c0, 0x27ef);
+  ADD_RANGE(0x02b0, 0x02ff);
+  ADD_RANGE(0x27f0, 0x27ff);
+  ADD_RANGE(0x0300, 0x036f);
+  ADD_RANGE(0x2800, 0x28ff);
+  ADD_RANGE(0x0370, 0x03ff);
+  ADD_RANGE(0x2900, 0x297f);
+  ADD_RANGE(0x0400, 0x04ff);
+  ADD_RANGE(0x2980, 0x29ff);
+  ADD_RANGE(0x0500, 0x052f);
+  ADD_RANGE(0x2a00, 0x2aff);
+  ADD_RANGE(0x0530, 0x058f);
+  ADD_RANGE(0x2b00, 0x2bff);
+  ADD_RANGE(0x0590, 0x05ff);
+  ADD_RANGE(0x2e80, 0x2eff);
+  ADD_RANGE(0x0600, 0x06ff);
+  ADD_RANGE(0x2f00, 0x2fdf);
+  ADD_RANGE(0x0700, 0x074f);
+  ADD_RANGE(0x2ff0, 0x2fff);
+  ADD_RANGE(0x0780, 0x07bf);
+  ADD_RANGE(0x3000, 0x303f);
+  ADD_RANGE(0x0900, 0x097f);
+  ADD_RANGE(0x3040, 0x309f);
+  ADD_RANGE(0x0980, 0x09ff);
+  ADD_RANGE(0x30a0, 0x30ff);
+  ADD_RANGE(0x0a00, 0x0a7f);
+  ADD_RANGE(0x3100, 0x312f);
+  ADD_RANGE(0x0a80, 0x0aff);
+  ADD_RANGE(0x3130, 0x318f);
+  ADD_RANGE(0x0b00, 0x0b7f);
+  ADD_RANGE(0x3190, 0x319f);
+  ADD_RANGE(0x0b80, 0x0bff);
+  ADD_RANGE(0x31a0, 0x31bf);
+  ADD_RANGE(0x0c00, 0x0c7f);
+  ADD_RANGE(0x31f0, 0x31ff);
+  ADD_RANGE(0x0c80, 0x0cff);
+  ADD_RANGE(0x3200, 0x32ff);
+  ADD_RANGE(0x0d00, 0x0d7f);
+  ADD_RANGE(0x3300, 0x33ff);
+  ADD_RANGE(0x0d80, 0x0dff);
+  ADD_RANGE(0x3400, 0x4dbf);
+  ADD_RANGE(0x0e00, 0x0e7f);
+  ADD_RANGE(0x4dc0, 0x4dff);
+  ADD_RANGE(0x0e80, 0x0eff);
+  ADD_RANGE(0x4e00, 0x9fff);
+  ADD_RANGE(0x0f00, 0x0fff);
+
+  /* for (int i = 0; i < 190; i++)
+   *   printf("v [%d] = %d\n", i, *(v+i)); */
+
+  dfa *dfap = dfa_from_bytes(vlen, v);
+
+  free(v);
+
+  if (dfap == NULL)
+    printf("returned null\n");
+  else
+    print_dfa(dfap);
+
+  /* char *s = MYALLOC(char, 5);
+   * str *strp = new_str(s, 5);
+   *
+   * for (int i = 0; i < 3; i++) {
+   *   printf("\n\n");
+   *
+   *   encode(v[i], strp);
+   *
+   *   for (int j = 0; j < str_length(strp)-1;)
+   *     printf("%d, ", *(s+j++)+(1<<8));
+   *
+   *   printf("%d\n", *(s+str_length(strp)-1)+(1<<8));
+   * }
+   *
+   * destroy_str(strp, 1); */
+
+  /* BOOL result = run_dfa(dfap, 32239);
+   *
+   * printf("the result = %d\n", result); */
+
+  if (dfap) destroy_dfa(dfap);
+  return 0;
+}