tregress.c

正则表达式库

/* tregress.c: reported bugs.  The `t' just makes the filename not have
   a common prefix with `regex.c', so completion works better.  */

#include "test.h"


boolean pause_at_error = true;

char *
itoa (i)
    int i;
{
  char *a = xmalloc (21); /* sign + 19 digits (enough for 64 bits) + null */
  
  sprintf (a, "%d", i);
  return a;
}


static void
simple_fail (routine, pat, buf, str, ret)
    const char *routine;
    const char *pat;
    struct re_pattern_buffer *buf;
    const char *str;
    char *ret;
{
  fprintf (stderr, "Failed %s (return = %s).\n", routine, ret);
  if (str && *str) fprintf (stderr, "   String = %s\n", str);
  fprintf (stderr, "  Pattern = %s\n", pat);
  print_compiled_pattern (buf);
  
  if (pause_at_error)
    {
      fprintf (stderr, "RET to continue: ");
      (void) getchar ();
    }
}


/* Abbreviate the most common calls.  */

static void
simple_compile (pat, buf)
    const char *pat;
    struct re_pattern_buffer *buf;
{
  const char *ret = re_compile_pattern (pat, strlen (pat), buf);
  
  if (ret != NULL) simple_fail ("compile", pat, buf, NULL, ret);
}


static void
simple_fastmap (pat)
    const char *pat;
{
  struct re_pattern_buffer buf;
  char fastmap[256];
  int ret;
  
  buf.allocated = 0;
  buf.buffer = buf.translate = NULL;
  buf.fastmap = fastmap;
  
  simple_compile (pat, &buf);
  
  ret = re_compile_fastmap (&buf);

  if (ret != 0) simple_fail ("fastmap compile", pat, &buf, NULL, itoa (ret));
}


#define SIMPLE_MATCH(pat, str) do_match (pat, str, strlen (str))
#define SIMPLE_NONMATCH(pat, str) do_match (pat, str, -1)

static void
do_match (pat, str, expected)
    const char *pat, *str;
    int expected;
{
  int ret;
  unsigned len;
  struct re_pattern_buffer buf;

  buf.allocated = 0;
  buf.buffer = buf.translate = buf.fastmap = NULL;
  
  simple_compile (pat, &buf);

  len = strlen (str);
  
  ret = re_match_2 (&buf, NULL, 0, str, len, 0, NULL, len);
  
  if (ret != expected) simple_fail ("match", pat, &buf, str, itoa (ret));
}


static void
simple_search (pat, str, correct_startpos)
    const char *pat, *str;
    int correct_startpos;
{
  int ret;
  unsigned len;
  struct re_pattern_buffer buf;

  buf.allocated = 0;
  buf.buffer = buf.translate = buf.fastmap = NULL;
  
  simple_compile (pat, &buf);

  len = strlen (str);
  
  ret = re_search_2 (&buf, NULL, 0, str, len, 0, len, NULL, len);
  
  if (ret != correct_startpos)
    simple_fail ("match", pat, &buf, str, itoa (ret));
}

/* Past bugs people have reported.  */

void
test_regress ()
{
  extern char upcase[];
  struct re_pattern_buffer buf;
  unsigned len;
  struct re_registers regs;
  int ret;
  char *fastmap = xmalloc (256);  
  
  buf.translate = NULL;
  buf.fastmap = NULL;
  buf.allocated = 0;
  buf.buffer = NULL;

  printf ("\nStarting regression tests.\n");
  t = regress_test;

  test_should_match = true;
  re_set_syntax (RE_SYNTAX_EMACS);

  /* [e-mail]  10 Nov 92 15:19:02 JST  */
  buf.translate = upcase;
  SIMPLE_MATCH ("[A-[]", "A");
  buf.translate = NULL;
  
  /* [e-mail]  Nov  6 22:34:41 1992  */
  simple_search ("\\w+", "a", 0);
   
  /* [e-mail]  10 Sep 92 00:42:33  */
  buf.translate = upcase;
  SIMPLE_MATCH ("[\001-\377]", "\001");
  SIMPLE_MATCH ("[\001-\377]", "a");
  SIMPLE_MATCH ("[\001-\377]", "\377");
  buf.translate = NULL;

  /* [e-mail]  1 Sep 92 01:45:22  */
  SIMPLE_MATCH ("^^$", "^");
  
  /* [e-mail]  Sep  7 22:42:36 1992  */
  re_set_syntax (RE_INTERVALS);
  SIMPLE_MATCH ("^a\\{3\\}$", "aaa");
  SIMPLE_NONMATCH ("^a\\{3\\}$", "aa");
  re_set_syntax (RE_SYNTAX_EMACS);
  
  /* [e-mail], 31 Aug 92.  (conjecture) */
  re_set_syntax (RE_INTERVALS);
  simple_search ("a\\{1,3\\}b", "aaab", 0);
  simple_search ("a\\{1,3\\}b", "aaaab", 1);
  re_set_syntax (RE_SYNTAX_EMACS);

  /* [e-mail], 31 Aug 92.  (simplified) */
  simple_fastmap ("^.*\n[  ]*");
  
  /* wind![e-mail], 25 Aug 92.  (simplified) */
  re_set_syntax (RE_INTERVALS);
  SIMPLE_MATCH ("[a-zA-Z]*.\\{5\\}", "xN0000");
  SIMPLE_MATCH ("[a-zA-Z]*.\\{5\\}$", "systemxN0000");
  SIMPLE_MATCH ("\\([a-zA-Z]*\\).\\{5\\}$", "systemxN0000");
  re_set_syntax (RE_SYNTAX_EMACS);
  
  /* jimb, 18 Aug 92.  Don't use \000, so `strlen' (in our testing
     routines) will work.  (This still tickles the bug jimb reported.)  */
  SIMPLE_MATCH ("[\001-\377]", "\001");
  SIMPLE_MATCH ("[\001-\377]", "a");
  SIMPLE_MATCH ("[\001-\377]", "\377");

  /* jimb, 13 Aug 92.  */
  SIMPLE_MATCH ("[\001-\177]", "\177");

  /* Tests based on bwoelfel's below.  */
  SIMPLE_MATCH ("\\(a\\|ab\\)*", "aab");
  SIMPLE_MATCH ("\\(a\\|ab\\)+", "aab");
  SIMPLE_MATCH ("\\(a*\\|ab\\)+", "aab");
  SIMPLE_MATCH ("\\(a+\\|ab\\)+", "aab");
  SIMPLE_MATCH ("\\(a?\\|ab\\)+", "aab");

  /* [e-mail], 25 Jul 92.  */
  SIMPLE_MATCH ("^\\([ab]+\\|bc\\)+", "abc");

  /* jla, 3 Jul 92.  Core dump in re_search_2.  */
  buf.fastmap = fastmap;
  buf.translate = upcase;
#define DATEDUMP_PATTERN " *[0-9]*:"
  if (re_compile_pattern (DATEDUMP_PATTERN, strlen (DATEDUMP_PATTERN), &buf)
      != NULL)
    printf ("date dump compile failed.\n");
  regs.num_regs = 0;
  regs.start = regs.end = NULL;
  if (re_search_2 (&buf, NULL, 0, "Thu Jul  2 18:34:18 1992",
                   24, 3, 21, &regs, 24) != 10)
    printf ("date dump search failed.\n");
  buf.fastmap = 0;
  buf.translate = 0;


  /* rms, 4 Jul 1992.  Pattern is much slower in Emacs 19.  Fastmap
     should be only a backslash.  */
#define BEGINEND_PATTERN "\\(\\\\begin\\s *{\\)\\|\\(\\\\end\\s *{\\)"
  test_fastmap (BEGINEND_PATTERN, "\\", false, 0);


  /* [e-mail], 27 Jun 1992.  Code for [a-z] (in regex.c)
     should translate the whole set.  */
  buf.translate = upcase;
#define CASE_SET_PATTERN "[ -`]"
  if (re_compile_pattern (CASE_SET_PATTERN, strlen (CASE_SET_PATTERN), &buf)
      != NULL)
    printf ("case set compile failed.\n");
  if (re_match_2 (&buf, "K", 1, "", 0, 0, NULL, 1) != 1)
    printf ("case set match failed.\n");