psx-interf.c

正则表达式库

/* psx-interf.c: test POSIX interface.  */

#include <string.h>
#include <assert.h>

#include "test.h"

#define ERROR_CODE_LENGTH 20
#define TEST_ERRBUF_SIZE 15
 

void test_compile ();


/* ANSWER should be at least ERROR_CODE_LENGTH long.  */

static char *
get_error_string (error_code, answer)
  int error_code;
  char answer[];
{
  switch (error_code)
    {
      case 0: strcpy (answer, "No error"); break;
      case REG_NOMATCH: strcpy (answer, "REG_NOMATCH"); break;
      case REG_BADPAT: strcpy (answer, "REG_BADPAT"); break;
      case REG_EPAREN: strcpy (answer, "REG_EPAREN"); break;
      case REG_ESPACE: strcpy (answer, "REG_ESPACE"); break;
      case REG_ECOLLATE: strcpy (answer, "REG_ECOLLATE"); break;
      case REG_ECTYPE: strcpy (answer, "REG_ECTYPE"); break;
      case REG_EESCAPE: strcpy (answer, "REG_EESCAPE"); break;
      case REG_ESUBREG: strcpy (answer, "REG_ESUBREG"); break;
      case REG_EBRACK: strcpy (answer, "REG_EBRACK"); break;
      case REG_EBRACE: strcpy (answer, "REG_EBRACE"); break;
      case REG_BADBR: strcpy (answer, "REG_BADBR"); break;
      case REG_ERANGE: strcpy (answer, "REG_ERANGE"); break;
      case REG_BADRPT: strcpy (answer, "REG_BADRPT"); break;
      case REG_EEND: strcpy (answer, "REG_EEND"); break;
      default: strcpy (answer, "Bad error code");
    }
  return answer;
}


/* I don't think we actually need to initialize all these things.
   --karl  */

void
init_pattern_buffer (pattern_buffer_ptr)
    regex_t *pattern_buffer_ptr;
{
  pattern_buffer_ptr->buffer = NULL;
  pattern_buffer_ptr->allocated = 0;
  pattern_buffer_ptr->used = 0;
  pattern_buffer_ptr->fastmap = NULL;
  pattern_buffer_ptr->fastmap_accurate = 0;
  pattern_buffer_ptr->translate = NULL;
  pattern_buffer_ptr->can_be_null = 0;
  pattern_buffer_ptr->re_nsub = 0;
  pattern_buffer_ptr->no_sub = 0;
  pattern_buffer_ptr->not_bol = 0;
  pattern_buffer_ptr->not_eol = 0;
}


void
test_compile (valid_pattern, error_code_expected, pattern, 
	      pattern_buffer_ptr, cflags)
    unsigned valid_pattern;
    int error_code_expected;
    const char *pattern;
    regex_t *pattern_buffer_ptr;
    int cflags;
{    
  int error_code_returned;
  boolean error = false;
  char errbuf[TEST_ERRBUF_SIZE];
  
  init_pattern_buffer (pattern_buffer_ptr);
  error_code_returned = regcomp (pattern_buffer_ptr, pattern, cflags);

  if (valid_pattern && error_code_returned)
    {
      printf ("\nShould have been a valid pattern but wasn't.\n");
      regerror (error_code_returned, pattern_buffer_ptr, errbuf, 
	        TEST_ERRBUF_SIZE);
      printf ("%s", errbuf);
      error = true;
    }
    
  if (!valid_pattern && !error_code_returned)
    {
      printf ("\n\nInvalid pattern compiled as valid:\n");
      error = true;
    }
 
  if (error_code_returned != error_code_expected)
    {
      char expected_error_string[ERROR_CODE_LENGTH];
      char returned_error_string[ERROR_CODE_LENGTH];
      
      get_error_string (error_code_expected, expected_error_string), 
      get_error_string (error_code_returned, returned_error_string);
      
      printf ("  Expected error code %s but got `%s'.\n", 
              expected_error_string, returned_error_string);

      error = true;
    }
  
  if (error)
    print_pattern_info (pattern, pattern_buffer_ptr);
}


static void
test_nsub (sub_count, pattern, cflags)
  unsigned sub_count;
  char *pattern;
  int cflags;

{
  regex_t pattern_buffer;
  
  test_compile (1, 0, pattern, &pattern_buffer, cflags);

  if (pattern_buffer.re_nsub != sub_count)
    {
      printf ("\nShould have counted %d subexpressions but counted %d \
instead.\n", sub_count, pattern_buffer.re_nsub);
    }

  regfree (&pattern_buffer);
}
  

static void
test_regcomp ()
{
  regex_t pattern_buffer;
  int cflags = 0;

  
  printf ("\nStarting regcomp tests.\n");
  
  cflags = 0;
  test_compile (0, REG_ESUBREG, "\\(a\\)\\2", &pattern_buffer, cflags);
  test_compile (0, REG_EBRACE, "a\\{", &pattern_buffer, cflags);
  test_compile (0, REG_BADBR, "a\\{-1\\}", &pattern_buffer, cflags);
  test_compile (0, REG_EBRACE, "a\\{", &pattern_buffer, cflags);
  test_compile (0, REG_EBRACE, "a\\{1", &pattern_buffer, cflags);

  cflags = REG_EXTENDED;
  test_compile (0, REG_ECTYPE, "[[:alpo:]]", &pattern_buffer, cflags);
  test_compile (0, REG_EESCAPE, "\\", &pattern_buffer, cflags);
  test_compile (0, REG_EBRACK, "[a", &pattern_buffer, cflags);
  test_compile (0, REG_EPAREN, "(", &pattern_buffer, cflags);
  test_compile (0, REG_ERANGE, "[z-a]", &pattern_buffer, cflags);

  test_nsub (1, "(a)", cflags);
  test_nsub (2, "((a))", cflags);
  test_nsub (2, "(a)(b)", cflags);
  
  cflags = REG_EXTENDED | REG_NOSUB;
  test_nsub (1, "(a)", cflags);

  regfree (&pattern_buffer);

  printf ("\nFinished regcomp tests.\n");
}


static void
fill_pmatch (pmatch, start0, end0, start1, end1, start2, end2)
  regmatch_t pmatch[];
  regoff_t start0, end0, start1, end1, start2, end2;
{
  pmatch[0].rm_so = start0;
  pmatch[0].rm_eo = end0;
  pmatch[1].rm_so = start1;
  pmatch[1].rm_eo = end1;
  pmatch[2].rm_so = start2;
  pmatch[2].rm_eo = end2;
}


static void
test_pmatch (pattern, string, nmatch, pmatch, correct_pmatch, cflags)
  char *pattern;
  char *string;
  unsigned nmatch;
  regmatch_t pmatch[];
  regmatch_t correct_pmatch[];
  int cflags;
{
  regex_t pattern_buffer;
  unsigned this_match;
  int error_code_returned;
  boolean found_nonmatch = false;
  
  test_compile (1, 0, pattern, &pattern_buffer, cflags);
  error_code_returned = regexec (&pattern_buffer, string, nmatch, pmatch, 0);
  
  if (error_code_returned == REG_NOMATCH)
    printf ("Matching failed in test_pmatch.\n");
  else    
    {
      for (this_match = 0; this_match < nmatch; this_match++)
        {
          if (pmatch[this_match].rm_so != correct_pmatch[this_match].rm_so)
            {
              if (found_nonmatch == false)
                printf ("\n");

             printf ("Pmatch start %d wrong: was %d when should have \
been %d.\n", this_match, pmatch[this_match].rm_so, 
		      correct_pmatch[this_match].rm_so);
              found_nonmatch = true;
            }
          if (pmatch[this_match].rm_eo != correct_pmatch[this_match].rm_eo)
            {
              if (found_nonmatch == false)
                printf ("\n");

              printf ("Pmatch end   %d wrong: was %d when should have been \
%d.\n", this_match, pmatch[this_match].rm_eo, 
                       correct_pmatch[this_match].rm_eo);
              found_nonmatch = true;
            }
        }

      if (found_nonmatch)
        {
          printf ("  The number of pmatches requested was: %d.\n", nmatch);
          printf ("  The string to match was:  `%s'.\n", string);
          print_pattern_info (pattern, &pattern_buffer);
        }
    }  /* error_code_returned == REG_NOMATCH  */

  regfree (&pattern_buffer);
}


static void
test_eflags (must_match_bol, must_match_eol, pattern, string, cflags, eflags)
  boolean must_match_bol;
  boolean must_match_eol;
  char *pattern;
  char *string;
  int cflags;
  int eflags;
{
  regex_t pattern_buffer;
  int error_code_returned;
  boolean was_error = false;

  test_compile (1, 0, pattern, &pattern_buffer, cflags);
  error_code_returned = regexec (&pattern_buffer, string, 0, 0, eflags);

  if (error_code_returned == REG_NOMATCH)
    {  
      /* If wasn't true that both 1) the anchored part of the pattern
         had to match this string and 2) this string was a proper
	 substring...  */

      if (!( (must_match_bol && (eflags & REG_NOTBOL)) 
	     || (must_match_eol && (eflags & REG_NOTEOL)) ))
        {
          printf ("\nEflags test failed:  didn't match when should have.\n");
          was_error = true;
	}
    }
  else  /* We got a match.  */
    {  
      /* If wasn't true that either 1) the anchored part of the pattern
         didn't have to match this string or 2) this string wasn't a
         proper substring...  */

      if ((must_match_bol == (eflags & REG_NOTBOL))
          || (must_match_eol == (eflags & REG_NOTEOL)))
        {
          printf ("\nEflags test failed:  matched when shouldn't have.\n");
	  was_error = true;
        }
    }

  if (was_error)
    {
      printf ("  The string to match was:  `%s'.\n", string);
      print_pattern_info (pattern, &pattern_buffer);

      if (eflags & REG_NOTBOL)
        printf ("  The eflag REG_BOL was set.\n");
      if (eflags & REG_NOTEOL)
        printf ("  The eflag REG_EOL was set.\n");
    }

  regfree (&pattern_buffer);
}


static void
test_ignore_case (should_match, pattern, string, cflags)
  boolean should_match;
  char *pattern;
  char *string;
  int cflags;
{
  regex_t pattern_buffer;
  int error_code_returned;

  test_compile (1, 0, pattern, &pattern_buffer, cflags);
  error_code_returned = regexec (&pattern_buffer, string, 0, 0, 0);