php_mbregex.c

php-4.4.7学习linux时下载的源代码

			}
			n = regs.end[0];
			if (pos < n) {
				pos = n;
			} else {
				if (pos < string_len) {
					_php_mb_regex_strbuf_ncat(&outdev, (const unsigned char *)&string[pos], 1);
				}
				pos++;
			}
		} else { /* nomatch */
			/* stick that last bit of string on our output */
			if (pos < string_len) {
				_php_mb_regex_strbuf_ncat(&outdev, (const unsigned char *)&string[pos], string_len - pos);
			}
		}
	}

	if (description) {
		efree(description);
	}
	mbre_free_registers(&regs);
	if (evaldev.buffer) {
		efree((void*)evaldev.buffer);
	}
	n = outdev.pos;
	_php_mb_regex_strbuf_ncat(&outdev, (const unsigned char *)"\0", 1);
	if (err <= -2) {
		if (outdev.buffer) {
			efree((void*)outdev.buffer);
		}
		RETVAL_FALSE;
	} else {
		RETVAL_STRINGL((char *)outdev.buffer, n, 0);
	}
}
/* }}} */
/* {{{ proto string mb_ereg_replace(string pattern, string replacement, string string [, string option])
   Replace regular expression for multibyte string */
PHP_FUNCTION(mb_ereg_replace)
{
	_php_mb_regex_ereg_replace_exec(INTERNAL_FUNCTION_PARAM_PASSTHRU, 0);
}
/* }}} */

/* {{{ proto string mb_eregi_replace(string pattern, string replacement, string string)
   Case insensitive replace regular expression for multibyte string */
PHP_FUNCTION(mb_eregi_replace)
{
	_php_mb_regex_ereg_replace_exec(INTERNAL_FUNCTION_PARAM_PASSTHRU, MBRE_OPTION_IGNORECASE);
}
/* }}} */


/* {{{ proto array mb_split(string pattern, string string [, int limit])
   split multibyte string into array by regular expression */
PHP_FUNCTION(mb_split)
{
	char *arg_pattern;
	int arg_pattern_len;
	mb_regex_t re;
	struct mbre_registers regs = {0, 0, 0, 0};
	char *string;
	int string_len;

	int n, err, pos;
	long count = -1;

	if (zend_parse_parameters(ZEND_NUM_ARGS() TSRMLS_CC, "ss|l", &arg_pattern, &arg_pattern_len, &string, &string_len, &count) == FAILURE) {
		RETURN_FALSE;
	} 

	if (count == 0) {
		count = 1;
	}

	if (array_init(return_value) == FAILURE) {
		RETURN_FALSE;
	}

	/* create regex pattern buffer */
	err = php_mbregex_compile_pattern(
	     &re,
	     arg_pattern,
	     arg_pattern_len,
	     MBSTRG(regex_default_options), MBSTRG(current_mbctype) TSRMLS_CC);
	if (err) {
		RETURN_FALSE;
	}

	pos = 0;
	err = 0;
	/* churn through str, generating array entries as we go */
	while ((--count != 0) &&
		   (err = mbre_search(&re, string, string_len, pos, string_len - pos, &regs)) >= 0) {
		if ( regs.beg[0] == regs.end[0] ) {
			php_error_docref(NULL TSRMLS_CC, E_WARNING, "Empty regular expression");
			break;
		}


		/* add it to the array */
		if ( regs.beg[0] < string_len && regs.beg[0] >= pos) {
			add_next_index_stringl(return_value, (char *)&string[pos], regs.beg[0]-pos, 1);
		} else {
			err = -2;
			break;
		}
		/* point at our new starting point */
		n = regs.end[0];
		if (pos < n) {
			pos = n;
		}
		if (count < 0) {
			count = 0;
		}
	}

	mbre_free_registers(&regs);

	/* see if we encountered an error */
	if (err <= -2) {
		php_error_docref(NULL TSRMLS_CC, E_WARNING, "mbregex search failure in mbsplit()");
		zval_dtor(return_value);
		RETURN_FALSE;
	}

	/* otherwise we just have one last element to add to the array */
	n = string_len - pos;
	if (n > 0) {
		add_next_index_stringl(return_value, (char *)&string[pos], n, 1);
	} else {
		add_next_index_stringl(return_value, (char *)empty_string, 0, 1);
	}
}
/* }}} */


/* {{{ proto bool mb_ereg_match(string pattern, string string [,string option])
   Regular expression match for multibyte string */
PHP_FUNCTION(mb_ereg_match)
{
	char *arg_pattern;
	int arg_pattern_len;

	char *string;
	int string_len;

	mb_regex_t re;
	int option, err;

	option = 0;	
	{
		char *option_str = NULL;
		int option_str_len = 0;

		if (zend_parse_parameters(ZEND_NUM_ARGS() TSRMLS_CC, "ss|s",
		                          &arg_pattern, &arg_pattern_len, &string, &string_len,
		                          &option_str, &option_str_len)==FAILURE) {
			RETURN_FALSE;
		}

		if (option_str != NULL) {
			_php_mb_regex_init_options(option_str, option_str_len, &option, NULL);
		} else {
			option |= MBSTRG(regex_default_options);
		}
	}

	err = php_mbregex_compile_pattern(
	    &re,
	    arg_pattern,
		arg_pattern_len,
	    option, MBSTRG(current_mbctype) TSRMLS_CC);

	if (err) {
		RETURN_FALSE;
	}

	/* match */
	err = mbre_match(&re, string, string_len, 0, NULL);
	if (err >= 0) {
		RETVAL_TRUE;
	} else {
		RETVAL_FALSE;
	}
}
/* }}} */


/* regex search */
static void
_php_mb_regex_ereg_search_exec(INTERNAL_FUNCTION_PARAMETERS, int mode)
{
	zval **arg_pattern, **arg_options;
	int n, i, err, pos, len, beg, end, option;
	unsigned char *str;

	option = MBSTRG(regex_default_options);
	switch (ZEND_NUM_ARGS()) {
	case 0:
		break;
	case 1:
		if (zend_get_parameters_ex(1, &arg_pattern) == FAILURE) {
			WRONG_PARAM_COUNT;
		}
		break;
	case 2:
		if (zend_get_parameters_ex(2, &arg_pattern, &arg_options) == FAILURE) {
			WRONG_PARAM_COUNT;
		}
		convert_to_string_ex(arg_options);
		option = 0;
		_php_mb_regex_init_options(Z_STRVAL_PP(arg_options), Z_STRLEN_PP(arg_options), &option, NULL);
		break;
	default:
		WRONG_PARAM_COUNT;
		break;
	}
	if (ZEND_NUM_ARGS() > 0) {
		/* create regex pattern buffer */
		convert_to_string_ex(arg_pattern);
		if (!MBSTRG(search_re)) {
			MBSTRG(search_re) = (mb_regex_t*)ecalloc(1, sizeof(mb_regex_t));
		}
		err = php_mbregex_compile_pattern(
		    MBSTRG(search_re),
		    Z_STRVAL_PP(arg_pattern),
		    Z_STRLEN_PP(arg_pattern),
		    option, MBSTRG(current_mbctype) TSRMLS_CC);
		if (err) {
			efree(MBSTRG(search_re));
			MBSTRG(search_re) = (mb_regex_t*)0;
			RETURN_FALSE;
		}
	}

	pos = MBSTRG(search_pos);
	str = NULL;
	len = 0;
	if (MBSTRG(search_str) != NULL && Z_TYPE_PP(MBSTRG(search_str)) == IS_STRING){
		str = (unsigned char *)Z_STRVAL_PP(MBSTRG(search_str));
		len = Z_STRLEN_PP(MBSTRG(search_str));
	}

	if (!MBSTRG(search_re)) {
		php_error_docref(NULL TSRMLS_CC, E_WARNING, "No regex given");
		RETURN_FALSE;
	}
	if (!str) {
		php_error_docref(NULL TSRMLS_CC, E_WARNING, "No string given");
		RETURN_FALSE;
	}
	if (MBSTRG(search_regs)) {
		mbre_free_registers(MBSTRG(search_regs));
		memset(MBSTRG(search_regs), 0, sizeof(struct mbre_registers));
	} else {
		MBSTRG(search_regs) = (struct mbre_registers*)ecalloc(1, sizeof(struct mbre_registers));
	}

	err = mbre_search(MBSTRG(search_re), (const char *)str, len, pos, len - pos, MBSTRG(search_regs));
	if (err <= -2) {
		php_error_docref(NULL TSRMLS_CC, E_WARNING, "mbregex search failure in mbregex_search()");
		RETVAL_FALSE;
	} else if (err < 0) {
		MBSTRG(search_pos) = len;
		RETVAL_FALSE;
	} else {
		if (MBSTRG(search_regs)->beg[0] == MBSTRG(search_regs)->end[0]) {
			php_error_docref(NULL TSRMLS_CC, E_WARNING, "Empty regular expression");
		}
		switch (mode) {
		case 1:
			if (array_init(return_value) != FAILURE) {
				beg = MBSTRG(search_regs)->beg[0];
				end = MBSTRG(search_regs)->end[0];
				add_next_index_long(return_value, beg);
				add_next_index_long(return_value, end - beg);
			} else {
				RETVAL_FALSE;
			}
			break;
		case 2:
			if (array_init(return_value) != FAILURE) {
				n = MBSTRG(search_regs)->num_regs;
				for (i = 0; i < n; i++) {
					beg = MBSTRG(search_regs)->beg[i];
					end = MBSTRG(search_regs)->end[i];
					if (beg >= 0 && beg <= end && end <= len) {
						add_index_stringl(return_value, i, (char *)&str[beg], end - beg, 1);
					} else {
						add_index_bool(return_value, i, 0);
					}
				}
			} else {
				RETVAL_FALSE;
			}
			break;
		default:
			RETVAL_TRUE;
			break;
		}
		end = MBSTRG(search_regs)->end[0];
		if (pos < end) {
			MBSTRG(search_pos) = end;
		} else {
			MBSTRG(search_pos) = pos + 1;
		}
	}

	if (err < 0) {
		mbre_free_registers(MBSTRG(search_regs));
		efree(MBSTRG(search_regs));
		MBSTRG(search_regs) = (struct mbre_registers*)0;
	}
}
/* }}} */


/* {{{ proto bool mb_ereg_search([string pattern[, string option]])
   Regular expression search for multibyte string */
PHP_FUNCTION(mb_ereg_search)
{
	_php_mb_regex_ereg_search_exec(INTERNAL_FUNCTION_PARAM_PASSTHRU, 0);
}
/* }}} */


/* {{{ proto array mb_ereg_search_pos([string pattern[, string option]])
   Regular expression search for multibyte string */
PHP_FUNCTION(mb_ereg_search_pos)
{
	_php_mb_regex_ereg_search_exec(INTERNAL_FUNCTION_PARAM_PASSTHRU, 1);
}
/* }}} */


/* {{{ proto array mb_ereg_search_regs([string pattern[, string option]])
   Regular expression search for multibyte string */
PHP_FUNCTION(mb_ereg_search_regs)
{
	_php_mb_regex_ereg_search_exec(INTERNAL_FUNCTION_PARAM_PASSTHRU, 2);
}
/* }}} */


/* {{{ proto bool mb_ereg_search_init(string string [, string pattern[, string option]])
   Initialize string and regular expression for search. */
PHP_FUNCTION(mb_ereg_search_init)
{
	zval **arg_str, **arg_pattern, **arg_options;
	int err, option;

	option = MBSTRG(regex_default_options);
	switch (ZEND_NUM_ARGS()) {
	case 1:
		if (zend_get_parameters_ex(1, &arg_str) == FAILURE) {
			WRONG_PARAM_COUNT;
		}
		break;
	case 2:
		if (zend_get_parameters_ex(2, &arg_str, &arg_pattern) == FAILURE) {
			WRONG_PARAM_COUNT;
		}
		break;
	case 3:
		if (zend_get_parameters_ex(3, &arg_str, &arg_pattern, &arg_options) == FAILURE) {
			WRONG_PARAM_COUNT;
		}
		convert_to_string_ex(arg_options);
		option = 0;
		_php_mb_regex_init_options(Z_STRVAL_PP(arg_options), Z_STRLEN_PP(arg_options), &option, NULL);
		break;
	default:
		WRONG_PARAM_COUNT;
		break;
	}
	if (ZEND_NUM_ARGS() > 1) {
		/* create regex pattern buffer */
		convert_to_string_ex(arg_pattern);
		if (!MBSTRG(search_re)) {
			MBSTRG(search_re) = (mb_regex_t*)ecalloc(1, sizeof(mb_regex_t));
		}
		err = php_mbregex_compile_pattern(
		    MBSTRG(search_re),
		    Z_STRVAL_PP(arg_pattern),
		    Z_STRLEN_PP(arg_pattern),
		    option, MBSTRG(current_mbctype) TSRMLS_CC);
		if (err) {
			efree(MBSTRG(search_re));
			MBSTRG(search_re) = (mb_regex_t*)0;
			RETURN_FALSE;
		}
	}

	if (MBSTRG(search_str)) {
		if (ZVAL_REFCOUNT(*MBSTRG(search_str)) > 1) {
			ZVAL_DELREF(*MBSTRG(search_str));
		} else {
			zval_dtor(*MBSTRG(search_str));
			FREE_ZVAL(*MBSTRG(search_str));
		}
		MBSTRG(search_str) = (zval **)0;
		MBSTRG(search_str_val) = (zval *)0;
	}

	if (PZVAL_IS_REF(*arg_str)) {
		ZVAL_ADDREF(*arg_str);
		MBSTRG(search_str_val) = *arg_str;
		MBSTRG(search_str) = &MBSTRG(search_str_val);
	} else {
		MAKE_STD_ZVAL(MBSTRG(search_str_val));
		*MBSTRG(search_str_val) = **arg_str;
		zval_copy_ctor(MBSTRG(search_str_val));
		MBSTRG(search_str_val)->refcount = 1;
		MBSTRG(search_str_val)->is_ref = 0;
		MBSTRG(search_str) = &MBSTRG(search_str_val);
		convert_to_string_ex(MBSTRG(search_str));
	}

	MBSTRG(search_pos) = 0;

	if (MBSTRG(search_regs)) {
		mbre_free_registers(MBSTRG(search_regs));
		efree(MBSTRG(search_regs));
		MBSTRG(search_regs) = (struct mbre_registers*)0;
	}

	RETURN_TRUE;
}
/* }}} */


/* {{{ proto array mb_ereg_search_getregs(void)
   Get matched substring of the last time */
PHP_FUNCTION(mb_ereg_search_getregs)
{
	int n, i, len, beg, end;
	unsigned char *str;

	if (MBSTRG(search_regs) && Z_TYPE_PP(MBSTRG(search_str)) == IS_STRING &&
	    Z_STRVAL_PP(MBSTRG(search_str)) && array_init(return_value) != FAILURE) {
		str = (unsigned char *)Z_STRVAL_PP(MBSTRG(search_str));
		len = Z_STRLEN_PP(MBSTRG(search_str));
		n = MBSTRG(search_regs)->num_regs;
		for (i = 0; i < n; i++) {
			beg = MBSTRG(search_regs)->beg[i];
			end = MBSTRG(search_regs)->end[i];
			if (beg >= 0 && beg <= end && end <= len) {
				add_index_stringl(return_value, i, (char *)&str[beg], end - beg, 1);
			} else {
				add_index_bool(return_value, i, 0);
			}
		}
	} else {
		RETVAL_FALSE;
	}
}
/* }}} */


/* {{{ proto int mb_ereg_search_getpos(void)
   Get search start position */
PHP_FUNCTION(mb_ereg_search_getpos)
{
	RETVAL_LONG(MBSTRG(search_pos));
}
/* }}} */


/* {{{ proto bool mb_ereg_search_setpos(int position)
   Set search start position */
PHP_FUNCTION(mb_ereg_search_setpos)
{
	zval **arg_pos;
	int n;

	if (ZEND_NUM_ARGS() != 1 || zend_get_parameters_ex(1, &arg_pos) == FAILURE) {
		WRONG_PARAM_COUNT;
	}
	convert_to_long_ex(arg_pos);
	n = Z_LVAL_PP(arg_pos);
	if (n < 0
	    || ( MBSTRG(search_str) != NULL && *MBSTRG(search_str) != NULL &&
	         Z_TYPE_PP(MBSTRG(search_str)) == IS_STRING && 
	         n >= Z_STRLEN_PP(MBSTRG(search_str)) ) ) {
		php_error_docref(NULL TSRMLS_CC, E_WARNING, "Position is out of range");
		MBSTRG(search_pos) = 0;
		RETVAL_FALSE;
	} else {
		MBSTRG(search_pos) = n;
		RETVAL_TRUE;
	}
}
/* }}} */

/* {{{ php_mb_regex_set_options */
int php_mb_regex_set_options( int options TSRMLS_DC) 
{
	int prev_opt = MBSTRG(regex_default_options);
	MBSTRG(regex_default_options) = options;
	return prev_opt;
}
/* }}} */

/* {{{ php_mb_regex_set_options_by_string */
int php_mb_regex_set_options_by_string( const char *opt_str, int len TSRMLS_DC)
{
	int new_opt = 0;
	_php_mb_regex_init_options( opt_str, len, &new_opt, NULL);
	return php_mb_regex_set_options( new_opt TSRMLS_CC);
}
/* }}} */

/* {{{ proto string mb_regex_set_options([string options])
   Set or get the default options for mbregex functions */
PHP_FUNCTION(mb_regex_set_options)
{
	int opt;
	char *string = NULL;
	int string_len;
	char buf[16];

	if ( zend_parse_parameters( ZEND_NUM_ARGS() TSRMLS_CC, "|s",
	                            &string, &string_len ) == FAILURE ) {
		RETURN_FALSE;
	}
	if (string != NULL) {
		opt = php_mb_regex_set_options_by_string( (const char*)string,
		                                          string_len TSRMLS_CC );
	} else {
		opt = MBSTRG(regex_default_options);
	}
	_php_mb_regex_get_option_string(buf, sizeof(buf), opt);

	RETVAL_STRING(buf, 1);
}
/* }}} */

#endif	/* HAVE_MBREGEX */

/*
 * Local variables:
 * tab-width: 4
 * c-basic-offset: 4
 * End:
 * vim600: fdm=marker
 * vim: noet sw=4 ts=4
 */