html.c

php-4.4.7学习linux时下载的源代码

	}

	/* try to detect the charset for the locale */
#if HAVE_NL_LANGINFO && HAVE_LOCALE_H && defined(CODESET)
	charset_hint = nl_langinfo(CODESET);
	if (charset_hint != NULL && (len=strlen(charset_hint)) != 0) {
		goto det_charset;
	}
#endif

#if HAVE_LOCALE_H
	/* try to figure out the charset from the locale */
	{
		char *localename;
		char *dot, *at;

		/* lang[_territory][.codeset][@modifier] */
		localename = setlocale(LC_CTYPE, NULL);

		dot = strchr(localename, '.');
		if (dot) {
			dot++;
			/* locale specifies a codeset */
			at = strchr(dot, '@');
			if (at)
				len = at - dot;
			else
				len = strlen(dot);
			charset_hint = dot;
		} else {
			/* no explicit name; see if the name itself
			 * is the charset */
			charset_hint = localename;
			len = strlen(charset_hint);
		}
	}
#endif

det_charset:

	if (charset_hint) {
		int found = 0;
		
		/* now walk the charset map and look for the codeset */
		for (i = 0; charset_map[i].codeset; i++) {
			if (strncasecmp(charset_hint, charset_map[i].codeset, len) == 0) {
				charset = charset_map[i].charset;
				found = 1;
				break;
			}
		}
		if (!found) {
			php_error_docref(NULL TSRMLS_CC, E_WARNING, "charset `%s' not supported, assuming iso-8859-1",
					charset_hint);
		}
	}
	if (uf_result != NULL) {
		zval_ptr_dtor(&uf_result);
	}
	return charset;
}
/* }}} */

/* {{{ php_unescape_html_entities
 */
PHPAPI char *php_unescape_html_entities(unsigned char *old, int oldlen, int *newlen, int all, int quote_style, char *hint_charset TSRMLS_DC)
{
	int retlen;
	int j, k;
	char *replaced, *ret;
	enum entity_charset charset = determine_charset(hint_charset TSRMLS_CC);
	unsigned char replacement[15];
	
	ret = estrndup(old, oldlen);
	retlen = oldlen;
	if (!retlen) {
		goto empty_source;
	}
	
	if (all) {
		/* look for a match in the maps for this charset */
		for (j = 0; entity_map[j].charset != cs_terminator; j++) {
			if (entity_map[j].charset != charset)
				continue;

			for (k = entity_map[j].basechar; k <= entity_map[j].endchar; k++) {
				unsigned char entity[32];
				int entity_length = 0;

				if (entity_map[j].table[k - entity_map[j].basechar] == NULL)
					continue;
			
				
				entity[0] = '&';
				entity_length = strlen(entity_map[j].table[k - entity_map[j].basechar]);
				strncpy(&entity[1], entity_map[j].table[k - entity_map[j].basechar], sizeof(entity) - 2);
				entity[entity_length+1] = ';';
				entity[entity_length+2] = '\0';
				entity_length += 2;

				/* When we have MBCS entities in the tables above, this will need to handle it */
				if (k > 0xff) {
					zend_error(E_WARNING, "cannot yet handle MBCS in html_entity_decode()!");
				}
				replacement[0] = k;
				replacement[1] = '\0';

				replaced = php_str_to_str(ret, retlen, entity, entity_length, replacement, 1, &retlen);
				efree(ret);
				ret = replaced;
			}
		}
	}

	for (j = 0; basic_entities[j].charcode != 0; j++) {

		if (basic_entities[j].flags && (quote_style & basic_entities[j].flags) == 0)
			continue;
		
		replacement[0] = (unsigned char)basic_entities[j].charcode;
		replacement[1] = '\0';
		
		replaced = php_str_to_str(ret, retlen, basic_entities[j].entity, basic_entities[j].entitylen, replacement, 1, &retlen);
		efree(ret);
		ret = replaced;
	}
empty_source:	
	*newlen = retlen;
	return ret;
}
/* }}} */




/* {{{ php_escape_html_entities
 */
PHPAPI char *php_escape_html_entities(unsigned char *old, int oldlen, int *newlen, int all, int quote_style, char *hint_charset TSRMLS_DC)
{
	int i, j, maxlen, len;
	char *replaced;
	enum entity_charset charset = determine_charset(hint_charset TSRMLS_CC);
	int matches_map;

	maxlen = 2 * oldlen;
	if (maxlen < 128)
		maxlen = 128;
	replaced = emalloc (maxlen);
	len = 0;

	i = 0;
	while (i < oldlen) {
		unsigned char mbsequence[16];	/* allow up to 15 characters in a multibyte sequence */
		int mbseqlen = sizeof(mbsequence);
		unsigned short this_char = get_next_char(charset, old, &i, mbsequence, &mbseqlen);

		matches_map = 0;

		if (len + 16 > maxlen)
			replaced = erealloc (replaced, maxlen += 128);

		if (all) {
			/* look for a match in the maps for this charset */
			unsigned char *rep = NULL;


			for (j = 0; entity_map[j].charset != cs_terminator; j++) {
				if (entity_map[j].charset == charset
						&& this_char >= entity_map[j].basechar
						&& this_char <= entity_map[j].endchar) {
					rep = (unsigned char*)entity_map[j].table[this_char - entity_map[j].basechar];
					if (rep == NULL) {
						/* there is no entity for this position; fall through and
						 * just output the character itself */
						break;
					}

					matches_map = 1;
					break;
				}
			}

			if (matches_map) {
				int l = strlen(rep);
				/* increase the buffer size */
				if (len + 2 + l >= maxlen) {
					replaced = erealloc(replaced, maxlen += 128);
				}

				replaced[len++] = '&';
				strcpy(replaced + len, rep);
				len += l;
				replaced[len++] = ';';
			}
		}
		if (!matches_map) {	
			int is_basic = 0;

			for (j = 0; basic_entities[j].charcode != 0; j++) {
				if ((basic_entities[j].charcode != this_char) ||
					(basic_entities[j].flags && (quote_style & basic_entities[j].flags) == 0))
					continue;

				memcpy(replaced + len, basic_entities[j].entity, basic_entities[j].entitylen);
				len += basic_entities[j].entitylen;
				
				is_basic = 1;
				break;

			}
			if (!is_basic) {
				if (mbseqlen > 1) {
					/* a wide char without a named entity; pass through the original sequence */
					memcpy(replaced + len, mbsequence, mbseqlen);
					len += mbseqlen;

				} else {
					replaced[len++] = (unsigned char)this_char;
				}
			}
		}
	}
	replaced[len] = '\0';
	*newlen = len;

	return replaced;


}
/* }}} */

/* {{{ php_html_entities
 */
static void php_html_entities(INTERNAL_FUNCTION_PARAMETERS, int all)
{
	char *str, *hint_charset = NULL;
	int str_len, hint_charset_len = 0;
	int len;
	long quote_style = ENT_COMPAT;
	char *replaced;

	if (zend_parse_parameters(ZEND_NUM_ARGS() TSRMLS_CC, "s|ls", &str, &str_len,
							  &quote_style, &hint_charset, &hint_charset_len) == FAILURE) {
		return;
	}

	replaced = php_escape_html_entities(str, str_len, &len, all, quote_style, hint_charset TSRMLS_CC);
	RETVAL_STRINGL(replaced, len, 0);
}
/* }}} */

#define HTML_SPECIALCHARS 	0
#define HTML_ENTITIES	 	1

/* {{{ register_html_constants
 */
void register_html_constants(INIT_FUNC_ARGS)
{
	REGISTER_LONG_CONSTANT("HTML_SPECIALCHARS", HTML_SPECIALCHARS, CONST_PERSISTENT|CONST_CS);
	REGISTER_LONG_CONSTANT("HTML_ENTITIES", HTML_ENTITIES, CONST_PERSISTENT|CONST_CS);
	REGISTER_LONG_CONSTANT("ENT_COMPAT", ENT_COMPAT, CONST_PERSISTENT|CONST_CS);
	REGISTER_LONG_CONSTANT("ENT_QUOTES", ENT_QUOTES, CONST_PERSISTENT|CONST_CS);
	REGISTER_LONG_CONSTANT("ENT_NOQUOTES", ENT_NOQUOTES, CONST_PERSISTENT|CONST_CS);
}
/* }}} */

/* {{{ proto string htmlspecialchars(string string [, int quote_style][, string charset])
   Convert special characters to HTML entities */
PHP_FUNCTION(htmlspecialchars)
{
	php_html_entities(INTERNAL_FUNCTION_PARAM_PASSTHRU, 0);
}
/* }}} */

/* {{{ proto string html_entity_decode(string string [, int quote_style][, string charset])
   Convert all HTML entities to their applicable characters */
PHP_FUNCTION(html_entity_decode)
{
	char *str, *hint_charset = NULL;
	int str_len, hint_charset_len, len; 
	long quote_style = ENT_COMPAT;
	char *replaced;

	if (zend_parse_parameters(ZEND_NUM_ARGS() TSRMLS_CC, "s|ls", &str, &str_len,
							  &quote_style, &hint_charset, &hint_charset_len) == FAILURE) {
		return;
	}

	replaced = php_unescape_html_entities(str, str_len, &len, 1, quote_style, hint_charset TSRMLS_CC);
	RETVAL_STRINGL(replaced, len, 0);
}
/* }}} */


/* {{{ proto string htmlentities(string string [, int quote_style][, string charset])
   Convert all applicable characters to HTML entities */
PHP_FUNCTION(htmlentities)
{
	php_html_entities(INTERNAL_FUNCTION_PARAM_PASSTHRU, 1);
}
/* }}} */

/* {{{ proto array get_html_translation_table([int table [, int quote_style]])
   Returns the internal translation table used by htmlspecialchars and htmlentities */
PHP_FUNCTION(get_html_translation_table)
{
	long which = HTML_SPECIALCHARS, quote_style = ENT_COMPAT;
	int i, j;
	char ind[2];
	enum entity_charset charset = determine_charset(NULL TSRMLS_CC);

	if (zend_parse_parameters(ZEND_NUM_ARGS() TSRMLS_CC, "|ll", &which, &quote_style) == FAILURE) {
		return;
	}

	array_init(return_value);

	ind[1] = 0;

	switch (which) {
		case HTML_ENTITIES:
			for (j=0; entity_map[j].charset != cs_terminator; j++) {
				if (entity_map[j].charset != charset)
					continue;
				for (i = 0; i <= entity_map[j].endchar - entity_map[j].basechar; i++) {
					char buffer[16];

					if (entity_map[j].table[i] == NULL)
						continue;
					/* what about wide chars here ?? */
					ind[0] = i + entity_map[j].basechar;
					sprintf(buffer, "&%s;", entity_map[j].table[i]);
					add_assoc_string(return_value, ind, buffer, 1);

				}
			}
			/* break thru */

		case HTML_SPECIALCHARS:
			for (j = 0; basic_entities[j].charcode != 0; j++) {

				if (basic_entities[j].flags && (quote_style & basic_entities[j].flags) == 0)
					continue;
				
				ind[0] = (unsigned char)basic_entities[j].charcode;
				add_assoc_string(return_value, ind, basic_entities[j].entity, 1);
			}
			break;
	}
}
/* }}} */

/*
 * Local variables:
 * tab-width: 4
 * c-basic-offset: 4
 * End:
 * vim600: sw=4 ts=4 fdm=marker
 * vim<600: sw=4 ts=4
 */