/* +----------------------------------------------------------------------+ | PHP Version 5 | +----------------------------------------------------------------------+ | Copyright (c) 1997-2015 The PHP Group | +----------------------------------------------------------------------+ | This source file is subject to version 3.01 of the PHP license, | | that is bundled with this package in the file LICENSE, and is | | available through the world-wide-web at the following url: | | http://www.php.net/license/3_01.txt | | If you did not receive a copy of the PHP license and are unable to | | obtain it through the world-wide-web, please send a note to | | license@php.net so we can mail you a copy immediately. | +----------------------------------------------------------------------+ | Authors: Rasmus Lerdorf | | Jaakko Hyvätti | | Wez Furlong | | Gustavo Lopes | +----------------------------------------------------------------------+ */ /* $Id$ */ /* * HTML entity resources: * * http://www.unicode.org/Public/MAPPINGS/OBSOLETE/UNI2SGML.TXT * * XHTML 1.0 DTD * http://www.w3.org/TR/2002/REC-xhtml1-20020801/dtds.html#h-A2 * * From HTML 4.01 strict DTD: * http://www.w3.org/TR/html4/HTMLlat1.ent * http://www.w3.org/TR/html4/HTMLsymbol.ent * http://www.w3.org/TR/html4/HTMLspecial.ent * * HTML 5: * http://dev.w3.org/html5/spec/Overview.html#named-character-references */ #include "php.h" #if PHP_WIN32 #include "config.w32.h" #else #include #endif #include "php_standard.h" #include "php_string.h" #include "SAPI.h" #if HAVE_LOCALE_H #include #endif #if HAVE_LANGINFO_H #include #endif #include #include "html_tables.h" /* Macro for disabling flag of translation of non-basic entities where this isn't supported. * Not appropriate for html_entity_decode/htmlspecialchars_decode */ #define LIMIT_ALL(all, doctype, charset) do { \ (all) = (all) && !CHARSET_PARTIAL_SUPPORT((charset)) && ((doctype) != ENT_HTML_DOC_XML1); \ } while (0) #define MB_FAILURE(pos, advance) do { \ *cursor = pos + (advance); \ *status = FAILURE; \ return 0; \ } while (0) #define CHECK_LEN(pos, chars_need) ((str_len - (pos)) >= (chars_need)) /* valid as single byte character or leading byte */ #define utf8_lead(c) ((c) < 0x80 || ((c) >= 0xC2 && (c) <= 0xF4)) /* whether it's actually valid depends on other stuff; * this macro cannot check for non-shortest forms, surrogates or * code points above 0x10FFFF */ #define utf8_trail(c) ((c) >= 0x80 && (c) <= 0xBF) #define gb2312_lead(c) ((c) != 0x8E && (c) != 0x8F && (c) != 0xA0 && (c) != 0xFF) #define gb2312_trail(c) ((c) >= 0xA1 && (c) <= 0xFE) #define sjis_lead(c) ((c) != 0x80 && (c) != 0xA0 && (c) < 0xFD) #define sjis_trail(c) ((c) >= 0x40 && (c) != 0x7F && (c) < 0xFD) /* {{{ get_default_charset */ static char *get_default_charset(TSRMLS_D) { if (PG(internal_encoding) && PG(internal_encoding)[0]) { return PG(internal_encoding); } else if (SG(default_charset) && SG(default_charset)[0] ) { return SG(default_charset); } return NULL; } /* }}} */ /* {{{ get_next_char */ static inline unsigned int get_next_char( enum entity_charset charset, const unsigned char *str, size_t str_len, size_t *cursor, int *status) { size_t pos = *cursor; unsigned int this_char = 0; *status = SUCCESS; assert(pos <= str_len); if (!CHECK_LEN(pos, 1)) MB_FAILURE(pos, 1); switch (charset) { case cs_utf_8: { /* We'll follow strategy 2. from section 3.6.1 of UTR #36: * "In a reported illegal byte sequence, do not include any * non-initial byte that encodes a valid character or is a leading * byte for a valid sequence." */ unsigned char c; c = str[pos]; if (c < 0x80) { this_char = c; pos++; } else if (c < 0xc2) { MB_FAILURE(pos, 1); } else if (c < 0xe0) { if (!CHECK_LEN(pos, 2)) MB_FAILURE(pos, 1); if (!utf8_trail(str[pos + 1])) { MB_FAILURE(pos, utf8_lead(str[pos + 1]) ? 1 : 2); } this_char = ((c & 0x1f) << 6) | (str[pos + 1] & 0x3f); if (this_char < 0x80) { /* non-shortest form */ MB_FAILURE(pos, 2); } pos += 2; } else if (c < 0xf0) { size_t avail = str_len - pos; if (avail < 3 || !utf8_trail(str[pos + 1]) || !utf8_trail(str[pos + 2])) { if (avail < 2 || utf8_lead(str[pos + 1])) MB_FAILURE(pos, 1); else if (avail < 3 || utf8_lead(str[pos + 2])) MB_FAILURE(pos, 2); else MB_FAILURE(pos, 3); } this_char = ((c & 0x0f) << 12) | ((str[pos + 1] & 0x3f) << 6) | (str[pos + 2] & 0x3f); if (this_char < 0x800) { /* non-shortest form */ MB_FAILURE(pos, 3); } else if (this_char >= 0xd800 && this_char <= 0xdfff) { /* surrogate */ MB_FAILURE(pos, 3); } pos += 3; } else if (c < 0xf5) { size_t avail = str_len - pos; if (avail < 4 || !utf8_trail(str[pos + 1]) || !utf8_trail(str[pos + 2]) || !utf8_trail(str[pos + 3])) { if (avail < 2 || utf8_lead(str[pos + 1])) MB_FAILURE(pos, 1); else if (avail < 3 || utf8_lead(str[pos + 2])) MB_FAILURE(pos, 2); else if (avail < 4 || utf8_lead(str[pos + 3])) MB_FAILURE(pos, 3); else MB_FAILURE(pos, 4); } this_char = ((c & 0x07) << 18) | ((str[pos + 1] & 0x3f) << 12) | ((str[pos + 2] & 0x3f) << 6) | (str[pos + 3] & 0x3f); if (this_char < 0x10000 || this_char > 0x10FFFF) { /* non-shortest form or outside range */ MB_FAILURE(pos, 4); } pos += 4; } else { MB_FAILURE(pos, 1); } } break; case cs_big5: /* reference http://demo.icu-project.org/icu-bin/convexp?conv=big5 */ { unsigned char c = str[pos]; if (c >= 0x81 && c <= 0xFE) { unsigned char next; if (!CHECK_LEN(pos, 2)) MB_FAILURE(pos, 1); next = str[pos + 1]; if ((next >= 0x40 && next <= 0x7E) || (next >= 0xA1 && next <= 0xFE)) { this_char = (c << 8) | next; } else { MB_FAILURE(pos, 1); } pos += 2; } else { this_char = c; pos += 1; } } break; case cs_big5hkscs: { unsigned char c = str[pos]; if (c >= 0x81 && c <= 0xFE) { unsigned char next; if (!CHECK_LEN(pos, 2)) MB_FAILURE(pos, 1); next = str[pos + 1]; if ((next >= 0x40 && next <= 0x7E) || (next >= 0xA1 && next <= 0xFE)) { this_char = (c << 8) | next; } else if (next != 0x80 && next != 0xFF) { MB_FAILURE(pos, 1); } else { MB_FAILURE(pos, 2); } pos += 2; } else { this_char = c; pos += 1; } } break; case cs_gb2312: /* EUC-CN */ { unsigned char c = str[pos]; if (c >= 0xA1 && c <= 0xFE) { unsigned char next; if (!CHECK_LEN(pos, 2)) MB_FAILURE(pos, 1); next = str[pos + 1]; if (gb2312_trail(next)) { this_char = (c << 8) | next; } else if (gb2312_lead(next)) { MB_FAILURE(pos, 1); } else { MB_FAILURE(pos, 2); } pos += 2; } else if (gb2312_lead(c)) { this_char = c; pos += 1; } else { MB_FAILURE(pos, 1); } } break; case cs_sjis: { unsigned char c = str[pos]; if ((c >= 0x81 && c <= 0x9F) || (c >= 0xE0 && c <= 0xFC)) { unsigned char next; if (!CHECK_LEN(pos, 2)) MB_FAILURE(pos, 1); next = str[pos + 1]; if (sjis_trail(next)) { this_char = (c << 8) | next; } else if (sjis_lead(next)) { MB_FAILURE(pos, 1); } else { MB_FAILURE(pos, 2); } pos += 2; } else if (c < 0x80 || (c >= 0xA1 && c <= 0xDF)) { this_char = c; pos += 1; } else { MB_FAILURE(pos, 1); } } break; case cs_eucjp: { unsigned char c = str[pos]; if (c >= 0xA1 && c <= 0xFE) { unsigned next; if (!CHECK_LEN(pos, 2)) MB_FAILURE(pos, 1); next = str[pos + 1]; if (next >= 0xA1 && next <= 0xFE) { /* this a jis kanji char */ this_char = (c << 8) | next; } else { MB_FAILURE(pos, (next != 0xA0 && next != 0xFF) ? 1 : 2); } pos += 2; } else if (c == 0x8E) { unsigned next; if (!CHECK_LEN(pos, 2)) MB_FAILURE(pos, 1); next = str[pos + 1]; if (next >= 0xA1 && next <= 0xDF) { /* JIS X 0201 kana */ this_char = (c << 8) | next; } else { MB_FAILURE(pos, (next != 0xA0 && next != 0xFF) ? 1 : 2); } pos += 2; } else if (c == 0x8F) { size_t avail = str_len - pos; if (avail < 3 || !(str[pos + 1] >= 0xA1 && str[pos + 1] <= 0xFE) || !(str[pos + 2] >= 0xA1 && str[pos + 2] <= 0xFE)) { if (avail < 2 || (str[pos + 1] != 0xA0 && str[pos + 1] != 0xFF)) MB_FAILURE(pos, 1); else if (avail < 3 || (str[pos + 2] != 0xA0 && str[pos + 2] != 0xFF)) MB_FAILURE(pos, 2); else MB_FAILURE(pos, 3); } else { /* JIS X 0212 hojo-kanji */ this_char = (c << 16) | (str[pos + 1] << 8) | str[pos + 2]; } pos += 3; } else if (c != 0xA0 && c != 0xFF) { /* character encoded in 1 code unit */ this_char = c; pos += 1; } else { MB_FAILURE(pos, 1); } } break; default: /* single-byte charsets */ this_char = str[pos++]; break; } *cursor = pos; return this_char; } /* }}} */ /* {{{ php_next_utf8_char * Public interface for get_next_char used with UTF-8 */ PHPAPI unsigned int php_next_utf8_char( const unsigned char *str, size_t str_len, size_t *cursor, int *status) { return get_next_char(cs_utf_8, str, str_len, cursor, status); } /* }}} */ /* {{{ entity_charset determine_charset * returns the charset identifier based on current locale or a hint. * defaults to UTF-8 */ static enum entity_charset determine_charset(char *charset_hint TSRMLS_DC) { int i; enum entity_charset charset = cs_utf_8; int len = 0; const zend_encoding *zenc; /* Default is now UTF-8 */ if (charset_hint == NULL) return cs_utf_8; if ((len = strlen(charset_hint)) != 0) { goto det_charset; } zenc = zend_multibyte_get_internal_encoding(TSRMLS_C); if (zenc != NULL) { charset_hint = (char *)zend_multibyte_get_encoding_name(zenc); if (charset_hint != NULL && (len=strlen(charset_hint)) != 0) { if ((len == 4) /* sizeof (none|auto|pass) */ && (!memcmp("pass", charset_hint, 4) || !memcmp("auto", charset_hint, 4) || !memcmp("none", charset_hint, 4))) { charset_hint = NULL; len = 0; } else { goto det_charset; } } } charset_hint = SG(default_charset); if (charset_hint != NULL && (len=strlen(charset_hint)) != 0) { goto det_charset; } /* 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 (len == strlen(charset_map[i].codeset) && 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 utf-8", charset_hint); } } return charset; } /* }}} */ /* {{{ php_utf32_utf8 */ static inline size_t php_utf32_utf8(unsigned char *buf, unsigned k) { size_t retval = 0; /* assert(0x0 <= k <= 0x10FFFF); */ if (k < 0x80) { buf[0] = k; retval = 1; } else if (k < 0x800) { buf[0] = 0xc0 | (k >> 6); buf[1] = 0x80 | (k & 0x3f); retval = 2; } else if (k < 0x10000) { buf[0] = 0xe0 | (k >> 12); buf[1] = 0x80 | ((k >> 6) & 0x3f); buf[2] = 0x80 | (k & 0x3f); retval = 3; } else { buf[0] = 0xf0 | (k >> 18); buf[1] = 0x80 | ((k >> 12) & 0x3f); buf[2] = 0x80 | ((k >> 6) & 0x3f); buf[3] = 0x80 | (k & 0x3f); retval = 4; } /* UTF-8 has been restricted to max 4 bytes since RFC 3629 */ return retval; } /* }}} */ /* {{{ php_mb2_int_to_char * Convert back big endian int representation of sequence of one or two 8-bit code units. */ static inline size_t php_mb2_int_to_char(unsigned char *buf, unsigned k) { assert(k <= 0xFFFFU); /* one or two bytes */ if (k <= 0xFFU) { /* 1 */ buf[0] = k; return 1U; } else { /* 2 */ buf[0] = k >> 8; buf[1] = k & 0xFFU; return 2U; } } /* }}} */ /* {{{ php_mb3_int_to_char * Convert back big endian int representation of sequence of one to three 8-bit code units. * For EUC-JP. */ static inline size_t php_mb3_int_to_char(unsigned char *buf, unsigned k) { assert(k <= 0xFFFFFFU); /* one to three bytes */ if (k <= 0xFFU) { /* 1 */ buf[0] = k; return 1U; } else if (k <= 0xFFFFU) { /* 2 */ buf[0] = k >> 8; buf[1] = k & 0xFFU; return 2U; } else { buf[0] = k >> 16; buf[1] = (k >> 8) & 0xFFU; buf[2] = k & 0xFFU; return 3U; } } /* }}} */ /* {{{ unimap_bsearc_cmp * Binary search of unicode code points in unicode <--> charset mapping. * Returns the code point in the target charset (whose mapping table was given) or 0 if * the unicode code point is not in the table. */ static inline unsigned char unimap_bsearch(const uni_to_enc *table, unsigned code_key_a, size_t num) { const uni_to_enc *l = table, *h = &table[num-1], *m; unsigned short code_key; /* we have no mappings outside the BMP */ if (code_key_a > 0xFFFFU) return 0; code_key = (unsigned short) code_key_a; while (l <= h) { m = l + (h - l) / 2; if (code_key < m->un_code_point) h = m - 1; else if (code_key > m->un_code_point) l = m + 1; else return m->cs_code; } return 0; } /* }}} */ /* {{{ map_from_unicode */ static inline int map_from_unicode(unsigned code, enum entity_charset charset, unsigned *res) { unsigned char found; const uni_to_enc *table; size_t table_size; switch (charset) { case cs_8859_1: /* identity mapping of code points to unicode */ if (code > 0xFF) { return FAILURE; } *res = code; break; case cs_8859_5: if (code <= 0xA0 || code == 0xAD /* soft hyphen */) { *res = code; } else if (code == 0x2116) { *res = 0xF0; /* numero sign */ } else if (code == 0xA7) { *res = 0xFD; /* section sign */ } else if (code >= 0x0401 && code <= 0x044F) { if (code == 0x040D || code == 0x0450 || code == 0x045D) return FAILURE; *res = code - 0x360; } else { return FAILURE; } break; case cs_8859_15: if (code < 0xA4 || (code > 0xBE && code <= 0xFF)) { *res = code; } else { /* between A4 and 0xBE */ found = unimap_bsearch(unimap_iso885915, code, sizeof(unimap_iso885915) / sizeof(*unimap_iso885915)); if (found) *res = found; else return FAILURE; } break; case cs_cp1252: if (code <= 0x7F || (code >= 0xA0 && code <= 0xFF)) { *res = code; } else { found = unimap_bsearch(unimap_win1252, code, sizeof(unimap_win1252) / sizeof(*unimap_win1252)); if (found) *res = found; else return FAILURE; } break; case cs_macroman: if (code == 0x7F) return FAILURE; table = unimap_macroman; table_size = sizeof(unimap_macroman) / sizeof(*unimap_macroman); goto table_over_7F; case cs_cp1251: table = unimap_win1251; table_size = sizeof(unimap_win1251) / sizeof(*unimap_win1251); goto table_over_7F; case cs_koi8r: table = unimap_koi8r; table_size = sizeof(unimap_koi8r) / sizeof(*unimap_koi8r); goto table_over_7F; case cs_cp866: table = unimap_cp866; table_size = sizeof(unimap_cp866) / sizeof(*unimap_cp866); table_over_7F: if (code <= 0x7F) { *res = code; } else { found = unimap_bsearch(table, code, table_size); if (found) *res = found; else return FAILURE; } break; /* from here on, only map the possible characters in the ASCII range. * to improve support here, it's a matter of building the unicode mappings. * See */ case cs_sjis: case cs_eucjp: /* we interpret 0x5C as the Yen symbol. This is not universal. * See */ if (code >= 0x20 && code <= 0x7D) { if (code == 0x5C) return FAILURE; *res = code; } else { return FAILURE; } break; case cs_big5: case cs_big5hkscs: case cs_gb2312: if (code >= 0x20 && code <= 0x7D) { *res = code; } else { return FAILURE; } break; default: return FAILURE; } return SUCCESS; } /* }}} */ /* {{{ */ static inline void map_to_unicode(unsigned code, const enc_to_uni *table, unsigned *res) { /* only single byte encodings are currently supported; assumed code <= 0xFF */ *res = table->inner[ENT_ENC_TO_UNI_STAGE1(code)]->uni_cp[ENT_ENC_TO_UNI_STAGE2(code)]; } /* }}} */ /* {{{ unicode_cp_is_allowed */ static inline int unicode_cp_is_allowed(unsigned uni_cp, int document_type) { /* XML 1.0 HTML 4.01 HTML 5 * 0x09..0x0A 0x09..0x0A 0x09..0x0A * 0x0D 0x0D 0x0C..0x0D * 0x0020..0xD7FF 0x20..0x7E 0x20..0x7E * 0x00A0..0xD7FF 0x00A0..0xD7FF * 0xE000..0xFFFD 0xE000..0x10FFFF 0xE000..0xFDCF * 0x010000..0x10FFFF 0xFDF0..0x10FFFF (*) * * (*) exclude code points where ((code & 0xFFFF) >= 0xFFFE) * * References: * XML 1.0: * HTML 4.01: * HTML 5: * * Not sure this is the relevant part for HTML 5, though. I opted to * disallow the characters that would result in a parse error when * preprocessing of the input stream. See also section 8.1.3. * * It's unclear if XHTML 1.0 allows C1 characters. I'll opt to apply to * XHTML 1.0 the same rules as for XML 1.0. * See . */ switch (document_type) { case ENT_HTML_DOC_HTML401: return (uni_cp >= 0x20 && uni_cp <= 0x7E) || (uni_cp == 0x0A || uni_cp == 0x09 || uni_cp == 0x0D) || (uni_cp >= 0xA0 && uni_cp <= 0xD7FF) || (uni_cp >= 0xE000 && uni_cp <= 0x10FFFF); case ENT_HTML_DOC_HTML5: return (uni_cp >= 0x20 && uni_cp <= 0x7E) || (uni_cp >= 0x09 && uni_cp <= 0x0D && uni_cp != 0x0B) || /* form feed U+0C allowed */ (uni_cp >= 0xA0 && uni_cp <= 0xD7FF) || (uni_cp >= 0xE000 && uni_cp <= 0x10FFFF && ((uni_cp & 0xFFFF) < 0xFFFE) && /* last two of each plane (nonchars) disallowed */ (uni_cp < 0xFDD0 || uni_cp > 0xFDEF)); /* U+FDD0-U+FDEF (nonchars) disallowed */ case ENT_HTML_DOC_XHTML: case ENT_HTML_DOC_XML1: return (uni_cp >= 0x20 && uni_cp <= 0xD7FF) || (uni_cp == 0x0A || uni_cp == 0x09 || uni_cp == 0x0D) || (uni_cp >= 0xE000 && uni_cp <= 0x10FFFF && uni_cp != 0xFFFE && uni_cp != 0xFFFF); default: return 1; } } /* }}} */ /* {{{ unicode_cp_is_allowed */ static inline int numeric_entity_is_allowed(unsigned uni_cp, int document_type) { /* less restrictive than unicode_cp_is_allowed */ switch (document_type) { case ENT_HTML_DOC_HTML401: /* all non-SGML characters (those marked with UNUSED in DESCSET) should be * representable with numeric entities */ return uni_cp <= 0x10FFFF; case ENT_HTML_DOC_HTML5: /* 8.1.4. The numeric character reference forms described above are allowed to * reference any Unicode code point other than U+0000, U+000D, permanently * undefined Unicode characters (noncharacters), and control characters other * than space characters (U+0009, U+000A, U+000C and U+000D) */ /* seems to allow surrogate characters, then */ return (uni_cp >= 0x20 && uni_cp <= 0x7E) || (uni_cp >= 0x09 && uni_cp <= 0x0C && uni_cp != 0x0B) || /* form feed U+0C allowed, but not U+0D */ (uni_cp >= 0xA0 && uni_cp <= 0x10FFFF && ((uni_cp & 0xFFFF) < 0xFFFE) && /* last two of each plane (nonchars) disallowed */ (uni_cp < 0xFDD0 || uni_cp > 0xFDEF)); /* U+FDD0-U+FDEF (nonchars) disallowed */ case ENT_HTML_DOC_XHTML: case ENT_HTML_DOC_XML1: /* OTOH, XML 1.0 requires "character references to match the production for Char * See */ return unicode_cp_is_allowed(uni_cp, document_type); default: return 1; } } /* }}} */ /* {{{ process_numeric_entity * Auxiliary function to traverse_for_entities. * On input, *buf should point to the first character after # and on output, it's the last * byte read, no matter if there was success or insuccess. */ static inline int process_numeric_entity(const char **buf, unsigned *code_point) { long code_l; int hexadecimal = (**buf == 'x' || **buf == 'X'); /* TODO: XML apparently disallows "X" */ char *endptr; if (hexadecimal && (**buf != '\0')) (*buf)++; /* strtol allows whitespace and other stuff in the beginning * we're not interested */ if ((hexadecimal && !isxdigit(**buf)) || (!hexadecimal && !isdigit(**buf))) { return FAILURE; } code_l = strtol(*buf, &endptr, hexadecimal ? 16 : 10); /* we're guaranteed there were valid digits, so *endptr > buf */ *buf = endptr; if (**buf != ';') return FAILURE; /* many more are invalid, but that depends on whether it's HTML * (and which version) or XML. */ if (code_l > 0x10FFFFL) return FAILURE; if (code_point != NULL) *code_point = (unsigned)code_l; return SUCCESS; } /* }}} */ /* {{{ process_named_entity */ static inline int process_named_entity_html(const char **buf, const char **start, size_t *length) { *start = *buf; /* "&" is represented by a 0x26 in all supported encodings. That means * the byte after represents a character or is the leading byte of an * sequence of 8-bit code units. If in the ranges below, it represents * necessarily a alpha character because none of the supported encodings * has an overlap with ASCII in the leading byte (only on the second one) */ while ((**buf >= 'a' && **buf <= 'z') || (**buf >= 'A' && **buf <= 'Z') || (**buf >= '0' && **buf <= '9')) { (*buf)++; } if (**buf != ';') return FAILURE; /* cast to size_t OK as the quantity is always non-negative */ *length = *buf - *start; if (*length == 0) return FAILURE; return SUCCESS; } /* }}} */ /* {{{ resolve_named_entity_html */ static inline int resolve_named_entity_html(const char *start, size_t length, const entity_ht *ht, unsigned *uni_cp1, unsigned *uni_cp2) { const entity_cp_map *s; ulong hash = zend_inline_hash_func(start, length); s = ht->buckets[hash % ht->num_elems]; while (s->entity) { if (s->entity_len == length) { if (memcmp(start, s->entity, length) == 0) { *uni_cp1 = s->codepoint1; *uni_cp2 = s->codepoint2; return SUCCESS; } } s++; } return FAILURE; } /* }}} */ static inline size_t write_octet_sequence(unsigned char *buf, enum entity_charset charset, unsigned code) { /* code is not necessarily a unicode code point */ switch (charset) { case cs_utf_8: return php_utf32_utf8(buf, code); case cs_8859_1: case cs_cp1252: case cs_8859_15: case cs_koi8r: case cs_cp1251: case cs_8859_5: case cs_cp866: case cs_macroman: /* single byte stuff */ *buf = code; return 1; case cs_big5: case cs_big5hkscs: case cs_sjis: case cs_gb2312: /* we don't have complete unicode mappings for these yet in entity_decode, * and we opt to pass through the octet sequences for these in htmlentities * instead of converting to an int and then converting back. */ #if 0 return php_mb2_int_to_char(buf, code); #else #if ZEND_DEBUG assert(code <= 0xFFU); #endif *buf = code; return 1; #endif case cs_eucjp: #if 0 /* idem */ return php_mb2_int_to_char(buf, code); #else #if ZEND_DEBUG assert(code <= 0xFFU); #endif *buf = code; return 1; #endif default: assert(0); return 0; } } /* {{{ traverse_for_entities * Auxiliary function to php_unescape_html_entities(). * - The argument "all" determines if all numeric entities are decode or only those * that correspond to quotes (depending on quote_style). */ /* maximum expansion (factor 1.2) for HTML 5 with ≫⃒ and ≪⃒ */ /* +2 is 1 because of rest (probably unnecessary), 1 because of terminating 0 */ #define TRAVERSE_FOR_ENTITIES_EXPAND_SIZE(oldlen) ((oldlen) + (oldlen) / 5 + 2) static void traverse_for_entities( const char *old, size_t oldlen, char *ret, /* should have allocated TRAVERSE_FOR_ENTITIES_EXPAND_SIZE(olden) */ size_t *retlen, int all, int flags, const entity_ht *inv_map, enum entity_charset charset) { const char *p, *lim; char *q; int doctype = flags & ENT_HTML_DOC_TYPE_MASK; lim = old + oldlen; /* terminator address */ assert(*lim == '\0'); for (p = old, q = ret; p < lim;) { unsigned code, code2 = 0; const char *next = NULL; /* when set, next > p, otherwise possible inf loop */ /* Shift JIS, Big5 and HKSCS use multi-byte encodings where an * ASCII range byte can be part of a multi-byte sequence. * However, they start at 0x40, therefore if we find a 0x26 byte, * we're sure it represents the '&' character. */ /* assumes there are no single-char entities */ if (p[0] != '&' || (p + 3 >= lim)) { *(q++) = *(p++); continue; } /* now p[3] is surely valid and is no terminator */ /* numerical entity */ if (p[1] == '#') { next = &p[2]; if (process_numeric_entity(&next, &code) == FAILURE) goto invalid_code; /* If we're in htmlspecialchars_decode, we're only decoding entities * that represent &, <, >, " and '. Is this one of them? */ if (!all && (code > 63U || stage3_table_be_apos_00000[code].data.ent.entity == NULL)) goto invalid_code; /* are we allowed to decode this entity in this document type? * HTML 5 is the only that has a character that cannot be used in * a numeric entity but is allowed literally (U+000D). The * unoptimized version would be ... || !numeric_entity_is_allowed(code) */ if (!unicode_cp_is_allowed(code, doctype) || (doctype == ENT_HTML_DOC_HTML5 && code == 0x0D)) goto invalid_code; } else { const char *start; size_t ent_len; next = &p[1]; start = next; if (process_named_entity_html(&next, &start, &ent_len) == FAILURE) goto invalid_code; if (resolve_named_entity_html(start, ent_len, inv_map, &code, &code2) == FAILURE) { if (doctype == ENT_HTML_DOC_XHTML && ent_len == 4 && start[0] == 'a' && start[1] == 'p' && start[2] == 'o' && start[3] == 's') { /* uses html4 inv_map, which doesn't include apos;. This is a * hack to support it */ code = (unsigned) '\''; } else { goto invalid_code; } } } assert(*next == ';'); if (((code == '\'' && !(flags & ENT_HTML_QUOTE_SINGLE)) || (code == '"' && !(flags & ENT_HTML_QUOTE_DOUBLE))) /* && code2 == '\0' always true for current maps */) goto invalid_code; /* UTF-8 doesn't need mapping (ISO-8859-1 doesn't either, but * the call is needed to ensure the codepoint <= U+00FF) */ if (charset != cs_utf_8) { /* replace unicode code point */ if (map_from_unicode(code, charset, &code) == FAILURE || code2 != 0) goto invalid_code; /* not representable in target charset */ } q += write_octet_sequence(q, charset, code); if (code2) { q += write_octet_sequence(q, charset, code2); } /* jump over the valid entity; may go beyond size of buffer; np */ p = next + 1; continue; invalid_code: for (; p < next; p++) { *(q++) = *p; } } *q = '\0'; *retlen = (size_t)(q - ret); } /* }}} */ /* {{{ unescape_inverse_map */ static const entity_ht *unescape_inverse_map(int all, int flags) { int document_type = flags & ENT_HTML_DOC_TYPE_MASK; if (all) { switch (document_type) { case ENT_HTML_DOC_HTML401: case ENT_HTML_DOC_XHTML: /* but watch out for '...*/ return &ent_ht_html4; case ENT_HTML_DOC_HTML5: return &ent_ht_html5; default: return &ent_ht_be_apos; } } else { switch (document_type) { case ENT_HTML_DOC_HTML401: return &ent_ht_be_noapos; default: return &ent_ht_be_apos; } } } /* }}} */ /* {{{ determine_entity_table * Entity table to use. Note that entity tables are defined in terms of * unicode code points */ static entity_table_opt determine_entity_table(int all, int doctype) { entity_table_opt retval = {NULL}; assert(!(doctype == ENT_HTML_DOC_XML1 && all)); if (all) { retval.ms_table = (doctype == ENT_HTML_DOC_HTML5) ? entity_ms_table_html5 : entity_ms_table_html4; } else { retval.table = (doctype == ENT_HTML_DOC_HTML401) ? stage3_table_be_noapos_00000 : stage3_table_be_apos_00000; } return retval; } /* }}} */ /* {{{ php_unescape_html_entities * The parameter "all" should be true to decode all possible entities, false to decode * only the basic ones, i.e., those in basic_entities_ex + the numeric entities * that correspond to quotes. */ PHPAPI char *php_unescape_html_entities(unsigned char *old, size_t oldlen, size_t *newlen, int all, int flags, char *hint_charset TSRMLS_DC) { size_t retlen; char *ret; enum entity_charset charset; const entity_ht *inverse_map = NULL; size_t new_size = TRAVERSE_FOR_ENTITIES_EXPAND_SIZE(oldlen); if (all) { charset = determine_charset(hint_charset TSRMLS_CC); } else { charset = cs_8859_1; /* charset shouldn't matter, use ISO-8859-1 for performance */ } /* don't use LIMIT_ALL! */ if (oldlen > new_size) { /* overflow, refuse to do anything */ ret = estrndup((char*)old, oldlen); retlen = oldlen; goto empty_source; } ret = emalloc(new_size); *ret = '\0'; retlen = oldlen; if (retlen == 0) { goto empty_source; } inverse_map = unescape_inverse_map(all, flags); /* replace numeric entities */ traverse_for_entities(old, oldlen, ret, &retlen, all, flags, inverse_map, charset); empty_source: *newlen = retlen; return ret; } /* }}} */ PHPAPI char *php_escape_html_entities(unsigned char *old, size_t oldlen, size_t *newlen, int all, int flags, char *hint_charset TSRMLS_DC) { return php_escape_html_entities_ex(old, oldlen, newlen, all, flags, hint_charset, 1 TSRMLS_CC); } /* {{{ find_entity_for_char */ static inline void find_entity_for_char( unsigned int k, enum entity_charset charset, const entity_stage1_row *table, const unsigned char **entity, size_t *entity_len, unsigned char *old, size_t oldlen, size_t *cursor) { unsigned stage1_idx = ENT_STAGE1_INDEX(k); const entity_stage3_row *c; if (stage1_idx > 0x1D) { *entity = NULL; *entity_len = 0; return; } c = &table[stage1_idx][ENT_STAGE2_INDEX(k)][ENT_STAGE3_INDEX(k)]; if (!c->ambiguous) { *entity = (const unsigned char *)c->data.ent.entity; *entity_len = c->data.ent.entity_len; } else { /* peek at next char */ size_t cursor_before = *cursor; int status = SUCCESS; unsigned next_char; if (!(*cursor < oldlen)) goto no_suitable_2nd; next_char = get_next_char(charset, old, oldlen, cursor, &status); if (status == FAILURE) goto no_suitable_2nd; { const entity_multicodepoint_row *s, *e; s = &c->data.multicodepoint_table[1]; e = s - 1 + c->data.multicodepoint_table[0].leading_entry.size; /* we could do a binary search but it's not worth it since we have * at most two entries... */ for ( ; s <= e; s++) { if (s->normal_entry.second_cp == next_char) { *entity = s->normal_entry.entity; *entity_len = s->normal_entry.entity_len; return; } } } no_suitable_2nd: *cursor = cursor_before; *entity = (const unsigned char *) c->data.multicodepoint_table[0].leading_entry.default_entity; *entity_len = c->data.multicodepoint_table[0].leading_entry.default_entity_len; } } /* }}} */ /* {{{ find_entity_for_char_basic */ static inline void find_entity_for_char_basic( unsigned int k, const entity_stage3_row *table, const unsigned char **entity, size_t *entity_len) { if (k >= 64U) { *entity = NULL; *entity_len = 0; return; } *entity = table[k].data.ent.entity; *entity_len = table[k].data.ent.entity_len; } /* }}} */ /* {{{ php_escape_html_entities */ PHPAPI char *php_escape_html_entities_ex(unsigned char *old, size_t oldlen, size_t *newlen, int all, int flags, char *hint_charset, zend_bool double_encode TSRMLS_DC) { size_t cursor, maxlen, len; char *replaced; enum entity_charset charset = determine_charset(hint_charset TSRMLS_CC); int doctype = flags & ENT_HTML_DOC_TYPE_MASK; entity_table_opt entity_table; const enc_to_uni *to_uni_table = NULL; const entity_ht *inv_map = NULL; /* used for !double_encode */ /* only used if flags includes ENT_HTML_IGNORE_ERRORS or ENT_HTML_SUBSTITUTE_DISALLOWED_CHARS */ const unsigned char *replacement = NULL; size_t replacement_len = 0; if (all) { /* replace with all named entities */ if (CHARSET_PARTIAL_SUPPORT(charset)) { php_error_docref0(NULL TSRMLS_CC, E_STRICT, "Only basic entities " "substitution is supported for multi-byte encodings other than UTF-8; " "functionality is equivalent to htmlspecialchars"); } LIMIT_ALL(all, doctype, charset); } entity_table = determine_entity_table(all, doctype); if (all && !CHARSET_UNICODE_COMPAT(charset)) { to_uni_table = enc_to_uni_index[charset]; } if (!double_encode) { /* first arg is 1 because we want to identify valid named entities * even if we are only encoding the basic ones */ inv_map = unescape_inverse_map(1, flags); } if (flags & (ENT_HTML_SUBSTITUTE_ERRORS | ENT_HTML_SUBSTITUTE_DISALLOWED_CHARS)) { if (charset == cs_utf_8) { replacement = (const unsigned char*)"\xEF\xBF\xBD"; replacement_len = sizeof("\xEF\xBF\xBD") - 1; } else { replacement = (const unsigned char*)"�"; replacement_len = sizeof("�") - 1; } } /* initial estimate */ if (oldlen < 64) { maxlen = 128; } else { maxlen = 2 * oldlen; if (maxlen < oldlen) { zend_error_noreturn(E_ERROR, "Input string is too long"); return NULL; } } replaced = emalloc(maxlen + 1); /* adding 1 is safe: maxlen is even */ len = 0; cursor = 0; while (cursor < oldlen) { const unsigned char *mbsequence = NULL; size_t mbseqlen = 0, cursor_before = cursor; int status = SUCCESS; unsigned int this_char = get_next_char(charset, old, oldlen, &cursor, &status); /* guarantee we have at least 40 bytes to write. * In HTML5, entities may take up to 33 bytes */ if (len > maxlen - 40) { /* maxlen can never be smaller than 128 */ replaced = safe_erealloc(replaced, maxlen , 1, 128 + 1); maxlen += 128; } if (status == FAILURE) { /* invalid MB sequence */ if (flags & ENT_HTML_IGNORE_ERRORS) { continue; } else if (flags & ENT_HTML_SUBSTITUTE_ERRORS) { memcpy(&replaced[len], replacement, replacement_len); len += replacement_len; continue; } else { efree(replaced); *newlen = 0; return STR_EMPTY_ALLOC(); } } else { /* SUCCESS */ mbsequence = &old[cursor_before]; mbseqlen = cursor - cursor_before; } if (this_char != '&') { /* no entity on this position */ const unsigned char *rep = NULL; size_t rep_len = 0; if (((this_char == '\'' && !(flags & ENT_HTML_QUOTE_SINGLE)) || (this_char == '"' && !(flags & ENT_HTML_QUOTE_DOUBLE)))) goto pass_char_through; if (all) { /* false that CHARSET_PARTIAL_SUPPORT(charset) */ if (to_uni_table != NULL) { /* !CHARSET_UNICODE_COMPAT therefore not UTF-8; since UTF-8 * is the only multibyte encoding with !CHARSET_PARTIAL_SUPPORT, * we're using a single byte encoding */ map_to_unicode(this_char, to_uni_table, &this_char); if (this_char == 0xFFFF) /* no mapping; pass through */ goto pass_char_through; } /* the cursor may advance */ find_entity_for_char(this_char, charset, entity_table.ms_table, &rep, &rep_len, old, oldlen, &cursor); } else { find_entity_for_char_basic(this_char, entity_table.table, &rep, &rep_len); } if (rep != NULL) { replaced[len++] = '&'; memcpy(&replaced[len], rep, rep_len); len += rep_len; replaced[len++] = ';'; } else { /* we did not find an entity for this char. * check for its validity, if its valid pass it unchanged */ if (flags & ENT_HTML_SUBSTITUTE_DISALLOWED_CHARS) { if (CHARSET_UNICODE_COMPAT(charset)) { if (!unicode_cp_is_allowed(this_char, doctype)) { mbsequence = replacement; mbseqlen = replacement_len; } } else if (to_uni_table) { if (!all) /* otherwise we already did this */ map_to_unicode(this_char, to_uni_table, &this_char); if (!unicode_cp_is_allowed(this_char, doctype)) { mbsequence = replacement; mbseqlen = replacement_len; } } else { /* not a unicode code point, unless, coincidentally, it's in * the 0x20..0x7D range (except 0x5C in sjis). We know nothing * about other code points, because we have no tables. Since * Unicode code points in that range are not disallowed in any * document type, we could do nothing. However, conversion * tables frequently map 0x00-0x1F to the respective C0 code * points. Let's play it safe and admit that's the case */ if (this_char <= 0x7D && !unicode_cp_is_allowed(this_char, doctype)) { mbsequence = replacement; mbseqlen = replacement_len; } } } pass_char_through: if (mbseqlen > 1) { memcpy(replaced + len, mbsequence, mbseqlen); len += mbseqlen; } else { replaced[len++] = mbsequence[0]; } } } else { /* this_char == '&' */ if (double_encode) { encode_amp: memcpy(&replaced[len], "&", sizeof("&") - 1); len += sizeof("&") - 1; } else { /* no double encode */ /* check if entity is valid */ size_t ent_len; /* not counting & or ; */ /* peek at next char */ if (old[cursor] == '#') { /* numeric entity */ unsigned code_point; int valid; char *pos = (char*)&old[cursor+1]; valid = process_numeric_entity((const char **)&pos, &code_point); if (valid == FAILURE) goto encode_amp; if (flags & ENT_HTML_SUBSTITUTE_DISALLOWED_CHARS) { if (!numeric_entity_is_allowed(code_point, doctype)) goto encode_amp; } ent_len = pos - (char*)&old[cursor]; } else { /* named entity */ /* check for vality of named entity */ const char *start = &old[cursor], *next = start; unsigned dummy1, dummy2; if (process_named_entity_html(&next, &start, &ent_len) == FAILURE) goto encode_amp; if (resolve_named_entity_html(start, ent_len, inv_map, &dummy1, &dummy2) == FAILURE) { if (!(doctype == ENT_HTML_DOC_XHTML && ent_len == 4 && start[0] == 'a' && start[1] == 'p' && start[2] == 'o' && start[3] == 's')) { /* uses html4 inv_map, which doesn't include apos;. This is a * hack to support it */ goto encode_amp; } } } /* checks passed; copy entity to result */ /* entity size is unbounded, we may need more memory */ /* at this point maxlen - len >= 40 */ if (maxlen - len < ent_len + 2 /* & and ; */) { /* ent_len < oldlen, which is certainly <= SIZE_MAX/2 */ replaced = safe_erealloc(replaced, maxlen, 1, ent_len + 128 + 1); maxlen += ent_len + 128; } replaced[len++] = '&'; memcpy(&replaced[len], &old[cursor], ent_len); len += ent_len; replaced[len++] = ';'; cursor += ent_len + 1; } } } 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; size_t new_len; long flags = ENT_COMPAT; char *replaced; zend_bool double_encode = 1; if (zend_parse_parameters(ZEND_NUM_ARGS() TSRMLS_CC, "s|ls!b", &str, &str_len, &flags, &hint_charset, &hint_charset_len, &double_encode) == FAILURE) { return; } if (!hint_charset) { hint_charset = get_default_charset(TSRMLS_C); } replaced = php_escape_html_entities_ex(str, str_len, &new_len, all, (int) flags, hint_charset, double_encode TSRMLS_CC); RETVAL_STRINGL(replaced, (int)new_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); REGISTER_LONG_CONSTANT("ENT_IGNORE", ENT_IGNORE, CONST_PERSISTENT|CONST_CS); REGISTER_LONG_CONSTANT("ENT_SUBSTITUTE", ENT_SUBSTITUTE, CONST_PERSISTENT|CONST_CS); REGISTER_LONG_CONSTANT("ENT_DISALLOWED", ENT_DISALLOWED, CONST_PERSISTENT|CONST_CS); REGISTER_LONG_CONSTANT("ENT_HTML401", ENT_HTML401, CONST_PERSISTENT|CONST_CS); REGISTER_LONG_CONSTANT("ENT_XML1", ENT_XML1, CONST_PERSISTENT|CONST_CS); REGISTER_LONG_CONSTANT("ENT_XHTML", ENT_XHTML, CONST_PERSISTENT|CONST_CS); REGISTER_LONG_CONSTANT("ENT_HTML5", ENT_HTML5, CONST_PERSISTENT|CONST_CS); } /* }}} */ /* {{{ proto string htmlspecialchars(string string [, int quote_style[, string charset[, bool double_encode]]]) Convert special characters to HTML entities */ PHP_FUNCTION(htmlspecialchars) { php_html_entities(INTERNAL_FUNCTION_PARAM_PASSTHRU, 0); } /* }}} */ /* {{{ proto string htmlspecialchars_decode(string string [, int quote_style]) Convert special HTML entities back to characters */ PHP_FUNCTION(htmlspecialchars_decode) { char *str; int str_len; size_t new_len = 0; long quote_style = ENT_COMPAT; char *replaced; if (zend_parse_parameters(ZEND_NUM_ARGS() TSRMLS_CC, "s|l", &str, &str_len, "e_style) == FAILURE) { return; } replaced = php_unescape_html_entities(str, str_len, &new_len, 0 /*!all*/, quote_style, NULL TSRMLS_CC); if (replaced) { RETURN_STRINGL(replaced, (int)new_len, 0); } RETURN_FALSE; } /* }}} */ /* {{{ 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; size_t new_len = 0; long quote_style = ENT_COMPAT; char *replaced; if (zend_parse_parameters(ZEND_NUM_ARGS() TSRMLS_CC, "s|ls", &str, &str_len, "e_style, &hint_charset, &hint_charset_len) == FAILURE) { return; } if (!hint_charset) { hint_charset = get_default_charset(TSRMLS_C); } replaced = php_unescape_html_entities(str, str_len, &new_len, 1 /*all*/, quote_style, hint_charset TSRMLS_CC); if (replaced) { RETURN_STRINGL(replaced, (int)new_len, 0); } RETURN_FALSE; } /* }}} */ /* {{{ proto string htmlentities(string string [, int quote_style[, string charset[, bool double_encode]]]) Convert all applicable characters to HTML entities */ PHP_FUNCTION(htmlentities) { php_html_entities(INTERNAL_FUNCTION_PARAM_PASSTHRU, 1); } /* }}} */ /* {{{ write_s3row_data */ static inline void write_s3row_data( const entity_stage3_row *r, unsigned orig_cp, enum entity_charset charset, zval *arr) { char key[9] = ""; /* two unicode code points in UTF-8 */ char entity[LONGEST_ENTITY_LENGTH + 2] = {'&'}; size_t written_k1; written_k1 = write_octet_sequence(key, charset, orig_cp); if (!r->ambiguous) { size_t l = r->data.ent.entity_len; memcpy(&entity[1], r->data.ent.entity, l); entity[l + 1] = ';'; add_assoc_stringl_ex(arr, key, written_k1 + 1, entity, l + 2, 1); } else { unsigned i, num_entries; const entity_multicodepoint_row *mcpr = r->data.multicodepoint_table; if (mcpr[0].leading_entry.default_entity != NULL) { size_t l = mcpr[0].leading_entry.default_entity_len; memcpy(&entity[1], mcpr[0].leading_entry.default_entity, l); entity[l + 1] = ';'; add_assoc_stringl_ex(arr, key, written_k1 + 1, entity, l + 2, 1); } num_entries = mcpr[0].leading_entry.size; for (i = 1; i <= num_entries; i++) { size_t l, written_k2; unsigned uni_cp, spe_cp; uni_cp = mcpr[i].normal_entry.second_cp; l = mcpr[i].normal_entry.entity_len; if (!CHARSET_UNICODE_COMPAT(charset)) { if (map_from_unicode(uni_cp, charset, &spe_cp) == FAILURE) continue; /* non representable in this charset */ } else { spe_cp = uni_cp; } written_k2 = write_octet_sequence(&key[written_k1], charset, spe_cp); memcpy(&entity[1], mcpr[i].normal_entry.entity, l); entity[l + 1] = ';'; entity[l + 1] = '\0'; add_assoc_stringl_ex(arr, key, written_k1 + written_k2 + 1, entity, l + 1, 1); } } } /* }}} */ /* {{{ proto array get_html_translation_table([int table [, int flags [, string charset_hint]]]) Returns the internal translation table used by htmlspecialchars and htmlentities */ PHP_FUNCTION(get_html_translation_table) { long all = HTML_SPECIALCHARS, flags = ENT_COMPAT; int doctype; entity_table_opt entity_table; const enc_to_uni *to_uni_table = NULL; char *charset_hint = NULL; int charset_hint_len; enum entity_charset charset; /* in this function we have to jump through some loops because we're * getting the translated table from data structures that are optimized for * random access, not traversal */ if (zend_parse_parameters(ZEND_NUM_ARGS() TSRMLS_CC, "|lls", &all, &flags, &charset_hint, &charset_hint_len) == FAILURE) { return; } charset = determine_charset(charset_hint TSRMLS_CC); doctype = flags & ENT_HTML_DOC_TYPE_MASK; LIMIT_ALL(all, doctype, charset); array_init(return_value); entity_table = determine_entity_table(all, doctype); if (all && !CHARSET_UNICODE_COMPAT(charset)) { to_uni_table = enc_to_uni_index[charset]; } if (all) { /* HTML_ENTITIES (actually, any non-zero value for 1st param) */ const entity_stage1_row *ms_table = entity_table.ms_table; if (CHARSET_UNICODE_COMPAT(charset)) { unsigned i, j, k, max_i, max_j, max_k; /* no mapping to unicode required */ if (CHARSET_SINGLE_BYTE(charset)) { /* ISO-8859-1 */ max_i = 1; max_j = 4; max_k = 64; } else { max_i = 0x1E; max_j = 64; max_k = 64; } for (i = 0; i < max_i; i++) { if (ms_table[i] == empty_stage2_table) continue; for (j = 0; j < max_j; j++) { if (ms_table[i][j] == empty_stage3_table) continue; for (k = 0; k < max_k; k++) { const entity_stage3_row *r = &ms_table[i][j][k]; unsigned code; if (r->data.ent.entity == NULL) continue; code = ENT_CODE_POINT_FROM_STAGES(i, j, k); if (((code == '\'' && !(flags & ENT_HTML_QUOTE_SINGLE)) || (code == '"' && !(flags & ENT_HTML_QUOTE_DOUBLE)))) continue; write_s3row_data(r, code, charset, return_value); } } } } else { /* we have to iterate through the set of code points for this * encoding and map them to unicode code points */ unsigned i; for (i = 0; i <= 0xFF; i++) { const entity_stage3_row *r; unsigned uni_cp; /* can be done before mapping, they're invariant */ if (((i == '\'' && !(flags & ENT_HTML_QUOTE_SINGLE)) || (i == '"' && !(flags & ENT_HTML_QUOTE_DOUBLE)))) continue; map_to_unicode(i, to_uni_table, &uni_cp); r = &ms_table[ENT_STAGE1_INDEX(uni_cp)][ENT_STAGE2_INDEX(uni_cp)][ENT_STAGE3_INDEX(uni_cp)]; if (r->data.ent.entity == NULL) continue; write_s3row_data(r, i, charset, return_value); } } } else { /* we could use sizeof(stage3_table_be_apos_00000) as well */ unsigned j, numelems = sizeof(stage3_table_be_noapos_00000) / sizeof(*stage3_table_be_noapos_00000); for (j = 0; j < numelems; j++) { const entity_stage3_row *r = &entity_table.table[j]; if (r->data.ent.entity == NULL) continue; if (((j == '\'' && !(flags & ENT_HTML_QUOTE_SINGLE)) || (j == '"' && !(flags & ENT_HTML_QUOTE_DOUBLE)))) continue; /* charset is indifferent, used cs_8859_1 for efficiency */ write_s3row_data(r, j, cs_8859_1, return_value); } } } /* }}} */ /* * Local variables: * tab-width: 4 * c-basic-offset: 4 * End: * vim600: sw=4 ts=4 fdm=marker * vim<600: sw=4 ts=4 */