From: Michael Wallner Date: Wed, 25 Feb 2015 21:14:57 +0000 (+0100) Subject: use builtin uidna_IDNToASCII on OSX X-Git-Tag: RELEASE_2_4_0~14 X-Git-Url: https://git.m6w6.name/?p=m6w6%2Fext-http;a=commitdiff_plain;h=f513f2b7d0b226d56d0fdb45f1f9f0463e139219 use builtin uidna_IDNToASCII on OSX --- diff --git a/config9.m4 b/config9.m4 index af3d55f..ec202b6 100644 --- a/config9.m4 +++ b/config9.m4 @@ -122,6 +122,11 @@ dnl ---- done if test "x$IDNA_DIR" = "x"; then AC_MSG_RESULT([not found]) + case $host_os in + darwin*) + AC_CHECK_HEADERS(unicode/uidna.h) + PHP_CHECK_FUNC(uidna_IDNToASCII, icucore);; + esac else AC_MSG_RESULT([found in $IDNA_DIR]) AC_DEFINE([PHP_HTTP_HAVE_IDN], [1], [Have libidn support]) diff --git a/php_http_url.c b/php_http_url.c index 6bfca24..3c82e61 100644 --- a/php_http_url.c +++ b/php_http_url.c @@ -661,8 +661,9 @@ static size_t parse_mb_loc(unsigned *wc, const char *ptr, const char *end) wchar_t wchar; size_t consumed = 0; #if defined(HAVE_MBRTOWC) - mbstate_t ps = {0}; + mbstate_t ps; + memset(&ps, 0, sizeof(ps)); consumed = mbrtowc(&wchar, ptr, end - ptr, &ps); #elif defined(HAVE_MBTOWC) consumed = mbtowc(&wchar, ptr, end - ptr); @@ -818,6 +819,124 @@ static STATUS parse_userinfo(struct parse_state *state, const char *ptr) return SUCCESS; } +#ifdef PHP_HTTP_HAVE_IDN +static STATUS parse_idn(struct parse_state *state) +{ + char *idn = NULL; + int rv = -1; + + if (state->flags & PHP_HTTP_URL_PARSE_MBUTF8) { + rv = idna_to_ascii_8z(state->url.host, &idn, IDNA_ALLOW_UNASSIGNED|IDNA_USE_STD3_ASCII_RULES); + } +# ifdef PHP_HTTP_HAVE_WCHAR + else if (state->flags & PHP_HTTP_URL_PARSE_MBLOC) { + rv = idna_to_ascii_lz(state->url.host, &idn, IDNA_ALLOW_UNASSIGNED|IDNA_USE_STD3_ASCII_RULES); + } +# endif + if (rv != IDNA_SUCCESS) { + php_error_docref(NULL TSRMLS_CC, E_WARNING, "Failed to parse IDN; %s", idna_strerror(rv)); + return FAILURE; + } else { + size_t idnlen = strlen(idn); + memcpy(state->url.host, idn, idnlen + 1); + free(idn); + state->offset += idnlen - len; + return SUCCESS; + } +} +#endif + +#ifdef HAVE_UIDNA_IDNTOASCII +# if HAVE_UNICODE_UIDNA_H +# include +# else +typedef uint16_t UChar; +typedef enum { U_ZERO_ERROR = 0 } UErrorCode; +int32_t uidna_IDNToASCII(const UChar *src, int32_t srcLength, UChar *dest, int32_t destCapacity, int32_t options, void *parseError, UErrorCode *status); +# endif +typedef size_t (*parse_mb_func)(unsigned *wc, const char *ptr, const char *end); +static STATUS toutf16(parse_mb_func fn, const char *u8, uint16_t **u16, size_t *len) +{ + size_t offset = 0, u8_len = strlen(u8); + + *u16 = ecalloc(4 * sizeof(uint16_t), u8_len + 1); + *len = 0; + + while (offset < u8_len) { + unsigned wc; + uint16_t buf[2], *ptr = buf; + size_t consumed = fn(&wc, &u8[offset], &u8[u8_len]); + + if (!consumed) { + efree(*u16); + return FAILURE; + } else { + offset += consumed; + } + + switch (wctoutf16(buf, wc)) { + case 2: + (*u16)[(*len)++] = *ptr++; + /* no break */ + case 1: + (*u16)[(*len)++] = *ptr++; + break; + case 0: + default: + efree(*u16); + return FAILURE; + } + } + + return SUCCESS; +} +static STATUS parse_uidn(struct parse_state *state) +{ + char *host_ptr; + uint16_t *uhost_str = NULL, *ahost_str, *ahost_ptr; + size_t uhost_len, ahost_len; + UErrorCode error = U_ZERO_ERROR; + + if (state->flags & PHP_HTTP_URL_PARSE_MBUTF8) { + if (SUCCESS != toutf16(parse_mb_utf8, state->url.host, &uhost_str, &uhost_len)) { + php_error_docref(NULL TSRMLS_CC, E_WARNING, "Failed to parse IDN"); + return FAILURE; + } +#ifdef PHP_HTTP_HAVE_WCHAR + } else if (state->flags & PHP_HTTP_URL_PARSE_MBLOC) { + if (SUCCESS != toutf16(parse_mb_loc, state->url.host, &uhost_str, &uhost_len)) { + php_error_docref(NULL TSRMLS_CC, E_WARNING, "Failed to parse IDN"); + return FAILURE; + } +#endif + } else { + php_error_docref(NULL TSRMLS_CC, E_WARNING, "Failed to parse IDN"); + return FAILURE; + } + + ahost_len = uhost_len * 3; + ahost_str = ecalloc(sizeof(uint16_t), ahost_len); + + ahost_len = uidna_IDNToASCII(uhost_str, uhost_len, ahost_str, ahost_len, 3, NULL, &error); + efree(uhost_str); + + if (error != U_ZERO_ERROR) { + php_error_docref(NULL TSRMLS_CC, E_WARNING, "Failed to parse IDN"); + return FAILURE; + } + + host_ptr = state->url.host; + ahost_ptr = ahost_str; + PHP_HTTP_DUFF(ahost_len, *host_ptr++ = *ahost_ptr++); + efree(ahost_str); + + *host_ptr = '\0'; + state->offset += host_ptr - state->url.host; + + return SUCCESS; +} +#endif + static STATUS parse_hostinfo(struct parse_state *state, const char *ptr) { size_t mb, len; @@ -932,30 +1051,14 @@ static STATUS parse_hostinfo(struct parse_state *state, const char *ptr) state->buffer[state->offset++] = 0; } -#ifdef PHP_HTTP_HAVE_IDN if (state->flags & PHP_HTTP_URL_PARSE_TOIDN) { - char *idn = NULL; - int rv = -1; - - if (state->flags & PHP_HTTP_URL_PARSE_MBUTF8) { - rv = idna_to_ascii_8z(state->url.host, &idn, IDNA_ALLOW_UNASSIGNED|IDNA_USE_STD3_ASCII_RULES); - } -# ifdef PHP_HTTP_HAVE_WCHAR - else if (state->flags & PHP_HTTP_URL_PARSE_MBLOC) { - rv = idna_to_ascii_lz(state->url.host, &idn, IDNA_ALLOW_UNASSIGNED|IDNA_USE_STD3_ASCII_RULES); - } -# endif - if (rv != IDNA_SUCCESS) { - php_error_docref(NULL TSRMLS_CC, E_WARNING, "Failed to parse IDN; %s", idna_strerror(rv)); - return FAILURE; - } else { - size_t idnlen = strlen(idn); - memcpy(state->url.host, idn, idnlen + 1); - free(idn); - state->offset += idnlen - len; - } - } +#ifdef PHP_HTTP_HAVE_IDN + return parse_idn(state); +#endif +#ifdef HAVE_UIDNA_IDNTOASCII + return parse_uidn(state); #endif + } return SUCCESS; } @@ -1494,7 +1597,7 @@ PHP_MINIT_FUNCTION(http_url) zend_declare_class_constant_long(php_http_url_class_entry, ZEND_STRL("PARSE_MBLOC"), PHP_HTTP_URL_PARSE_MBLOC TSRMLS_CC); #endif zend_declare_class_constant_long(php_http_url_class_entry, ZEND_STRL("PARSE_MBUTF8"), PHP_HTTP_URL_PARSE_MBUTF8 TSRMLS_CC); -#ifdef PHP_HTTP_HAVE_IDN +#if defined(PHP_HTTP_HAVE_IDN) || defined(HAVE_UIDNA_IDNTOASCII) zend_declare_class_constant_long(php_http_url_class_entry, ZEND_STRL("PARSE_TOIDN"), PHP_HTTP_URL_PARSE_TOIDN TSRMLS_CC); #endif zend_declare_class_constant_long(php_http_url_class_entry, ZEND_STRL("PARSE_TOPCT"), PHP_HTTP_URL_PARSE_TOPCT TSRMLS_CC); diff --git a/php_http_utf8.h b/php_http_utf8.h index 2503f06..f900803 100644 --- a/php_http_utf8.h +++ b/php_http_utf8.h @@ -625,6 +625,39 @@ static inline zend_bool isualnum(unsigned ch) return isualpha(ch); } +static inline size_t wctoutf16(unsigned short u16[2], unsigned wc) +{ + if (wc > 0x10ffff || (wc >= 0xd800 && wc <= 0xdfff)) { + return 0; + } + + if (wc <= 0xffff) { + u16[0] = (unsigned short) wc; + return 1; + } + + wc -= 0x10000; + u16[0] = (unsigned short) ((wc >> 10) + 0xd800); + u16[1] = (unsigned short) ((wc & 0x3ff) + 0xdc00); + return 2; +} + +static inline size_t utf16towc(unsigned *wc, unsigned short *u16_str, size_t u16_len) +{ + if (u16_len < 1) { + return 0; + } + if (u16_str[0] - 0xd800 >= 0x800) { + *wc = u16_str[0]; + return 1; + } + if (u16_len < 2 || (u16_str[0] & 0xfffffc00) != 0xd800 || (u16_str[1] & 0xfffffc00) != 0xdc00) { + return 0; + } + *wc = (u16_str[0] << 10) + u16_str[1] - 0x35fdc00; + return 2; +} + #endif /* PHP_HTTP_UTF8_H */ /*