fix for bug #69076, fix handling of URLs with lone '?' as last character
[m6w6/ext-http] / php_http_url.c
index 173bf475723daf1e729090b023354ee9ae94f9ab..14ad6f7067b6444577571e8046e17e0f7b71979c 100644 (file)
@@ -12,7 +12,9 @@
 
 #include "php_http_api.h"
 
-#ifdef PHP_HTTP_HAVE_IDN
+#if PHP_HTTP_HAVE_IDN2
+#      include <idn2.h>
+#elif PHP_HTTP_HAVE_IDN
 #      include <idna.h>
 #endif
 
@@ -550,7 +552,7 @@ HashTable *php_http_url_to_struct(const php_http_url_t *url, zval *strct TSRMLS_
        return Z_ARRVAL(arr);
 }
 
-STATUS php_http_url_encode_hash(HashTable *hash, const char *pre_encoded_str, size_t pre_encoded_len, char **encoded_str, size_t *encoded_len TSRMLS_DC)
+ZEND_RESULT_CODE php_http_url_encode_hash(HashTable *hash, const char *pre_encoded_str, size_t pre_encoded_len, char **encoded_str, size_t *encoded_len TSRMLS_DC)
 {
        const char *arg_sep_str;
        size_t arg_sep_len;
@@ -569,7 +571,7 @@ STATUS php_http_url_encode_hash(HashTable *hash, const char *pre_encoded_str, si
        return SUCCESS;
 }
 
-STATUS php_http_url_encode_hash_ex(HashTable *hash, php_http_buffer_t *qstr, const char *arg_sep_str, size_t arg_sep_len, const char *val_sep_str, size_t val_sep_len, const char *pre_encoded_str, size_t pre_encoded_len TSRMLS_DC)
+ZEND_RESULT_CODE php_http_url_encode_hash_ex(HashTable *hash, php_http_buffer_t *qstr, const char *arg_sep_str, size_t arg_sep_len, const char *val_sep_str, size_t val_sep_len, const char *pre_encoded_str, size_t pre_encoded_len TSRMLS_DC)
 {
        if (pre_encoded_len && pre_encoded_str) {
                php_http_buffer_append(qstr, pre_encoded_str, pre_encoded_len);
@@ -760,7 +762,7 @@ static size_t parse_mb(struct parse_state *state, parse_mb_what_t what, const ch
        return 0;
 }
 
-static STATUS parse_userinfo(struct parse_state *state, const char *ptr)
+static ZEND_RESULT_CODE parse_userinfo(struct parse_state *state, const char *ptr)
 {
        size_t mb;
        const char *password = NULL, *end = state->ptr, *tmp = ptr;
@@ -827,7 +829,7 @@ static STATUS parse_userinfo(struct parse_state *state, const char *ptr)
 
 #if defined(PHP_WIN32) || defined(HAVE_UIDNA_IDNTOASCII)
 typedef size_t (*parse_mb_func)(unsigned *wc, const char *ptr, const char *end);
-static STATUS to_utf16(parse_mb_func fn, const char *u8, uint16_t **u16, size_t *len)
+static ZEND_RESULT_CODE to_utf16(parse_mb_func fn, const char *u8, uint16_t **u16, size_t *len TSRMLS_DC)
 {
        size_t offset = 0, u8_len = strlen(u8);
 
@@ -870,11 +872,38 @@ static STATUS to_utf16(parse_mb_func fn, const char *u8, uint16_t **u16, size_t
 #      define MAXHOSTNAMELEN 256
 #endif
 
-#ifdef PHP_HTTP_HAVE_IDN
-static STATUS parse_idn(struct parse_state *state)
+#if PHP_HTTP_HAVE_IDN2
+static ZEND_RESULT_CODE parse_idn2(struct parse_state *state, size_t prev_len)
 {
        char *idn = NULL;
        int rv = -1;
+       TSRMLS_FETCH_FROM_CTX(state->ts);
+
+       if (state->flags & PHP_HTTP_URL_PARSE_MBUTF8) {
+               rv = idn2_lookup_u8((const unsigned char *) state->url.host, (unsigned char **) &idn, IDN2_NFC_INPUT);
+       }
+#      ifdef PHP_HTTP_HAVE_WCHAR
+       else if (state->flags & PHP_HTTP_URL_PARSE_MBLOC) {
+               rv = idn2_lookup_ul(state->url.host, &idn, 0);
+       }
+#      endif
+       if (rv != IDN2_OK) {
+               php_error_docref(NULL TSRMLS_CC, E_WARNING, "Failed to parse IDN; %s", idn2_strerror(rv));
+               return FAILURE;
+       } else {
+               size_t idnlen = strlen(idn);
+               memcpy(state->url.host, idn, idnlen + 1);
+               free(idn);
+               state->offset += idnlen - prev_len;
+               return SUCCESS;
+       }
+}
+#elif PHP_HTTP_HAVE_IDN
+static ZEND_RESULT_CODE parse_idn(struct parse_state *state, size_t prev_len)
+{
+       char *idn = NULL;
+       int rv = -1;
+       TSRMLS_FETCH_FROM_CTX(state->ts);
 
        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);
@@ -891,7 +920,7 @@ static STATUS parse_idn(struct parse_state *state)
                size_t idnlen = strlen(idn);
                memcpy(state->url.host, idn, idnlen + 1);
                free(idn);
-               state->offset += idnlen - len;
+               state->offset += idnlen - prev_len;
                return SUCCESS;
        }
 }
@@ -905,20 +934,21 @@ 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
-static STATUS parse_uidn(struct parse_state *state)
+static ZEND_RESULT_CODE parse_uidn(struct parse_state *state)
 {
        char *host_ptr;
        uint16_t *uhost_str, ahost_str[MAXHOSTNAMELEN], *ahost_ptr;
        size_t uhost_len, ahost_len;
        UErrorCode error = U_ZERO_ERROR;
+       TSRMLS_FETCH_FROM_CTX(state->ts);
 
        if (state->flags & PHP_HTTP_URL_PARSE_MBUTF8) {
-               if (SUCCESS != to_utf16(parse_mb_utf8, state->url.host, &uhost_str, &uhost_len)) {
+               if (SUCCESS != to_utf16(parse_mb_utf8, state->url.host, &uhost_str, &uhost_len TSRMLS_CC)) {
                        return FAILURE;
                }
 #ifdef PHP_HTTP_HAVE_WCHAR
        } else if (state->flags & PHP_HTTP_URL_PARSE_MBLOC) {
-               if (SUCCESS != to_utf16(parse_mb_loc, state->url.host, &uhost_str, &uhost_len)) {
+               if (SUCCESS != to_utf16(parse_mb_loc, state->url.host, &uhost_str, &uhost_len TSRMLS_CC)) {
                        return FAILURE;
                }
 #endif
@@ -947,11 +977,12 @@ static STATUS parse_uidn(struct parse_state *state)
 #endif
 
 #if 0 && defined(PHP_WIN32)
-static STATUS parse_widn(struct parse_state *state)
+static ZEND_RESULT_CODE parse_widn(struct parse_state *state)
 {
        char *host_ptr;
        uint16_t *uhost_str, ahost_str[MAXHOSTNAMELEN], *ahost_ptr;
        size_t uhost_len;
+       TSRMLS_FETCH_FROM_CTX(state->ts);
 
        if (state->flags & PHP_HTTP_URL_PARSE_MBUTF8) {
                if (SUCCESS != to_utf16(parse_mb_utf8, state->url.host, &uhost_str, &uhost_len)) {
@@ -989,7 +1020,7 @@ static STATUS parse_widn(struct parse_state *state)
 }
 #endif
 
-static STATUS parse_hostinfo(struct parse_state *state, const char *ptr)
+static ZEND_RESULT_CODE parse_hostinfo(struct parse_state *state, const char *ptr)
 {
        size_t mb, len;
        const char *end = state->ptr, *tmp = ptr, *port = NULL;
@@ -1103,11 +1134,16 @@ static STATUS parse_hostinfo(struct parse_state *state, const char *ptr)
        }
 
        if (state->flags & PHP_HTTP_URL_PARSE_TOIDN) {
-#ifdef PHP_HTTP_HAVE_IDN
-               return parse_idn(state);
+#if PHP_HTTP_HAVE_IDN2
+               return parse_idn2(state, len);
+#elif PHP_HTTP_HAVE_IDN
+               return parse_idn(state, len);
 #endif
 #ifdef HAVE_UIDNA_IDNTOASCII
                return parse_uidn(state);
+#endif
+#if 0 && defined(PHP_WIN32)
+               return parse_widn(state);
 #endif
        }
 
@@ -1235,7 +1271,7 @@ static const char *parse_query(struct parse_state *state)
        tmp = ++state->ptr;
        state->url.query = &state->buffer[state->offset];
 
-       do {
+       while(state->ptr < state->end) {
                switch (*state->ptr) {
                case '#':
                        goto done;
@@ -1287,7 +1323,9 @@ static const char *parse_query(struct parse_state *state)
                        }
                        state->ptr += mb - 1;
                }
-       } while (++state->ptr < state->end);
+
+               ++state->ptr;
+       }
 
        done:
        state->buffer[state->offset++] = 0;
@@ -1648,7 +1686,7 @@ 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);
-#if defined(PHP_HTTP_HAVE_IDN) || defined(HAVE_UIDNA_IDNTOASCII)
+#if defined(PHP_HTTP_HAVE_IDN2) || 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);