- fixed a gotcha in http_chunked_decode (-size_t is always > 0)
[m6w6/ext-http] / http_api.c
index 9960f2bb77064aa765cfd34eb1095bb9ee345955..eebfeb2ef80840c9f2f53bb46b3782413941dcea 100644 (file)
 #ifdef HAVE_CONFIG_H
 #      include "config.h"
 #endif
-
-#include <ctype.h>
-
 #include "php.h"
-#include "php_output.h"
-#include "ext/standard/md5.h"
+
+#include "SAPI.h"
+#include "ext/standard/url.h"
 
 #include "php_http.h"
 #include "php_http_std_defs.h"
 #include "php_http_api.h"
-#include "php_http_send_api.h"
-#include "php_http_cache_api.h"
 #include "php_http_headers_api.h"
+#include "php_http_send_api.h"
 
 #ifdef ZEND_ENGINE_2
+#      include "zend_exceptions.h"
 #      include "php_http_exception_object.h"
 #endif
 
+#include <ctype.h>
+
+#ifdef HTTP_HAVE_MAGIC
+#      if defined(PHP_WIN32) && !defined(USE_MAGIC_DLL) && !defined(USE_MAGIC_STATIC)
+#              define USE_MAGIC_STATIC
+#      endif
+#      include <magic.h>
+#endif
+
 ZEND_EXTERN_MODULE_GLOBALS(http);
 
-/* char *pretty_key(char *, size_t, zend_bool, zebd_bool) */
+/* char *pretty_key(char *, size_t, zend_bool, zend_bool) */
 char *_http_pretty_key(char *key, size_t key_len, zend_bool uctitle, zend_bool xhyphen)
 {
        if (key && key_len) {
-               unsigned i, wasalpha;
-               if (wasalpha = isalpha(key[0])) {
-                       key[0] = uctitle ? toupper(key[0]) : tolower(key[0]);
+               size_t i;
+               int wasalpha;
+               if (wasalpha = isalpha((int) key[0])) {
+                       key[0] = (char) (uctitle ? toupper((int) key[0]) : tolower((int) key[0]));
                }
                for (i = 1; i < key_len; i++) {
-                       if (isalpha(key[i])) {
-                               key[i] = ((!wasalpha) && uctitle) ? toupper(key[i]) : tolower(key[i]);
+                       if (isalpha((int) key[i])) {
+                               key[i] = (char) (((!wasalpha) && uctitle) ? toupper((int) key[i]) : tolower((int) key[i]));
                                wasalpha = 1;
                        } else {
                                if (xhyphen && (key[i] == '_')) {
@@ -62,233 +70,349 @@ char *_http_pretty_key(char *key, size_t key_len, zend_bool uctitle, zend_bool x
 }
 /* }}} */
 
+/* {{{ */
+void _http_key_list_default_decoder(const char *encoded, size_t encoded_len, char **decoded, size_t *decoded_len TSRMLS_DC)
+{
+       *decoded = estrndup(encoded, encoded_len);
+       *decoded_len = (size_t) php_url_decode(*decoded, encoded_len);
+}
+/* }}} */
+
+/* {{{ */
+STATUS _http_parse_key_list(const char *list, HashTable *items, char separator, http_key_list_decode_t decode, zend_bool first_entry_is_name_value_pair TSRMLS_DC)
+{
+       const char *key = list, *val = NULL;
+       int vallen = 0, keylen = 0, done = 0;
+       zval array;
+
+       INIT_ZARR(array, items);
+
+       if (!(val = strchr(list, '='))) {
+               return FAILURE;
+       }
+
+#define HTTP_KEYLIST_VAL(array, k, str, len) \
+       { \
+               char *decoded; \
+               size_t decoded_len; \
+               if (decode) { \
+                       decode(str, len, &decoded, &decoded_len TSRMLS_CC); \
+               } else { \
+                       decoded_len = len; \
+                       decoded = estrndup(str, decoded_len); \
+               } \
+               add_assoc_stringl(array, k, decoded, decoded_len, 0); \
+       }
+#define HTTP_KEYLIST_FIXKEY() \
+       { \
+                       while (isspace(*key)) ++key; \
+                       keylen = val - key; \
+                       while (isspace(key[keylen - 1])) --keylen; \
+       }
+#define HTTP_KEYLIST_FIXVAL() \
+       { \
+                       ++val; \
+                       while (isspace(*val)) ++val; \
+                       vallen = key - val; \
+                       while (isspace(val[vallen - 1])) --vallen; \
+       }
+
+       HTTP_KEYLIST_FIXKEY();
+
+       if (first_entry_is_name_value_pair) {
+               HTTP_KEYLIST_VAL(&array, "name", key, keylen);
+
+               /* just one name=value */
+               if (!(key = strchr(val, separator))) {
+                       key = val + strlen(val);
+                       HTTP_KEYLIST_FIXVAL();
+                       HTTP_KEYLIST_VAL(&array, "value", val, vallen);
+                       return SUCCESS;
+               }
+               /* additional info appended */
+               else {
+                       HTTP_KEYLIST_FIXVAL();
+                       HTTP_KEYLIST_VAL(&array, "value", val, vallen);
+               }
+       }
+
+       do {
+               char *keydup = NULL;
+
+               if (!(val = strchr(key, '='))) {
+                       break;
+               }
+
+               /* start at 0 if first_entry_is_name_value_pair==0 */
+               if (zend_hash_num_elements(items)) {
+                       ++key;
+               }
+
+               HTTP_KEYLIST_FIXKEY();
+               keydup = estrndup(key, keylen);
+               if (!(key = strchr(val, separator))) {
+                       done = 1;
+                       key = val + strlen(val);
+               }
+               HTTP_KEYLIST_FIXVAL();
+               HTTP_KEYLIST_VAL(&array, keydup, val, vallen);
+               efree(keydup);
+       } while (!done);
+
+       return SUCCESS;
+}
+/* }}} */
+
 /* {{{ void http_error(long, long, char*) */
-void _http_error_ex(long type, long code, const char *format, ...)
+void _http_error_ex(long type TSRMLS_DC, long code, const char *format, ...)
 {
        va_list args;
-       TSRMLS_FETCH();
-
+       
        va_start(args, format);
-       if (type == E_THROW) {
 #ifdef ZEND_ENGINE_2
+       if ((type == E_THROW) || (PG(error_handling) == EH_THROW)) {
                char *message;
+               
                vspprintf(&message, 0, format, args);
-               zend_throw_exception(http_exception_get_default(), message, code TSRMLS_CC);
-#else
-               type = E_WARNING;
+               zend_throw_exception(http_exception_get_for_code(code), message, code TSRMLS_CC);
+       } else
 #endif
-       }
-       if (type != E_THROW) {
-               php_verror(NULL, "", type, format, args TSRMLS_CC);
-       }
+       php_verror(NULL, "", type, format, args TSRMLS_CC);
        va_end(args);
 }
 /* }}} */
 
-/* {{{ static STATUS http_ob_stack_get(php_ob_buffer *, php_ob_buffer **) */
-static STATUS http_ob_stack_get(php_ob_buffer *o, php_ob_buffer **s)
+/* {{{ void http_log(char *, char *, char *) */
+void _http_log_ex(char *file, const char *ident, const char *message TSRMLS_DC)
 {
-       static int i = 0;
-       php_ob_buffer *b = emalloc(sizeof(php_ob_buffer));
-       b->handler_name = estrdup(o->handler_name);
-       b->buffer = estrndup(o->buffer, o->text_length);
-       b->text_length = o->text_length;
-       b->chunk_size = o->chunk_size;
-       b->erase = o->erase;
-       s[i++] = b;
-       return SUCCESS;
+       time_t now;
+       struct tm nowtm;
+       char datetime[128];
+       
+       time(&now);
+       strftime(datetime, sizeof(datetime), "%Y-%m-%d %H:%M:%S", php_localtime_r(&now, &nowtm));
+
+#define HTTP_LOG_WRITE(file, type, msg) \
+       if (file && strlen(file)) { \
+               php_stream *log = php_stream_open_wrapper(file, "ab", REPORT_ERRORS|ENFORCE_SAFE_MODE, NULL); \
+                \
+               if (log) { \
+                       php_stream_printf(log TSRMLS_CC, "%s\t[%s]\t%s\t<%s>%s", datetime, type, msg, SG(request_info).request_uri, PHP_EOL); \
+                       php_stream_close(log); \
+               } \
+        \
+       }
+       
+       HTTP_LOG_WRITE(file, ident, message);
+       HTTP_LOG_WRITE(HTTP_G(log).composite, ident, message);
 }
 /* }}} */
 
-/* {{{ zval *http_get_server_var_ex(char *, size_t) */
-PHP_HTTP_API zval *_http_get_server_var_ex(const char *key, size_t key_size, zend_bool check TSRMLS_DC)
+/* {{{ STATUS http_exit(int, char*, char*) */
+STATUS _http_exit_ex(int status, char *header, char *body, zend_bool send_header TSRMLS_DC)
 {
-       zval **var;
-       if (SUCCESS == zend_hash_find(HTTP_SERVER_VARS, (char *) key, key_size, (void **) &var)) {
-               if (check) {
-                       return Z_STRVAL_PP(var) && Z_STRLEN_PP(var) ? *var : NULL;
-               } else {
-                       return *var;
+       if (status || send_header) {
+               if (SUCCESS != http_send_status_header(status, send_header ? header : NULL)) {
+                       http_error_ex(HE_WARNING, HTTP_E_HEADER, "Failed to exit with status/header: %d - %s", status, header ? header : "");
+                       STR_FREE(header);
+                       STR_FREE(body);
+                       return FAILURE;
                }
        }
-       return NULL;
+       
+       if (body) {
+               PHPWRITE(body, strlen(body));
+       }
+       
+       switch (status)
+       {
+               case 301:       http_log(HTTP_G(log).redirect, "301-REDIRECT", header);                 break;
+               case 302:       http_log(HTTP_G(log).redirect, "302-REDIRECT", header);                 break;
+               case 304:       http_log(HTTP_G(log).cache, "304-CACHE", header);                               break;
+               case 405:       http_log(HTTP_G(log).allowed_methods, "405-ALLOWED", header);   break;
+               default:        http_log(NULL, header, body);                                                                   break;
+       }
+       
+       STR_FREE(header);
+       STR_FREE(body);
+       
+       zend_bailout();
+       /* fake */
+       return SUCCESS;
 }
 /* }}} */
 
-/* {{{ void http_ob_etaghandler(char *, uint, char **, uint *, int) */
-PHP_HTTP_API void _http_ob_etaghandler(char *output, uint output_len,
-       char **handled_output, uint *handled_output_len, int mode TSRMLS_DC)
+/* {{{ STATUS http_check_method(char *) */
+STATUS _http_check_method_ex(const char *method, const char *methods)
 {
-       char etag[33] = { 0 };
-       unsigned char digest[16];
-
-       if (mode & PHP_OUTPUT_HANDLER_START) {
-               PHP_MD5Init(&HTTP_G(etag_md5));
-       }
-
-       PHP_MD5Update(&HTTP_G(etag_md5), output, output_len);
-
-       if (mode & PHP_OUTPUT_HANDLER_END) {
-               PHP_MD5Final(digest, &HTTP_G(etag_md5));
-
-               /* just do that if desired */
-               if (HTTP_G(etag_started)) {
-                       make_digest(etag, digest);
+       const char *found;
 
-                       if (http_etag_match("HTTP_IF_NONE_MATCH", etag)) {
-                               http_send_status(304);
-                               zend_bailout();
-                       } else {
-                               http_send_etag(etag, 32);
-                       }
-               }
+       if (    (found = strstr(methods, method)) &&
+                       (found == method || !isalpha(found[-1])) &&
+                       (!isalpha(found[strlen(method) + 1]))) {
+               return SUCCESS;
        }
-
-       *handled_output_len = output_len;
-       *handled_output = estrndup(output, output_len);
+       return FAILURE;
 }
 /* }}} */
 
-/* {{{ STATUS http_start_ob_handler(php_output_handler_func_t, char *, uint, zend_bool) */
-PHP_HTTP_API STATUS _http_start_ob_handler(php_output_handler_func_t handler_func,
-       char *handler_name, uint chunk_size, zend_bool erase TSRMLS_DC)
+/* {{{ zval *http_get_server_var_ex(char *, size_t) */
+PHP_HTTP_API zval *_http_get_server_var_ex(const char *key, size_t key_size, zend_bool check TSRMLS_DC)
 {
-       php_ob_buffer **stack;
-       int count, i;
-
-       if (count = OG(ob_nesting_level)) {
-               stack = ecalloc(count, sizeof(php_ob_buffer *));
-
-               if (count > 1) {
-                       zend_stack_apply_with_argument(&OG(ob_buffers), ZEND_STACK_APPLY_BOTTOMUP,
-                               (int (*)(void *elem, void *)) http_ob_stack_get, stack);
-               }
-
-               if (count > 0) {
-                       http_ob_stack_get(&OG(active_ob_buffer), stack);
-               }
-
-               while (OG(ob_nesting_level)) {
-                       php_end_ob_buffer(0, 0 TSRMLS_CC);
-               }
+       zval **hsv;
+       zval **var;
+       
+       if (SUCCESS != zend_hash_find(&EG(symbol_table), "_SERVER", sizeof("_SERVER"), (void **) &hsv)) {
+               return NULL;
        }
-
-       php_ob_set_internal_handler(handler_func, chunk_size, handler_name, erase TSRMLS_CC);
-
-       for (i = 0; i < count; i++) {
-               php_ob_buffer *s = stack[i];
-               if (strcmp(s->handler_name, "default output handler")) {
-                       php_start_ob_buffer_named(s->handler_name, s->chunk_size, s->erase TSRMLS_CC);
-               }
-               php_body_write(s->buffer, s->text_length TSRMLS_CC);
-               efree(s->handler_name);
-               efree(s->buffer);
-               efree(s);
+       if (SUCCESS != zend_hash_find(Z_ARRVAL_PP(hsv), (char *) key, key_size, (void **) &var)) {
+               return NULL;
        }
-       if (count) {
-               efree(stack);
+       if (check && !(Z_STRVAL_PP(var) && Z_STRLEN_PP(var))) {
+               return NULL;
        }
+       return *var;
+}
+/* }}} */
 
-       return SUCCESS;
+/* {{{ STATUS http_get_request_body(char **, size_t *) */
+PHP_HTTP_API STATUS _http_get_request_body_ex(char **body, size_t *length, zend_bool dup TSRMLS_DC)
+{
+       *length = 0;
+       *body = NULL;
+
+       if (SG(request_info).raw_post_data) {
+               *length = SG(request_info).raw_post_data_length;
+               *body = (char *) (dup ? estrndup(SG(request_info).raw_post_data, *length) : SG(request_info).raw_post_data);
+               return SUCCESS;
+       }
+       return FAILURE;
 }
 /* }}} */
 
-/* {{{ STATUS http_chunked_decode(char *, size_t, char **, size_t *) */
-PHP_HTTP_API STATUS _http_chunked_decode(const char *encoded, size_t encoded_len,
-       char **decoded, size_t *decoded_len TSRMLS_DC)
+/* {{{ char *http_chunked_decode(char *, size_t, char **, size_t *) */
+PHP_HTTP_API const char *_http_chunked_decode(const char *encoded, size_t encoded_len, char **decoded, size_t *decoded_len TSRMLS_DC)
 {
        const char *e_ptr;
        char *d_ptr;
-
+       long rest;
+       
        *decoded_len = 0;
        *decoded = ecalloc(1, encoded_len);
        d_ptr = *decoded;
        e_ptr = encoded;
 
-       while (((e_ptr - encoded) - encoded_len) > 0) {
-               char hex_len[9] = {0};
-               size_t chunk_len = 0;
-               int i = 0;
-
-               /* read in chunk size */
-               while (isxdigit(*e_ptr)) {
-                       if (i == 9) {
-                               http_error_ex(E_WARNING, HTTP_E_PARSE, "Chunk size is too long: 0x%s...", hex_len);
+       while ((rest = encoded + encoded_len - e_ptr) > 0) {
+               long chunk_len = 0;
+               int EOL_len = 0, eol_mismatch = 0;
+               char *n_ptr;
+
+               chunk_len = strtol(e_ptr, &n_ptr, 16);
+
+               /* check if:
+                * - we could not read in chunk size
+                * - we got a negative chunk size
+                * - chunk size is greater then remaining size
+                * - chunk size is not followed by (CR)LF|NUL
+                */
+               if (    (n_ptr == e_ptr) ||     (chunk_len < 0) || (chunk_len > rest) || 
+                               (*n_ptr && (eol_mismatch = (n_ptr != http_locate_eol(e_ptr, &EOL_len))))) {
+                       /* don't fail on apperently not encoded data */
+                       if (e_ptr == encoded) {
+                               memcpy(*decoded, encoded, encoded_len);
+                               *decoded_len = encoded_len;
+                               return encoded + encoded_len;
+                       } else {
                                efree(*decoded);
-                               return FAILURE;
+                               if (eol_mismatch) {
+                                       if (EOL_len == 2) {
+                                               http_error_ex(HE_WARNING, HTTP_E_ENCODING, "Invalid character (expected 0x0D 0x0A; got: 0x%X 0x%X)", *n_ptr, *(n_ptr + 1));
+                                       } else {
+                                               http_error_ex(HE_WARNING, HTTP_E_ENCODING, "Invalid character (expected 0x0A; got: 0x%X)", *n_ptr);
+                                       }
+                               } else {
+                                       char *error = estrndup(n_ptr, strcspn(n_ptr, "\r\n "));
+                                       http_error_ex(HE_WARNING, HTTP_E_ENCODING, "Invalid chunk size: '%s' at pos %d", error, n_ptr - encoded);
+                                       efree(error);
+                               }
+                               return NULL;
                        }
-                       hex_len[i++] = *e_ptr++;
+               } else {
+                       e_ptr = n_ptr;
                }
 
                /* reached the end */
-               if (!strcmp(hex_len, "0")) {
+               if (!chunk_len) {
                        break;
                }
 
-               /* new line */
-               if (strncmp(e_ptr, HTTP_CRLF, 2)) {
-                       http_error_ex(E_WARNING, HTTP_E_PARSE, "Invalid character (expected 0x0D 0x0A; got: %x %x)", *e_ptr, *(e_ptr + 1));
-                       efree(*decoded);
-                       return FAILURE;
-               }
-
-               /* hex to long */
-               {
-                       char *error = NULL;
-                       chunk_len = strtol(hex_len, &error, 16);
-                       if (error == hex_len) {
-                               http_error_ex(E_WARNING, HTTP_E_PARSE, "Invalid chunk size string: '%s'", hex_len);
-                               efree(*decoded);
-                               return FAILURE;
-                       }
-               }
-
-               memcpy(d_ptr, e_ptr += 2, chunk_len);
+               memcpy(d_ptr, e_ptr += EOL_len, chunk_len);
                d_ptr += chunk_len;
-               e_ptr += chunk_len + 2;
+               e_ptr += chunk_len + EOL_len;
                *decoded_len += chunk_len;
        }
 
-       return SUCCESS;
-}
-/* }}} */
-
-/* {{{ STATUS http_split_response(zval *, zval *, zval *) */
-PHP_HTTP_API STATUS _http_split_response(zval *response, zval *headers, zval *body TSRMLS_DC)
-{
-       char *b = NULL;
-       size_t l = 0;
-       STATUS status = http_split_response_ex(Z_STRVAL_P(response), Z_STRLEN_P(response), Z_ARRVAL_P(headers), &b, &l);
-       ZVAL_STRINGL(body, b, l, 0);
-       return status;
+       return e_ptr;
 }
 /* }}} */
 
-/* {{{ STATUS http_split_response(char *, size_t, HashTable *, char **, size_t *) */
-PHP_HTTP_API STATUS _http_split_response_ex(char *response, size_t response_len,
-       HashTable *headers, char **body, size_t *body_len TSRMLS_DC)
+/* {{{ char *http_guess_content_type(char *magic_file, long magic_mode, void *data, size_t size, http_send_mode mode) */
+PHP_HTTP_API char *_http_guess_content_type(const char *magicfile, long magicmode, void *data_ptr, size_t data_len, http_send_mode data_mode TSRMLS_DC)
 {
-       char *header = response, *real_body = NULL;
-
-       while (0 < (response_len - (response - header + 4))) {
-               if (    (*response++ == '\r') &&
-                               (*response++ == '\n') &&
-                               (*response++ == '\r') &&
-                               (*response++ == '\n')) {
-                       real_body = response;
+       char *ct = NULL;
+
+#ifdef HTTP_HAVE_MAGIC
+       /*      magic_load() fails if MAGIC_MIME is set because it 
+               cowardly adds .mime to the file name */
+       struct magic_set *magic = magic_open(magicmode &~ MAGIC_MIME);
+       
+       if (!magic) {
+               http_error_ex(HE_WARNING, HTTP_E_INVALID_PARAM, "Invalid magic mode: %ld", magicmode);
+       } else if (-1 == magic_load(magic, magicfile)) {
+               http_error_ex(HE_WARNING, HTTP_E_RUNTIME, "Failed to load magic database '%s' (%s)", magicfile, magic_error(magic));
+       } else {
+               const char *ctype = NULL;
+               
+               magic_setflags(magic, magicmode);
+               
+               switch (data_mode)
+               {
+                       case SEND_RSRC:
+                       {
+                               char *buffer;
+                               size_t b_len;
+                               
+                               b_len = php_stream_copy_to_mem(data_ptr, &buffer, 65536, 0);
+                               ctype = magic_buffer(magic, buffer, b_len);
+                               efree(buffer);
+                       }
+                       break;
+                       
+                       case SEND_DATA:
+                               ctype = magic_buffer(magic, data_ptr, data_len);
+                       break;
+                       
+                       default:
+                               ctype = magic_file(magic, data_ptr);
                        break;
                }
+               
+               if (ctype) {
+                       ct = estrdup(ctype);
+               } else {
+                       http_error_ex(HE_WARNING, HTTP_E_RUNTIME, "Failed to guess Content-Type: %s", magic_error(magic));
+               }
        }
-
-       if (real_body && (*body_len = (response_len - (real_body - header)))) {
-               *body = ecalloc(1, *body_len + 1);
-               memcpy(*body, real_body, *body_len);
+       if (magic) {
+               magic_close(magic);
        }
-
-       return http_parse_headers_ex(header, real_body ? response_len - *body_len : response_len, headers, 1);
+#else
+       http_error(HE_WARNING, HTTP_E_RUNTIME, "Cannot guess Content-Type; libmagic not available");
+#endif
+       
+       return ct;
 }
 /* }}} */
-
 /*
  * Local variables:
  * tab-width: 4