git.m6w6.name Git - m6w6/ext-http/blob - php_http_header_parser.c

   1 /*
   2     +--------------------------------------------------------------------+
   3     | PECL :: http                                                       |
   4     +--------------------------------------------------------------------+
   5     | Redistribution and use in source and binary forms, with or without |
   6     | modification, are permitted provided that the conditions mentioned |
   7     | in the accompanying LICENSE file are met.                          |
   8     +--------------------------------------------------------------------+
   9     | Copyright (c) 2004-2014, Michael Wallner <mike@php.net>            |
  10     +--------------------------------------------------------------------+
  11 */
  12
  13 #include "php_http_api.h"
  14
  15 typedef struct php_http_header_parser_state_spec {
  16         php_http_header_parser_state_t state;
  17         unsigned need_data:1;
  18 } php_http_header_parser_state_spec_t;
  19
  20 static const php_http_header_parser_state_spec_t php_http_header_parser_states[] = {
  21                 {PHP_HTTP_HEADER_PARSER_STATE_START,            1},
  22                 {PHP_HTTP_HEADER_PARSER_STATE_KEY,                      1},
  23                 {PHP_HTTP_HEADER_PARSER_STATE_VALUE,            1},
  24                 {PHP_HTTP_HEADER_PARSER_STATE_VALUE_EX,         1},
  25                 {PHP_HTTP_HEADER_PARSER_STATE_HEADER_DONE,      0},
  26                 {PHP_HTTP_HEADER_PARSER_STATE_DONE,                     0}
  27 };
  28
  29 php_http_header_parser_t *php_http_header_parser_init(php_http_header_parser_t *parser TSRMLS_DC)
  30 {
  31         if (!parser) {
  32                 parser = emalloc(sizeof(*parser));
  33         }
  34         memset(parser, 0, sizeof(*parser));
  35
  36         TSRMLS_SET_CTX(parser->ts);
  37
  38         return parser;
  39 }
  40
  41 php_http_header_parser_state_t php_http_header_parser_state_push(php_http_header_parser_t *parser, unsigned argc, ...)
  42 {
  43         va_list va_args;
  44         unsigned i;
  45         php_http_header_parser_state_t state = 0;
  46
  47         /* short circuit */
  48         ZEND_PTR_STACK_RESIZE_IF_NEEDED((&parser->stack), argc);
  49
  50         va_start(va_args, argc);
  51         for (i = 0; i < argc; ++i) {
  52                 state = va_arg(va_args, php_http_header_parser_state_t);
  53                 zend_ptr_stack_push(&parser->stack, (void *) state);
  54         }
  55         va_end(va_args);
  56
  57         return state;
  58 }
  59
  60 php_http_header_parser_state_t php_http_header_parser_state_is(php_http_header_parser_t *parser)
  61 {
  62         if (parser->stack.top) {
  63                 return (php_http_header_parser_state_t) parser->stack.elements[parser->stack.top - 1];
  64         }
  65
  66         return PHP_HTTP_HEADER_PARSER_STATE_START;
  67 }
  68
  69 php_http_header_parser_state_t php_http_header_parser_state_pop(php_http_header_parser_t *parser)
  70 {
  71         if (parser->stack.top) {
  72                 return (php_http_header_parser_state_t) zend_ptr_stack_pop(&parser->stack);
  73         }
  74
  75         return PHP_HTTP_HEADER_PARSER_STATE_START;
  76 }
  77
  78 void php_http_header_parser_dtor(php_http_header_parser_t *parser)
  79 {
  80         zend_ptr_stack_destroy(&parser->stack);
  81         php_http_info_dtor(&parser->info);
  82         PTR_FREE(parser->_key.str);
  83         PTR_FREE(parser->_val.str);
  84 }
  85
  86 void php_http_header_parser_free(php_http_header_parser_t **parser)
  87 {
  88         if (*parser) {
  89                 php_http_header_parser_dtor(*parser);
  90                 efree(*parser);
  91                 *parser = NULL;
  92         }
  93 }
  94
  95 STATUS php_http_header_parser_parse(php_http_header_parser_t *parser, php_http_buffer_t *buffer, unsigned flags, HashTable *headers, php_http_info_callback_t callback_func, void *callback_arg)
  96 {
  97         TSRMLS_FETCH_FROM_CTX(parser->ts);
  98
  99         while (buffer->used || !php_http_header_parser_states[php_http_header_parser_state_is(parser)].need_data) {
 100 #if 0
 101                 const char *state[] = {"START", "KEY", "VALUE", "HEADER_DONE", "DONE"};
 102                 fprintf(stderr, "#HP: %s (avail:%zu, num:%d)\n", php_http_header_parser_state_is(parser) < 0 ? "FAILURE" : state[php_http_header_parser_state_is(parser)], buffer->used, headers?zend_hash_num_elements(headers):0);
 103                 _dpf(0, buffer->data, buffer->used);
 104 #endif
 105                 switch (php_http_header_parser_state_pop(parser)) {
 106                         case PHP_HTTP_HEADER_PARSER_STATE_FAILURE:
 107                                 return php_http_header_parser_state_push(parser, 1, PHP_HTTP_HEADER_PARSER_STATE_FAILURE);
 108
 109                         case PHP_HTTP_HEADER_PARSER_STATE_START: {
 110                                 char *ptr = buffer->data;
 111
 112                                 while (ptr - buffer->data < buffer->used && PHP_HTTP_IS_CTYPE(space, *ptr)) {
 113                                         ++ptr;
 114                                 }
 115
 116                                 php_http_buffer_cut(buffer, 0, ptr - buffer->data);
 117                                 php_http_header_parser_state_push(parser, 1, PHP_HTTP_HEADER_PARSER_STATE_KEY);
 118                                 break;
 119                         }
 120
 121                         case PHP_HTTP_HEADER_PARSER_STATE_KEY: {
 122                                 const char *colon, *eol_str = NULL;
 123                                 int eol_len = 0;
 124
 125                                 if (buffer->data == (eol_str = php_http_locate_bin_eol(buffer->data, buffer->used, &eol_len))) {
 126                                         /* end of headers */
 127                                         php_http_buffer_cut(buffer, 0, eol_len);
 128                                         php_http_header_parser_state_push(parser, 1, PHP_HTTP_HEADER_PARSER_STATE_DONE);
 129                                 } else if (php_http_info_parse(&parser->info, php_http_buffer_fix(buffer)->data TSRMLS_CC)) {
 130                                         /* new message starting with request/response line */
 131                                         if (callback_func) {
 132                                                 callback_func(callback_arg, &headers, &parser->info TSRMLS_CC);
 133                                         }
 134                                         php_http_info_dtor(&parser->info);
 135                                         php_http_buffer_cut(buffer, 0, eol_str + eol_len - buffer->data);
 136                                         php_http_header_parser_state_push(parser, 1, PHP_HTTP_HEADER_PARSER_STATE_HEADER_DONE);
 137                                 } else if ((colon = memchr(buffer->data, ':', buffer->used)) && (!eol_str || eol_str > colon)) {
 138                                         /* header: string */
 139                                         parser->_key.str = estrndup(buffer->data, parser->_key.len = colon - buffer->data);
 140                                         while (PHP_HTTP_IS_CTYPE(space, *++colon) && *colon != '\n' && *colon != '\r');
 141                                         php_http_buffer_cut(buffer, 0, colon - buffer->data);
 142                                         php_http_header_parser_state_push(parser, 1, PHP_HTTP_HEADER_PARSER_STATE_VALUE);
 143                                 } else {
 144                                         /* neither reqeust/response line nor header: string */
 145                                         return php_http_header_parser_state_push(parser, 1, PHP_HTTP_HEADER_PARSER_STATE_FAILURE);
 146                                 }
 147                                 break;
 148                         }
 149
 150                         case PHP_HTTP_HEADER_PARSER_STATE_VALUE: {
 151                                 const char *eol_str;
 152                                 int eol_len;
 153
 154 #define SET_ADD_VAL(slen, eol_len) \
 155         do { \
 156                 const char *ptr = buffer->data; \
 157                 size_t len = slen; \
 158                  \
 159                 while (len > 0 && PHP_HTTP_IS_CTYPE(space, *ptr)) { \
 160                         ++ptr; \
 161                         --len; \
 162                 } \
 163                 while (len > 0 && PHP_HTTP_IS_CTYPE(space, ptr[len - 1])) { \
 164                         --len; \
 165                 } \
 166                  \
 167                 if (len > 0) { \
 168                         if (parser->_val.str) { \
 169                                 parser->_val.str = erealloc(parser->_val.str, parser->_val.len + len + 2); \
 170                                 parser->_val.str[parser->_val.len++] = ' '; \
 171                                 memcpy(&parser->_val.str[parser->_val.len], ptr, len); \
 172                                 parser->_val.len += len; \
 173                                 parser->_val.str[parser->_val.len] = '\0'; \
 174                         } else { \
 175                                 parser->_val.len = len; \
 176                                 parser->_val.str = estrndup(ptr, len); \
 177                         } \
 178                 } \
 179                 php_http_buffer_cut(buffer, 0, slen + eol_len); \
 180         } while (0)
 181
 182                                 if ((eol_str = php_http_locate_bin_eol(buffer->data, buffer->used, &eol_len))) {
 183                                         SET_ADD_VAL(eol_str - buffer->data, eol_len);
 184
 185                                         if (buffer->used) {
 186                                                 if (*buffer->data != '\t' && *buffer->data != ' ') {
 187                                                         php_http_header_parser_state_push(parser, 1, PHP_HTTP_HEADER_PARSER_STATE_HEADER_DONE);
 188                                                         break;
 189                                                 } else {
 190                                                         php_http_header_parser_state_push(parser, 1, PHP_HTTP_HEADER_PARSER_STATE_VALUE);
 191                                                         break;
 192                                                 }
 193                                         }
 194                                 }
 195
 196                                 if (flags & PHP_HTTP_HEADER_PARSER_CLEANUP) {
 197                                         if (buffer->used) {
 198                                                 SET_ADD_VAL(buffer->used, 0);
 199                                         }
 200                                         php_http_header_parser_state_push(parser, 1, PHP_HTTP_HEADER_PARSER_STATE_HEADER_DONE);
 201                                 } else {
 202                                         return php_http_header_parser_state_push(parser, 1, PHP_HTTP_HEADER_PARSER_STATE_VALUE_EX);
 203                                 }
 204                                 break;
 205                         }
 206
 207                         case PHP_HTTP_HEADER_PARSER_STATE_VALUE_EX:
 208                                 if (*buffer->data == ' ' || *buffer->data == '\t') {
 209                                         php_http_header_parser_state_push(parser, 1, PHP_HTTP_HEADER_PARSER_STATE_VALUE);
 210                                 } else {
 211                                         php_http_header_parser_state_push(parser, 1, PHP_HTTP_HEADER_PARSER_STATE_HEADER_DONE);
 212                                 }
 213                                 break;
 214
 215                         case PHP_HTTP_HEADER_PARSER_STATE_HEADER_DONE:
 216                                 if (parser->_key.str && parser->_val.str) {
 217                                         zval array, **exist;
 218
 219                                         if (!headers && callback_func) {
 220                                                 callback_func(callback_arg, &headers, NULL TSRMLS_CC);
 221                                         }
 222
 223                                         INIT_PZVAL_ARRAY(&array, headers);
 224                                         php_http_pretty_key(parser->_key.str, parser->_key.len, 1, 1);
 225                                         if (SUCCESS == zend_symtable_find(headers, parser->_key.str, parser->_key.len + 1, (void *) &exist)) {
 226                                                 convert_to_array(*exist);
 227                                                 add_next_index_stringl(*exist, parser->_val.str, parser->_val.len, 0);
 228                                         } else {
 229                                                 add_assoc_stringl_ex(&array, parser->_key.str, parser->_key.len + 1, parser->_val.str, parser->_val.len, 0);
 230                                         }
 231                                         parser->_val.str = NULL;
 232                                 }
 233
 234                                 PTR_SET(parser->_key.str, NULL);
 235                                 PTR_SET(parser->_val.str, NULL);
 236
 237                                 php_http_header_parser_state_push(parser, 1, PHP_HTTP_HEADER_PARSER_STATE_KEY);
 238                                 break;
 239
 240                         case PHP_HTTP_HEADER_PARSER_STATE_DONE:
 241                                 return PHP_HTTP_HEADER_PARSER_STATE_DONE;
 242                 }
 243         }
 244
 245         return php_http_header_parser_state_is(parser);
 246 }
 247
 248 /*
 249  * Local variables:
 250  * tab-width: 4
 251  * c-basic-offset: 4
 252  * End:
 253  * vim600: noet sw=4 ts=4 fdm=marker
 254  * vim<600: noet sw=4 ts=4
 255  */
 256