back to dev
[m6w6/ext-http] / php_http_message_parser.c
index 23cf4f8b79357779c82be746f16b8d1e063d76e5..20ef3acc776b664e6cc4d78d7a440cd89f384d5d 100644 (file)
@@ -6,12 +6,16 @@
     | modification, are permitted provided that the conditions mentioned |
     | in the accompanying LICENSE file are met.                          |
     +--------------------------------------------------------------------+
-    | Copyright (c) 2004-2011, Michael Wallner <mike@php.net>            |
+    | Copyright (c) 2004-2014, Michael Wallner <mike@php.net>            |
     +--------------------------------------------------------------------+
 */
 
 #include "php_http_api.h"
 
+#ifndef DBG_PARSER
+#      define DBG_PARSER 0
+#endif
+
 typedef struct php_http_message_parser_state_spec {
        php_http_message_parser_state_t state;
        unsigned need_data:1;
@@ -29,7 +33,18 @@ static const php_http_message_parser_state_spec_t php_http_message_parser_states
                {PHP_HTTP_MESSAGE_PARSER_STATE_DONE,                    0}
 };
 
-PHP_HTTP_API php_http_message_parser_t *php_http_message_parser_init(php_http_message_parser_t *parser TSRMLS_DC)
+#if DBG_PARSER
+const char *php_http_message_parser_state_name(php_http_message_parser_state_t state) {
+       const char *states[] = {"START", "HEADER", "HEADER_DONE", "BODY", "BODY_DUMB", "BODY_LENGTH", "BODY_CHUNK", "BODY_DONE", "DONE"};
+       
+       if (state < 0 || state > (sizeof(states)/sizeof(char*))-1) {
+               return "FAILURE";
+       }
+       return states[state];
+}
+#endif
+
+php_http_message_parser_t *php_http_message_parser_init(php_http_message_parser_t *parser TSRMLS_DC)
 {
        if (!parser) {
                parser = emalloc(sizeof(*parser));
@@ -39,52 +54,49 @@ PHP_HTTP_API php_http_message_parser_t *php_http_message_parser_init(php_http_me
        TSRMLS_SET_CTX(parser->ts);
 
        php_http_header_parser_init(&parser->header TSRMLS_CC);
-       zend_stack_init(&parser->stack);
 
        return parser;
 }
 
-PHP_HTTP_API php_http_message_parser_state_t php_http_message_parser_state_push(php_http_message_parser_t *parser, unsigned argc, ...)
+php_http_message_parser_state_t php_http_message_parser_state_push(php_http_message_parser_t *parser, unsigned argc, ...)
 {
+       php_http_message_parser_state_t state;
        va_list va_args;
        unsigned i;
-       va_start(va_args, argc);
-       php_http_message_parser_state_t state;
 
+       /* short circuit */
+       ZEND_PTR_STACK_RESIZE_IF_NEEDED((&parser->stack), argc);
+
+       va_start(va_args, argc);
        for (i = 0; i < argc; ++i) {
                state  = va_arg(va_args, php_http_message_parser_state_t);
-               zend_stack_push(&parser->stack, &state, sizeof(state));
+               zend_ptr_stack_push(&parser->stack, (void *) state);
        }
        va_end(va_args);
 
        return state;
 }
 
-PHP_HTTP_API php_http_message_parser_state_t php_http_message_parser_state_is(php_http_message_parser_t *parser)
+php_http_message_parser_state_t php_http_message_parser_state_is(php_http_message_parser_t *parser)
 {
-       php_http_message_parser_state_t *state;
-
-       if (SUCCESS == zend_stack_top(&parser->stack, (void *) &state)) {
-               return *state;
+       if (parser->stack.top) {
+               return (php_http_message_parser_state_t) parser->stack.elements[parser->stack.top - 1];
        }
        return PHP_HTTP_MESSAGE_PARSER_STATE_START;
 }
 
-PHP_HTTP_API php_http_message_parser_state_t php_http_message_parser_state_pop(php_http_message_parser_t *parser)
+php_http_message_parser_state_t php_http_message_parser_state_pop(php_http_message_parser_t *parser)
 {
-       php_http_message_parser_state_t state, *state_ptr;
-       if (SUCCESS == zend_stack_top(&parser->stack, (void *) &state_ptr)) {
-               state = *state_ptr;
-               zend_stack_del_top(&parser->stack);
-               return state;
+       if (parser->stack.top) {
+               return (php_http_message_parser_state_t) zend_ptr_stack_pop(&parser->stack);
        }
        return PHP_HTTP_MESSAGE_PARSER_STATE_START;
 }
 
-PHP_HTTP_API void php_http_message_parser_dtor(php_http_message_parser_t *parser)
+void php_http_message_parser_dtor(php_http_message_parser_t *parser)
 {
        php_http_header_parser_dtor(&parser->header);
-       zend_stack_destroy(&parser->stack);
+       zend_ptr_stack_destroy(&parser->stack);
        if (parser->dechunk) {
                php_http_encoding_stream_free(&parser->dechunk);
        }
@@ -93,7 +105,7 @@ PHP_HTTP_API void php_http_message_parser_dtor(php_http_message_parser_t *parser
        }
 }
 
-PHP_HTTP_API void php_http_message_parser_free(php_http_message_parser_t **parser)
+void php_http_message_parser_free(php_http_message_parser_t **parser)
 {
        if (*parser) {
                php_http_message_parser_dtor(*parser);
@@ -102,55 +114,52 @@ PHP_HTTP_API void php_http_message_parser_free(php_http_message_parser_t **parse
        }
 }
 
-PHP_HTTP_API php_http_message_parser_state_t php_http_message_parser_parse_stream(php_http_message_parser_t *parser, php_stream *s, php_http_message_t **message)
+php_http_message_parser_state_t php_http_message_parser_parse_stream(php_http_message_parser_t *parser, php_stream *s, unsigned flags, php_http_message_t **message)
 {
        php_http_buffer_t buf;
+       php_http_message_parser_state_t state = PHP_HTTP_MESSAGE_PARSER_STATE_START;
        TSRMLS_FETCH_FROM_CTX(parser->ts);
 
        php_http_buffer_init_ex(&buf, 0x1000, PHP_HTTP_BUFFER_INIT_PREALLOC);
 
        while (!php_stream_eof(s)) {
                size_t len = 0;
-
-               switch (php_http_message_parser_state_is(parser)) {
+#if DBG_PARSER
+               fprintf(stderr, "#SP: %s (f:%u)\n", php_http_message_parser_state_name(state), flags);
+#endif
+               switch (state) {
                        case PHP_HTTP_MESSAGE_PARSER_STATE_START:
                        case PHP_HTTP_MESSAGE_PARSER_STATE_HEADER:
                        case PHP_HTTP_MESSAGE_PARSER_STATE_HEADER_DONE:
                                /* read line */
-                               php_stream_get_line(s, buf.data, buf.free, &len);
-                               buf.used += len;
-                               buf.free -= len;
+                               php_stream_get_line(s, buf.data + buf.used, buf.free, &len);
+                               php_http_buffer_account(&buf, len);
                                break;
 
                        case PHP_HTTP_MESSAGE_PARSER_STATE_BODY_DUMB:
                                /* read all */
-                               len = php_stream_read(s, buf.data, buf.free);
-                               buf.used += len;
-                               buf.free -= len;
+                               php_http_buffer_account(&buf, php_stream_read(s, buf.data + buf.used, buf.free));
                                break;
 
                        case PHP_HTTP_MESSAGE_PARSER_STATE_BODY_LENGTH:
                                /* read body_length */
-                               len = php_stream_read(s, buf.data, MIN(buf.free, parser->body_length));
-                               buf.used += len;
-                               buf.free -= len;
+                               php_http_buffer_account(&buf, php_stream_read(s, buf.data + buf.used, MIN(buf.free, parser->body_length)));
                                break;
 
                        case PHP_HTTP_MESSAGE_PARSER_STATE_BODY_CHUNKED:
                                /* duh, this is very naive */
                                if (len) {
-                                       size_t read = php_stream_read(s, buf.data, MIN(len, buf.free));
+                                       size_t read = php_stream_read(s, buf.data + buf.used, MIN(len, buf.free));
 
-                                       buf.used += read;
-                                       buf.free -= read;
+                                       php_http_buffer_account(&buf, read);
 
                                        len -= read;
                                } else {
+                                       php_http_buffer_resize(&buf, 24);
                                        php_stream_get_line(s, buf.data, buf.free, &len);
-                                       buf.used += len;
-                                       buf.free -= len;
+                                       php_http_buffer_account(&buf, len);
 
-                                       len = strtoul(buf.data - len, NULL, 16);
+                                       len = strtoul(buf.data + buf.used - len, NULL, 16);
                                }
                                break;
 
@@ -166,7 +175,7 @@ PHP_HTTP_API php_http_message_parser_state_t php_http_message_parser_parse_strea
                                return php_http_message_parser_state_is(parser);
                }
 
-               php_http_message_parser_parse(parser, &buf, 0, message);
+               state = php_http_message_parser_parse(parser, &buf, flags, message);
        }
 
        php_http_buffer_dtor(&buf);
@@ -174,17 +183,21 @@ PHP_HTTP_API php_http_message_parser_state_t php_http_message_parser_parse_strea
 }
 
 
-PHP_HTTP_API php_http_message_parser_state_t php_http_message_parser_parse(php_http_message_parser_t *parser, php_http_buffer_t *buffer, unsigned flags, php_http_message_t **message)
+php_http_message_parser_state_t php_http_message_parser_parse(php_http_message_parser_t *parser, php_http_buffer_t *buffer, unsigned flags, php_http_message_t **message)
 {
-       TSRMLS_FETCH_FROM_CTX(parser->ts);
        char *str = NULL;
        size_t len = 0;
        size_t cut = 0;
+       TSRMLS_FETCH_FROM_CTX(parser->ts);
 
        while (buffer->used || !php_http_message_parser_states[php_http_message_parser_state_is(parser)].need_data) {
-#if 0
-               const char *state[] = {"START", "HEADER", "HEADER_DONE", "BODY", "BODY_DUMB", "BODY_LENGTH", "BODY_CHUNK", "BODY_DONE", "DONE"};
-               fprintf(stderr, "#MP: %s (%d)\n", php_http_message_parser_state_is(parser) < 0 ? "FAILURE" : state[php_http_message_parser_state_is(parser)], message && *message ? (*message)->type : -1);
+#if DBG_PARSER
+               fprintf(stderr, "#MP: %s (f: %u, t:%d, l:%zu)\n", 
+                       php_http_message_parser_state_name(php_http_message_parser_state_is(parser)),
+                       flags, 
+                       message && *message ? (*message)->type : -1, 
+                       buffer->used
+               );
                _dpf(0, buffer->data, buffer->used);
 #endif
 
@@ -305,10 +318,16 @@ PHP_HTTP_API php_http_message_parser_state_t php_http_message_parser_parse(php_h
                                        if (h_cl) {
                                                char *stop;
 
-                                               parser->body_length = strtoul(Z_STRVAL_PP(h_cl), &stop, 10);
+                                               if (Z_TYPE_PP(h_cl) == IS_STRING) {
+                                                       parser->body_length = strtoul(Z_STRVAL_PP(h_cl), &stop, 10);
 
-                                               if (stop != Z_STRVAL_PP(h_cl)) {
-                                                       php_http_message_parser_state_push(parser, 1, PHP_HTTP_MESSAGE_PARSER_STATE_BODY_LENGTH);
+                                                       if (stop != Z_STRVAL_PP(h_cl)) {
+                                                               php_http_message_parser_state_push(parser, 1, !parser->body_length?PHP_HTTP_MESSAGE_PARSER_STATE_BODY_DONE:PHP_HTTP_MESSAGE_PARSER_STATE_BODY_LENGTH);
+                                                               break;
+                                                       }
+                                               } else if (Z_TYPE_PP(h_cl) == IS_LONG) {
+                                                       parser->body_length = Z_LVAL_PP(h_cl);
+                                                       php_http_message_parser_state_push(parser, 1, !parser->body_length?PHP_HTTP_MESSAGE_PARSER_STATE_BODY_DONE:PHP_HTTP_MESSAGE_PARSER_STATE_BODY_LENGTH);
                                                        break;
                                                }
                                        }
@@ -334,7 +353,7 @@ PHP_HTTP_API php_http_message_parser_state_t php_http_message_parser_parse(php_h
 
                                                                if (end >= start && (!total || end < total)) {
                                                                        parser->body_length = end + 1 - start;
-                                                                       php_http_message_parser_state_push(parser, 1, PHP_HTTP_MESSAGE_PARSER_STATE_BODY_LENGTH);
+                                                                       php_http_message_parser_state_push(parser, 1, !parser->body_length?PHP_HTTP_MESSAGE_PARSER_STATE_BODY_DONE:PHP_HTTP_MESSAGE_PARSER_STATE_BODY_LENGTH);
                                                                        break;
                                                                }
                                                        }
@@ -371,11 +390,11 @@ PHP_HTTP_API php_http_message_parser_state_t php_http_message_parser_parse(php_h
                                                len = dec_len;
                                        }
 
-                                       php_stream_write(php_http_message_body_stream(&(*message)->body), str, len);
+                                       php_stream_write(php_http_message_body_stream((*message)->body), str, len);
 
                                        /* keep track */
                                        MAKE_STD_ZVAL(zcl);
-                                       ZVAL_LONG(zcl, php_http_message_body_size(&(*message)->body));
+                                       ZVAL_LONG(zcl, php_http_message_body_size((*message)->body));
                                        zend_hash_update(&(*message)->hdrs, "Content-Length", sizeof("Content-Length"), &zcl, sizeof(zval *), NULL);
                                }
 
@@ -429,14 +448,14 @@ PHP_HTTP_API php_http_message_parser_state_t php_http_message_parser_parse(php_h
                                size_t dec_len;
 
                                if (SUCCESS != php_http_encoding_stream_update(parser->dechunk, buffer->data, buffer->used, &dec_str, &dec_len)) {
-                                       return FAILURE;
+                                       return PHP_HTTP_MESSAGE_PARSER_STATE_FAILURE;
                                }
 
                                str = dec_str;
                                len = dec_len;
 
                                if (php_http_encoding_stream_done(parser->dechunk)) {
-                                       cut = buffer->used - PHP_HTTP_BUFFER_LEN(parser->dechunk->ctx);
+                                       cut = buffer->used - PHP_HTTP_BUFFER(parser->dechunk->ctx)->used;
                                        php_http_message_parser_state_push(parser, 2, PHP_HTTP_MESSAGE_PARSER_STATE_BODY_DONE, PHP_HTTP_MESSAGE_PARSER_STATE_BODY);
                                } else {
                                        cut = buffer->used;
@@ -477,6 +496,10 @@ PHP_HTTP_API php_http_message_parser_state_t php_http_message_parser_parse(php_h
                                }
 
                                php_http_buffer_cut(buffer, 0, ptr - buffer->data);
+                               
+                               if (!(flags & PHP_HTTP_MESSAGE_PARSER_GREEDY)) {
+                                       return PHP_HTTP_MESSAGE_PARSER_STATE_DONE;
+                               }
                                break;
                        }
                }