X-Git-Url: https://git.m6w6.name/?p=m6w6%2Fext-http;a=blobdiff_plain;f=php_http_header_parser.c;h=da02ff5f765237b76ba79d477be637c588691f7c;hp=2ad902280897ee852982238bd435d7668289fe3b;hb=629c7b270b047582160c87b7c688c2b942a75d60;hpb=29a54250b58e444974ae19840194e214cab80bd5 diff --git a/php_http_header_parser.c b/php_http_header_parser.c index 2ad9022..da02ff5 100644 --- a/php_http_header_parser.c +++ b/php_http_header_parser.c @@ -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 | + | Copyright (c) 2004-2014, Michael Wallner | +--------------------------------------------------------------------+ */ #include "php_http_api.h" +#ifndef DBG_PARSER +# define DBG_PARSER 0 +#endif + typedef struct php_http_header_parser_state_spec { php_http_header_parser_state_t state; unsigned need_data:1; @@ -21,12 +25,12 @@ static const php_http_header_parser_state_spec_t php_http_header_parser_states[] {PHP_HTTP_HEADER_PARSER_STATE_START, 1}, {PHP_HTTP_HEADER_PARSER_STATE_KEY, 1}, {PHP_HTTP_HEADER_PARSER_STATE_VALUE, 1}, + {PHP_HTTP_HEADER_PARSER_STATE_VALUE_EX, 0}, {PHP_HTTP_HEADER_PARSER_STATE_HEADER_DONE, 0}, {PHP_HTTP_HEADER_PARSER_STATE_DONE, 0} }; - -PHP_HTTP_API php_http_header_parser_t *php_http_header_parser_init(php_http_header_parser_t *parser TSRMLS_DC) +php_http_header_parser_t *php_http_header_parser_init(php_http_header_parser_t *parser TSRMLS_DC) { if (!parser) { parser = emalloc(sizeof(*parser)); @@ -38,53 +42,52 @@ PHP_HTTP_API php_http_header_parser_t *php_http_header_parser_init(php_http_head return parser; } - -PHP_HTTP_API php_http_header_parser_state_t php_http_header_parser_state_push(php_http_header_parser_t *parser, unsigned argc, ...) +php_http_header_parser_state_t php_http_header_parser_state_push(php_http_header_parser_t *parser, unsigned argc, ...) { va_list va_args; unsigned i; php_http_header_parser_state_t state = 0; + /* 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_header_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_header_parser_state_t php_http_header_parser_state_is(php_http_header_parser_t *parser) +php_http_header_parser_state_t php_http_header_parser_state_is(php_http_header_parser_t *parser) { - php_http_header_parser_state_t *state; - - if (SUCCESS == zend_stack_top(&parser->stack, (void *) &state)) { - return *state; + if (parser->stack.top) { + return (php_http_header_parser_state_t) parser->stack.elements[parser->stack.top - 1]; } + return PHP_HTTP_HEADER_PARSER_STATE_START; } -PHP_HTTP_API php_http_header_parser_state_t php_http_header_parser_state_pop(php_http_header_parser_t *parser) +php_http_header_parser_state_t php_http_header_parser_state_pop(php_http_header_parser_t *parser) { - php_http_header_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_header_parser_state_t) zend_ptr_stack_pop(&parser->stack); } + return PHP_HTTP_HEADER_PARSER_STATE_START; } -PHP_HTTP_API void php_http_header_parser_dtor(php_http_header_parser_t *parser) +void php_http_header_parser_dtor(php_http_header_parser_t *parser) { - zend_stack_destroy(&parser->stack); + zend_ptr_stack_destroy(&parser->stack); php_http_info_dtor(&parser->info); - STR_FREE(parser->_key.str); - STR_FREE(parser->_val.str); + PTR_FREE(parser->_key.str); + PTR_FREE(parser->_val.str); } -PHP_HTTP_API void php_http_header_parser_free(php_http_header_parser_t **parser) +void php_http_header_parser_free(php_http_header_parser_t **parser) { if (*parser) { php_http_header_parser_dtor(*parser); @@ -93,14 +96,14 @@ PHP_HTTP_API void php_http_header_parser_free(php_http_header_parser_t **parser) } } -PHP_HTTP_API 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) +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) { TSRMLS_FETCH_FROM_CTX(parser->ts); while (buffer->used || !php_http_header_parser_states[php_http_header_parser_state_is(parser)].need_data) { -#if 0 - const char *state[] = {"START", "KEY", "VALUE", "HEADER_DONE", "DONE"}; - fprintf(stderr, "#HP: %s (%d)\n", php_http_header_parser_state_is(parser) < 0 ? "FAILURE" : state[php_http_header_parser_state_is(parser)], zend_hash_num_elements(headers)); +#if DBG_PARSER + const char *state[] = {"START", "KEY", "VALUE", "VALUE_EX", "HEADER_DONE", "DONE"}; + fprintf(stderr, "#HP: %s (avail:%zu, num:%d cleanup:%u)\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, flags); _dpf(0, buffer->data, buffer->used); #endif switch (php_http_header_parser_state_pop(parser)) { @@ -120,8 +123,8 @@ PHP_HTTP_API STATUS php_http_header_parser_parse(php_http_header_parser_t *parse } case PHP_HTTP_HEADER_PARSER_STATE_KEY: { - const char *colon, *eol_str; - int eol_len; + const char *colon, *eol_str = NULL; + int eol_len = 0; if (buffer->data == (eol_str = php_http_locate_bin_eol(buffer->data, buffer->used, &eol_len))) { /* end of headers */ @@ -138,12 +141,15 @@ PHP_HTTP_API STATUS php_http_header_parser_parse(php_http_header_parser_t *parse } else if ((colon = memchr(buffer->data, ':', buffer->used)) && (!eol_str || eol_str > colon)) { /* header: string */ parser->_key.str = estrndup(buffer->data, parser->_key.len = colon - buffer->data); - while (PHP_HTTP_IS_CTYPE(space, *++colon)); + while (PHP_HTTP_IS_CTYPE(space, *++colon) && *colon != '\n' && *colon != '\r'); php_http_buffer_cut(buffer, 0, colon - buffer->data); php_http_header_parser_state_push(parser, 1, PHP_HTTP_HEADER_PARSER_STATE_VALUE); - } else { + } else if (flags & PHP_HTTP_HEADER_PARSER_CLEANUP) { /* neither reqeust/response line nor header: string */ return php_http_header_parser_state_push(parser, 1, PHP_HTTP_HEADER_PARSER_STATE_FAILURE); + } else { + /* keep feeding */ + return php_http_header_parser_state_push(parser, 1, PHP_HTTP_HEADER_PARSER_STATE_KEY); } break; } @@ -152,38 +158,67 @@ PHP_HTTP_API STATUS php_http_header_parser_parse(php_http_header_parser_t *parse const char *eol_str; int eol_len; - do { - if ((eol_str = php_http_locate_bin_eol(buffer->data, buffer->used, &eol_len))) { - if (eol_str + eol_len - buffer->data < buffer->used) { - char nextline = *(eol_str + eol_len); - - if (nextline == '\t' || nextline == ' ') { - php_http_buffer_cut(buffer, eol_str - buffer->data, eol_len); - continue; - } - } - - parser->_val.str = estrndup(buffer->data, parser->_val.len = eol_str - buffer->data); - php_http_buffer_cut(buffer, 0, eol_str + eol_len - buffer->data); - php_http_header_parser_state_push(parser, 1, PHP_HTTP_HEADER_PARSER_STATE_HEADER_DONE); - } else if (flags & PHP_HTTP_HEADER_PARSER_CLEANUP) { - if (buffer->used) { - parser->_val.str = estrndup(buffer->data, parser->_val.len = buffer->used); - php_http_buffer_reset(buffer); - } - php_http_header_parser_state_push(parser, 1, PHP_HTTP_HEADER_PARSER_STATE_HEADER_DONE); - } else { - return php_http_header_parser_state_push(parser, 1, PHP_HTTP_HEADER_PARSER_STATE_VALUE); +#define SET_ADD_VAL(slen, eol_len) \ + do { \ + const char *ptr = buffer->data; \ + size_t len = slen; \ + \ + while (len > 0 && PHP_HTTP_IS_CTYPE(space, *ptr)) { \ + ++ptr; \ + --len; \ + } \ + while (len > 0 && PHP_HTTP_IS_CTYPE(space, ptr[len - 1])) { \ + --len; \ + } \ + \ + if (len > 0) { \ + if (parser->_val.str) { \ + parser->_val.str = erealloc(parser->_val.str, parser->_val.len + len + 2); \ + parser->_val.str[parser->_val.len++] = ' '; \ + memcpy(&parser->_val.str[parser->_val.len], ptr, len); \ + parser->_val.len += len; \ + parser->_val.str[parser->_val.len] = '\0'; \ + } else { \ + parser->_val.len = len; \ + parser->_val.str = estrndup(ptr, len); \ + } \ + } \ + php_http_buffer_cut(buffer, 0, slen + eol_len); \ + } while (0) + + if ((eol_str = php_http_locate_bin_eol(buffer->data, buffer->used, &eol_len))) { + SET_ADD_VAL(eol_str - buffer->data, eol_len); + php_http_header_parser_state_push(parser, 1, PHP_HTTP_HEADER_PARSER_STATE_VALUE_EX); + } else if (flags & PHP_HTTP_HEADER_PARSER_CLEANUP) { + if (buffer->used) { + SET_ADD_VAL(buffer->used, 0); } - } while (0); - + php_http_header_parser_state_push(parser, 1, PHP_HTTP_HEADER_PARSER_STATE_HEADER_DONE); + } else { + return php_http_header_parser_state_push(parser, 1, PHP_HTTP_HEADER_PARSER_STATE_VALUE); + } break; } + case PHP_HTTP_HEADER_PARSER_STATE_VALUE_EX: + if (buffer->used && (*buffer->data == ' ' || *buffer->data == '\t')) { + php_http_header_parser_state_push(parser, 1, PHP_HTTP_HEADER_PARSER_STATE_VALUE); + } else if (buffer->used || (flags & PHP_HTTP_HEADER_PARSER_CLEANUP)) { + php_http_header_parser_state_push(parser, 1, PHP_HTTP_HEADER_PARSER_STATE_HEADER_DONE); + } else { + /* keep feeding */ + return php_http_header_parser_state_push(parser, 1, PHP_HTTP_HEADER_PARSER_STATE_VALUE_EX); + } + break; + case PHP_HTTP_HEADER_PARSER_STATE_HEADER_DONE: if (parser->_key.str && parser->_val.str) { zval array, **exist; + if (!headers && callback_func) { + callback_func(callback_arg, &headers, NULL TSRMLS_CC); + } + INIT_PZVAL_ARRAY(&array, headers); php_http_pretty_key(parser->_key.str, parser->_key.len, 1, 1); if (SUCCESS == zend_symtable_find(headers, parser->_key.str, parser->_key.len + 1, (void *) &exist)) { @@ -195,8 +230,8 @@ PHP_HTTP_API STATUS php_http_header_parser_parse(php_http_header_parser_t *parse parser->_val.str = NULL; } - STR_SET(parser->_key.str, NULL); - STR_SET(parser->_val.str, NULL); + PTR_SET(parser->_key.str, NULL); + PTR_SET(parser->_val.str, NULL); php_http_header_parser_state_push(parser, 1, PHP_HTTP_HEADER_PARSER_STATE_KEY); break; @@ -209,6 +244,119 @@ PHP_HTTP_API STATUS php_http_header_parser_parse(php_http_header_parser_t *parse return php_http_header_parser_state_is(parser); } + +zend_class_entry *php_http_header_parser_class_entry; +static zend_object_handlers php_http_header_parser_object_handlers; + +zend_object_value php_http_header_parser_object_new(zend_class_entry *ce TSRMLS_DC) +{ + return php_http_header_parser_object_new_ex(ce, NULL, NULL TSRMLS_CC); +} + +zend_object_value php_http_header_parser_object_new_ex(zend_class_entry *ce, php_http_header_parser_t *parser, php_http_header_parser_object_t **ptr TSRMLS_DC) +{ + php_http_header_parser_object_t *o; + + o = ecalloc(1, sizeof(php_http_header_parser_object_t)); + zend_object_std_init((zend_object *) o, ce TSRMLS_CC); + object_properties_init((zend_object *) o, ce); + + if (ptr) { + *ptr = o; + } + + if (parser) { + o->parser = parser; + } else { + o->parser = php_http_header_parser_init(NULL TSRMLS_CC); + } + o->buffer = php_http_buffer_new(); + + o->zv.handle = zend_objects_store_put((zend_object *) o, NULL, php_http_header_parser_object_free, NULL TSRMLS_CC); + o->zv.handlers = &php_http_header_parser_object_handlers; + + return o->zv; +} + +void php_http_header_parser_object_free(void *object TSRMLS_DC) +{ + php_http_header_parser_object_t *o = (php_http_header_parser_object_t *) object; + + if (o->parser) { + php_http_header_parser_free(&o->parser); + } + if (o->buffer) { + php_http_buffer_free(&o->buffer); + } + zend_object_std_dtor((zend_object *) o TSRMLS_CC); + efree(o); +} + +ZEND_BEGIN_ARG_INFO_EX(ai_HttpHeaderParser_getState, 0, 0, 0) +ZEND_END_ARG_INFO(); +static PHP_METHOD(HttpHeaderParser, getState) +{ + php_http_header_parser_object_t *parser_obj = zend_object_store_get_object(getThis() TSRMLS_CC); + + zend_parse_parameters_none(); + /* always return the real state */ + RETVAL_LONG(php_http_header_parser_state_is(parser_obj->parser)); +} + +ZEND_BEGIN_ARG_INFO_EX(ai_HttpHeaderParser_parse, 0, 0, 3) + ZEND_ARG_INFO(0, data) + ZEND_ARG_INFO(0, flags) + ZEND_ARG_ARRAY_INFO(1, headers, 1) +ZEND_END_ARG_INFO(); +static PHP_METHOD(HttpHeaderParser, parse) +{ + php_http_header_parser_object_t *parser_obj; + zval *zmsg; + char *data_str; + int data_len; + long flags; + + php_http_expect(SUCCESS == zend_parse_parameters(ZEND_NUM_ARGS() TSRMLS_CC, "slz", &data_str, &data_len, &flags, &zmsg), invalid_arg, return); + + if (Z_TYPE_P(zmsg) != IS_ARRAY) { + zval_dtor(zmsg); + array_init(zmsg); + } + parser_obj = zend_object_store_get_object(getThis() TSRMLS_CC); + php_http_buffer_append(parser_obj->buffer, data_str, data_len); + RETVAL_LONG(php_http_header_parser_parse(parser_obj->parser, parser_obj->buffer, flags, Z_ARRVAL_P(zmsg), NULL, NULL)); +} + + +static zend_function_entry php_http_header_parser_methods[] = { + PHP_ME(HttpHeaderParser, getState, ai_HttpHeaderParser_getState, ZEND_ACC_PUBLIC) + PHP_ME(HttpHeaderParser, parse, ai_HttpHeaderParser_parse, ZEND_ACC_PUBLIC) + {NULL, NULL, NULL} +}; + +PHP_MINIT_FUNCTION(http_header_parser) +{ + zend_class_entry ce; + + INIT_NS_CLASS_ENTRY(ce, "http\\Header", "Parser", php_http_header_parser_methods); + php_http_header_parser_class_entry = zend_register_internal_class(&ce TSRMLS_CC); + memcpy(&php_http_header_parser_object_handlers, zend_get_std_object_handlers(), sizeof(zend_object_handlers)); + php_http_header_parser_class_entry->create_object = php_http_header_parser_object_new; + php_http_header_parser_object_handlers.clone_obj = NULL; + + zend_declare_class_constant_long(php_http_header_parser_class_entry, ZEND_STRL("CLEANUP"), PHP_HTTP_HEADER_PARSER_CLEANUP TSRMLS_CC); + + zend_declare_class_constant_long(php_http_header_parser_class_entry, ZEND_STRL("STATE_FAILURE"), PHP_HTTP_HEADER_PARSER_STATE_FAILURE TSRMLS_CC); + zend_declare_class_constant_long(php_http_header_parser_class_entry, ZEND_STRL("STATE_START"), PHP_HTTP_HEADER_PARSER_STATE_START TSRMLS_CC); + zend_declare_class_constant_long(php_http_header_parser_class_entry, ZEND_STRL("STATE_KEY"), PHP_HTTP_HEADER_PARSER_STATE_KEY TSRMLS_CC); + zend_declare_class_constant_long(php_http_header_parser_class_entry, ZEND_STRL("STATE_VALUE"), PHP_HTTP_HEADER_PARSER_STATE_VALUE TSRMLS_CC); + zend_declare_class_constant_long(php_http_header_parser_class_entry, ZEND_STRL("STATE_VALUE_EX"), PHP_HTTP_HEADER_PARSER_STATE_VALUE_EX TSRMLS_CC); + zend_declare_class_constant_long(php_http_header_parser_class_entry, ZEND_STRL("STATE_HEADER_DONE"), PHP_HTTP_HEADER_PARSER_STATE_HEADER_DONE TSRMLS_CC); + zend_declare_class_constant_long(php_http_header_parser_class_entry, ZEND_STRL("STATE_DONE"), PHP_HTTP_HEADER_PARSER_STATE_DONE TSRMLS_CC); + + return SUCCESS; +} + /* * Local variables: * tab-width: 4