--- parser3/src/targets/cgi/parser3.C 2003/03/21 07:08:28 1.217 +++ parser3/src/targets/cgi/parser3.C 2026/10/27 23:38:52 1.388 @@ -1,598 +1,798 @@ /** @file Parser: scripting and CGI main. - Copyright(c) 2001, 2003 ArtLebedev Group (http://www.artlebedev.com) - Author: Alexandr Petrosian (http://paf.design.ru) + Copyright (c) 2001-2026 Art. Lebedev Studio (https://www.artlebedev.com) + Authors: Konstantin Morshnev , Alexandr Petrosian */ -static const char* IDENT_PARSER3_C="$Date: 2003/03/21 07:08:28 $"; +volatile const char * IDENT_PARSER3_C="$Id: parser3.C,v 1.388 2026/10/27 23:38:52 moko Exp $"; #include "pa_config_includes.h" -#if _MSC_VER -# include -# include -#endif - #include "pa_sapi.h" #include "classes.h" #include "pa_common.h" #include "pa_request.h" -#include "pa_socks.h" #include "pa_version.h" +#include "pa_threads.h" +#include "pa_vconsole.h" +#include "pa_sapi_info.h" +#include "pa_code_cache.h" +#include "pa_websocket.h" +#include "pa_os.h" -#include "pool_storage.h" +#ifdef _MSC_VER +#include +#include +#include + +extern "C" HANDLE WINAPI GC_CreateThread(LPSECURITY_ATTRIBUTES, DWORD, LPTHREAD_START_ROUTINE, LPVOID, DWORD, LPDWORD); -#ifdef WIN32 -# include -# include "getopt.h" #else -# include + +extern "C" int GC_pthread_create(pthread_t *, const pthread_attr_t *, void *(*)(void *), void * /* arg */); + #endif -//#define DEBUG_POOL_MALLOC -//#define DEBUG_STRING_APPENDS_VS_EXPANDS -//#define DEBUG_MAILRECEIVE "mailreceive.eml" - -#ifdef DEBUG_STRING_APPENDS_VS_EXPANDS -extern ulong - string_piece_appends, - wcontext_result_size, - total_alloc_size, - string_string_shortcut_economy; +// defines + +// comment remove me after debugging +//#define PA_DEBUG_CGI_ENTRY_EXIT + +#if defined(_MSC_VER) && !defined(_DEBUG) +# define PA_SUPPRESS_SYSTEM_EXCEPTION #endif // consts #define REDIRECT_PREFIX "REDIRECT_" #define PARSER_CONFIG_ENV_NAME "CGI_PARSER_CONFIG" +#define PARSER_LOG_ENV_NAME "CGI_PARSER_LOG" -/// IIS refuses to read bigger chunks -const size_t READ_POST_CHUNK_SIZE=0x400*0x400; // 1M +SAPI_Info sapi_console; +SAPI_Info_CGI sapi_cgi; -static const char *argv0; -static const char *config_filespec_cstr=0; -static bool fail_on_config_read_problem=true; - -static Pool_storage global_pool_storage; ///< global pool storage -static Pool global_pool(&global_pool_storage); ///< global pool -static bool cgi; ///< we were started as CGI? -static bool mail_received=false; ///< we were started with -m option? [asked to parse incoming message to $mail:received] +static SAPI_Info *sapi_info = &sapi_cgi; +static THREAD_LOCAL SAPI_Info *sapi_info_4log = NULL; // global for correct send error in die() -// for signal handlers -Request *request=0; +const char* parser3_mode = "cgi"; // $status:mode +static const char* filespec_to_process = 0; // [file] +static const char* httpd_host_port = 0; // -p option +static const char* config_filespec = 0; // -f option or from env or next to the executable if exists +static const char* log_filespec = 0; // -l option or from env, full disk path +static bool mail_received = false; // -m option? [asked to parse incoming message to $mail:received] +static const char* parser3_filespec = 0; // argv[0] +static char** argv_extra = NULL; + +// for error logging +static THREAD_LOCAL Request_info *request_info_4log = NULL; // global for correct log() reporting +static const char* filespec_4log = NULL; // null only if system-wide auto.p used; full disk path, parser3.log goes next to it + +const char *parser3_log_filespec() { // $status:log-filename + const char* slog=log_filespec; + + if(!slog) { + static THREAD_LOCAL char log_spec[MAX_STRING + 12 /* '/parser3.log' */]; + pa_strncpy(log_spec, filespec_4log, MAX_STRING); + + if(char* log_dir_pos=strrchr(log_spec, '/')){ + strcpy(log_dir_pos, "/parser3.log"); + } else { + // system-wide auto.p used: the current directory + strcpy(log_spec, "./parser3.log"); + } + slog=log_spec; + } + return slog; +} // SAPI -static void log(const char *fmt, va_list args) { - bool opened=false; - FILE *f=0; - - if(config_filespec_cstr) { - char beside_config_path[MAX_STRING]; - strncpy(beside_config_path, config_filespec_cstr, MAX_STRING-1); beside_config_path[MAX_STRING-1]=0; - if(!( - rsplit(beside_config_path, '/') || - rsplit(beside_config_path, '\\'))) { // strip filename - // no path, just filename - beside_config_path[0]='.'; beside_config_path[1]=0; - } - - char file_spec[MAX_STRING]; - snprintf(file_spec, MAX_STRING, - "%s/parser3.log", beside_config_path); - f=fopen(file_spec, "at"); - opened=f!=0; - } - // fallback to stderr - if(!opened) - f=stderr; +#ifndef O_CLOEXEC +#define O_CLOEXEC 0 +#endif +#ifndef STDERR_FILENO +#define STDERR_FILENO 2 +#endif - // use no memory [so that we could log out-of-memory error] - setbuf(f, 0); // stderr stream is unbuffered by default, but still... +static void pa_log(const char* stamp, const char* fmt, va_list args) { + // use no memory (to log out-of-memory error): the line is built on the stack and written at once + char line[64 + MAX_LOG_STRING + 1024 + 32 + 100]; // prefix + message + request info suffix: uri + method and the rest of it + size_t size=0; // prefix - time_t t=time(0); - if(const char *stamp=ctime(&t)) { // never saw that - if(size_t len=strlen(stamp)) // saw once stamp being ="" - fprintf(f, "[%.*s] ", len-1, stamp); - } - // message + if(!stamp) + stamp=pa_ctime(time(0)); + if(size_t len=strlen(stamp)) // saw once stamp being ="" + size=pa_snprintf(line, 64, "[%.*s] [%u] ", (int)len-1 /* exclude \n */, stamp, (unsigned int)pa_get_thread_id() ); - char buf[MAX_STRING]; - size_t size=vsnprintf(buf, MAX_STRING, fmt, args); - remove_crlf(buf, buf+size); - - fwrite(buf, size, 1, f); - // newline - fprintf(f, "\n"); + // message + size+=remove_crlf(line+size, line+size+pa_vsnprintf(line+size, MAX_LOG_STRING, fmt, args)); - if(opened) - fclose(f); + // suffix + if(request_info_4log && request_info_4log->method) + size+=pa_snprintf(line+size, sizeof(line)-size, " [uri=%.1024s, method=%.32s, cl=%lu]\n", request_info_4log->uri ? request_info_4log->uri : "", request_info_4log->method, (unsigned long)request_info_4log->content_length); else - fflush(f); + size+=pa_snprintf(line+size, sizeof(line)-size, " [no request info]\n"); + + const char* slog=parser3_log_filespec(); + int fd=strcmp(slog,"-") ? pa_open(slog, O_WRONLY | O_APPEND | O_CREAT | O_CLOEXEC, 0666) : -1; + if(fd<0) + fd=STDERR_FILENO; // fallback to stderr + PA_UNUSED int ignore_result=write(fd, line, (unsigned int)size); + if(fd!=STDERR_FILENO) + close(fd); } -// appends to parser3.log located beside my binary if openable, to stderr otherwize -void SAPI::log(Pool& , const char *fmt, ...) { - va_list args; +void pa_log(const char* fmt, ...) { + va_list args; va_start(args,fmt); - ::log(fmt, args); + pa_log(0, fmt, args); va_end(args); } -void SAPI::die(const char *fmt, ...) { -#ifdef DEBUG_POOL_MALLOC - extern void log_pool_stats(Pool& pool); - log_pool_stats(global_pool); -#endif +// for signal handlers, where pa_ctime can't be called: the stamp is prepared in advance +static void pa_log_at(const char* stamp, const char* fmt, ...) { + va_list args; + va_start(args,fmt); + pa_log(stamp, fmt, args); + va_end(args); +} - va_list args; +// appends to parser3.log located next to the config file if openable, to stderr otherwize +void SAPI::log(SAPI_Info&, const char* fmt, ...) { + va_list args; va_start(args,fmt); - // log + pa_log(0, fmt, args); + va_end(args); +} - // logging is more important than user - // she can cancel download, we'd get SIGPIPE, - // nothing would be logged then - ::log(fmt, args); +void SAPI::die(const char* fmt, ...) { + va_list args; - // inform user + // logging first, first vsnprintf + va_start(args,fmt); + pa_log(0, fmt, args); + va_end(args); - char body[MAX_STRING]; - int content_length=vsnprintf(body, MAX_STRING, fmt, args); + // inform user, second vsnprintf + va_start(args, fmt); + char message[MAX_STRING]; + pa_vsnprintf(message, MAX_STRING, fmt, args); - va_end(args); + SAPI::send_error(sapi_info_4log ? *sapi_info_4log : *sapi_info, message); + exit(1); +// va_end(args); +} - // prepare header - // let's be honest, that's bad we couldn't produce valid output - SAPI::add_header_attribute(global_pool, "status", "500"); - SAPI::add_header_attribute(global_pool, "content-type", "text/plain"); - char content_length_cstr[MAX_NUMBER]; - snprintf(content_length_cstr, sizeof(content_length_cstr), "%u", content_length); - SAPI::add_header_attribute(global_pool, "content-length", content_length_cstr); +void SAPI::send_error(SAPI_Info& info, const char *exception_cstr, const char *status){ + info.send_error(exception_cstr, status); +} - // send header - SAPI::send_header(global_pool); +char* SAPI::Env::get(SAPI_Info& info, const char* name) { + return info.get_env(name); +} - // body - SAPI::send_body(global_pool, body, content_length); +bool SAPI::Env::set(SAPI_Info& info, const char* name, const char* value) { + return info.set_env(name, value); +} -#ifdef WIN32 - // IIS with abort failes to show STDOUT, it just barks "abnormal program termination" - exit(1); -#else - // exit & try to produce core dump - abort(); -#endif +const char* const *SAPI::Env::get(SAPI_Info& info) { + return info.get_env(); } -const char *SAPI::get_env(Pool& , const char *name) { - return getenv(name); +size_t SAPI::read_post(SAPI_Info& info, char* buf, size_t max_bytes) { + return info.read_post(buf, max_bytes); } -const char *const *SAPI::environment(Pool&) { -#ifdef _MSC_VER - extern char **_environ; - return _environ; -#else - extern char **environ; - return environ; -#endif +void SAPI::add_header_attribute(SAPI_Info& info, const char* dont_store_key, const char* dont_store_value) { + info.add_header(dont_store_key, dont_store_value); } -size_t SAPI::read_post(Pool& , char *buf, size_t max_bytes) { - size_t read_size=0; - do { - ssize_t chunk_size=read(fileno(stdin), - buf+read_size, min(READ_POST_CHUNK_SIZE, max_bytes-read_size)); - if(chunk_size<=0) - break; - read_size+=chunk_size; - } while(read_sizeexpired && timeout->kill_process){ + // the request is still running after the grace period + pa_log_at(timeout->stamp, "response timeout (%d s) exceeded, process killed", timeout->seconds); +#ifndef _MSC_VER + kill(getpid(), SIGKILL); #endif + return; + } + + timeout->expired=1; + if(thread_request) + thread_request->set_skip(Request::SKIP_INTERRUPTED); + } +}; + +THREAD_LOCAL Response_timeout *Response_timeout::current=0; + +void SAPI::set_response_timeout(SAPI_Info&, int seconds) { + if(Response_timeout::current) + Response_timeout::current->set(seconds); } -static void log_signal(const char *signal_name) { - SAPI::log(global_pool, request? "%s received. uri=%s, qs=%s" - :"%s received. no request being processed", - signal_name, - request && request->info.uri?request->info.uri:"-", - request && request->info.query_string?request->info.query_string:"-"); +class RequestInfoController { +public: + RequestInfoController(Request_info* rinfo, SAPI_Info* sinfo){ + request_info_4log=rinfo; + sapi_info_4log=sinfo; + } + ~RequestInfoController(){ + request_info_4log=0; + sapi_info_4log=0; + } +}; + +/** httpd support */ +static const String httpd_class_name("httpd"); + +static void config_handler(SAPI_Info &info) { + Request_info request_info; + RequestInfoController ric(&request_info, &info); + + request_info.set_document_root(full_disk_path()); + request_info.uri = ""; + request_info.argv = argv_extra; + + // prepare to process request + Request r(info, request_info, String::Language(String::L_HTML|String::L_OPTIMIZE_BIT)); + // only once + config_filespec = locate_config(config_filespec, parser3_filespec); + // process main auto.p only +#ifdef CODE_CACHE + Code_cache_manager::start(); +#endif + r.core(config_filespec, String::Empty); +#ifdef CODE_CACHE + Code_cache_manager::finish(r); +#endif } -#ifdef SIGUSR1 -static void SIGUSR1_handler(int /*sig*/){ - log_signal("SIGUSR1"); +static void connection_handler(SAPI_Info_HTTPD &info, HTTPD_Connection &connection) { + Request_info request_info; + RequestInfoController ric(&request_info, &info); + + try { + if(!connection.read_header()) + return; // ignore "void" connections + info.populate_env(); + + request_info.set_document_root(full_disk_path()); + request_info.path_translated = filespec_to_process; + request_info.method = connection.method(); + request_info.query_string = connection.query(); + request_info.uri = request_info.strip_absolute_uri(connection.uri()); + request_info.content_type = connection.content_type(); + request_info.content_length = (size_t)connection.content_length(); + request_info.cookie = info.get_env("HTTP_COOKIE"); + request_info.mail_received = false; + request_info.argv = argv_extra; + + // process can be killed on timeout only in parallel mode + Response_timeout response_timeout(HTTPD_Server::mode == HTTPD_Server::PARALLEL); + + // prepare to process request + Request r(info, request_info, String::Language(String::L_HTML|String::L_OPTIMIZE_BIT)); + WebSocket websocket(connection); + r.websocket=&websocket; + // process the request + r.core(config_filespec, main_method_name, &httpd_class_name); + } catch(const Exception& e) { // exception in connection handling + SAPI::log(info, "%s", e.comment()); + const char* status = info.exception_http_status(e.type()); + if(*status) + SAPI::send_error(info, e.comment(), status); + } } + +#ifdef _MSC_VER +DWORD WINAPI connection_thread(void *arg){ +#else +static void *connection_thread(void *arg){ #endif + Temp_gc_references gc_references; // for pointers GC does not see, the very start of the thread -#ifdef SIGPIPE -static void SIGPIPE_handler(int /*sig*/){ - log_signal("SIGPIPE"); - if(request) - request->interrupt(); + HTTPD_Connection &connection=*(HTTPD_Connection*)arg; + SAPI_Info_HTTPD info(connection); + + try { + connection_handler(info, connection); + } catch(const Exception& e) { // exception in send_error + pa_log("%s", e.comment()); + } + + delete(&connection); + return 0; } + +static void reject_connection(HTTPD_Connection &connection) { + // main thread only + static unsigned long total=0; + static time_t logged=0; + + total++; + time_t now=time(0); + if(now!=logged){ // not more than once a second + logged=now; + pa_log("max-connections limit %d reached, rejecting connection from %s (%lu rejected in total)", HTTPD_Server::max_connections, connection.remote_addr, total); + } + + SAPI_Info_HTTPD info(connection); + SAPI::send_error(info, "too many connections", "503"); +} + +static void httpd_mode() { + config_handler(*sapi_info); + + SOCKET sock = HTTPD_Server::bind(httpd_host_port); + + while(1){ +#ifndef _MSC_VER + pid_t pid=1; #endif + try { + HTTPD_Connection connection; + bool accepted=connection.accept(sock, 500); +#ifndef _MSC_VER + if(HTTPD_Server::mode == HTTPD_Server::PARALLEL) + while (waitpid((pid_t)(-1), 0, WNOHANG) > 0) + HTTPD_Server::track_connection(-1); // the child has exited, some of them while we were waiting +#endif + if(!accepted) + continue; -/** -main workhorse + if(HTTPD_Server::too_many_connections()){ + reject_connection(connection); // no thread or process is started + continue; + } - @todo - IIS: remove trailing default-document[index.html] from $request.uri. - to do that we need to consult metabase, - wich is tested but seems slow. - IIS5 todo find out proper 'illegal call' check -*/ -static void real_parser_handler( - const char *filespec_to_process, - const char *request_method, bool header_only) { - // init socks - init_socks(global_pool); - - // init global classes - init_methoded_array(global_pool); - // init global variables - pa_globals_init(global_pool); - - // request pool, must be different ptr from global [used in VStateless_class.add_method] - Pool request_pool(&global_pool_storage); + switch (HTTPD_Server::mode) { + case HTTPD_Server::MULTITHREADED: +#ifdef _MSC_VER + if (!GC_CreateThread(0, 0, connection_thread, new HTTPD_Connection(connection), 0, 0)) + throw Exception("httpd.fork", 0, "thread creation failed"); + connection.sock=INVALID_SOCKET; // thread owns the connection now + break; +#else +#ifdef HAVE_TLS + pthread_t thread; + pthread_attr_t attr; + pthread_attr_init(&attr); + pthread_attr_setdetachstate(&attr, PTHREAD_CREATE_DETACHED); + + if(int result=GC_pthread_create(&thread, &attr, connection_thread, new HTTPD_Connection(connection))) + throw Exception("httpd.fork", 0, "thread creation failed (%d)", result); + connection.sock=INVALID_SOCKET; // thread owns the connection now + break; +#endif + case HTTPD_Server::PARALLEL: + pid=fork(); + if(pid<0) + throw Exception("httpd.fork", 0, "fork failed: %s (%d)", strerror(errno), errno); + if(pid>0){ + closesocket(connection.sock); // parent should close connection.sock as well + connection.sock=INVALID_SOCKET; // child owns the connection now + continue; + } +#endif + case HTTPD_Server::SEQUENTIAL: // and fork child - if(!filespec_to_process || !*filespec_to_process) - SAPI::die("Parser/%s", PARSER_VERSION); - - // Request info - Request::Info request_info; - char document_root_buf[MAX_STRING]; - if(cgi) { - if(const char *env_document_root=SAPI::get_env(request_pool, "DOCUMENT_ROOT")) - request_info.document_root=env_document_root; - else if(const char *path_info=SAPI::get_env(request_pool, "PATH_INFO")) { - // IIS - size_t len=min(sizeof(document_root_buf)-1, strlen(filespec_to_process)-strlen(path_info)); - memcpy(document_root_buf, filespec_to_process, len); document_root_buf[len]=0; - request_info.document_root=document_root_buf; - } else - throw Exception("parser.runtime", - 0, - "CGI: no PATH_INFO defined(in reinventing DOCUMENT_ROOT)"); - } else { - full_file_spec("", document_root_buf, sizeof(document_root_buf)); - request_info.document_root=document_root_buf; + SAPI_Info_HTTPD info(connection); + connection_handler(info, connection); + } + // SEQUENTIAL or error: closing connection socket in HTTPD_Connection destructor + } catch(const Exception& e) { // exception in accept or send_error + pa_log("%s", e.comment()); + } + +#ifndef _MSC_VER + if(pid==0) // fork child + exit(0); +#endif } - request_info.path_translated=filespec_to_process; - request_info.method=request_method ? request_method : "GET"; - const char *query_string=SAPI::get_env(request_pool, "QUERY_STRING"); - request_info.query_string=query_string; +} + +/** main workhorse */ + +static void real_parser_handler(bool cgi) { + // init libraries + pa_globals_init(); + + if(httpd_host_port){ + httpd_mode(); + } + + const char* request_method=getenv("REQUEST_METHOD"); + + if(!filespec_to_process){ + SAPI::send_error(*sapi_info, "Parser/" PARSER_VERSION_WITH_FEATURES); + exit(1); + } + + // global request info + Request_info request_info; + RequestInfoController ric(&request_info, sapi_info); + + request_info.path_translated = filespec_to_process; + request_info.method = request_method ? request_method : "GET"; + request_info.query_string = MAYBE_RECONSTRUCT_IIS_STATUS_IN_QS(getenv("QUERY_STRING")); + if(cgi) { - // few absolute obligatory - const char *path_info=SAPI::get_env(request_pool, "PATH_INFO"); + // obligatory + const char* path_info=getenv("PATH_INFO"); if(!path_info) - SAPI::die("CGI: illegal call (missing PATH_INFO)"); - const char *script_name=SAPI::get_env(request_pool, "SCRIPT_NAME"); - if(!script_name) - SAPI::die("CGI: illegal call (missing SCRIPT_NAME)"); - - const char *env_request_uri=SAPI::get_env(request_pool, "REQUEST_URI"); - if(env_request_uri) - request_info.uri=env_request_uri; - else - if(query_string) { - char *reconstructed_uri=(char *)request_pool.malloc( - strlen(path_info)+1/*'?'*/+ - strlen(query_string)+1/*0*/); - strcpy(reconstructed_uri, path_info); - strcat(reconstructed_uri, "?"); - strcat(reconstructed_uri, query_string); - request_info.uri=reconstructed_uri; - } else - request_info.uri=path_info; + SAPI::die("parser3: illegal CGI call (missing PATH_INFO)"); + + const char* document_root = getenv("DOCUMENT_ROOT"); + if(!document_root) { + // IIS or fcgiwrap minimalistic setup + ssize_t prefix_len = strlen(filespec_to_process) - strlen(path_info); + if(prefix_len < 0 || strcmp(filespec_to_process + prefix_len, path_info) != 0) + SAPI::die("parser3: illegal CGI call (invalid PATH_INFO in reinventing DOCUMENT_ROOT)"); + + document_root = pa_strdup(filespec_to_process, prefix_len); + } + request_info.set_document_root(document_root); - if(env_request_uri) { // apache & others stuck to standards + request_info.uri = request_info.strip_absolute_uri(getenv("REQUEST_URI")); + if(request_info.uri) { // apache & others stuck to standards + // another obligatory + const char* script_name = getenv("SCRIPT_NAME"); + if(!script_name) + SAPI::die("parser3: illegal CGI call (missing SCRIPT_NAME)"); /* http://parser3/env.html?123 =OK $request:uri=/env.html?123 REQUEST_URI='/env.html?123' SCRIPT_NAME='/cgi-bin/parser3' PATH_INFO='/env.html' - + http://parser3/cgi-bin/parser3/env.html?123 =ERROR $request:uri=/cgi-bin/parser3/env.html?123 REQUEST_URI='/cgi-bin/parser3/env.html?123' SCRIPT_NAME='/cgi-bin/parser3' PATH_INFO='/env.html' */ - size_t script_name_len=strlen(script_name); - size_t uri_len=strlen(env_request_uri); - if(strncmp(env_request_uri, script_name, script_name_len)==0 && - script_name_len != uri_len) // under IIS they are the same - SAPI::die("CGI: illegal call (1)"); - } else { // seen on IIS5 - /* - http://nestle/env.html?123 =OK - $request:uri=/env.html?123 - REQUEST_URI='' - SCRIPT_NAME='/env.html' - PATH_INFO='/env.html' - - http://nestle/cgi-bin/parser3.exe/env.html =ERROR - $request:uri=/env.html - REQUEST_URI='' - SCRIPT_NAME='/cgi-bin/parser3.exe' - PATH_INFO='/env.html' - */ - if(strcmp(script_name, path_info)!=0) - SAPI::die("CGI: illegal call (2)"); - } - } else - request_info.uri=""; - - request_info.content_type=SAPI::get_env(request_pool, "CONTENT_TYPE"); - const char *content_length=SAPI::get_env(request_pool, "CONTENT_LENGTH"); - request_info.content_length=(content_length?atoi(content_length):0); - request_info.cookie=SAPI::get_env(request_pool, "HTTP_COOKIE"); - request_info.mail_received=mail_received; - - - // prepare to process request - Request request(request_pool, - request_info, -/*#ifdef _DEBUG - String::UL_HTML|String::UL_OPTIMIZE_BIT -#else*/ - cgi ? String::UL_HTML|String::UL_OPTIMIZE_BIT : String::UL_AS_IS -/*#endif*/ - , - true /* status_allowed */); - - // get request ptr for signal handlers - ::request=&request; - - char config_filespec_buf[MAX_STRING]; - if(!config_filespec_cstr) { - const char *config_by_env=getenv(PARSER_CONFIG_ENV_NAME); - if(!config_by_env) - config_by_env=getenv(REDIRECT_PREFIX PARSER_CONFIG_ENV_NAME); - if(config_by_env) - config_filespec_cstr=config_by_env; - else { - // beside by binary - char beside_binary_path[MAX_STRING]; - strncpy(beside_binary_path, argv0, MAX_STRING-1); beside_binary_path[MAX_STRING-1]=0; // filespec of my binary - if(!( - rsplit(beside_binary_path, '/') || - rsplit(beside_binary_path, '\\'))) { // strip filename - // no path, just filename - // @todo full path, not ./! - beside_binary_path[0]='.'; beside_binary_path[1]=0; - } - snprintf(config_filespec_buf, MAX_STRING, - "%s/%s", - beside_binary_path, AUTO_FILE_NAME); - config_filespec_cstr=config_filespec_buf; - fail_on_config_read_problem=entry_exists(config_filespec_cstr); + size_t script_name_len = strlen(script_name); + size_t uri_len = strlen(request_info.uri); + if(strncmp(request_info.uri, script_name, script_name_len)==0 && script_name_len != uri_len) // under IIS they are the same + SAPI::die("parser3: illegal CGI call (REQUEST_URI starts with SCRIPT_NAME)"); + } else { // fcgiwrap minimalistic setup + request_info.uri = request_info.query_string && *request_info.query_string ? pa_strcat(path_info, "?", request_info.query_string) : path_info; } + } else{ + request_info.set_document_root(full_disk_path()); + request_info.uri = ""; } - // process the request - request.core( - config_filespec_cstr, fail_on_config_read_problem, - header_only); + request_info.content_type = getenv("CONTENT_TYPE"); + request_info.content_length = cgi ? (size_t)pa_atoul(getenv("CONTENT_LENGTH")) : 0; // only SAPI_Info_CGI can read POST + request_info.cookie = getenv("HTTP_COOKIE"); + request_info.mail_received = mail_received; + + request_info.argv = argv_extra; + +#ifdef PA_DEBUG_CGI_ENTRY_EXIT + pa_log("request_info: method=%s, uri=%s, q=%s, dr=%s, pt=%s", request_info.method, request_info.uri, request_info.query_string, request_info.document_root, request_info.path_translated); +#endif + + { + // CGI process can be killed on timeout + Response_timeout response_timeout(true); + + // prepare to process request + Request r(*sapi_info, request_info, cgi ? String::Language(String::L_HTML|String::L_OPTIMIZE_BIT) : String::L_AS_IS); + // process the request + r.core(locate_config(config_filespec, parser3_filespec)); + // then the request is destroyed, before the libraries are unloaded: its pool cleanups call them (memcached) + } - // no request [prevent signal handlers from accessing invalid memory] - ::request=0; - - // - done_socks(); - -#ifdef DEBUG_POOL_MALLOC - extern void log_pool_stats(Pool& pool); - log_pool_stats(request_pool); -#endif + // finalize libraries + pa_globals_done(); +} -#ifdef DEBUG_STRING_APPENDS_VS_EXPANDS - SAPI::log(global_pool, - "string piece appends=%lu, wcontext_result_size=%lu, string_string_shortcut_economy_closer=%lu, total_alloc_size=%lu", - string_piece_appends, - wcontext_result_size, - string_string_shortcut_economy, - total_alloc_size); -#endif +#ifdef PA_SUPPRESS_SYSTEM_EXCEPTION +static const Exception call_real_parser_handler__do_PEH_return_it(bool cgi) { + try { + real_parser_handler(cgi); + } catch(const Exception& e) { + return e; + } + return Exception(); } -static void call_real_parser_handler__do_SEH( - const char *filespec_to_process, - const char *request_method, bool header_only) { -#if _MSC_VER && !defined(_DEBUG) +static void call_real_parser_handler__supress_system_exception(bool cgi) { + Exception parser_exception; LPEXCEPTION_POINTERS system_exception=0; + __try { -#endif - real_parser_handler( - filespec_to_process, - request_method, header_only); - -#if _MSC_VER && !defined(_DEBUG) - } __except ( - (system_exception=GetExceptionInformation()), - EXCEPTION_EXECUTE_HANDLER) { - + parser_exception=call_real_parser_handler__do_PEH_return_it(cgi); + } __except ( (system_exception=GetExceptionInformation()), EXCEPTION_EXECUTE_HANDLER) { if(system_exception) if(_EXCEPTION_RECORD *er=system_exception->ExceptionRecord) - throw Exception(0, - 0, - "Exception 0x%08X at 0x%08X", er->ExceptionCode, er->ExceptionAddress); - else - throw Exception(0, 0, "Exception "); + throw Exception("system", 0, "0x%08X at %p", er->ExceptionCode, er->ExceptionAddress); else - throw Exception(0, 0, "Exception "); + throw Exception("system", 0, ""); + else + throw Exception("system", 0, ""); } -#endif -} -#if _MSC_VER -int failed_new(size_t size) { - SAPI::die("out of memory in 'new', failed to allocated %u bytes", size); - return 0; // not reached + if(parser_exception) + throw Exception(parser_exception); } -#endif -#ifdef HAVE_SET_NEW_HANDLER -static void failed_new() { - SAPI::die("out of memory in 'new'"); -} +#define REAL_PARSER_HANDLER call_real_parser_handler__supress_system_exception +#else +#define REAL_PARSER_HANDLER real_parser_handler #endif -static void usage(const char *program) { +static void usage(const char* message=NULL) { + if(message){ + fprintf(stderr, message, parser3_filespec); + } + printf( - "Parser/%s Copyright(c) 2001, 2003 ArtLebedev Group (http://www.artlebedev.com)\n" - "Author: Alexandr Petrosian (http://paf.design.ru)\n" + "Parser/%s\n" + "Copyright (c) 2001-2026 Art. Lebedev Studio (https://www.artlebedev.com)\n" + "Authors: Konstantin Morshnev , Alexandr Petrosian \n" "\n" - "Usage: %s [options] file\n" + "Usage: %s [options] [file]\n" "Options are:\n" #ifdef WITH_MAILRECEIVE " -m Parse mail, put received letter to $mail:received\n" #endif " -f config_file Use this config file (/path/to/auto.p)\n" - " -h Display usage information (this message)\n" - , PARSER_VERSION, - program); + " -l log_file Use this log file (/path/to/parser3.log)\n" + " -p [host:]port Start web server on this port\n" + " -h Display usage information (this message)\n", + PARSER_VERSION_WITH_FEATURES, + parser3_filespec ? parser3_filespec : "parser3" ); exit(EINVAL); } - -class C { -public: - C() {} -}; -C *trick() { - return new C; -} +#define ARG_REQUIRED \ + if(c[1] || !*(++carg)){ \ + fprintf(stderr, "%s: option '%c' requires an argument\n", parser3_filespec, *c); \ + usage(); \ + } int main(int argc, char *argv[]) { - delete trick(); + Temp_gc_references gc_references; // for pointers GC does not see -#ifdef SIGUSR1 - if(signal(SIGUSR1, SIGUSR1_handler)==SIG_ERR) - SAPI::die("Can not set handler for SIGUSR1"); -#endif -#ifdef SIGPIPE - if(signal(SIGPIPE, SIGPIPE_handler)==SIG_ERR) - SAPI::die("Can not set handler for SIGPIPE"); +#ifdef PA_DEBUG_CGI_ENTRY_EXIT + pa_log("main: entry"); #endif - -#ifdef DEBUG_MAILRECEIVE - if(FILE *fake_in=fopen(DEBUG_MAILRECEIVE, "rt")) { - dup2(fake_in->_file, 0/*STDIN_FILENO*/); - } -#endif - -#ifdef _DEBUG -// _crtBreakAlloc=33112; -#endif -// _asm int 3; - argv0=argv[0]; + if(!argc || !argv[0]) + usage(); + parser3_filespec = backslashes_to_slashes(argv[0]); umask(2); // were we started as CGI? - cgi= - getenv("SERVER_SOFTWARE") || - getenv("SERVER_NAME") || - getenv("GATEWAY_INTERFACE") || - getenv("REQUEST_METHOD"); - - char *raw_filespec_to_process; + bool cgi=(getenv("SERVER_SOFTWARE") || getenv("SERVER_NAME") || getenv("GATEWAY_INTERFACE") || getenv("REQUEST_METHOD")) && !getenv("PARSER_VERSION"); + if(!cgi){ + sapi_info = &sapi_console; + parser3_mode = "console"; + } + +#ifdef SIGPIPE + signal(SIGPIPE, SIGPIPE_handler); +#endif + char* raw_filespec_to_process = NULL; if(cgi) { raw_filespec_to_process=getenv("PATH_TRANSLATED"); - if(raw_filespec_to_process && !*raw_filespec_to_process) - raw_filespec_to_process=0; + argv_extra=argv + 1; } else { - optind = 1; - opterr = 0; - int c; - while((c = getopt(argc, argv, "hf:" -#ifdef WITH_MAILRECEIVE - "m" -#endif - )) > 0) { - switch (c) { - case 'h': - usage(argv[0]); - break; - case 'f': - config_filespec_cstr=optarg; + char** carg = argv + 1; + for(;*carg; carg++){ + if((*carg)[0] != '-') break; + + for(char* c=(*carg)+1; *c; c++){ + switch (*c) { + case 'h': + usage(); + break; + case 'f': + ARG_REQUIRED; + config_filespec=*carg; + break; + case 'l': + ARG_REQUIRED; + log_filespec=*carg; + break; + case 'p': + ARG_REQUIRED; + httpd_host_port=*carg; + parser3_mode="httpd"; + break; #ifdef WITH_MAILRECEIVE - case 'm': - mail_received=true; - break; -#endif - default: - fprintf(stderr, "%s: invalid option '%c'\n", argv[0], optopt); - usage(argv[0]); - break; + case 'm': + mail_received=true; + parser3_mode="mail"; + break; +#endif + default: + fprintf(stderr, "%s: invalid option '%c'\n", parser3_filespec, *c); + usage(); + break; + } } } - if (optind != argc - 1) { - fprintf(stderr, "%s: file not specified\n", argv[0]); - usage(argv[0]); - } - raw_filespec_to_process=argv[optind++]; + if (*carg) { + raw_filespec_to_process=*carg; + } else { + if(!httpd_host_port) + usage("%s: file not specified\n"); + } + + if (httpd_host_port && mail_received) + usage("%s: -p and -m options should not be used together\n"); + + argv_extra=carg; } -#ifdef WIN32 + // prepare full disk path log_filespec here, as logging must not allocate + if(!log_filespec) + log_filespec=getenv(PARSER_LOG_ENV_NAME); + if(!log_filespec) + log_filespec=getenv(REDIRECT_PREFIX PARSER_LOG_ENV_NAME); + if(log_filespec && strcmp(log_filespec, "-")) // "-" is stderr + log_filespec=full_disk_path(log_filespec); + +#ifdef _MSC_VER setmode(fileno(stdin), _O_BINARY); setmode(fileno(stdout), _O_BINARY); setmode(fileno(stderr), _O_BINARY); #endif -#if _MSC_VER - _set_new_handler(failed_new); - -#ifdef _DEBUG +#if defined(_MSC_VER) && defined(_DEBUG) // Get current flag int tmpFlag = _CrtSetDbgFlag( _CRTDBG_REPORT_FLAG ); @@ -601,38 +801,24 @@ int main(int argc, char *argv[]) { // Set flag to the new value _CrtSetDbgFlag( tmpFlag ); -// _CrtSetBreakAlloc(471); - -#endif + _CrtSetReportMode( _CRT_WARN, _CRTDBG_MODE_FILE ); + _CrtSetReportFile( _CRT_WARN, _CRTDBG_FILE_STDERR ); #endif -#ifdef HAVE_SET_NEW_HANDLER - std::set_new_handler(failed_new); -#endif - - char filespec_to_process[MAX_STRING]; - full_file_spec(raw_filespec_to_process, filespec_to_process, sizeof(filespec_to_process)); - - const char *request_method=getenv("REQUEST_METHOD"); - bool header_only=request_method && strcasecmp(request_method, "HEAD")==0; - try { // global try - call_real_parser_handler__do_SEH( - filespec_to_process, - request_method, header_only); - } catch(const Exception& e) { // global problem - // don't allocate anything on pool here: - // possible pool' exception not catch-ed now - // and there could be out-of-memory exception - - SAPI::die("exception in request exception handler: %s", e.comment()); -#ifndef _DEBUG - } catch(...) { - SAPI::die(""); -#endif + if(raw_filespec_to_process && *raw_filespec_to_process){ + filespec_to_process=full_disk_path(raw_filespec_to_process); + } + + REAL_PARSER_HANDLER(cgi); + } catch(const Exception& e) { // exception in config_handler + SAPI::log(*sapi_info, "%s", e.comment()); + SAPI::send_error(*sapi_info, e.comment(), strcmp(e.type(), "file.missing") ? "500" : "404"); } - //_asm int 3; - return 0; +#ifdef PA_DEBUG_CGI_ENTRY_EXIT + pa_log("main: successful return"); +#endif + return sapi_info->http_response_code < 100 ? sapi_info->http_response_code : 0; }