--- parser3/src/targets/cgi/parser3.C 2007/08/17 08:16:11 1.243 +++ parser3/src/targets/cgi/parser3.C 2024/12/11 18:35:31 1.366 @@ -1,300 +1,228 @@ /** @file Parser: scripting and CGI main. - Copyright(c) 2001-2005 ArtLebedev Group (http://www.artlebedev.com) - Author: Alexandr Petrosian (http://paf.design.ru) + Copyright (c) 2001-2024 Art. Lebedev Studio (http://www.artlebedev.com) + Authors: Konstantin Morshnev , Alexandr Petrosian */ -static const char * const IDENT_PARSER3_C="$Date: 2007/08/17 08:16:11 $"; +volatile const char * IDENT_PARSER3_C="$Id: parser3.C,v 1.366 2024/12/11 18:35:31 moko Exp $"; #include "pa_config_includes.h" -#if _MSC_VER -# 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" + +#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 // defines // comment remove me after debugging -//#define PA_DEBUG_CGI_ENTRY_EXIT "c:\\parser\\debug-parser3.log" +//#define PA_DEBUG_CGI_ENTRY_EXIT -#if _MSC_VER && !defined(_DEBUG) +#if defined(_MSC_VER) && !defined(_DEBUG) # define PA_SUPPRESS_SYSTEM_EXCEPTION #endif -//#define DEBUG_MAILRECEIVE "mailreceive.eml" - // 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 - -static const char* argv0; -static const char* config_filespec_cstr=0; -static bool fail_on_config_read_problem=true; - -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] - -// for signal handlers -Request *request=0; -Request_info *request_info=0; -bool execution_canceled=false; - -// SAPI +SAPI_Info sapi_console; +SAPI_Info_CGI sapi_cgi; -class SAPI_Info{} SAPI_info; +static SAPI_Info *sapi_info = &sapi_cgi; +static THREAD_LOCAL SAPI_Info *sapi_info_4log = NULL; // global for correct send error in die() -static void log(const char* fmt, va_list args) { - bool opened=false; - FILE *f=0; - - const char* log_by_env=getenv(PARSER_LOG_ENV_NAME); - if(!log_by_env) - log_by_env=getenv(REDIRECT_PREFIX PARSER_LOG_ENV_NAME); - if(log_by_env) { - f=fopen(log_by_env, "at"); - opened=f!=0; - } -#ifdef PA_DEBUG_CGI_ENTRY_EXIT - f=fopen(PA_DEBUG_CGI_ENTRY_EXIT, "at"); - opened=f!=0; -#endif - - if(!opened && 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 +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 +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 + +template static T *dir_pos(T *fname){ + T *result=NULL; + while (fname=strpbrk(fname, "/\\")){ + result=fname; + fname++; + } + return result; +} + +const char *parser3_log_filespec() { // $status:log-filename + const char* slog=log_filespec; + + if(!slog) + slog=getenv(PARSER_LOG_ENV_NAME); + if(!slog) + slog=getenv(REDIRECT_PREFIX PARSER_LOG_ENV_NAME); + if(!slog) { + static char log_spec[MAX_STRING + 12 /* '/parser3.log' */]; + pa_strncpy(log_spec, filespec_4log, MAX_STRING); + + if(char* log_dir_pos=dir_pos(log_spec)){ + strcpy(log_dir_pos, "/parser3.log"); + } else { // no path, just filename - beside_config_path[0]='.'; beside_config_path[1]=0; + strcpy(log_spec, "./parser3.log"); } - - char file_spec[MAX_STRING]; - snprintf(file_spec, MAX_STRING, - "%s/parser3.log", beside_config_path); - f=fopen(file_spec, "at"); - opened=f!=0; + slog=log_spec; } + return slog; +} + +// SAPI + +static void pa_log(const char* fmt, va_list args) { + // use no memory [so that we could log out-of-memory error] + const char* slog=parser3_log_filespec(); + FILE *f=strcmp(slog,"-") ? fopen(slog, "at") : stderr; + // fallback to stderr - if(!opened) + if(!f) f=stderr; - // use no memory [so that we could log out-of-memory error] setbuf(f, 0); // stderr stream is unbuffered by default, but still... // 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] [%u] ", len-1, stamp, - (unsigned int)getpid() - ); + fprintf(f, "[%.*s] [%u] ", (int)len-1, stamp, (unsigned int)pa_get_thread_id() ); } - // message - char buf[MAX_STRING]; - size_t size=vsnprintf(buf, MAX_STRING, fmt, args); - remove_crlf(buf, buf+size); + // message + char buf[MAX_LOG_STRING]; + size_t size=vsnprintf(buf, MAX_LOG_STRING, fmt, args); + size=remove_crlf(buf, buf+size); fwrite(buf, size, 1, f); - if(request_info) - fprintf(f, " [uri=%s, method=%s, cl=%u]", - request_info->uri? request_info->uri: "", - request_info->method? request_info->method: "", - request_info->content_length); - else - fputs(" [no request info]", f); - - // newline - fputs("\n", f); + if(request_info_4log && request_info_4log->method) { + fprintf(f, " [uri=%s, method=%s, cl=%lu]\n", request_info_4log->uri ? request_info_4log->uri : "", request_info_4log->method, (unsigned long)request_info_4log->content_length); + } else + fputs(" [no request info]\n", f); - if(opened) + if(f!=stderr) fclose(f); else fflush(f); } -#ifdef PA_DEBUG_CGI_ENTRY_EXIT -static void log(const char* fmt, ...) { - va_list args; + +void pa_log(const char* fmt, ...) { + va_list args; va_start(args,fmt); - log(fmt, args); + pa_log(fmt, args); va_end(args); } -#endif -// appends to parser3.log located beside my binary if openable, to stderr otherwize +// 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_list args; va_start(args,fmt); - ::log(fmt, args); + pa_log(fmt, args); va_end(args); } -static void die_or_abort(const char* fmt, va_list args, bool write_core) { - // log +void SAPI::die(const char* fmt, ...) { + va_list args; - // logging is more important than user - // she can cancel download, we'd get SIGPIPE, - // nothing would be logged then - ::log(fmt, args); - - // inform user - - char body[MAX_STRING]; - int content_length=vsnprintf(body, MAX_STRING, fmt, args); - - // prepare header - // let's be honest, that's bad we couldn't produce valid output - SAPI::add_header_attribute(SAPI_info, "status", "500"); - SAPI::add_header_attribute(SAPI_info, "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(SAPI_info, "content-length", content_length_cstr); - - // send header - SAPI::send_header(SAPI_info); - - // body - SAPI::send_body(SAPI_info, body, content_length); - - // exit & try to produce core dump[unix] or invoke debugger[Win32 Debug version] - if(write_core) { -#if defined(WIN32) && !defined(_DEBUG) - // IIS with abort failes to show STDOUT, it just barks "abnormal program termination" - exit(1); -#else -#if _MSC_VER - _asm int 3; -#endif - abort(); -#endif - } - else - exit(1); -} + // logging first, first vsnprintf + va_start(args,fmt); + pa_log(fmt, args); + va_end(args); -void SAPI::die(const char* fmt, ...) { - va_list args; + // inform user, second vsnprintf va_start(args, fmt); - die_or_abort(fmt, args, false/*write core?*/); -//unreachable anyway va_end(args); -} + char message[MAX_STRING]; + vsnprintf(message, MAX_STRING, fmt, args); -void SAPI::abort(const char* fmt, ...) { - va_list args; - va_start(args, fmt); - die_or_abort(fmt, args, true/*write core?*/); -//unreachable anyway va_end(args); + SAPI::send_error(sapi_info_4log ? *sapi_info_4log : *sapi_info, message); + exit(1); +// va_end(args); } -char* SAPI::get_env(SAPI_Info& , const char* name) { - if(char *local=getenv(name)) - return pa_strdup(local); - else - return 0; +void SAPI::send_error(SAPI_Info& info, const char *exception_cstr, const char *status){ + info.send_error(exception_cstr, status); } -const char* const *SAPI::environment(SAPI_Info&) { -#ifdef _MSC_VER - extern char **_environ; - return _environ; -#else - extern char **environ; - return environ; -#endif +char* SAPI::Env::get(SAPI_Info& info, const char* name) { + return info.get_env(name); } -size_t SAPI::read_post(SAPI_Info& , 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_sizeconsole.was_used()) ) - printf("%s: %s\n", dont_store_key, dont_store_value); +size_t SAPI::read_post(SAPI_Info& info, char* buf, size_t max_bytes) { + return info.read_post(buf, max_bytes); } -/// @todo intelligent cache-control -void SAPI::send_header(SAPI_Info& ) { - if(cgi) { -// puts("expires: Fri, 23 Mar 2001 09:32:23 GMT"); +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); +} - // header | body delimiter - puts(""); - } +void SAPI::send_headers(SAPI_Info& info) { + info.send_headers(); } -size_t SAPI::send_body(SAPI_Info& , const void *buf, size_t size) { - return stdout_write(buf, size); +void SAPI::clear_headers(SAPI_Info& info) { + info.clear_headers(); } -// +size_t SAPI::send_body(SAPI_Info& info, const void *buf, size_t size) { + return info.send_body(buf, size); +} -static void full_file_spec(const char* file_name, char *buf, size_t buf_size) { - if(file_name) - if(file_name[0]=='/' +static const char* full_disk_path(const char* file_name = "") { + char* result; + if(file_name[0]=='/' #ifdef WIN32 - || file_name[0] && file_name[1]==':' + || file_name[0] && file_name[1]==':' #endif - ) - strncpy(buf, file_name, buf_size); - else { - char cwd[MAX_STRING]; getcwd(cwd, MAX_STRING); - snprintf(buf, buf_size, "%s/%s", cwd, file_name); - } - else - buf[0]=0; + ){ + result=pa_strdup(file_name); + } else { + char cwd[MAX_STRING]; + result=pa_strcat(getcwd(cwd, MAX_STRING) ? cwd : "", "/", file_name); + } #ifdef WIN32 - back_slashes_to_slashes(buf); + back_slashes_to_slashes(result); #endif + return result; } static void log_signal(const char* signal_name) { - if(request_info) - SAPI::log(SAPI_info, "%s received while %s. uri=%s, method=%s, cl=%u", - signal_name, - request?"executing code":"reading data", - request_info->uri, - request_info->method, - request_info->content_length); - else - SAPI::log(SAPI_info, "%s received before or after processing request", - signal_name); -} - -#ifdef SIGUSR1 -static void SIGUSR1_handler(int /*sig*/){ - log_signal("SIGUSR1"); + pa_log("%s received %s processing request", signal_name, request ? "while" : "before or after"); } -#endif #ifdef SIGPIPE #define SIGPIPE_NAME "SIGPIPE" @@ -302,34 +230,57 @@ static const String sigpipe_name(SIGPIPE static void SIGPIPE_handler(int /*sig*/){ Value* sigpipe=0; if(request) - sigpipe=request->main_class.get_element(sigpipe_name, request->main_class, false); + sigpipe=request->main_class.get_element(sigpipe_name); if(sigpipe && sigpipe->as_bool()) - log_signal("SIGPIPE"); + log_signal(SIGPIPE_NAME); - execution_canceled=true; if(request) - request->set_interrupted(true); + request->set_skip(Request::SKIP_INTERRUPTED); } #endif +// requires pa_thread_request() in entry_exists() under Windows +static const char* locate_config(const char* config_filespec_option, const char* executable_path){ + filespec_4log=config_filespec_option; + if(!filespec_4log) + filespec_4log=getenv(PARSER_CONFIG_ENV_NAME); + if(!filespec_4log) + filespec_4log=getenv(REDIRECT_PREFIX PARSER_CONFIG_ENV_NAME); + if(!filespec_4log){ + const char* exec_dir_pos = dir_pos(executable_path); +#ifdef SYSTEM_CONFIG_FILE + if(exec_dir_pos){ +#endif + // next to the executable + if(!exec_dir_pos || (exec_dir_pos==executable_path+1 && *executable_path=='.')){ + // when just parser3 or ./parser3 full path should be used to avoid "parser already configured" + filespec_4log=full_disk_path(AUTO_FILE_NAME); + } else { + filespec_4log=pa_strcat(pa_strdup(executable_path, exec_dir_pos - executable_path), "/" AUTO_FILE_NAME); + } + if(entry_exists(filespec_4log)) + return filespec_4log; +#ifdef SYSTEM_CONFIG_FILE + } + if(entry_exists(SYSTEM_CONFIG_FILE)){ + filespec_4log=NULL; + return SYSTEM_CONFIG_FILE; + } +#endif + return NULL; + } + return filespec_4log; +} + #ifdef WIN32 -const char* maybe_reconstruct_IIS_status_in_qs(const char* original) -{ +static const char* maybe_reconstruct_IIS_status_in_qs(const char* original) { // 404;http://servername/page[?param=value...] // ';' should be urlencoded by HTTP standard, so we shouldn't get it from browser // and can consider that as an indication that this is IIS way to report errors - if(original - && isdigit((unsigned char)original[0]) - && isdigit((unsigned char)original[1]) - && isdigit((unsigned char)original[2]) - && original[3]==';') - { + if(original && isdigit((unsigned char)original[0]) && isdigit((unsigned char)original[1]) && isdigit((unsigned char)original[2]) && original[3]==';'){ size_t original_len=strlen(original); - char* reconstructed=new(PointerFreeGC) char[original_len - +12/*IIS-STATUS=&*/ - +14/*IIS-DOCUMENT=&*/ - +1]; + char* reconstructed=new(PointerFreeGC) char[original_len +12/*IIS-STATUS=&*/ +14/*IIS-DOCUMENT=&*/ +1]; char* cur=reconstructed; memcpy(cur, "IIS-STATUS=", 11); cur+=11; memcpy(cur, original, 3); cur+=3; @@ -338,7 +289,7 @@ const char* maybe_reconstruct_IIS_status const char* qmark_at=strchr(original, '?'); memcpy(cur, "IIS-DOCUMENT=", 13); cur+=13; { - size_t value_len=(qmark_at? qmark_at-original: original_len)-4; + size_t value_len=(qmark_at ? qmark_at-original : original_len)-4; memcpy(cur, original+4, value_len); cur+=value_len; } @@ -353,359 +304,417 @@ const char* maybe_reconstruct_IIS_status return original; } + +static const char* maybe_back_slashes_to_slashes(const char* original){ + char *result=pa_strdup(original); + back_slashes_to_slashes(result); + return result; +} + +#define MAYBE_RECONSTRUCT_IIS_STATUS_IN_QS(s) maybe_reconstruct_IIS_status_in_qs(s) +#define MAYBE_BACK_SLASHES_TO_SLASHES(s) maybe_back_slashes_to_slashes(s) + +#else + +#define MAYBE_RECONSTRUCT_IIS_STATUS_IN_QS(s) s +#define MAYBE_BACK_SLASHES_TO_SLASHES(s) s + #endif -/** -main workhorse +class RequestController { +public: + RequestController(Request* r){ + request=r; + } + ~RequestController(){ + request=0; + } +}; - @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. -*/ -static void real_parser_handler(const char* filespec_to_process, - const char* request_method, bool header_only) -{ - // init socks - pa_socks_init(); +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; + } +}; - // init global variables - pa_globals_init(); - - if(!filespec_to_process || !*filespec_to_process) - SAPI::die("Parser/%s" -#ifdef PA_DEBUG_CGI_ENTRY_EXIT - " with entry/exit tracing" +/** 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.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 + r.core(config_filespec, false, String::Empty); +} + +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.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; + + // prepare to process request + Request r(info, request_info, String::Language(String::L_HTML|String::L_OPTIMIZE_BIT)); + // process the request + r.core(config_filespec, strcasecmp(request_info.method, "HEAD")==0, 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 - , PARSER_VERSION); - - // Request info - Request_info request_info; memset(&request_info, 0, sizeof(request_info)); - char document_root_buf[MAX_STRING]; - if(cgi) { - if(const char* env_document_root=getenv("DOCUMENT_ROOT")) - request_info.document_root=env_document_root; - else if(const char* path_info=getenv("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; + 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()); } - request_info.path_translated=filespec_to_process; - request_info.method=request_method ? request_method : "GET"; - const char* query_string= -#ifdef WIN32 - maybe_reconstruct_IIS_status_in_qs + + delete(&connection); + return 0; +} + +static void httpd_mode() { + config_handler(*sapi_info); + + SOCKET sock = HTTPD_Server::bind(httpd_host_port); + +#ifdef SIGPIPE + signal(SIGPIPE, SIG_IGN); +#endif + + while(1){ +#ifndef _MSC_VER + pid_t pid=1; + if(HTTPD_Server::mode == HTTPD_Server::PARALLEL) + while (waitpid((pid_t)(-1), 0, WNOHANG) > 0); +#endif + try { + HTTPD_Connection connection; + if(!connection.accept(sock, 500)) + continue; + + 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; + 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; + 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) + continue; // parent should close connection.sock as well +#endif + case HTTPD_Server::SEQUENTIAL: // and fork child + + SAPI_Info_HTTPD info(connection); + connection_handler(info, connection); + } + // 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 - (getenv("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::die("Parser/%s", PARSER_VERSION); + + // 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 + // obligatory const char* path_info=getenv("PATH_INFO"); if(!path_info) - SAPI::die("CGI: illegal call (missing PATH_INFO)"); - const char* script_name=getenv("SCRIPT_NAME"); - if(!script_name) - SAPI::die("CGI: illegal call (missing SCRIPT_NAME)"); - - const char* env_request_uri=getenv("REQUEST_URI"); - if(env_request_uri) - request_info.uri=env_request_uri; - else - if(query_string) { - char* reconstructed_uri=new(PointerFreeGC) char[ - 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)"); + + request_info.document_root = getenv("DOCUMENT_ROOT"); + if(!request_info.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)"); + + char* document_root = new(PointerFreeGC) char[prefix_len + 1/*0*/]; + memcpy(document_root, filespec_to_process, prefix_len); document_root[prefix_len] = 0; + request_info.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)"); + 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.uri=""; + } else{ + request_info.document_root = full_disk_path(); + request_info.uri = ""; + } - request_info.content_type=getenv("CONTENT_TYPE"); - const char* content_length=getenv("CONTENT_LENGTH"); - request_info.content_length=(content_length?atoi(content_length):0); - request_info.cookie=getenv("HTTP_COOKIE"); - request_info.mail_received=mail_received; - - // get request_info ptr for signal handlers - ::request_info=&request_info; - if(execution_canceled) - SAPI::die("Execution canceled"); + 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 - log("request_info: method=%s, uri=%s, q=%s, dr=%s, pt=%s, cookies=%s, cl=%u", - request_info.method, - request_info.uri, - request_info.query_string, - request_info.document_root, - request_info.path_translated, - request_info.cookie, - request_info.content_length); + 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 // prepare to process request - Request request(SAPI_info, request_info, - cgi ? String::Language(String::L_HTML|String::L_OPTIMIZE_BIT) : String::L_AS_IS, - 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); - } + Request r(*sapi_info, request_info, cgi ? String::Language(String::L_HTML|String::L_OPTIMIZE_BIT) : String::L_AS_IS); + { + // initing ::request ptr for signal handlers + RequestController rc(&r); + // process the request + r.core(locate_config(config_filespec, parser3_filespec), strcasecmp(request_info.method, "HEAD")==0); + // clearing ::request in RequestController destructor to prevent signal handlers from accessing invalid memory } - - // process the request - request.core( - config_filespec_cstr, fail_on_config_read_problem, - header_only); - - // no request [prevent signal handlers from accessing invalid memory] - ::request=0; - // finalize global variables + // finalize libraries pa_globals_done(); - - // - pa_socks_done(); } #ifdef PA_SUPPRESS_SYSTEM_EXCEPTION -static const Exception -call_real_parser_handler__do_PEH_return_it( - const char* filespec_to_process, - const char* request_method, bool header_only) -{ +static const Exception call_real_parser_handler__do_PEH_return_it(bool cgi) { try { - real_parser_handler( - filespec_to_process, - request_method, header_only); + real_parser_handler(cgi); } catch(const Exception& e) { return e; } return Exception(); } -static void call_real_parser_handler__supress_system_exception( - const char* filespec_to_process, - const char* request_method, bool header_only) -{ + +static void call_real_parser_handler__supress_system_exception(bool cgi) { Exception parser_exception; LPEXCEPTION_POINTERS system_exception=0; __try { - parser_exception=call_real_parser_handler__do_PEH_return_it( - filespec_to_process, - request_method, header_only); - } __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("system", - 0, - "0x%08X at 0x%08X", er->ExceptionCode, er->ExceptionAddress); + throw Exception("system", 0, "0x%08X at 0x%08X", er->ExceptionCode, er->ExceptionAddress); else - throw Exception("system", - 0, - ""); + throw Exception("system", 0, ""); else - throw Exception("system", - 0, - ""); + throw Exception("system", 0, ""); } if(parser_exception) throw Exception(parser_exception); } + +#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\n" - "Copyright(c) 2001-2007 ArtLebedev Group (http://www.artlebedev.com)\n" - "Author: Alexandr Petrosian (http://paf.design.ru)\n" + "Copyright (c) 2001-2024 Art. Lebedev Studio (http://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, + parser3_filespec ? parser3_filespec : "parser3" ); exit(EINVAL); } +#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[]) { #ifdef PA_DEBUG_CGI_ENTRY_EXIT - log("main: entry"); + pa_log("main: entry"); #endif - //_asm int 3; - GC_java_finalization=0; -#ifndef PA_DEBUG_DISABLE_GC - // Dont collect unless explicitly requested - // this is quicker (~30% ), but less memory-efficient(~8%) - // so deciding for speed - GC_dont_gc=1; -#endif -/* - - Array test; - test+=3; - test+=4; -// int a=test.count(); - int i=0; - scanf("%d", &i); - int b=test.get(i); -// int b=test.get(10); - printf("%d", b);//test.count());*/ - -#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"); -#endif + if(!argc || !argv[0]) + usage(); + parser3_filespec = MAYBE_BACK_SLASHES_TO_SLASHES(argv[0]); + umask(2); -#ifdef DEBUG_MAILRECEIVE - if(FILE *fake_in=fopen(DEBUG_MAILRECEIVE, "rt")) { - dup2(fake_in->_file, 0/*STDIN_FILENO*/); + // were we started as CGI? + 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"; } -#endif -#ifdef _DEBUG - //_crtBreakAlloc=46; +#ifdef SIGPIPE + signal(SIGPIPE, SIGPIPE_handler); #endif - argv0=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; + 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 +#ifdef _MSC_VER setmode(fileno(stdin), _O_BINARY); setmode(fileno(stdout), _O_BINARY); setmode(fileno(stderr), _O_BINARY); @@ -720,57 +729,24 @@ int main(int argc, char *argv[]) { // Set flag to the new value _CrtSetDbgFlag( tmpFlag ); -// _CrtSetBreakAlloc(61); _CrtSetReportMode( _CRT_WARN, _CRTDBG_MODE_FILE ); _CrtSetReportFile( _CRT_WARN, _CRTDBG_FILE_STDERR ); #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 -#ifdef PA_SUPPRESS_SYSTEM_EXCEPTION - call_real_parser_handler__supress_system_exception( -#else - real_parser_handler( -#endif - 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 - char buf[MAX_STRING]; - snprintf(buf, MAX_STRING, "Unhandled exception %s", - e.comment()); - // log it - SAPI::log(SAPI_info, "%s", buf); - - // - int content_length=strlen(buf); - - // prepare header - SAPI::add_header_attribute(SAPI_info, "content-type", "text/plain"); - char content_length_cstr[MAX_NUMBER]; - snprintf(content_length_cstr, MAX_NUMBER, "%u", content_length); - SAPI::add_header_attribute(SAPI_info, "content-length", content_length_cstr); - - // send header - SAPI::send_header(SAPI_info); - - // send body - if(!header_only) - SAPI::send_body(SAPI_info, buf, content_length); + if(raw_filespec_to_process && *raw_filespec_to_process){ + filespec_to_process=full_disk_path(raw_filespec_to_process); + } - // unsuccessful finish + 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"); } #ifdef PA_DEBUG_CGI_ENTRY_EXIT - log("main: successful return"); + pa_log("main: successful return"); #endif - return 0; + return sapi_info->http_response_code < 100 ? sapi_info->http_response_code : 0; }