--- parser3/src/targets/cgi/parser3.C 2002/01/31 10:20:21 1.151 +++ parser3/src/targets/cgi/parser3.C 2024/12/19 02:47:46 1.367 @@ -1,424 +1,728 @@ /** @file Parser: scripting and CGI main. - Copyright(c) 2001 ArtLebedev Group(http://www.artlebedev.com) - Author: Alexander Petrosyan (http://paf.design.ru) - - $Id: parser3.C,v 1.151 2002/01/31 10:20:21 paf Exp $ + Copyright (c) 2001-2024 Art. Lebedev Studio (http://www.artlebedev.com) + Authors: Konstantin Morshnev , Alexandr Petrosian */ -#include "pa_config_includes.h" +volatile const char * IDENT_PARSER3_C="$Id: parser3.C,v 1.367 2024/12/19 02:47:46 moko Exp $"; -#if _MSC_VER -# include -# include -#else -# include "pa_config_paths.h" -#endif +#include "pa_config_includes.h" #include "pa_sapi.h" #include "classes.h" #include "pa_common.h" #include "pa_request.h" -#include "pa_socks.h" #include "pa_version.h" -#include "pool_storage.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); + +#else + +extern "C" int GC_pthread_create(pthread_t *, const pthread_attr_t *, void *(*)(void *), void * /* arg */); -#ifdef WIN32 -# include #endif -//#define DEBUG_POOL_MALLOC +// 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 -extern const char *main_RCSIds[]; -#ifdef USE_SMTP -extern const char *smtp_RCSIds[]; -#endif -extern const char *gd_RCSIds[]; -extern const char *classes_RCSIds[]; -extern const char *types_RCSIds[]; -extern const char *parser3_RCSIds[]; -const char **RCSIds[]={ - main_RCSIds, -#ifdef USE_SMTP - smtp_RCSIds, -#endif - gd_RCSIds, - classes_RCSIds, - types_RCSIds, - parser3_RCSIds, - 0 -}; +#define REDIRECT_PREFIX "REDIRECT_" +#define PARSER_CONFIG_ENV_NAME "CGI_PARSER_CONFIG" +#define PARSER_LOG_ENV_NAME "CGI_PARSER_LOG" + +SAPI_Info sapi_console; +SAPI_Info_CGI sapi_cgi; + +static SAPI_Info *sapi_info = &sapi_cgi; +static THREAD_LOCAL SAPI_Info *sapi_info_4log = NULL; // global for correct send error in die() + +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; +} -/// IIS refuses to read bigger chunks -const size_t READ_POST_CHUNK_SIZE=0x400*0x400; // 1M +const char *parser3_log_filespec() { // $status:log-filename + const char* slog=log_filespec; -const char *argv0; -Pool_storage pool_storage; -Pool pool(&pool_storage); // global pool [dont describe to doxygen: it confuses it with param names] -bool cgi; ///< we were started as CGI? + 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 + strcpy(log_spec, "./parser3.log"); + } + slog=log_spec; + } + return slog; +} // SAPI -static void log(const char *fmt, va_list args) { - bool opened; - FILE *f=0; - - if(argv0) { - // beside by binary - char file_spec[MAX_STRING]; - strncpy(file_spec, argv0, MAX_STRING-1); file_spec[MAX_STRING-1]=0; // filespec of my binary - rsplit(file_spec, '/'); rsplit(file_spec, '\\');// strip filename - strcat(file_spec, "/parser3.log"); - f=fopen(file_spec, "at"); - } - opened=f!=0; - if(!opened) +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(!f) f=stderr; + setbuf(f, 0); // stderr stream is unbuffered by default, but still... + // prefix time_t t=time(0); - const char *stamp=ctime(&t); - fprintf(f, "[%.*s] ", strlen(stamp)-1, stamp); - // message - - char buf[MAX_STRING]; - size_t size=vsnprintf(buf, MAX_STRING, fmt, args); - remove_crlf(buf, buf+size); + if(const char* stamp=ctime(&t)) { // never saw that + if(size_t len=strlen(stamp)) // saw once stamp being ="" + fprintf(f, "[%.*s] [%u] ", (int)len-1, stamp, (unsigned int)pa_get_thread_id() ); + } + // 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); - // newline - fprintf(f, "\n"); - if(opened) + 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(f!=stderr) fclose(f); else fflush(f); } -// 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(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(pool); -#endif - - 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(fmt, args); + va_end(args); +} - // logging is more important than user - // she can cancel download, we'd get SIG_PIPE, - // 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(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]; + 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(pool, "status", "500"); - SAPI::add_header_attribute(pool, "content-type", "text/plain"); - char content_length_cstr[MAX_NUMBER]; - snprintf(content_length_cstr, MAX_NUMBER, "%u", content_length); - SAPI::add_header_attribute(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(pool); +char* SAPI::Env::get(SAPI_Info& info, const char* name) { + return info.get_env(name); +} - // body - SAPI::send_body(pool, body, content_length); +bool SAPI::Env::set(SAPI_Info& info, const char* name, const char* value) { + return info.set_env(name, value); +} - exit(1); +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); } -size_t SAPI::read_post(Pool& , char *buf, size_t max_bytes) { - size_t read_size=0; - do { - int 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_sizemain_class.get_element(sigpipe_name); + if(sigpipe && sigpipe->as_bool()) + log_signal(SIGPIPE_NAME); + + if(request) + request->set_skip(Request::SKIP_INTERRUPTED); +} +#endif -char *full_file_spec(char *file_name) { - if(file_name && !strchr(file_name, '/')) { - static char cwd[MAX_STRING]; getcwd(cwd, MAX_STRING); - static char buf[MAX_STRING]; - snprintf(buf, MAX_STRING, "%s/%s", cwd, file_name); - return buf; +// 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 file_name; + return filespec_4log; } -/** -main workhorse +#ifdef WIN32 +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]==';'){ + size_t original_len=strlen(original); + 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; + *cur++='&'; + + 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; + memcpy(cur, original+4, value_len); cur+=value_len; + } - @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 -*/ -void real_parser_handler( - const char *filespec_to_process, - const char *request_method, bool header_only) { - // init socks - init_socks(pool); - - // init global classes - init_methoded_array(pool); - // init global variables - pa_globals_init(pool); - - if(!filespec_to_process) - SAPI::die("Parser/%s", PARSER_VERSION); - - // Request info - Request::Info request_info; - if(cgi) { - if(const char *env_document_root=SAPI::get_env(pool, "DOCUMENT_ROOT")) - request_info.document_root=env_document_root; - else if(const char *path_info=SAPI::get_env(pool, "PATH_INFO")) { - // IIS - size_t len=strlen(filespec_to_process)-strlen(path_info); - char *buf=(char *)pool.malloc(len+1); - memcpy(buf, filespec_to_process, len); buf[len]=0; - request_info.document_root=buf; + if(qmark_at) { + *cur++='&'; + strcpy(cur, qmark_at+1/*skip ? itself*/); } else - throw Exception(0, 0, - 0, - "CGI: no PATH_INFO defined(in reinventing DOCUMENT_ROOT)"); - } else { - static char buf[MAX_STRING]; - strncpy(buf, filespec_to_process, MAX_STRING-1); buf[MAX_STRING-1]=0; - if(rsplit(buf, '/') || rsplit(buf, '\\')) // strip filename - request_info.document_root=buf; - else - request_info.document_root=""; + *cur=0; + + return reconstructed; } - request_info.path_translated=filespec_to_process; - request_info.method=request_method ? request_method : "GET"; - const char *query_string=SAPI::get_env(pool, "QUERY_STRING"); - request_info.query_string=query_string; - if(cgi) { - if(const char *env_request_uri=SAPI::get_env(pool, "REQUEST_URI")) - request_info.uri=env_request_uri; - else if(const char *path_info=SAPI::get_env(pool, "PATH_INFO")) - if(query_string) { - char *reconstructed_uri=(char *)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; - else - throw Exception(0, 0, - 0, - "CGI: no PATH_INFO defined(in reinventing REQUEST_URI)"); - -#ifndef WIN32 - // they've changed this under IIS5. - if(const char *script_name=SAPI::get_env(pool, "SCRIPT_NAME")) { - 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("CGI: illegal call"); - } -#endif - } else - request_info.uri=0; - - request_info.content_type=SAPI::get_env(pool, "CONTENT_TYPE"); - const char *content_length=SAPI::get_env(pool, "CONTENT_LENGTH"); - request_info.content_length=(content_length?atoi(content_length):0); - request_info.cookie=SAPI::get_env(pool, "HTTP_COOKIE"); - request_info.user_agent=SAPI::get_env(pool, "HTTP_USER_AGENT"); + 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 + +class RequestController { +public: + RequestController(Request* r){ + request=r; + } + ~RequestController(){ + request=0; + } +}; + +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.document_root = full_disk_path(); + request_info.uri = ""; + request_info.argv = argv_extra; + // prepare to process request - Request request(pool, - request_info, -#ifdef _DEBUG - String::UL_HTML|String::UL_OPTIMIZE_BIT + 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 - cgi ? String::UL_HTML|String::UL_OPTIMIZE_BIT : String::UL_AS_IS +static void *connection_thread(void *arg){ #endif - , - true /* status_allowed */); - - // some root-controlled location -#ifdef PARSER_ROOT_CONFIG_DIR - const char *root_config_filespec=PARSER_ROOT_CONFIG_DIR "/" CONFIG_FILE_NAME; + 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 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 WIN32 - // c:\windows - char root_config_path[MAX_STRING]; - GetWindowsDirectory(root_config_path, MAX_STRING); - - char root_config_filespec[MAX_STRING]; - snprintf(root_config_filespec, MAX_STRING, - "%s/%s", - root_config_path, CONFIG_FILE_NAME); -# else -#error must be compiled either configure/make or MSVC++ -# endif +#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 - - // beside by binary - // @todo full path, not ./! - static char site_config_path[MAX_STRING]; - strncpy(site_config_path, argv0, MAX_STRING-1); site_config_path[MAX_STRING-1]=0; // filespec of my binary - if(!( - rsplit(site_config_path, '/') || - rsplit(site_config_path, '\\'))) { // strip filename - // no path, just filename - site_config_path[0]='.'; site_config_path[1]=0; + } +} + +/** 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); + 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) { + // obligatory + const char* path_info=getenv("PATH_INFO"); + if(!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; + } + + 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(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.document_root = full_disk_path(); + request_info.uri = ""; } - char site_config_filespec[MAX_STRING]; - snprintf(site_config_filespec, MAX_STRING, - "%s/%s", - site_config_path, CONFIG_FILE_NAME); - - // process the request - request.core( - root_config_filespec, false, - site_config_filespec, false, - header_only); - - // - done_socks(); - -#ifdef DEBUG_POOL_MALLOC - extern void log_pool_stats(Pool& pool); - log_pool_stats(pool); + 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 + + // prepare to process request + 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 + } + + // finalize libraries + pa_globals_done(); } -void call_real_parser_handler__do_SEH( - const char *filespec_to_process, - const char *request_method, bool header_only) { -#if _MSC_VER && !defined(_DEBUG) +#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__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, - 0, - "Exception 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(0, 0, 0, "Exception "); - else - throw Exception(0, 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 -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* message=NULL) { + if(message){ + fprintf(stderr, message, parser3_filespec); + } + + printf( + "Parser/%s\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" + "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" + " -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[]) { -// _asm int 3; - argv0=argv[0]; +#ifdef PA_DEBUG_CGI_ENTRY_EXIT + pa_log("main: entry"); +#endif + + if(!argc || !argv[0]) + usage(); + parser3_filespec = MAYBE_BACK_SLASHES_TO_SLASHES(argv[0]); umask(2); // were we started as CGI? - cgi= - getenv("SERVER_SOFTWARE") || - getenv("SERVER_NAME") || - getenv("GATEWAY_INTERFACE") || - getenv("REQUEST_METHOD"); - - if(!cgi) { - if(argc<2) { - printf( - "Parser/%s Copyright(c) 2001 ArtLebedev Group(http://www.artlebedev.com)\n" - "Author: Alexander Petrosyan (http://paf.design.ru)\n" - "\n" - "Usage: %s \n", - PARSER_VERSION, - argv0?argv0:"parser3"); - return 1; + 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"); + argv_extra=argv + 1; + } else { + 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; + parser3_mode="mail"; + break; +#endif + default: + fprintf(stderr, "%s: invalid option '%c'\n", parser3_filespec, *c); + usage(); + break; + } + } } + + 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); #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 ); @@ -427,46 +731,24 @@ 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=cgi?getenv("PATH_TRANSLATED"):argv[1]; -#ifdef WIN32 - back_slashes_to_slashes(filespec_to_process); -#endif - filespec_to_process=full_file_spec(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"); } - -#ifndef WIN32 - // - if(!cgi) - SAPI::send_body(pool, "\n", 1); +#ifdef PA_DEBUG_CGI_ENTRY_EXIT + pa_log("main: successful return"); #endif - return 0; + return sapi_info->http_response_code < 100 ? sapi_info->http_response_code : 0; }