|
|
| version 1.167, 2002/04/04 08:04:28 | version 1.367, 2024/12/19 02:47:46 |
|---|---|
| Line 1 | Line 1 |
| /** @file | /** @file |
| Parser: scripting and CGI main. | Parser: scripting and CGI main. |
| Copyright(c) 2001, 2002 ArtLebedev Group (http://www.artlebedev.com) | Copyright (c) 2001-2024 Art. Lebedev Studio (http://www.artlebedev.com) |
| Author: Alexandr Petrosian <paf@design.ru> (http://paf.design.ru) | Authors: Konstantin Morshnev <moko@design.ru>, Alexandr Petrosian <paf@design.ru> |
| $Id$ | |
| */ | */ |
| #include "pa_config_includes.h" | volatile const char * IDENT_PARSER3_C="$Id$"; |
| #if _MSC_VER | #include "pa_config_includes.h" |
| # include <new.h> | |
| # include <crtdbg.h> | |
| #else | |
| # include "pa_config_paths.h" | |
| #endif | |
| #include "pa_sapi.h" | #include "pa_sapi.h" |
| #include "classes.h" | #include "classes.h" |
| #include "pa_common.h" | #include "pa_common.h" |
| #include "pa_request.h" | #include "pa_request.h" |
| #include "pa_socks.h" | |
| #include "pa_version.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 <crtdbg.h> | |
| #include <windows.h> | |
| #include <direct.h> | |
| 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 <windows.h> | |
| #endif | #endif |
| //#define DEBUG_POOL_MALLOC | // defines |
| //#define DEBUG_STRING_APPENDS_VS_EXPANDS | |
| // comment remove me after debugging | |
| //#define PA_DEBUG_CGI_ENTRY_EXIT | |
| #ifdef DEBUG_STRING_APPENDS_VS_EXPANDS | #if defined(_MSC_VER) && !defined(_DEBUG) |
| extern ulong | # define PA_SUPPRESS_SYSTEM_EXCEPTION |
| string_piece_appends, | |
| wcontext_result_size, | |
| total_alloc_size, | |
| string_string_shortcut_economy; | |
| #endif | #endif |
| // consts | // consts |
| extern const char *main_RCSIds[]; | #define REDIRECT_PREFIX "REDIRECT_" |
| #ifdef USE_SMTP | #define PARSER_CONFIG_ENV_NAME "CGI_PARSER_CONFIG" |
| extern const char *smtp_RCSIds[]; | #define PARSER_LOG_ENV_NAME "CGI_PARSER_LOG" |
| #endif | |
| extern const char *gd_RCSIds[]; | SAPI_Info sapi_console; |
| extern const char *classes_RCSIds[]; | SAPI_Info_CGI sapi_cgi; |
| extern const char *types_RCSIds[]; | |
| extern const char *parser3_RCSIds[]; | static SAPI_Info *sapi_info = &sapi_cgi; |
| const char **RCSIds[]={ | static THREAD_LOCAL SAPI_Info *sapi_info_4log = NULL; // global for correct send error in die() |
| main_RCSIds, | |
| #ifdef USE_SMTP | const char* parser3_mode = "cgi"; // $status:mode |
| smtp_RCSIds, | static const char* filespec_to_process = 0; // [file] |
| #endif | static const char* httpd_host_port = 0; // -p option |
| gd_RCSIds, | static const char* config_filespec = 0; // -f option or from env or next to the executable if exists |
| classes_RCSIds, | static const char* log_filespec = 0; // -l option |
| types_RCSIds, | static bool mail_received = false; // -m option? [asked to parse incoming message to $mail:received] |
| parser3_RCSIds, | static const char* parser3_filespec = 0; // argv[0] |
| 0 | static char** argv_extra = NULL; |
| }; | |
| // for error logging | |
| const char *PARSER_ROOT_CONFIG_ENV_NAME="HTTP_PARSER_ROOT_CONFIG"; | static THREAD_LOCAL Request_info *request_info_4log = NULL; // global for correct log() reporting |
| const char *PARSER_SITE_CONFIG_ENV_NAME="HTTP_PARSER_SITE_CONFIG"; | static const char* filespec_4log = NULL; // null only if system-wide auto.p used |
| template <typename T> 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 char *parser3_log_filespec() { // $status:log-filename |
| const size_t READ_POST_CHUNK_SIZE=0x400*0x400; // 1M | const char* slog=log_filespec; |
| const char *argv0; | if(!slog) |
| Pool_storage pool_storage; | slog=getenv(PARSER_LOG_ENV_NAME); |
| Pool pool(&pool_storage); // global pool [dont describe to doxygen: it confuses it with param names] | if(!slog) |
| bool cgi; ///< we were started as CGI? | 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 | // SAPI |
| static void log(const char *fmt, va_list args) { | static void pa_log(const char* fmt, va_list args) { |
| bool opened; | // use no memory [so that we could log out-of-memory error] |
| FILE *f=0; | const char* slog=parser3_log_filespec(); |
| FILE *f=strcmp(slog,"-") ? fopen(slog, "at") : stderr; | |
| if(argv0) { | |
| // beside by binary | // fallback to stderr |
| char file_spec[MAX_STRING]; | if(!f) |
| 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) | |
| f=stderr; | f=stderr; |
| setbuf(f, 0); // stderr stream is unbuffered by default, but still... | |
| // prefix | // prefix |
| time_t t=time(0); | time_t t=time(0); |
| const char *stamp=ctime(&t); | if(const char* stamp=ctime(&t)) { // never saw that |
| fprintf(f, "[%.*s] ", strlen(stamp)-1, stamp); | if(size_t len=strlen(stamp)) // saw once stamp being ="" |
| // message | fprintf(f, "[%.*s] [%u] ", (int)len-1, 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); | |
| // 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); | 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 : "<unknown>", request_info_4log->method, (unsigned long)request_info_4log->content_length); | |
| } else | |
| fputs(" [no request info]\n", f); | |
| if(f!=stderr) | |
| fclose(f); | fclose(f); |
| else | else |
| fflush(f); | fflush(f); |
| } | } |
| // appends to parser3.log located beside my binary if openable, to stderr otherwize | void pa_log(const char* fmt, ...) { |
| void SAPI::log(Pool& , const char *fmt, ...) { | va_list args; |
| va_list args; | |
| va_start(args,fmt); | va_start(args,fmt); |
| ::log(fmt, args); | pa_log(fmt, args); |
| va_end(args); | va_end(args); |
| } | } |
| void SAPI::die(const char *fmt, ...) { | // appends to parser3.log located next to the config file if openable, to stderr otherwize |
| #ifdef DEBUG_POOL_MALLOC | void SAPI::log(SAPI_Info&, const char* fmt, ...) { |
| extern void log_pool_stats(Pool& pool); | va_list args; |
| log_pool_stats(pool); | |
| #endif | |
| va_list args; | |
| va_start(args,fmt); | va_start(args,fmt); |
| // log | pa_log(fmt, args); |
| // logging is more important than user | |
| // she can cancel download, we'd get SIG_PIPE, | |
| // nothing would be logged then | |
| ::log(fmt, args); | |
| // inform user | |
| char body[MAX_STRING]; | |
| int content_length=vsnprintf(body, MAX_STRING, fmt, args); | |
| va_end(args); | va_end(args); |
| } | |
| // prepare header | void SAPI::die(const char* fmt, ...) { |
| // let's be honest, that's bad we couldn't produce valid output | va_list args; |
| 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); | |
| // send header | // logging first, first vsnprintf |
| SAPI::send_header(pool); | va_start(args,fmt); |
| pa_log(fmt, args); | |
| va_end(args); | |
| // body | // inform user, second vsnprintf |
| SAPI::send_body(pool, body, content_length); | va_start(args, fmt); |
| char message[MAX_STRING]; | |
| vsnprintf(message, MAX_STRING, fmt, args); | |
| SAPI::send_error(sapi_info_4log ? *sapi_info_4log : *sapi_info, message); | |
| exit(1); | exit(1); |
| // va_end(args); | |
| } | } |
| const char *SAPI::get_env(Pool& , const char *name) { | void SAPI::send_error(SAPI_Info& info, const char *exception_cstr, const char *status){ |
| return getenv(name); | info.send_error(exception_cstr, status); |
| } | } |
| size_t SAPI::read_post(Pool& , char *buf, size_t max_bytes) { | char* SAPI::Env::get(SAPI_Info& info, const char* name) { |
| size_t read_size=0; | return info.get_env(name); |
| 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_size<max_bytes); | |
| return read_size; | bool SAPI::Env::set(SAPI_Info& info, const char* name, const char* value) { |
| return info.set_env(name, value); | |
| } | } |
| void SAPI::add_header_attribute(Pool& , const char *key, const char *value) { | const char* const *SAPI::Env::get(SAPI_Info& info) { |
| if(cgi) | return info.get_env(); |
| printf("%s: %s\n", key, value); | |
| } | } |
| /// @todo intelligent cache-control | size_t SAPI::read_post(SAPI_Info& info, char* buf, size_t max_bytes) { |
| void SAPI::send_header(Pool& ) { | return info.read_post(buf, max_bytes); |
| if(cgi) { | } |
| // puts("expires: Fri, 23 Mar 2001 09:32:23 GMT"); | |
| // header | body delimiter | void SAPI::add_header_attribute(SAPI_Info& info, const char* dont_store_key, const char* dont_store_value) { |
| puts(""); | info.add_header(dont_store_key, dont_store_value); |
| } | |
| } | } |
| void SAPI::send_body(Pool& , const void *buf, size_t size) { | void SAPI::send_headers(SAPI_Info& info) { |
| stdout_write(buf, size); | info.send_headers(); |
| } | } |
| // | 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); | |
| } | |
| void full_file_spec(const char *file_name, char *buf, size_t buf_size) { | static const char* full_disk_path(const char* file_name = "") { |
| if(file_name) | char* result; |
| if(file_name[0]=='/' | if(file_name[0]=='/' |
| #ifdef WIN32 | #ifdef WIN32 |
| || (file_name[0] && file_name[1]==':') | || file_name[0] && file_name[1]==':' |
| #endif | #endif |
| ) | ){ |
| strncpy(buf, file_name, buf_size); | result=pa_strdup(file_name); |
| else { | } else { |
| char cwd[MAX_STRING]; getcwd(cwd, MAX_STRING); | char cwd[MAX_STRING]; |
| snprintf(buf, buf_size, "%s/%s", cwd, file_name); | result=pa_strcat(getcwd(cwd, MAX_STRING) ? cwd : "", "/", file_name); |
| } | } |
| else | |
| buf[0]=0; | |
| #ifdef WIN32 | #ifdef WIN32 |
| back_slashes_to_slashes(buf); | back_slashes_to_slashes(result); |
| #endif | #endif |
| return result; | |
| } | } |
| /** | static void log_signal(const char* signal_name) { |
| main workhorse | pa_log("%s received %s processing request", signal_name, request ? "while" : "before or after"); |
| } | |
| @todo | #ifdef SIGPIPE |
| IIS: remove trailing default-document[index.html] from $request.uri. | #define SIGPIPE_NAME "SIGPIPE" |
| to do that we need to consult metabase, | static const String sigpipe_name(SIGPIPE_NAME); |
| wich is tested but seems slow. | static void SIGPIPE_handler(int /*sig*/){ |
| IIS5 todo find out proper 'illegal call' check | Value* sigpipe=0; |
| */ | if(request) |
| void real_parser_handler( | sigpipe=request->main_class.get_element(sigpipe_name); |
| const char *filespec_to_process, | if(sigpipe && sigpipe->as_bool()) |
| const char *request_method, bool header_only) { | log_signal(SIGPIPE_NAME); |
| // init socks | |
| init_socks(pool); | if(request) |
| request->set_skip(Request::SKIP_INTERRUPTED); | |
| // init global classes | } |
| init_methoded_array(pool); | #endif |
| // init global variables | |
| pa_globals_init(pool); | // requires pa_thread_request() in entry_exists() under Windows |
| static const char* locate_config(const char* config_filespec_option, const char* executable_path){ | |
| if(!filespec_to_process) | filespec_4log=config_filespec_option; |
| SAPI::die("Parser/%s", PARSER_VERSION); | if(!filespec_4log) |
| filespec_4log=getenv(PARSER_CONFIG_ENV_NAME); | |
| // Request info | if(!filespec_4log) |
| Request::Info request_info; | filespec_4log=getenv(REDIRECT_PREFIX PARSER_CONFIG_ENV_NAME); |
| char document_root_buf[MAX_STRING]; | if(!filespec_4log){ |
| if(cgi) { | const char* exec_dir_pos = dir_pos(executable_path); |
| if(const char *env_document_root=SAPI::get_env(pool, "DOCUMENT_ROOT")) | #ifdef SYSTEM_CONFIG_FILE |
| request_info.document_root=env_document_root; | if(exec_dir_pos){ |
| else if(const char *path_info=SAPI::get_env(pool, "PATH_INFO")) { | #endif |
| // IIS | // next to the executable |
| size_t len=min(sizeof(document_root_buf)-1, strlen(filespec_to_process)-strlen(path_info)); | if(!exec_dir_pos || (exec_dir_pos==executable_path+1 && *executable_path=='.')){ |
| memcpy(document_root_buf, filespec_to_process, len); document_root_buf[len]=0; | // when just parser3 or ./parser3 full path should be used to avoid "parser already configured" |
| request_info.document_root=document_root_buf; | filespec_4log=full_disk_path(AUTO_FILE_NAME); |
| } else | } else { |
| throw Exception("parser.runtime", | filespec_4log=pa_strcat(pa_strdup(executable_path, exec_dir_pos - executable_path), "/" AUTO_FILE_NAME); |
| 0, | } |
| "CGI: no PATH_INFO defined(in reinventing DOCUMENT_ROOT)"); | if(entry_exists(filespec_4log)) |
| } else { | return filespec_4log; |
| full_file_spec("", document_root_buf, sizeof(document_root_buf)); | #ifdef SYSTEM_CONFIG_FILE |
| request_info.document_root=document_root_buf; | } |
| } | if(entry_exists(SYSTEM_CONFIG_FILE)){ |
| request_info.path_translated=filespec_to_process; | filespec_4log=NULL; |
| request_info.method=request_method ? request_method : "GET"; | return SYSTEM_CONFIG_FILE; |
| 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("parser.runtime", | |
| 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 | #endif |
| } else | return NULL; |
| request_info.uri=0; | } |
| return filespec_4log; | |
| 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); | #ifdef WIN32 |
| request_info.cookie=SAPI::get_env(pool, "HTTP_COOKIE"); | static const char* maybe_reconstruct_IIS_status_in_qs(const char* original) { |
| request_info.user_agent=SAPI::get_env(pool, "HTTP_USER_AGENT"); | // 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; | |
| } | |
| if(qmark_at) { | |
| *cur++='&'; | |
| strcpy(cur, qmark_at+1/*skip ? itself*/); | |
| } else | |
| *cur=0; | |
| return reconstructed; | |
| } | |
| 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 | // prepare to process request |
| Request request(pool, | Request r(info, request_info, String::Language(String::L_HTML|String::L_OPTIMIZE_BIT)); |
| request_info, | // only once |
| #ifdef _DEBUG | config_filespec = locate_config(config_filespec, parser3_filespec); |
| String::UL_HTML|String::UL_OPTIMIZE_BIT | // 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 | #else |
| cgi ? String::UL_HTML|String::UL_OPTIMIZE_BIT : String::UL_AS_IS | static void *connection_thread(void *arg){ |
| #endif | #endif |
| , | HTTPD_Connection &connection=*(HTTPD_Connection*)arg; |
| true /* status_allowed */); | SAPI_Info_HTTPD info(connection); |
| // some root-controlled location | try { |
| const char *root_config_filespec; | connection_handler(info, connection); |
| if(const char *root_config_by_env=getenv(PARSER_ROOT_CONFIG_ENV_NAME)) | } catch(const Exception& e) { // exception in send_error |
| root_config_filespec=root_config_by_env; | pa_log("%s", e.comment()); |
| else { | } |
| #ifdef ROOT_CONFIG_DIR | |
| root_config_filespec=ROOT_CONFIG_DIR "/" CONFIG_FILE_NAME; | 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 | #else |
| # ifdef WIN32 | #ifdef HAVE_TLS |
| // c:\windows | pthread_t thread; |
| char windows_dir[MAX_STRING]; | pthread_attr_t attr; |
| GetWindowsDirectory(windows_dir, MAX_STRING); | pthread_attr_init(&attr); |
| pthread_attr_setdetachstate(&attr, PTHREAD_CREATE_DETACHED); | |
| char buf[MAX_STRING]; | |
| snprintf(buf, MAX_STRING, | if(int result=GC_pthread_create(&thread, &attr, connection_thread, new HTTPD_Connection(connection))) |
| "%s/%s", | throw Exception("httpd.fork", 0, "thread creation failed (%d)", result); |
| windows_dir, CONFIG_FILE_NAME); | connection.sock=INVALID_SOCKET; |
| break; | |
| root_config_filespec=buf; | #endif |
| # else | case HTTPD_Server::PARALLEL: |
| #error must be compiled either configure/make or MSVC++ | pid=fork(); |
| # endif | 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 | #endif |
| } | } |
| } | |
| const char *site_config_filespec; | |
| if(const char *site_config_by_env=getenv(PARSER_SITE_CONFIG_ENV_NAME)) | /** main workhorse */ |
| site_config_filespec=site_config_by_env; | |
| else { | static void real_parser_handler(bool cgi) { |
| // beside by binary | // init libraries |
| // @todo full path, not ./! | pa_globals_init(); |
| 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(httpd_host_port){ |
| if(!( | httpd_mode(); |
| rsplit(beside_binary_path, '/') || | } |
| rsplit(beside_binary_path, '\\'))) { // strip filename | |
| // no path, just filename | const char* request_method=getenv("REQUEST_METHOD"); |
| beside_binary_path[0]='.'; beside_binary_path[1]=0; | |
| 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; | |
| } | } |
| char buf[MAX_STRING]; | |
| snprintf(buf, MAX_STRING, | request_info.uri = request_info.strip_absolute_uri(getenv("REQUEST_URI")); |
| "%s/%s", | if(request_info.uri) { // apache & others stuck to standards |
| beside_binary_path, CONFIG_FILE_NAME); | // another obligatory |
| site_config_filespec=buf; | 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 = ""; | |
| } | } |
| // process the request | request_info.content_type = getenv("CONTENT_TYPE"); |
| request.core( | request_info.content_length = cgi ? (size_t)pa_atoul(getenv("CONTENT_LENGTH")) : 0; // only SAPI_Info_CGI can read POST |
| root_config_filespec, false /*don't fail_on_read_problem*/, | request_info.cookie = getenv("HTTP_COOKIE"); |
| site_config_filespec, false /*don't fail_on_read_problem*/, | request_info.mail_received = mail_received; |
| header_only); | |
| request_info.argv = argv_extra; | |
| // | |
| done_socks(); | |
| #ifdef DEBUG_POOL_MALLOC | |
| extern void log_pool_stats(Pool& pool); | |
| log_pool_stats(pool); | |
| #endif | |
| #ifdef DEBUG_STRING_APPENDS_VS_EXPANDS | #ifdef PA_DEBUG_CGI_ENTRY_EXIT |
| SAPI::log(pool, | 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); |
| "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 | #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(); | |
| } | |
| #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(); | |
| } | } |
| void call_real_parser_handler__do_SEH( | static void call_real_parser_handler__supress_system_exception(bool cgi) { |
| const char *filespec_to_process, | Exception parser_exception; |
| const char *request_method, bool header_only) { | |
| #if _MSC_VER && !defined(_DEBUG) | |
| LPEXCEPTION_POINTERS system_exception=0; | LPEXCEPTION_POINTERS system_exception=0; |
| __try { | __try { |
| #endif | parser_exception=call_real_parser_handler__do_PEH_return_it(cgi); |
| real_parser_handler( | } __except ( (system_exception=GetExceptionInformation()), EXCEPTION_EXECUTE_HANDLER) { |
| filespec_to_process, | |
| request_method, header_only); | |
| #if _MSC_VER && !defined(_DEBUG) | |
| } __except ( | |
| (system_exception=GetExceptionInformation()), | |
| EXCEPTION_EXECUTE_HANDLER) { | |
| if(system_exception) | if(system_exception) |
| if(_EXCEPTION_RECORD *er=system_exception->ExceptionRecord) | if(_EXCEPTION_RECORD *er=system_exception->ExceptionRecord) |
| throw Exception(0, | throw Exception("system", 0, "0x%08X at 0x%08X", er->ExceptionCode, er->ExceptionAddress); |
| 0, | |
| "Exception 0x%08X at 0x%08X", er->ExceptionCode, er->ExceptionAddress); | |
| else | else |
| throw Exception(0, 0, "Exception <no exception record>"); | throw Exception("system", 0, "<no exception record>"); |
| else | else |
| throw Exception(0, 0, "Exception <no exception information>"); | throw Exception("system", 0, "<no exception information>"); |
| } | } |
| #endif | |
| } | |
| #if _MSC_VER | if(parser_exception) |
| int failed_new(size_t size) { | throw Exception(parser_exception); |
| SAPI::die("out of memory in 'new', failed to allocated %u bytes", size); | |
| return 0; // not reached | |
| } | } |
| #endif | |
| #ifdef HAVE_SET_NEW_HANDLER | #define REAL_PARSER_HANDLER call_real_parser_handler__supress_system_exception |
| void failed_new() { | #else |
| SAPI::die("out of memory in 'new'"); | #define REAL_PARSER_HANDLER real_parser_handler |
| } | |
| #endif | #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 <moko@design.ru>, Alexandr Petrosian <paf@design.ru>\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[]) { | int main(int argc, char *argv[]) { |
| // _asm int 3; | #ifdef PA_DEBUG_CGI_ENTRY_EXIT |
| argv0=argv[0]; | pa_log("main: entry"); |
| #endif | |
| if(!argc || !argv[0]) | |
| usage(); | |
| parser3_filespec = MAYBE_BACK_SLASHES_TO_SLASHES(argv[0]); | |
| umask(2); | umask(2); |
| // were we started as CGI? | // were we started as CGI? |
| cgi= | bool cgi=(getenv("SERVER_SOFTWARE") || getenv("SERVER_NAME") || getenv("GATEWAY_INTERFACE") || getenv("REQUEST_METHOD")) && !getenv("PARSER_VERSION"); |
| getenv("SERVER_SOFTWARE") || | if(!cgi){ |
| getenv("SERVER_NAME") || | sapi_info = &sapi_console; |
| getenv("GATEWAY_INTERFACE") || | parser3_mode = "console"; |
| getenv("REQUEST_METHOD"); | } |
| if(!cgi) { | #ifdef SIGPIPE |
| if(argc<2) { | signal(SIGPIPE, SIGPIPE_handler); |
| printf( | #endif |
| "Parser/%s Copyright(c) 2001, 2002 ArtLebedev Group (http://www.artlebedev.com)\n" | char* raw_filespec_to_process = NULL; |
| "Author: Alexandr Petrosian <paf@design.ru> (http://paf.design.ru)\n" | if(cgi) { |
| "\n" | raw_filespec_to_process=getenv("PATH_TRANSLATED"); |
| "Usage: %s <file>\n", | argv_extra=argv + 1; |
| PARSER_VERSION, | } else { |
| argv0?argv0:"parser3"); | char** carg = argv + 1; |
| return 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(stdin), _O_BINARY); |
| setmode(fileno(stdout), _O_BINARY); | setmode(fileno(stdout), _O_BINARY); |
| setmode(fileno(stderr), _O_BINARY); | setmode(fileno(stderr), _O_BINARY); |
| #endif | #endif |
| #if _MSC_VER | #if defined(_MSC_VER) && defined(_DEBUG) |
| _set_new_handler(failed_new); | |
| #ifdef _DEBUG | |
| // Get current flag | // Get current flag |
| int tmpFlag = _CrtSetDbgFlag( _CRTDBG_REPORT_FLAG ); | int tmpFlag = _CrtSetDbgFlag( _CRTDBG_REPORT_FLAG ); |
| Line 466 int main(int argc, char *argv[]) { | Line 731 int main(int argc, char *argv[]) { |
| // Set flag to the new value | // Set flag to the new value |
| _CrtSetDbgFlag( tmpFlag ); | _CrtSetDbgFlag( tmpFlag ); |
| // _CrtSetBreakAlloc(471); | |
| _CrtSetReportMode( _CRT_WARN, _CRTDBG_MODE_FILE ); | |
| _CrtSetReportFile( _CRT_WARN, _CRTDBG_FILE_STDERR ); | |
| #endif | #endif |
| #endif | |
| #ifdef HAVE_SET_NEW_HANDLER | |
| std::set_new_handler(failed_new); | |
| #endif | |
| char *raw_filespec_to_process=cgi?getenv("PATH_TRANSLATED"):argv[1]; | |
| 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 | try { // global try |
| call_real_parser_handler__do_SEH( | if(raw_filespec_to_process && *raw_filespec_to_process){ |
| filespec_to_process, | filespec_to_process=full_disk_path(raw_filespec_to_process); |
| request_method, header_only); | } |
| } catch(const Exception& e) { // global problem | |
| // don't allocate anything on pool here: | REAL_PARSER_HANDLER(cgi); |
| // possible pool' exception not catch-ed now | } catch(const Exception& e) { // exception in config_handler |
| // and there could be out-of-memory exception | SAPI::log(*sapi_info, "%s", e.comment()); |
| SAPI::send_error(*sapi_info, e.comment(), strcmp(e.type(), "file.missing") ? "500" : "404"); | |
| SAPI::die("exception in request exception handler: %s", e.comment()); | |
| #ifndef _DEBUG | |
| } catch(...) { | |
| SAPI::die("<unknown exception>"); | |
| #endif | |
| } | } |
| #ifndef WIN32 | #ifdef PA_DEBUG_CGI_ENTRY_EXIT |
| // | pa_log("main: successful return"); |
| if(!cgi) | |
| SAPI::send_body(pool, "\n", 1); | |
| #endif | #endif |
| //_asm int 3; | return sapi_info->http_response_code < 100 ? sapi_info->http_response_code : 0; |
| return 0; | |
| } | } |