--- parser3/src/targets/cgi/parser3.C	2001/03/14 08:53:58	1.6
+++ parser3/src/targets/cgi/parser3.C	2002/04/04 07:57:50	1.166
@@ -1,161 +1,503 @@
-/*
-	Parser
-	Copyright (c) 2001 ArtLebedev Group (http://www.artlebedev.com)
-	Author: Alexander Petrosyan <paf@design.ru> (http://design.ru/paf)
+/** @file
+	Parser: scripting and CGI main.
 
-	$Id: parser3.C,v 1.6 2001/03/14 08:53:58 paf Exp $
+	Copyright(c) 2001, 2002 ArtLebedev Group (http://www.artlebedev.com)
+	Author: Alexandr Petrosian <paf@design.ru> (http://paf.design.ru)
+
+	$Id: parser3.C,v 1.166 2002/04/04 07:57:50 paf Exp $
 */
 
-#ifdef HAVE_CONFIG_H
-#	include "pa_config.h"
+#include "pa_config_includes.h"
+
+#if _MSC_VER
+#	include <new.h>
+#	include <crtdbg.h>
+#else
+#	include "pa_config_paths.h"
 #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 "pool_storage.h"
 
 #ifdef WIN32
 #	include <windows.h>
-#	include <io.h>
 #endif
-#include <stdlib.h>
-#include <stdio.h>
-#include <string.h>
-#include <fcntl.h>
 
-#include "pa_globals.h"
-#include "pa_request.h"
-#include "pa_common.h"
+//#define DEBUG_POOL_MALLOC
+//#define DEBUG_STRING_APPENDS_VS_EXPANDS
 
-Pool pool; // global pool
+#ifdef DEBUG_STRING_APPENDS_VS_EXPANDS
+extern ulong 
+	string_piece_appends, 
+	wcontext_result_size,
+	total_alloc_size,
+	string_string_shortcut_economy;
+#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
+};
+
+const char *PARSER_ROOT_CONFIG_ENV_NAME="HTTP_PARSER_ROOT_CONFIG";
+const char *PARSER_SITE_CONFIG_ENV_NAME="HTTP_PARSER_SITE_CONFIG";
+
+/// IIS refuses to read bigger chunks
+const size_t READ_POST_CHUNK_SIZE=0x400*0x400; // 1M 
+
+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?
+
+// 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)
+		f=stderr;
+
+	// prefix
+	time_t t=time(0);
+	const char *stamp=ctime(&t);
+	fprintf(f, "[%.*s] ", strlen(stamp)-1, stamp);
+	// message
 
-#ifdef WIN32
-// intercept global system errors
-LONG WINAPI TopLevelExceptionFilter (
-									 struct _EXCEPTION_POINTERS *ExceptionInfo
-									 ) {
 	char buf[MAX_STRING];
-	if(ExceptionInfo && ExceptionInfo->ExceptionRecord) {
-		struct _EXCEPTION_RECORD *r=ExceptionInfo->ExceptionRecord;
+	size_t size=vsnprintf(buf, MAX_STRING, fmt, args);
+	remove_crlf(buf, buf+size);
+
+	fwrite(buf, size, 1, f);
+	// newline
+	fprintf(f, "\n");
+
+	if(opened)
+		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;
+	va_start(args,fmt);
+	::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;
+	va_start(args,fmt);
+	// log
+
+	// 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);
+
+	// 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);
+
+	// send header
+	SAPI::send_header(pool);
+
+	// body
+	SAPI::send_body(pool, body, content_length);
+
+	exit(1);
+}
+
+const char *SAPI::get_env(Pool& , const char *name) {
+	return getenv(name);
+}
+
+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_size<max_bytes);
+
+	return read_size;
+}
+
+void SAPI::add_header_attribute(Pool& , const char *key, const char *value) {
+	if(cgi)
+		printf("%s: %s\n", key, value);
+}
+
+/// @todo intelligent cache-control
+void SAPI::send_header(Pool& ) {
+	if(cgi) {
+//		puts("expires: Fri, 23 Mar 2001 09:32:23 GMT");
+
+		// header | body  delimiter
+		puts("");
+	}
+}
+
+void SAPI::send_body(Pool& , const void *buf, size_t size) {
+	stdout_write(buf, size);
+}
+
+//
+
+void full_file_spec(const char *file_name, char *buf, size_t buf_size) {
+	if(file_name && !strchr(file_name, '/')) {
+		char cwd[MAX_STRING];  getcwd(cwd, MAX_STRING);
+		snprintf(buf, buf_size, "%s/%s", cwd, file_name);
+	} else {
+		strncpy(buf, file_name, buf_size);
+	}
+#ifdef WIN32
+	back_slashes_to_slashes(buf);
+#endif
+}
+
+/**
+main workhorse
+
+  @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;
+	char document_root_buf[MAX_STRING];
+	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=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;
+	}
+	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("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
+	} 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");
+	
+	// prepare to process request
+	Request request(pool,
+		request_info,
+#ifdef _DEBUG
+		String::UL_HTML|String::UL_OPTIMIZE_BIT
+#else
+		cgi ? String::UL_HTML|String::UL_OPTIMIZE_BIT : String::UL_AS_IS
+#endif
+		,
+		true /* status_allowed */);
+	
+	// some root-controlled location
+	const char *root_config_filespec;
+	if(const char *root_config_by_env=getenv(PARSER_ROOT_CONFIG_ENV_NAME))
+		root_config_filespec=root_config_by_env;
+	else {
+#ifdef ROOT_CONFIG_DIR
+		root_config_filespec=ROOT_CONFIG_DIR "/" CONFIG_FILE_NAME;
+#else
+#	ifdef WIN32
+		// c:\windows
+		char windows_dir[MAX_STRING];
+		GetWindowsDirectory(windows_dir, MAX_STRING);
+
+		char buf[MAX_STRING];
+		snprintf(buf, MAX_STRING, 
+			"%s/%s", 
+			windows_dir, CONFIG_FILE_NAME);
 		
-		int printed=0;
-		printed+=snprintf(buf+printed, MAX_STRING-printed, "Exception 0x%X at 0x%p", 
-			r->ExceptionCode, 
-			r->ExceptionAddress);
-		for(unsigned int i=0; i<r->NumberParameters; i++)
-			printed+=snprintf(buf+printed, MAX_STRING-printed, ", 0x%X", 
-				r->ExceptionInformation[i]);
-	} else 
-		strcpy(buf, "Exception <unknown>");
+		root_config_filespec=buf;
+#	else
+#error must be compiled either configure/make or MSVC++
+#	endif
+#endif
+	}
+	
+	const char *site_config_filespec;
+	if(const char *site_config_by_env=getenv(PARSER_SITE_CONFIG_ENV_NAME))
+		site_config_filespec=site_config_by_env;
+	else {
+	// beside by binary
+	// @todo full path, not ./!
+		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
+			beside_binary_path[0]='.'; beside_binary_path[1]=0;
+		}
+		char buf[MAX_STRING];
+		snprintf(buf, MAX_STRING, 
+			"%s/%s", 
+			beside_binary_path, CONFIG_FILE_NAME);
+		site_config_filespec=buf;
+	}
+	
+	// process the request
+	request.core(
+		root_config_filespec, false /*don't fail_on_read_problem*/,
+		site_config_filespec, false /*don't fail_on_read_problem*/,
+		header_only);
 	
-	PTHROW(0, 0,
-		0,
-		buf);
+	//
+	done_socks();
+	
+#ifdef DEBUG_POOL_MALLOC
+	extern void log_pool_stats(Pool& pool);
+	log_pool_stats(pool);
+#endif
 
-	return EXCEPTION_EXECUTE_HANDLER; // never reached
-}
+#ifdef DEBUG_STRING_APPENDS_VS_EXPANDS
+	SAPI::log(pool, 
+		"string piece appends=%lu, wcontext_result_size=%lu, string_string_shortcut_economy_closer=%lu, total_alloc_size=%lu", 
+		string_piece_appends,
+		wcontext_result_size,
+		string_string_shortcut_economy,
+		total_alloc_size);
 #endif
 
-void fill_vform_fields(Pool& pool, bool cgi, Hash& fields) {
-	String& ename=*new(pool) String(pool);
-	ename.APPEND_CONST("test");
+}
 
-	String& evalue=*new(pool) String(pool);
-	evalue.APPEND_TAINTED("<value>", 0, "form", 0);
+void call_real_parser_handler__do_SEH(
+								 const char *filespec_to_process,
+								 const char *request_method, bool header_only) {
+#if _MSC_VER && !defined(_DEBUG)
+	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) {
+		
+		if(system_exception)
+			if(_EXCEPTION_RECORD *er=system_exception->ExceptionRecord)
+				throw Exception(0,
+					0,
+					"Exception 0x%08X at 0x%08X", er->ExceptionCode,  er->ExceptionAddress);
+			else
+				throw Exception(0, 0, "Exception <no exception record>");
+			else
+				throw Exception(0, 0, "Exception <no exception information>");
+	}
+#endif
+}
 
-	fields.put(ename, new(pool) VString(evalue));
+#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
 }
+#endif
+
+#ifdef HAVE_SET_NEW_HANDLER
+void failed_new() {
+    SAPI::die("out of memory in 'new'");
+}
+#endif
 
 int main(int argc, char *argv[]) {
+//	_asm int 3;
+	argv0=argv[0];
+
+	umask(2);
+
 	// were we started as CGI?
-	bool cgi=
+	cgi=
 		getenv("SERVER_SOFTWARE") || 
 		getenv("SERVER_NAME") || 
 		getenv("GATEWAY_INTERFACE") || 
 		getenv("REQUEST_METHOD");
 	
-	char *result;  char error[MAX_STRING];  error[0]=0;
-	PTRY { // global try
-		// must be first in PTRY{}PCATCH
+	if(!cgi) {
+		if(argc<2) {
+			printf(
+				"Parser/%s Copyright(c) 2001, 2002 ArtLebedev Group (http://www.artlebedev.com)\n"
+				"Author: Alexandr Petrosian <paf@design.ru> (http://paf.design.ru)\n"
+				"\n"
+				"Usage: %s <file>\n",
+				PARSER_VERSION, 
+				argv0?argv0:"parser3");
+			return 1;
+		}
+	}
+
 #ifdef WIN32
-		SetUnhandledExceptionFilter(&TopLevelExceptionFilter);
-		//TODO: initSocks();
+	setmode(fileno(stdin), _O_BINARY);
+	setmode(fileno(stdout), _O_BINARY);
+	setmode(fileno(stderr), _O_BINARY);
 #endif
 
-		fill_globals(pool);
-		
-		// TODO: ifdef WIN32 flip \\ to /
-		const char *document_root="Y:/parser3/src/";
-		const char *page_filespec="Y:/parser3/src/test.p";
-		
-		// prepare to process request
-		Request request(Pool(),
-			cgi ? String::Untaint_lang::HTML_TYPO : String::Untaint_lang::NO,
-			document_root,
-			page_filespec
-			);
-		
-		// fill user passed forms
-		fill_vform_fields(pool, cgi, request.form_class.fields());
-		
-		// some root-controlled location
-		char *sys_auto_path1;
-#ifdef WIN32
-		sys_auto_path1=(char *)pool.malloc(MAX_STRING);
-		GetWindowsDirectory(sys_auto_path1, MAX_STRING-1/*for \*/);
-		strcat(sys_auto_path1, "\\");
-#else
-		sys_auto_path1=getenv("HOME");
+#if _MSC_VER
+	_set_new_handler(failed_new);
+
+#ifdef _DEBUG
+	// Get current flag
+	int tmpFlag = _CrtSetDbgFlag( _CRTDBG_REPORT_FLAG );
+
+	// Turn on leak-checking bit
+	tmpFlag |= _CRTDBG_LEAK_CHECK_DF;
+
+	// Set flag to the new value
+	_CrtSetDbgFlag( tmpFlag );
+//	_CrtSetBreakAlloc(471);
+
 #endif
-		
-		// beside by binary
-		char *sys_auto_path2=(char *)pool.malloc(MAX_STRING);
-		strncpy(sys_auto_path2, argv[0], MAX_STRING);  // filespec of my binary
-		// strip filename
-#ifdef WIN32
-		rsplit(sys_auto_path2, '\\');  
-#else
-		rsplit(sys_auto_path2, '/'); 
+
 #endif
-		
-		// process the request
-		result=request.core(
-			sys_auto_path1, 
-			sys_auto_path2);
-		// set error, will be reported in case result==0
-		strcpy(error, "exception occured in request exception handler");
 
-		// must be last in PTRY{}PCATCH
-#ifdef WIN32
-		SetUnhandledExceptionFilter(0);
+#ifdef HAVE_SET_NEW_HANDLER
+	std::set_new_handler(failed_new);
 #endif
-	} PCATCH(e) { // global problem [creating Request, preparing to .core()]
-		result=0;
-		strcpy(error, e.comment());
-	}
-	PEND_CATCH
 
-	// write out the result	
-	if(cgi) {
-		if(result) {
-			const char *content_type="text/html";
-			printf(
-				"Content-type: %s\n"
-				"Content-length: %d\n"
-				"\n", 
-				content_type,
-				strlen(result));
-			stdout_write(result);
-		} else {
-			printf(
-				"Content-type: text/plain\n"
-				"Content-length: %d\n"
-				"\n", 
-				strlen(error));
-			stdout_write(error);
-		}
-	} else
-		if(result)
-			printf("%s", result);
-		else
-			fputs(error, stderr);
+	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
+		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("<unknown exception>");
+#endif
+	}
+	
 
+#ifndef WIN32
+	// 
+	if(!cgi)
+		SAPI::send_body(pool, "\n", 1);
+#endif
+//_asm int 3;
 	return 0;
 }