websrvr.c

/* websrvr.c */

/* Synchronet Web Server */

/* $Id$ */

/****************************************************************************
 * @format.tab-size 4		(Plain Text/Source Code File Header)			*
 * @format.use-tabs true	(see http://www.synchro.net/ptsc_hdr.html)		*
 *																			*
 * Copyright 2004 Rob Swindell - http://www.synchro.net/copyright.html		*
 *																			*
 * This program is free software; you can redistribute it and/or			*
 * modify it under the terms of the GNU General Public License				*
 * as published by the Free Software Foundation; either version 2			*
 * of the License, or (at your option) any later version.					*
 * See the GNU General Public License for more details: gpl.txt or			*
 * http://www.fsf.org/copyleft/gpl.html										*
 *																			*
 * Anonymous FTP access to the most recent released source is available at	*
 * ftp://vert.synchro.net, ftp://cvs.synchro.net and ftp://ftp.synchro.net	*
 *																			*
 * Anonymous CVS access to the development source and modification history	*
 * is available at cvs.synchro.net:/cvsroot/sbbs, example:					*
 * cvs -d :pserver:anonymous@cvs.synchro.net:/cvsroot/sbbs login			*
 *     (just hit return, no password is necessary)							*
 * cvs -d :pserver:anonymous@cvs.synchro.net:/cvsroot/sbbs checkout src		*
 *																			*
 * For Synchronet coding style and modification guidelines, see				*
 * http://www.synchro.net/source.html										*
 *																			*
 * You are encouraged to submit any modifications (preferably in Unix diff	*
 * format) via e-mail to mods@synchro.net									*
 *																			*
 * Note: If this box doesn't appear square, then you need to fix your tabs.	*
 ****************************************************************************/

/*
 * General notes: (ToDo stuff)
 * strtok() is used a LOT in here... notice that there is a strtok_r() for reentrant...
 * this may imply that strtok() is NOT thread-safe... if in fact it isn't this HAS
 * to be fixed before any real production-level quality is achieved with this web server
 * however, strtok_r() may not be a standard function.
 *
 * RE: not sending the headers if an nph scrpit is detected.  (The headers buffer could
 * just be free()ed and NULLed)
 *
 * Currently, all SSJS requests for a session are ran in the same context without clearing the context in
 * any way.  This behaviour should not be relied on as it may disappear in the future... this will require
 * some thought as it COULD be handy in some circumstances and COULD cause weird bugs in others.
 *
 * Dynamic content is always resent on an If-Modified-Since request... this may not be optimal behaviour
 * for GET requests...
 *
 * Should support RFC2617 Digest auth.
 *
 * Fix up all the logging stuff.
 *
 * SSJS stuff could work using three different methods:
 * 1) Temporary file as happens currently
 *		Advantages:
 *			Allows to keep current connection (keep-alive works)
 *			write() doesn't need to be "special"
 *		Disadvantages:
 *			Depends on the temp dir being writable and capable of holding
 *				the full reply
 *			Everything goes throug the disk, so probobly some performance
 *				penalty is involved
 *			No way of sending directly to the remote system
 * 2) nph- style
 *		Advantages:
 *			No file I/O involved
 *			Can do magic tricks (ala my perl web wrapper)
 *		Disadvantages:
 *			Pretty much everything needs to be handled by the script.
 * 3) Return body in http_reply object
 *		All the advantages of 1)
 *		Could use a special write() to make everything just great.
 *		Still doesn't allow page to be sent until fully composed (ie: long
 *			delays)
 * 4) Type three with a callback that sends the header and current body, then
 *		converts write() to send directly to remote.
 *
 * Add in support to pass connections through to a different webserver...
 *      probobly in access.ars... with like a simplified mod_rewrite.
 *      This would allow people to run apache and Synchronet as the same site.
 */

/* Headers for CGI stuff */
#if defined(__unix__)
	#include <sys/wait.h>		/* waitpid() */
	#include <sys/types.h>
	#include <signal.h>			/* kill() */
#endif

#ifndef JAVASCRIPT
#define JAVASCRIPT
#endif

#undef SBBS	/* this shouldn't be defined unless building sbbs.dll/libsbbs.so */
#include "sbbs.h"
#include "sockwrap.h"		/* sendfilesocket() */
#include "threadwrap.h"		/* pthread_mutex_t */
#include "semwrap.h"
#include "websrvr.h"
#include "base64.h"

static const char*	server_name="Synchronet Web Server";
static const char*	newline="\r\n";
static const char*	http_scheme="http://";
static const size_t	http_scheme_len=7;
static const char*	error_404="404 Not Found";
static const char*	error_500="500 Internal Server Error";
static const char*	unknown="<unknown>";

/* Is this not in a header somewhere? */
extern const uchar* nular;

#define TIMEOUT_THREAD_WAIT		60		/* Seconds */
#define MAX_REQUEST_LINE		1024	/* NOT including terminator */
#define MAX_HEADERS_SIZE		16384	/* Maximum total size of all headers 
										   (Including terminator )*/
#define MAX_REDIR_LOOPS			20		/* Max. times to follow internal redirects for a single request */

static scfg_t	scfg;
static BOOL		scfg_reloaded=TRUE;
static BOOL		http_logging_thread_running=FALSE;
static ulong	active_clients=0;
static ulong	sockets=0;
static BOOL		terminate_server=FALSE;
static BOOL		terminate_http_logging_thread=FALSE;
static uint		thread_count=0;
static SOCKET	server_socket=INVALID_SOCKET;
static char		revision[16];
static char		root_dir[MAX_PATH+1];
static char		error_dir[MAX_PATH+1];
static char		cgi_dir[MAX_PATH+1];
static time_t	uptime=0;
static DWORD	served=0;
static web_startup_t* startup=NULL;
static js_server_props_t js_server_props;
static link_list_t recycle_semfiles;
static link_list_t shutdown_semfiles;

static named_string_t** mime_types;

/* Logging stuff */
sem_t	log_sem;
pthread_mutex_t	log_mutex;
link_list_t	log_list;
struct log_data {
	char	*hostname;
	char	*ident;
	char	*user;
	char	*request;
	char	*referrer;
	char	*agent;
	int		status;
	unsigned int	size;
	struct tm completed;
};

typedef struct  {
	int			method;
	char		virtual_path[MAX_PATH+1];
	char		physical_path[MAX_PATH+1];
	BOOL		parsed_headers;
	BOOL    	expect_go_ahead;
	time_t		if_modified_since;
	BOOL		keep_alive;
	char		ars[256];
	char    	auth[128];				/* UserID:Password */
	char		host[128];				/* The requested host. (virtual hosts) */
	int			send_location;
	const char*	mime_type;
	link_list_t	headers;
	char		status[MAX_REQUEST_LINE+1];
	char *		post_data;
	size_t		post_len;
	int			dynamic;
	struct log_data	*ld;

	/* CGI parameters */
	char		query_str[MAX_REQUEST_LINE+1];
	char		extra_path_info[MAX_REQUEST_LINE+1];
	link_list_t	cgi_env;
	link_list_t	dynamic_heads;

	/* Dynamically (sever-side JS) generated HTML parameters */
	FILE*	fp;

} http_request_t;

typedef struct  {
	SOCKET			socket;
	SOCKADDR_IN		addr;
	http_request_t	req;
	char			host_ip[64];
	char			host_name[128];	/* Resolved remote host */
	int				http_ver;       /* HTTP version.  0 = HTTP/0.9, 1=HTTP/1.0, 2=HTTP/1.1 */