htaccess.c

用于linux和其他unix下面的

/*		Access Manager					HTAccess.c
**		==============
**
**  Authors
**	TBL	Tim Berners-Lee [e-mail]
**	JFG	Jean-Francois Groff [e-mail]
**	DD	Denis DeLaRoca [phone]  <[e-mail]>
**	FM	Foteos Macrides [e-mail]
**	PDM	Danny Mayer [e-mail]
**
**  History
**	 8 Jun 92 Telnet hopping prohibited as telnet is not secure TBL
**	26 Jun 92 When over DECnet, suppressed FTP, Gopher and News. JFG
**	 6 Oct 92 Moved HTClientHost and logfile into here. TBL
**	17 Dec 92 Tn3270 added, bug fix. DD
**	 4 Feb 93 Access registration, Search escapes bad chars TBL
**		  PARAMETERS TO HTSEARCH AND HTLOADRELATIVE CHANGED
**	28 May 93 WAIS gateway explicit if no WAIS library linked in.
**	31 May 94 Added DIRECT_WAIS support for VMS. FM
**	27 Jan 95 Fixed proxy support to use NNTPSERVER for checking
**		  whether or not to use the proxy server. PDM
**	27 Jan 95 Ensured that proxy service will be overridden for files
**		  on the local host (because HTLoadFile() doesn't try ftp
**		  for those) and will substitute ftp for remote files. FM
**	28 Jan 95 Tweeked PDM's proxy override mods to handle port info
**		  for news and wais URL's. FM
**
**  Bugs
**	This module assumes that that the graphic object is hypertext, as it
**	needs to select it when it has been loaded.  A superclass needs to be
**	defined which accepts select and select_anchor.
*/

#ifdef VMS
#define DIRECT_WAIS
#endif /* VMS */

#include <HTUtils.h>
#include <HTTP.h>
#include <HTAlert.h>
/*
**  Implements:
*/
#include <HTAccess.h>

/*
**  Uses:
*/
#include <HTParse.h>
#include <HTML.h>		/* SCW */

#ifndef NO_RULES
#include <HTRules.h>
#endif

#include <HTList.h>
#include <HText.h>	/* See bugs above */
#include <HTCJK.h>
#include <UCMap.h>
#include <GridText.h>

#include <LYGlobalDefs.h>
#include <LYexit.h>
#include <LYUtils.h>
#include <LYLeaks.h>

/*
**  These flags may be set to modify the operation of this module
*/
PUBLIC char * HTClientHost = NULL; /* Name of remote login host if any */
PUBLIC FILE * HTlogfile = NULL;    /* File to which to output one-liners */
PUBLIC BOOL HTSecure = NO;	   /* Disable access for telnet users? */
PUBLIC BOOL HTPermitRedir = NO;	   /* Always allow redirection in getfile()? */

PUBLIC BOOL using_proxy = NO; /* are we using a proxy gateway? */

/*
**  To generate other things, play with these:
*/
PUBLIC HTFormat HTOutputFormat = NULL;
PUBLIC HTStream* HTOutputStream = NULL; /* For non-interactive, set this */

PRIVATE HTList * protocols = NULL; /* List of registered protocol descriptors */

PUBLIC char *use_this_url_instead = NULL;

PRIVATE int pushed_assume_LYhndl = -1; /* see LYUC* functions below - kw */
PRIVATE char * pushed_assume_MIMEname = NULL;

#ifdef LY_FIND_LEAKS
PRIVATE void free_protocols NOARGS
{
    HTList_delete(protocols);
    protocols = NULL;
    FREE(pushed_assume_MIMEname); /* shouldn't happen, just in case - kw */
}
#endif /* LY_FIND_LEAKS */

/*	Register a Protocol.				HTRegisterProtocol()
**	--------------------
*/
PUBLIC BOOL HTRegisterProtocol ARGS1(
	HTProtocol *,	protocol)
{
    if (!protocols) {
	protocols = HTList_new();
#ifdef LY_FIND_LEAKS
	atexit(free_protocols);
#endif
    }
    HTList_addObject(protocols, protocol);
    return YES;
}


/*	Register all known protocols.			HTAccessInit()
**	-----------------------------
**
**	Add to or subtract from this list if you add or remove protocol
**	modules.  This routine is called the first time the protocol list
**	is needed, unless any protocols are already registered, in which
**	case it is not called.	Therefore the application can override
**	this list.
**
**	Compiling with NO_INIT prevents all known protocols from being
**	forced in at link time.
*/
#ifndef NO_INIT
#ifdef GLOBALREF_IS_MACRO
extern GLOBALREF (HTProtocol, HTTP);
extern GLOBALREF (HTProtocol, HTTPS);
extern GLOBALREF (HTProtocol, HTFile);
extern GLOBALREF (HTProtocol, HTTelnet);
extern GLOBALREF (HTProtocol, HTTn3270);
extern GLOBALREF (HTProtocol, HTRlogin);
#ifndef DECNET
#ifndef DISABLE_FTP
extern GLOBALREF (HTProtocol, HTFTP);
#endif /* DISABLE_FTP */
#ifndef DISABLE_NEWS
extern GLOBALREF (HTProtocol, HTNews);
extern GLOBALREF (HTProtocol, HTNNTP);
extern GLOBALREF (HTProtocol, HTNewsPost);
extern GLOBALREF (HTProtocol, HTNewsReply);
extern GLOBALREF (HTProtocol, HTSNews);
extern GLOBALREF (HTProtocol, HTSNewsPost);
extern GLOBALREF (HTProtocol, HTSNewsReply);
#endif /* not DISABLE_NEWS */
#ifndef DISABLE_GOPHER
extern GLOBALREF (HTProtocol, HTGopher);
extern GLOBALREF (HTProtocol, HTCSO);
#endif /* not DISABLE_GOPHER */
#ifndef DISABLE_FINGER
extern GLOBALREF (HTProtocol, HTFinger);
#endif /* not DISABLE_FINGER */
#ifdef DIRECT_WAIS
extern GLOBALREF (HTProtocol, HTWAIS);
#endif /* DIRECT_WAIS */
#endif /* !DECNET */
#else
GLOBALREF HTProtocol HTTP, HTTPS, HTFile, HTTelnet, HTTn3270, HTRlogin;
#ifndef DECNET
#ifndef DISABLE_FTP
GLOBALREF HTProtocol HTFTP;
#endif /* DISABLE_FTP */
#ifndef DISABLE_NEWS
GLOBALREF HTProtocol HTNews, HTNNTP, HTNewsPost, HTNewsReply;
GLOBALREF HTProtocol HTSNews, HTSNewsPost, HTSNewsReply;
#endif /* not DISABLE_NEWS */
#ifndef DISABLE_GOPHER
GLOBALREF HTProtocol HTGopher, HTCSO;
#endif /* not DISABLE_GOPHER */
#ifndef DISABLE_FINGER
GLOBALREF HTProtocol HTFinger;
#endif /* not DISABLE_FINGER */
#ifdef DIRECT_WAIS
GLOBALREF  HTProtocol HTWAIS;
#endif /* DIRECT_WAIS */
#endif /* !DECNET */
#endif /* GLOBALREF_IS_MACRO */

PRIVATE void HTAccessInit NOARGS			/* Call me once */
{
    HTRegisterProtocol(&HTTP);
    HTRegisterProtocol(&HTTPS);
    HTRegisterProtocol(&HTFile);
    HTRegisterProtocol(&HTTelnet);
    HTRegisterProtocol(&HTTn3270);
    HTRegisterProtocol(&HTRlogin);
#ifndef DECNET
#ifndef DISABLE_FTP
    HTRegisterProtocol(&HTFTP);
#endif /* DISABLE_FTP */
#ifndef DISABLE_NEWS
    HTRegisterProtocol(&HTNews);
    HTRegisterProtocol(&HTNNTP);
    HTRegisterProtocol(&HTNewsPost);
    HTRegisterProtocol(&HTNewsReply);
    HTRegisterProtocol(&HTSNews);
    HTRegisterProtocol(&HTSNewsPost);
    HTRegisterProtocol(&HTSNewsReply);
#endif /* not DISABLE_NEWS */
#ifndef DISABLE_GOPHER
    HTRegisterProtocol(&HTGopher);
    HTRegisterProtocol(&HTCSO);
#endif /* not DISABLE_GOPHER */
#ifndef DISABLE_FINGER
    HTRegisterProtocol(&HTFinger);
#endif /* not DISABLE_FINGER */
#ifdef DIRECT_WAIS
    HTRegisterProtocol(&HTWAIS);
#endif /* DIRECT_WAIS */
#endif /* !DECNET */
    LYRegisterLynxProtocols();
}
#endif /* !NO_INIT */

/*	Check for proxy override.			override_proxy()
**	-------------------------
**
**	Check the no_proxy environment variable to get the list
**	of hosts for which proxy server is not consulted.
**
**	no_proxy is a comma- or space-separated list of machine
**	or domain names, with optional :port part.  If no :port
**	part is present, it applies to all ports on that domain.
**
**  Example:
**	    no_proxy="cern.ch,some.domain:8001"
**
**  Use "*" to override all proxy service:
**	     no_proxy="*"
*/
PUBLIC BOOL override_proxy ARGS1(
	CONST char *,	addr)
{
    CONST char * no_proxy = getenv("no_proxy");
    char * p = NULL;
    char * at = NULL;
    char * host = NULL;
    char * Host = NULL;
    char * acc_method = NULL;
    int port = 0;
    int h_len = 0;

    /*
     * Check for global override.
     */
    if (no_proxy) {
	if (!strcmp(no_proxy, "*"))
	    return YES;
    }

    /*
     *	Never proxy file:// URLs if they are on the local host.
     *	HTLoadFile() will not attempt ftp for those if direct
     *	access fails.  We'll check that first, in case no_proxy
     *	hasn't been defined. - FM
     */
    if (!addr)
	return NO;
    if (!(host = HTParse(addr, "", PARSE_HOST)))
	return NO;
    if (!*host) {
	FREE(host);
	return NO;
    }
    Host = (((at = strchr(host, '@')) != NULL) ? (at+1) : host);

#ifdef VMS
#define CompareHostname(a,b) strcasecomp(a, b)
#else
#define CompareHostname(a,b) strcmp(a, b)
#endif /* VMS */

    if ((acc_method = HTParse(addr, "", PARSE_ACCESS))) {
	if (!strcmp("file", acc_method) &&
	    (!strcmp(Host, "localhost") ||
	     !CompareHostname(Host, HTHostName()))) {
	    FREE(host);
	    FREE(acc_method);
	    return YES;
	}
	FREE(acc_method);
    }

    if (!no_proxy) {
	FREE(host);
	return NO;
    }

    if (NULL != (p = strrchr(Host, ':'))) {	/* Port specified */
	*p++ = 0;				/* Chop off port */
	port = atoi(p);
    } else {					/* Use default port */
	acc_method = HTParse(addr, "", PARSE_ACCESS);
	if (acc_method != NULL) {
	    if	    (!strcmp(acc_method, "http"))	port = 80;
	    else if (!strcmp(acc_method, "https"))	port = 443;
	    else if (!strcmp(acc_method, "ftp"))	port = 21;
#ifndef DISABLE_GOPHER
	    else if (!strcmp(acc_method, "gopher"))	port = 70;
#endif
	    else if (!strcmp(acc_method, "cso"))	port = 105;
#ifndef DISABLE_NEWS
	    else if (!strcmp(acc_method, "news"))	port = 119;
	    else if (!strcmp(acc_method, "nntp"))	port = 119;
	    else if (!strcmp(acc_method, "newspost"))	port = 119;
	    else if (!strcmp(acc_method, "newsreply"))	port = 119;
	    else if (!strcmp(acc_method, "snews"))	port = 563;
	    else if (!strcmp(acc_method, "snewspost"))	port = 563;
	    else if (!strcmp(acc_method, "snewsreply")) port = 563;
#endif
	    else if (!strcmp(acc_method, "wais"))	port = 210;
#ifndef DISABLE_FINGER
	    else if (!strcmp(acc_method, "finger"))	port = 79;
#endif
	    else if (!strcmp(acc_method, "telnet"))	port = 23;
	    else if (!strcmp(acc_method, "tn3270"))	port = 23;
	    else if (!strcmp(acc_method, "rlogin"))	port = 513;
	    FREE(acc_method);
	}
    }
    if (!port)
	port = 80;		    /* Default */
    h_len = strlen(Host);

    while (*no_proxy) {
	CONST char * end;
	CONST char * colon = NULL;
	int templ_port = 0;
	int t_len;

	while (*no_proxy && (WHITE(*no_proxy) || *no_proxy == ','))
	    no_proxy++;		    /* Skip whitespace and separators */

	end = no_proxy;
	while (*end && !WHITE(*end) && *end != ',') {	/* Find separator */
	    if (*end == ':') colon = end;		/* Port number given */
	    end++;
	}

	if (colon) {
	    templ_port = atoi(colon+1);
	    t_len = colon - no_proxy;
	}
	else {
	    t_len = end - no_proxy;
	}

	if ((!templ_port || templ_port == port)  &&
	    (t_len > 0	&&  t_len <= h_len  &&
	     !strncasecomp(Host + h_len - t_len, no_proxy, t_len))) {
	    FREE(host);
	    return YES;
	}
#ifdef CJK_EX	/* ASATAKU PROXY HACK */
	if ((!templ_port || templ_port == port)	 &&
	    (t_len > 0	&&  t_len <= h_len  &&
	     isdigit(UCH(*no_proxy)) && !strncmp(host, no_proxy, t_len))) {
	    FREE(host);
	    return YES;
	}
#endif	/* ASATAKU PROXY HACK */

	if (*end)
	    no_proxy = (end + 1);
	else
	    break;
    }

    FREE(host);
    return NO;
}

/*	Find physical name and access protocol		get_physical()
**	--------------------------------------
**
**  On entry,
**	addr		must point to the fully qualified hypertext reference.
**	anchor		a parent anchor with whose address is addr
**
**  On exit,
**	returns		HT_NO_ACCESS		Error has occurred.
**			HT_OK			Success
*/
PRIVATE int get_physical ARGS2(
	CONST char *,		addr,
	HTParentAnchor *,	anchor)
{
    char * acc_method = NULL;	/* Name of access method */
    char * physical = NULL;
    char * Server_addr = NULL;
    BOOL override_flag = NO;

    /*
    **	Make sure the using_proxy variable is FALSE.
    */
    using_proxy = NO;

#ifndef NO_RULES
    physical = HTTranslate(addr);
    if (!physical) {
	if (redirecting_url) {
	    return HT_REDIRECTING;
	}
	return HT_FORBIDDEN;
    }
    if (anchor->isISMAPScript == TRUE) {
	StrAllocCat(physical, "?0,0");
	CTRACE((tfp, "HTAccess: Appending '?0,0' coordinate pair.\n"));
    }
    if (!strncmp(physical, "Proxied=", 8)) {
	HTAnchor_setPhysical(anchor, physical + 8);
	using_proxy = YES;
    } else if (!strncmp(physical, "NoProxy=", 8)) {
	HTAnchor_setPhysical(anchor, physical + 8);
	override_flag = YES;
    } else {
	HTAnchor_setPhysical(anchor, physical);
    }
    FREE(physical);			/* free our copy */
#else
    if (anchor->isISMAPScript == TRUE) {
	StrAllocCopy(physical, addr);
	StrAllocCat(physical, "?0,0");
	CTRACE((tfp, "HTAccess: Appending '?0,0' coordinate pair.\n"));
	HTAnchor_setPhysical(anchor, physical);
	FREE(physical);			/* free our copy */
    } else {
	HTAnchor_setPhysical(anchor, addr);
    }
#endif /* NO_RULES */

    acc_method =  HTParse(HTAnchor_physical(anchor),
		"file:", PARSE_ACCESS);

    /*
    **	Check whether gateway access has been set up for this.
    **
    **	This function can be replaced by the rule system above.
    **
    **  If the rule system has already determined that we should
    **  use a proxy, or that we shouldn't, ignore proxy-related
    **  settings, don't use no_proxy either.
    */
#define USE_GATEWAYS
#ifdef USE_GATEWAYS

    if (!override_flag && !using_proxy) {   /* else ignore no_proxy env var */
	if (!strcasecomp(acc_method, "news")) {
	    /*
	    **  News is different, so we need to check the name of the server,
	    **  as well as the default port for selective exclusions.
	    */
	    char *host = NULL;
	    if ((host = HTParse(addr, "", PARSE_HOST))) {
		if (strchr(host, ':') == NULL) {
		    StrAllocCopy(Server_addr, "news://");
		    StrAllocCat(Server_addr, host);
		    StrAllocCat(Server_addr, ":119/");
		}
		FREE(host);
	    } else if (getenv("NNTPSERVER") != NULL) {
		StrAllocCopy(Server_addr, "news://");
		StrAllocCat(Server_addr, (char *)getenv("NNTPSERVER"));
		StrAllocCat(Server_addr, ":119/");
	    }
	} else if (!strcasecomp(acc_method, "wais")) {
	    /*
	    **  Wais also needs checking of the default port
	    **  for selective exclusions.