htparse.c

www工具包. 这是W3C官方支持的www支撑库. 其中提供通用目的的客户端的WebAPI: complete HTTP/1.1 (with caching, pipelining, PUT, POS

/*								      HTParse.c
**	URI MANAGEMENT
**
**	(c) COPYRIGHT MIT 1995.
**	Please first read the full copyright statement in the file COPYRIGH.
**	@(#) $Id: HTParse.c,v 2.80 1999/02/22 22:10:11 frystyk Exp $
**
** history:
**	May 12 94	TAB added as legal char in HTCleanTelnetString
**
*/

/* Library include files */
#include "wwwsys.h"
#include "WWWUtil.h"
#include "HTParse.h"					 /* Implemented here */

typedef struct _HTURI {
    char * access;		/* Now known as "scheme" */
    char * host;
    char * absolute;
    char * relative;
    char * fragment;
} HTURI;

/*	Scan a filename for its consituents
**	-----------------------------------
**
** On entry,
**	name	points to a document name which may be incomplete.
** On exit,
**      absolute or relative may be nonzero (but not both).
**	host, fragment and access may be nonzero if they were specified.
**	Any which are nonzero point to zero terminated strings.
*/
PRIVATE void scan (char * name, HTURI * parts)
{
    char * p;
    char * after_access = name;
    memset(parts, '\0', sizeof(HTURI));

    /* Look for fragment identifier */
    if ((p = strchr(name, '#')) != NULL) {
	*p++ = '\0';
	parts->fragment = p;
    }
    
    
    if ((p = strchr(name, ' ')) != NULL) *p++ = '\0';    
    
    for(p=name; *p; p++) {

	/* 
	** Look for any whitespace. This is very bad for pipelining as it
	** makes the request invalid
	*/
	if (isspace((int) *p)) {
	    char *orig=p, *dest=p+1;
	    while ((*orig++ = *dest++));
	    p = p-1;
	}
	if (*p=='/' || *p=='#' || *p=='?')
	    break;
	if (*p==':') {
		*p = 0;
		parts->access = after_access; /* Scheme has been specified */

/* The combination of gcc, the "-O" flag and the HP platform is
   unhealthy. The following three lines is a quick & dirty fix, but is
   not recommended. Rather, turn off "-O". */

/*		after_access = p;*/
/*		while (*after_access == 0)*/
/*		    after_access++;*/

		after_access = p+1;

		if (0==strcasecomp("URL", parts->access)) {
		    parts->access = NULL;  /* Ignore IETF's URL: pre-prefix */
		} else break;
	}
    }
    
    p = after_access;
    if (*p=='/'){
	if (p[1]=='/') {
	    parts->host = p+2;		/* host has been specified 	*/
	    *p=0;			/* Terminate access 		*/
	    p=strchr(parts->host,'/');	/* look for end of host name if any */
	    if(p) {
	        *p=0;			/* Terminate host */
	        parts->absolute = p+1;		/* Root has been found */
	    }
	} else {
	    parts->absolute = p+1;		/* Root found but no host */
	}	    
    } else {
        parts->relative = (*after_access) ? after_access : 0; /* zero for "" */
    }
}


/*	Parse a Name relative to another name
**	-------------------------------------
**
**	This returns those parts of a name which are given (and requested)
**	substituting bits from the related name where necessary.
**
** On entry,
**	aName		A filename given
**      relatedName     A name relative to which aName is to be parsed. Give
**                      it an empty string if aName is absolute.
**      wanted          A mask for the bits which are wanted.
**
** On exit,
**	returns		A pointer to a malloc'd string which MUST BE FREED
*/
PUBLIC char * HTParse (const char *aName, const char *relatedName, int wanted)
{
    char * result = 0;
    char * return_value = 0;
    int len;
    char * name = 0;
    char * rel = 0;
    char * p;
    char * access;
    HTURI given, related;
    
    if (!aName) return NULL;
    if (!relatedName)        /* HWL 23/8/94: dont dump due to NULL */
        relatedName = "";
    
    /* Make working copies of input strings to cut up: */
    len = strlen(aName)+strlen(relatedName)+10;
    if ((result=(char *) HT_MALLOC(len)) == NULL) /* Lots of space: more than enough */
	HT_OUTOFMEM("parse space");
    StrAllocCopy(name, aName);
    StrAllocCopy(rel, relatedName);
    
    scan(name, &given);
    scan(rel,  &related); 
    result[0]=0;		/* Clear string  */
    access = given.access ? given.access : related.access;
    if (wanted & PARSE_ACCESS)
        if (access) {
	    strcat(result, access);
	    if(wanted & PARSE_PUNCTUATION) strcat(result, ":");
	}
	
    if (given.access && related.access)	/* If different, inherit nothing. */
        if (strcmp(given.access, related.access)!=0) {
	    related.host=0;
	    related.absolute=0;
	    related.relative=0;
	    related.fragment=0;
	}
	
    if (wanted & PARSE_HOST)
        if(given.host || related.host) {
	    if(wanted & PARSE_PUNCTUATION) strcat(result, "//");
	    strcat(result, given.host ? given.host : related.host);
	}
	
    if (given.host && related.host)  /* If different hosts, inherit no path. */
        if (strcmp(given.host, related.host)!=0) {
	    related.absolute=0;
	    related.relative=0;
	    related.fragment=0;
	}
	
    if (wanted & PARSE_PATH) {
        if(given.absolute) {				/* All is given */
	    if(wanted & PARSE_PUNCTUATION) strcat(result, "/");
	    strcat(result, given.absolute);
	} else if(related.absolute) {	/* Adopt path not name */
	    strcat(result, "/");
	    strcat(result, related.absolute);
	    if (given.relative) {
		p = strchr(result, '?');	/* Search part? */
		if (!p) p=result+strlen(result)-1;
		for (; *p!='/'; p--);	/* last / */
		p[1]=0;					/* Remove filename */
		strcat(result, given.relative);		/* Add given one */
#if 0
		result = HTSimplify (&result);
#endif
	    }
	} else if(given.relative) {
	    strcat(result, given.relative);		/* what we've got */
	} else if(related.relative) {
	    strcat(result, related.relative);
	} else {  /* No inheritance */
	    strcat(result, "/");
	}
    }
		
    if (wanted & PARSE_VIEW)
	if(given.fragment || related.fragment) {
	    if(given.absolute && given.fragment) {   /*Fixes for relURLs...*/
		if(wanted & PARSE_PUNCTUATION) strcat(result, "#");
		strcat(result, given.fragment); 
	    } else if (!(given.absolute) && !(given.fragment)) {
		strcat(result, "");
	    } else {
		if(wanted & PARSE_PUNCTUATION) strcat(result, "#");
		strcat(result, given.fragment ? given.fragment : related.fragment); 
	    }
	}
    HT_FREE(rel);
    HT_FREE(name);
    
    StrAllocCopy(return_value, result);
    HT_FREE(result);
    return return_value;		/* exactly the right length */
}


/*
**	Canonicalizes the URL in the following manner starting from the host
**	pointer:
**
**	1) The host name is converted to lowercase
**	2) Chop off port if `:80' (http), `:70' (gopher), or `:21' (ftp)
**
**	Return: OK	The position of the current path part of the URL
**			which might be the old one or a new one.
*/
PRIVATE char * HTCanon (char ** filename, char * host)
{
    char *newname = NULL;
    char *port;
    char *strptr;
    char *path;
    char *access = host-3;

    while (access>*filename && *(access-1)!='/')       /* Find access method */
	access--;
    if ((path = strchr(host, '/')) == NULL)			/* Find path */
	path = host + strlen(host);
    if ((strptr = strchr(host, '@')) != NULL && strptr<path)	   /* UserId */
	host = strptr;
    if ((port = strchr(host, ':')) != NULL && port>path)      /* Port number */
	port = NULL;

    strptr = host;				    /* Convert to lower-case */
    while (strptr<path) {
	*strptr = TOLOWER(*strptr);
	strptr++;
    }
    
    /* Does the URL contain a full domain name? This also works for a
       numerical host name. The domain name is already made lower-case
       and without a trailing dot. */
#if 0
    if (((strptr = strchr(host, '.')) == NULL || strptr >= path) &&
	strncasecomp(host, "localhost", 9)) {
	const char *domain = HTGetDomainName();
	if (domain && *domain) {
	    if ((newname = (char *) HT_CALLOC(1, strlen(*filename) + strlen(domain)+2)) == NULL)
		HT_OUTOFMEM("HTCanon");
	    if (port)
		strncpy(newname, *filename, (int) (port-*filename));
	    else
		strncpy(newname, *filename, (int) (path-*filename));
	    strcat(newname, ".");
	    strcat(newname, domain);
	}
    } else 					  /* Look for a trailing dot */
#endif
    {
	char *dot = port ? port : path;
	if (dot > *filename && *--dot=='.') {