htanchor.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

	if (me->children) {
	    int cnt = 0;
	    for (; cnt<CHILD_HASH_SIZE; cnt++) {
		HTList * kids = me->children[cnt];
		if (kids) {
		    HTChildAnchor * child;
		    while ((child = (HTChildAnchor *) HTList_nextObject(kids)))
			delete_links((HTAnchor *) child);
		    return NO;	/* Parent not deleted */
		}
	    }
	}

	/*
	** No more incoming links : kill everything
	** First, recursively delete children
	*/
	if (me->children) {
	    int cnt = 0;
	    for (; cnt<CHILD_HASH_SIZE; cnt++) {
		HTList * kids = me->children[cnt];
		if (kids) {
		    HTChildAnchor * child;
		    while ((child=(HTChildAnchor *) HTList_removeLastObject(kids)))
			delete_links((HTAnchor *) child);
		    HT_FREE(child->tag);
		    HT_FREE(child);
		}
	    }
	}
    }

    /* 2001/03/06: Bug fix by Serge Adda <[e-mail]>
       HTAnchor_delete wasn't removing the reference to the deleted
       anchor. This caused a bug whenever requesting another anchor
       for the same URL.
    */
    if (adult_table) {
      int hash;
      const char *p;
      HTList * adults;
      HTList * grownups;
      HTList * last;
      HTParentAnchor * foundAnchor;

      /* Select list from hash table */
      for(p=me->address, hash=0; *p; p++)
	hash = (int) ((hash * 3 + (*(unsigned char*)p)) %
		      PARENT_HASH_SIZE);
      adults = adult_table[hash];

      /* Search list for anchor */
      grownups = adults;
      last = grownups;
      while ((foundAnchor = (HTParentAnchor *)
	      HTList_nextObject(grownups))){
	if (!strcmp(foundAnchor->address, me->address)) {
	  HTList_quickRemoveElement (grownups, last);
	  break;
	}
	last = grownups;
      }
    }

    /* Now kill myself */
    delete_parent(me);
    return YES;  /* Parent deleted */
#if 0
  if (! HTList_isEmpty (me->sources)) {  /* There are still incoming links */
    /* Delete all outgoing links from children, if any */
    HTList *kids = me->children;
    while ((child = (HTChildAnchor *) HTList_nextObject (kids)))
      delete_links ((HTAnchor *) child);
    return NO;  /* Parent not deleted */
  }

  /* No more incoming links : kill everything */
  /* First, recursively delete children */
  while ((child = (HTChildAnchor *) HTList_removeLastObject (me->children))) {
    delete_links ((HTAnchor *) child);
    HT_FREE(child->tag);
    HT_FREE(child);
  }
#endif
}

/*	FLATTEN ALL ANCHORS
**	-------------------
**	Flattens the anchor web structure into an array.
**	This is useful for calculating statistics, sorting
**	the parent anchors etc.
**
**      The caller can indicate the size of the array (total
**      number of anchors if known - otherwise 0).
**
**	Return an array that must be freed by the caller or
**      NULL if no anchors.
*/
PUBLIC HTArray * HTAnchor_getArray (int growby)
{
    int cnt;
    HTArray * array = NULL;
    HTList * cur = NULL;
    if (!adult_table) return NULL;

    /* Allocate an array for the anchors */
    if (growby <= 0) growby = PARENT_HASH_SIZE;
    array = HTArray_new(growby);

    /* Traverse anchor structure */
    for (cnt=0; cnt<PARENT_HASH_SIZE; cnt++) {
	if ((cur = adult_table[cnt])) { 
	    HTParentAnchor * pres = NULL;
	    while ((pres = (HTParentAnchor *) HTList_nextObject(cur)) != NULL) {
                if (HTArray_addObject(array, pres) == NO) {
                    HTTRACE(ANCH_TRACE, "Anchor...... Can't add object %p to array %p\n" _ 
				pres _ array);
                    break;
                }
	    }
	}
    }
    return array;
}

/* ------------------------------------------------------------------------- */
/*				Data Access Methods			     */
/* ------------------------------------------------------------------------- */

PUBLIC HTParentAnchor * HTAnchor_parent  (HTAnchor * me)
{
    return me ? me->parent : NULL;
}

PUBLIC void HTAnchor_setDocument  (HTParentAnchor * me, void * doc)
{
    if (me) me->document = doc;
}

PUBLIC void * HTAnchor_document  (HTParentAnchor * me)
{
    return me ? me->document : NULL;
}

PUBLIC char * HTAnchor_address  (HTAnchor * me) 
{ 
    char *addr = NULL;
    if (me) {
        if (((HTParentAnchor *) me == me->parent) ||
            !((HTChildAnchor *) me)->tag) { /* it's an adult or no tag */
            StrAllocCopy (addr, me->parent->address);
        }
        else {                  /* it's a named child */
            if ((addr = (char  *) HT_MALLOC(2 + strlen (me->parent->address) + \
strlen (((HTChildAnchor *) me)->tag))) == NULL)
                HT_OUTOFMEM("HTAnchor_address");
            sprintf (addr, "%s#%s", me->parent->address,
                     ((HTChildAnchor *) me)->tag);
        }
    }
    return addr;
}

/*
**	We resolve the child address with respect to either a base URL,
**	a content-location, or to the request-URI
*/
PUBLIC char * HTAnchor_expandedAddress  (HTAnchor * me)
{
    char *addr = NULL;
    if (me) {
	HTParentAnchor * parent = me->parent;
	char * base = HTAnchor_base(parent);
	if (((HTParentAnchor *) me == me->parent) ||
	    !((HTChildAnchor *) me)->tag) { /* it's an adult or no tag */
	    StrAllocCopy(addr, base);
	} else {			/* it's a named child */
	    if ((addr = (char *) HT_MALLOC(2 + strlen(base) + strlen(((HTChildAnchor *) me)->tag))) == NULL)
	        HT_OUTOFMEM("HTAnchor_address");
	    sprintf (addr, "%s#%s", base, ((HTChildAnchor *) me)->tag);
	}
    }
    return addr;
}

/*	Physical Address
**	----------------
*/
PUBLIC char * HTAnchor_physical (HTParentAnchor * me)
{
    return me ? me->physical ? me->physical : me->address : NULL;
}

PUBLIC void HTAnchor_setPhysical (HTParentAnchor * me, char * physical)
{
    if (!me || !physical) {
	HTTRACE(ANCH_TRACE, "HTAnchor.... setPhysical, called with null argument\n");
	return;
    }
    StrAllocCopy(me->physical, physical);
}

PUBLIC void HTAnchor_clearPhysical(HTParentAnchor * me)
{
    if (me) HT_FREE(me->physical);
}

/*
**	Children information
*/
PUBLIC BOOL HTAnchor_hasChildren  (HTParentAnchor * me)
{
    return (me && me->children);
}

/*
** Fix up a simple routine to see if this anchor is a (ChildAnchor *)
** Seem to be doing it all over the place, so simplify!
*/
PUBLIC BOOL HTAnchor_isChild (HTAnchor * me)
{
    return (me && (HTParentAnchor *) me != me->parent);
}

PUBLIC char * HTAnchor_view (HTAnchor * me)
{
    char * view = NULL;
    if (me && (HTParentAnchor *) me != me->parent && ((HTChildAnchor *) me)->tag)
	StrAllocCopy(view, ((HTChildAnchor *) me)->tag);
    return view;
}

/* ------------------------------------------------------------------------- */
/*			      Entity Header Information			     */
/* ------------------------------------------------------------------------- */

/*
**  Take the relevant infomration from the response object and cache it
**  in the anchor object. We inherit the information that is already
**  parsed in the response along with the unparsed headers.
*/
PUBLIC BOOL HTAnchor_update (HTParentAnchor * me, HTResponse * response)
{
    if (me && response) {
	HTCachable cachable = HTResponse_isCachable(response);

	if (cachable == HT_CACHE_ETAG) {
	    char * etag = HTResponse_etag(response);
	    HTTRACE(ANCH_TRACE, "HTAnchor.... Updating etag for %p\n" _ me);
	    if (etag) {
		HTAnchor_setEtag(me, etag);
		return YES;
	    }

	} else if (cachable == HT_CACHE_NOT_MODIFIED) {
	    HTTRACE(ANCH_TRACE, "HTAnchor.... Information is up to date for %p\n" _ me);
	    return YES;

	} else if (cachable == HT_CACHE_ALL) {
	    char * etag = HTResponse_etag(response);
	    HTTRACE(ANCH_TRACE, "HTAnchor.... Updating metainformation for %p\n" _ me);

	    /*
	    **  The content length and type is already parsed at this point
	    **  in time. We also check for format parameters like charset etc.
	    **  and copy the contents in the anchor object
	    */
	    me->content_length = HTResponse_length(response);
	    me->content_type = HTResponse_format(response);
	    me->type_parameters = HTResponse_formatParam(response);
	    me->content_encoding = HTResponse_encoding(response);
	
            /* Don't forget the etag as well */
       	    if (etag) HTAnchor_setEtag(me, etag);

	    /*
	    **  Inherit all the unparsed headers - we may need them later!
	    */
	    if (me->headers) HTAssocList_delete(me->headers);
	    me->headers = HTResponse_handOverHeader(response);

	    /*
	    **  Notifify the response object not to delete the lists that we
	    **  have inherited in the anchor object
	    */
	    HTResponse_isCached(response, YES);

	    /*
	    **  Set the datestamp of when the anchor was updated if we didn't
	    **  get any in the response
	    */
	    if (!HTAssocList_findObject(me->headers, "date"))
		HTAnchor_setDate(me, time(NULL));

	    return YES;
	}
    }
    return NO;
}

/*
**	Variants. If this anchor has any variants then keep them in a list
**	so that we can find them later. The list is simply a list of 
**	parent anchors.
*/
PUBLIC HTList * HTAnchor_variants (HTParentAnchor * me)
{
    return me ? me->variants : NULL;
}

PUBLIC BOOL HTAnchor_addVariant (HTParentAnchor * me,
				 HTParentAnchor * variant)
{
    if (me && variant) {
	if (!me->variants) me->variants = HTList_new();
	return HTList_addObject(me->variants, variant);
    }
    return NO;
}

PUBLIC BOOL HTAnchor_deleteVariant (HTParentAnchor * me,
				    HTParentAnchor * variant)
{
    return (me && variant) ? HTList_removeObject(me->variants, variant) : NO;
}

/*
**	Is this resource an index?
*/
PUBLIC void HTAnchor_clearIndex  (HTParentAnchor * me)
{
    if (me) me->isIndex = NO;
}

PUBLIC void HTAnchor_setIndex  (HTParentAnchor * me)
{
  if (me) me->isIndex = YES;
}

PUBLIC BOOL HTAnchor_isIndex  (HTParentAnchor * me)
{
    return me ? me->isIndex : NO;
}

/*	Content Base
**	------------
*/
PUBLIC char * HTAnchor_base (HTParentAnchor * me)
{
    if (me) {
	if (me->content_base) return me->content_base;
	if (me->headers) {
	    char * base = HTAssocList_findObject(me->headers, "content-base");
	    /*
	    **  If no base is found then take the content-location if this
	    **  is present and is absolute, else use the Request-URI.
	    */
	    if (base) StrAllocCopy(me->content_base, HTStrip(base));
	}

	/*
	**  Try the content location if any
	*/
	{
	    char * location = HTAnchor_location(me);
	    StrAllocCopy(me->content_base,
			 (location && HTURL_isAbsolute(location)) ?
			 location : me->address);
	}
	return me->content_base;
    }
    return NULL;
}

PUBLIC BOOL HTAnchor_setBase (HTParentAnchor * me, char * base)
{
    if (me && base) {
	StrAllocCopy(me->content_base, base);
	return YES;
    }
    return NO;
}

/*	Content Location
**	----------------
*/
PUBLIC char * HTAnchor_location (HTParentAnchor * me)
{
    if (me) {
	if (me->content_location)
	    return *me->content_location ? me->content_location : NULL;
	if (me->headers) {
	    char * location = HTAssocList_findObject(me->headers, "content-location");
	    StrAllocCopy(me->content_location, location ? HTStrip(location) : "");
	    return me->content_location;
	}
    }
    return NULL;
}

/*
**	Expand the location relative to the base URL if any, otherwise the 
**	anchor address it self
*/
PUBLIC BOOL HTAnchor_setLocation (HTParentAnchor * me, char * location)
{
    if (me && location) {
	char * base = HTAnchor_base(me);
	if (!base) base = me->address;
	me->content_location = HTParse(location, base, PARSE_ALL);
	return YES;
    }
    return NO;
}

/*	Meta tags
**	---------
*/
PUBLIC HTAssocList * HTAnchor_meta (HTParentAnchor * me)
{
    return me ? me->meta_tags : NULL;
}

PUBLIC BOOL HTAnchor_addMeta (HTParentAnchor * me,
			      const char * name, const char * value)
{
    if (me) {
	if (!me->meta_tags) me->meta_tags = HTAssocList_new();
	return HTAssocList_replaceObject(me->meta_tags, name, value);
    }
    return NO;
}

/*
**	robots meta tag
*/
PUBLIC char * HTAnchor_robots (HTParentAnchor * me)
{
    if (me && me->meta_tags) {
	char * robots = HTAssocList_findObject(me->meta_tags, "robots");
	return robots;
    }
    return NULL;
}

/*	Content-Type
**	------------
*/
PUBLIC HTFormat HTAnchor_format (HTParentAnchor * me)
{
    return me ? me->content_type : NULL;
}

PUBLIC void HTAnchor_setFormat (HTParentAnchor * me, HTFormat form)
{
    if (me) me->content_type = form;
}

PUBLIC HTAssocList * HTAnchor_formatParam (HTParentAnchor * me)
{
    return me ? me->type_parameters : NULL;
}

PUBLIC BOOL HTAnchor_addFormatParam (HTParentAnchor * me,
				     const char * name, const char * value)
{
    if (me) {
	if (!me->type_parameters) me->type_parameters = HTAssocList_new();
	return HTAssocList_replaceObject(me->type_parameters, name, value);
    }
    return NO;
}

/*
**	Charset parameter to Content-Type
*/
PUBLIC HTCharset HTAnchor_charset (HTParentAnchor * me)
{
    if (me && me->type_parameters) {