htanchor.c

elinks下lynx是最重要的二个文本浏览器, 在linux下非常实用, lynx比elinks早的多, 目前好像停止开发, 这是lynx源代码

/*	Hypertext "Anchor" Object				HTAnchor.c
**	==========================
**
** An anchor represents a region of a hypertext document which is linked to
** another anchor in the same or a different document.
**
** History
**
**	   Nov 1990  Written in Objective-C for the NeXT browser (TBL)
**	24-Oct-1991 (JFG), written in C, browser-independent
**	21-Nov-1991 (JFG), first complete version
**
**	(c) Copyright CERN 1991 - See Copyright.html
*/

#define HASH_SIZE 101		/* Arbitrary prime.  Memory/speed tradeoff */

#include <HTUtils.h>
#include <HTAnchor.h>
#include <HTParse.h>
#include <HTString.h>
#include <UCAux.h>
#include <UCMap.h>

#include <GridText.h>
#include <LYUtils.h>
#include <LYCharSets.h>
#include <LYLeaks.h>

#ifdef NOT_DEFINED
/*
 *	This is the hashing function used to determine which list in the
 *		adult_table a parent anchor should be put in.  This is a
 *		much simpler function than the original used.
 */
#define HASH_FUNCTION(cp_address) ((unsigned short int)strlen(cp_address) *\
	(unsigned short int)TOUPPER(*cp_address) % HASH_SIZE)
#endif /* NOT_DEFINED */
/*
 *	This is the original function.	We'll use it again. - FM
 */
PRIVATE int HASH_FUNCTION ARGS1(
	CONST char *,	cp_address)
{
    int hash;
    CONST unsigned char *p;

    for (p = (CONST unsigned char *)cp_address, hash = 0; *p; p++)
	hash = (int) (hash * 3 + (*(CONST unsigned char *)p)) % HASH_SIZE;

    return(hash);
}

typedef struct _HyperDoc Hyperdoc;
#ifdef VMS
struct _HyperDoc {
	int junk;	/* VMS cannot handle pointers to undefined structs */
};
#endif /* VMS */

/* Table of lists of all parents */
PRIVATE HTList adult_table[HASH_SIZE] = { {NULL, NULL} };


/*				Creation Methods
**				================
**
**	Do not use "new" by itself outside this module.  In order to enforce
**	consistency, we insist that you furnish more information about the
**	anchor you are creating : use newWithParent or newWithAddress.
*/
PRIVATE HTParentAnchor0 * HTParentAnchor0_new ARGS2(
	CONST char *,	address,
	short,		hash)
{
    HTParentAnchor0 *newAnchor = typecalloc(HTParentAnchor0);
    if (newAnchor == NULL)
	outofmem(__FILE__, "HTParentAnchor0_new");

    newAnchor->parent = newAnchor;	/* self */
    StrAllocCopy(newAnchor->address, address);
    newAnchor->adult_hash = hash;

    return(newAnchor);
}

PRIVATE HTParentAnchor * HTParentAnchor_new ARGS1(
	HTParentAnchor0 *,	parent)
{
    HTParentAnchor *newAnchor = typecalloc(HTParentAnchor);
    if (newAnchor == NULL)
	outofmem(__FILE__, "HTParentAnchor_new");

    newAnchor->parent = parent;		/* cross reference */
    parent->info = newAnchor;		/* cross reference */
    newAnchor->address = parent->address;  /* copy pointer */

    newAnchor->isISMAPScript = FALSE;	/* Lynx appends ?0,0 if TRUE. - FM */
    newAnchor->isHEAD = FALSE;		/* HEAD request if TRUE. - FM */
    newAnchor->safe = FALSE;		/* Safe. - FM */
    newAnchor->no_cache = FALSE;	/* no-cache? - FM */
    newAnchor->inBASE = FALSE;		/* duplicated from HTML.c/h */
    newAnchor->content_length = 0;	/* Content-Length. - FM */
    return(newAnchor);
}

PRIVATE HTChildAnchor * HTChildAnchor_new ARGS1(
	HTParentAnchor0 *,	parent)
{
    HTChildAnchor *p = typecalloc(HTChildAnchor);
    if (p == NULL)
	outofmem(__FILE__, "HTChildAnchor_new");

    p->parent = parent;		/* parent reference */
    return p;
}

PRIVATE HTChildAnchor * HText_pool_ChildAnchor_new ARGS1(
	HTParentAnchor *,	parent)
{
    HTChildAnchor *p = (HTChildAnchor *)HText_pool_calloc((HText*)(parent->document),
						sizeof(HTChildAnchor));
    if (p == NULL)
	outofmem(__FILE__, "HText_pool_ChildAnchor_new");

    p->parent = parent->parent;	/* parent reference */
    return p;
}

#ifdef CASE_INSENSITIVE_ANCHORS
/* Case insensitive string comparison */
#define HT_EQUIV(a,b) (TOUPPER(a) == TOUPPER(b))
#else
/* Case sensitive string comparison */
#define HT_EQUIV(a,b) ((a) == (b))
#endif

/*	Null-terminated string comparison
**	---------------------------------
** On entry,
**	s	Points to one string, null terminated
**	t	points to the other.
** On exit,
**	returns YES if the strings are equivalent
**		NO if they differ.
*/
PRIVATE BOOL HTSEquivalent ARGS2(
	CONST char *,	s,
	CONST char *,	t)
{
    if (s && t) {  /* Make sure they point to something */
	for (; *s && *t; s++, t++) {
	    if (!HT_EQUIV(*s, *t)) {
		return(NO);
	    }
	}
	return(HT_EQUIV(*s, *t));
    } else {
	return(s == t);		/* Two NULLs are equivalent, aren't they ? */
    }
}

/*	Binary string comparison
**	------------------------
** On entry,
**	s	Points to one bstring
**	t	points to the other.
** On exit,
**	returns YES if the strings are equivalent
**		NO if they differ.
*/
PRIVATE BOOL HTBEquivalent ARGS2(
	CONST bstring *,	s,
	CONST bstring *,	t)
{
    if (s && t && BStrLen(s) == BStrLen(t)) {
	int j;
	int len = BStrLen(s);
	for (j = 0; j < len; ++j) {
	    if (!HT_EQUIV(BStrData(s)[j], BStrData(t)[j])) {
		return(NO);
	    }
	}
	return(YES);
    } else {
	return(s == t);		/* Two NULLs are equivalent, aren't they ? */
    }
}

/*
 *  Three-way compare function
 */
PRIVATE int compare_anchors ARGS2(
	void *, l,
	void *, r)
{
    CONST char* a = ((HTChildAnchor *)l)->tag;
    CONST char* b = ((HTChildAnchor *)r)->tag;
    /* both tags are not NULL */

#ifdef CASE_INSENSITIVE_ANCHORS
	return strcasecomp(a, b); /* Case insensitive */
#else
	return strcmp(a, b);      /* Case sensitive - FM */
#endif /* CASE_INSENSITIVE_ANCHORS */
}


/*	Create new or find old sub-anchor
**	---------------------------------
**
**	This one is for a named child.
**	The parent anchor must already exist.
*/
PRIVATE HTChildAnchor * HTAnchor_findNamedChild ARGS2(
	HTParentAnchor0 *,	parent,
	CONST char *,		tag)
{
    HTChildAnchor *child;

    if (parent && tag && *tag) {  /* TBL */
	if (parent->children) {
	    /*
	    **  Parent has children.  Search them.
	    */
	    HTChildAnchor sample;
	    sample.tag = (char*)tag;    /* for compare_anchors() only */

	    child = (HTChildAnchor *)HTBTree_search(parent->children, &sample);
	    if (child != NULL) {
		CTRACE((tfp, "Child anchor %p of parent %p with name `%s' already exists.\n",
				(void *)child, (void *)parent, tag));
		return(child);
	    }
	} else {  /* parent doesn't have any children yet : create family */
	    parent->children = HTBTree_new(compare_anchors);
	}

	child = HTChildAnchor_new(parent);
	CTRACE((tfp, "HTAnchor: New Anchor %p named `%s' is child of %p\n",
		(void *)child,
		NonNull(tag),
		(void *)child->parent));

	StrAllocCopy(child->tag, tag);   /* should be set before HTBTree_add */
	HTBTree_add(parent->children, child);
	return(child);

    } else {
	CTRACE((tfp, "HTAnchor_findNamedChild called with NULL parent.\n"));
	return(NULL);
    }

}

/*
**	This one is for a new unnamed child being edited into an existing
**	document.  The parent anchor and the document must already exist.
**	(Just add new unnamed child).
*/
PRIVATE HTChildAnchor * HTAnchor_addChild ARGS1(
	HTParentAnchor *,	parent)
{
    HTChildAnchor *child;

    if (!parent) {
	CTRACE((tfp, "HTAnchor_addChild called with NULL parent.\n"));
	return(NULL);
    }

    child = HText_pool_ChildAnchor_new(parent);
    CTRACE((tfp, "HTAnchor: New unnamed Anchor %p is child of %p\n",
		(void *)child,
		(void *)child->parent));

    child->tag = 0;
    HTList_linkObject(&parent->children_notag, child, &child->_add_children_notag);

    return(child);
}


PRIVATE HTParentAnchor0 * HTAnchor_findAddress_in_adult_table PARAMS((
	CONST DocAddress *	newdoc));

PRIVATE BOOL HTAnchor_link PARAMS((
	HTChildAnchor *		child,
	HTAnchor *		destination,
	HTLinkType *		type));


/*	Create or find a child anchor with a possible link
**	--------------------------------------------------
**
**	Create new anchor with a given parent and possibly
**	a name, and possibly a link to a _relatively_ named anchor.
**	(Code originally in ParseHTML.h)
*/
PUBLIC HTChildAnchor * HTAnchor_findChildAndLink ARGS4(
	HTParentAnchor *,	parent, /* May not be 0   */
	CONST char *,		tag,	/* May be "" or 0 */
	CONST char *,		href,	/* May be "" or 0 */
	HTLinkType *,		ltype)	/* May be 0	  */
{
    HTChildAnchor * child;
    CTRACE((tfp,"Entered HTAnchor_findChildAndLink:  tag=`%s',%s href=`%s'\n",
	       NonNull(tag),
	       (ltype == HTInternalLink) ? " (internal link)" : "",
	       NonNull(href) ));

    if (tag && *tag) {
	child = HTAnchor_findNamedChild(parent->parent, tag);
    } else {
	child = HTAnchor_addChild(parent);
    }

    if (href && *href) {
	CONST char *fragment = NULL;
	HTParentAnchor0 * dest;

	if (ltype == HTInternalLink && *href == '#') {
	    dest = parent->parent;
	} else {
	    CONST char *relative_to = (parent->inBASE && *href != '#') ?
				parent->content_base : parent->address;
	    DocAddress parsed_doc;
	    parsed_doc.address = HTParse(href, relative_to,
					 PARSE_ALL_WITHOUT_ANCHOR);

	    parsed_doc.post_data = NULL;
	    parsed_doc.post_content_type = NULL;
	    if (ltype && parent->post_data && ltype == HTInternalLink) {
		/* for internal links, find a destination with the same
		   post data if the source of the link has post data. - kw
		   Example: LYNXIMGMAP: */
		parsed_doc.post_data = parent->post_data;
		parsed_doc.post_content_type = parent->post_content_type;
	    }
	    parsed_doc.bookmark = NULL;
	    parsed_doc.isHEAD = FALSE;
	    parsed_doc.safe = FALSE;

	    dest = HTAnchor_findAddress_in_adult_table(&parsed_doc);
	    FREE(parsed_doc.address);
	}

	/*
	** [from HTAnchor_findAddress()]
	** If the address represents a sub-anchor, we load its parent (above),
	** then we create a named child anchor within that parent.
	*/
	fragment = (*href == '#') ?  href+1 : HTParseAnchor(href);

	if (*fragment)
	    dest = (HTParentAnchor0 *)HTAnchor_findNamedChild(dest, fragment);


	if (tag && *tag) {
	    if (child->dest) { /* DUPLICATE_ANCHOR_NAME_WORKAROUND  - kw */
		CTRACE((tfp,
			   "*** Duplicate ChildAnchor %p named `%s'",
			   child, tag));
		if ((HTAnchor *)dest != child->dest || ltype != child->type) {
		    CTRACE((tfp,
			   ", different dest %p or type, creating unnamed child\n",
			   child->dest));
		    child = HTAnchor_addChild(parent);
		}
	    }
	}
	HTAnchor_link(child, (HTAnchor *)dest, ltype);
    }
    return child;
}


/*	Create new or find old parent anchor
**	------------------------------------
**
**	Me one is for a reference which is found in a document, and might
**	not be already loaded.
**	Note: You are not guaranteed a new anchor -- you might get an old one,
**	like with fonts.
*/
PUBLIC HTParentAnchor * HTAnchor_findAddress ARGS1(
	CONST DocAddress *,	newdoc)
{
    /* Anchor tag specified ? */
    CONST char *tag = HTParseAnchor(newdoc->address);

    CTRACE((tfp,"Entered HTAnchor_findAddress\n"));

    /*
    **	If the address represents a sub-anchor, we load its parent,
    **	then we create a named child anchor within that parent.
    */
    if (*tag) {
	DocAddress parsed_doc;
	HTParentAnchor0 * foundParent;
	HTChildAnchor * foundAnchor;

	parsed_doc.address = HTParse(newdoc->address, "",
				     PARSE_ALL_WITHOUT_ANCHOR);
	parsed_doc.post_data = newdoc->post_data;
	parsed_doc.post_content_type = newdoc->post_content_type;
	parsed_doc.bookmark = newdoc->bookmark;
	parsed_doc.isHEAD = newdoc->isHEAD;
	parsed_doc.safe = newdoc->safe;

	foundParent = HTAnchor_findAddress_in_adult_table(&parsed_doc);
	foundAnchor = HTAnchor_findNamedChild (foundParent, tag);
	FREE(parsed_doc.address);
	return HTAnchor_parent((HTAnchor *)foundParent);
    }
    return HTAnchor_parent((HTAnchor *)HTAnchor_findAddress_in_adult_table(newdoc));
}


/*  The address has no anchor tag, for sure.
 */
PRIVATE HTParentAnchor0 * HTAnchor_findAddress_in_adult_table ARGS1(
	CONST DocAddress *,	newdoc)
{
    /*
    **  Check whether we have this node.
    */
    int hash;
    HTList * adults;
    HTList *grownups;
    HTParentAnchor0 * foundAnchor;
    BOOL need_extra_info = (newdoc->post_data || newdoc->post_content_type ||
		newdoc->bookmark || newdoc->isHEAD || newdoc->safe);

    /*
     *  We need not free adult_table[] atexit -
     *  it should be perfectly empty after free'ing all HText's.
     *  (There is an error if it is not empty at exit). -LP
     */

    /*
    **  Select list from hash table,
    */
    hash = HASH_FUNCTION(newdoc->address);
    adults = &(adult_table[hash]);

    /*
    **  Search list for anchor.
    */
    grownups = adults;
    while (NULL != (foundAnchor =
		    (HTParentAnchor0 *)HTList_nextObject(grownups))) {
	if (HTSEquivalent(foundAnchor->address, newdoc->address) &&

	    ((!foundAnchor->info && !need_extra_info) ||
	     (foundAnchor->info &&
	      HTBEquivalent(foundAnchor->info->post_data, newdoc->post_data) &&
	      foundAnchor->info->isHEAD == newdoc->isHEAD)))
	{
	    CTRACE((tfp, "Anchor %p with address `%s' already exists.\n",
			(void *)foundAnchor, newdoc->address));
	    return foundAnchor;
	}
    }

    /*
    **  Node not found: create new anchor.
    */
    foundAnchor = HTParentAnchor0_new(newdoc->address, hash);
    CTRACE((tfp, "New anchor %p has hash %d and address `%s'\n",
		(void *)foundAnchor, hash, newdoc->address));

    if (need_extra_info) {
	/* rare case, create a big structure */
	HTParentAnchor *p = HTParentAnchor_new(foundAnchor);

	if (newdoc->post_data)
	    BStrCopy(p->post_data, newdoc->post_data);
	if (newdoc->post_content_type)
	    StrAllocCopy(p->post_content_type,
		     newdoc->post_content_type);
	if (newdoc->bookmark)
	    StrAllocCopy(p->bookmark, newdoc->bookmark);
	p->isHEAD = newdoc->isHEAD;
	p->safe = newdoc->safe;
    }
    HTList_linkObject(adults, foundAnchor, &foundAnchor->_add_adult);