bithtmldtd.c

一个很好的html网页分析工具

#include "BitHtml.h"


#define HTML_ENTITY(_name, _value) {#_name,_value},

BitEntity gHTMLEntity[101]={
HTML_ENTITY(quot	, 34)
HTML_ENTITY(amp	, 38)
HTML_ENTITY(lt	, 60)
HTML_ENTITY(gt	, 62)
HTML_ENTITY(nbsp	, 160)
HTML_ENTITY(iexcl, 161)
HTML_ENTITY(cent	, 162)
HTML_ENTITY(pound, 163)
HTML_ENTITY(curren, 164)
HTML_ENTITY(yen	, 165)
HTML_ENTITY(brvbar, 166)
HTML_ENTITY(sect	, 167)
HTML_ENTITY(uml	, 168)
HTML_ENTITY(copy	, 169)
HTML_ENTITY(ordf	, 170)
HTML_ENTITY(laquo, 171)
HTML_ENTITY(not	, 172)
HTML_ENTITY(shy	, 173)
HTML_ENTITY(reg	, 174)
HTML_ENTITY(macr	, 175)
HTML_ENTITY(deg	, 176)
HTML_ENTITY(plusmn, 177)
HTML_ENTITY(sup2	, 178)
HTML_ENTITY(sup3	, 179)
HTML_ENTITY(acute, 180)
HTML_ENTITY(micro, 181)
HTML_ENTITY(para	, 182)
HTML_ENTITY(middot, 183)
HTML_ENTITY(cedil, 184)
HTML_ENTITY(sup1	, 185)
HTML_ENTITY(ordm, 186)
HTML_ENTITY(raquo, 187)
HTML_ENTITY(frac14, 188)
HTML_ENTITY(frac12, 189)
HTML_ENTITY(frac34, 190)
HTML_ENTITY(iquest, 191)
HTML_ENTITY(Agrave, 192)
HTML_ENTITY(Aacute, 193)
HTML_ENTITY(Acirc, 194)
HTML_ENTITY(Atilde, 195)
HTML_ENTITY(Auml, 196)
HTML_ENTITY(Aring, 197)
HTML_ENTITY(AElig, 198)
HTML_ENTITY(Ccedil, 199)
HTML_ENTITY(Egrave, 200)
HTML_ENTITY(Eacute, 201)
HTML_ENTITY(Ecirc, 202)
HTML_ENTITY(Euml, 203)
HTML_ENTITY(Igrave, 204)
HTML_ENTITY(Iacute, 205)
HTML_ENTITY(Icirc, 206)
HTML_ENTITY(Iuml, 207)
HTML_ENTITY(ETH, 208)
HTML_ENTITY(Ntilde, 209)
HTML_ENTITY(Ograve, 210)
HTML_ENTITY(Oacute, 211)
HTML_ENTITY(Ocirc, 212)
HTML_ENTITY(Otilde, 213)
HTML_ENTITY(Ouml, 214)
HTML_ENTITY(times, 215)
HTML_ENTITY(Oslash, 216)
HTML_ENTITY(Ugrave, 217)
HTML_ENTITY(Uacute, 218)
HTML_ENTITY(Ucirc, 219)
HTML_ENTITY(Uuml, 220)
HTML_ENTITY(Yacute, 221)
HTML_ENTITY(THORN, 222)
HTML_ENTITY(szlig, 223)
HTML_ENTITY(agrave, 224)
HTML_ENTITY(aacute, 225)
HTML_ENTITY(acirc, 226)
HTML_ENTITY(atilde, 227)
HTML_ENTITY(auml, 228)
HTML_ENTITY(aring, 229)
HTML_ENTITY(aelig, 230)
HTML_ENTITY(ccedil, 231)
HTML_ENTITY(egrave, 232)
HTML_ENTITY(eacute, 233)
HTML_ENTITY(ecirc, 234)
HTML_ENTITY(euml, 235)
HTML_ENTITY(igrave, 236)
HTML_ENTITY(iacute, 237)
HTML_ENTITY(icirc, 238)
HTML_ENTITY(iuml, 239)
HTML_ENTITY(eth, 240)
HTML_ENTITY(ntilde, 241)
HTML_ENTITY(ograve, 242)
HTML_ENTITY(oacute, 243)
HTML_ENTITY(ocirc, 244)
HTML_ENTITY(otilde, 245)
HTML_ENTITY(ouml, 246)
HTML_ENTITY(divide, 247)
HTML_ENTITY(oslash, 248)
HTML_ENTITY(ugrave, 249)
HTML_ENTITY(uacute, 250)
HTML_ENTITY(ucirc, 251)
HTML_ENTITY(uuml, 252)
HTML_ENTITY(yacute, 253)
HTML_ENTITY(thorn, 254)
HTML_ENTITY(yuml, 255)
{0}};
#undef HTML_ENTITY

/*
**	ELEMENTS
**	Must match definitions in HTMLPDBit.html!
**	Must be in alphabetical order.
**Name	ID	
*/
#define	HTML_TAG(t)		{ #t,HTML_##t }

BitTag gHTMLTags[HTML_ELEMENTS+1] = {
	HTML_TAG(A),
	HTML_TAG(ABBREV),
	HTML_TAG(ACRONYM),
	HTML_TAG(ADDRESS),
	HTML_TAG(APPLET),
	HTML_TAG(AREA),
	HTML_TAG(AU),
	HTML_TAG(AUTHOR),
	HTML_TAG(B),
	HTML_TAG(BANNER),
	HTML_TAG(BASE),
	HTML_TAG(BASEFONT),
	HTML_TAG(BDO),
	HTML_TAG(BGSOUND),
	HTML_TAG(BIG),
	HTML_TAG(BLINK),
	HTML_TAG(BLOCKQUOTE),
	HTML_TAG(BODY),
	HTML_TAG(BODYTEXT),
	HTML_TAG(BQ),
	HTML_TAG(BR),
	HTML_TAG(BUTTON),
	HTML_TAG(CAPTION),
	HTML_TAG(CENTER),
	HTML_TAG(CITE),
	HTML_TAG(CODE),
	HTML_TAG(COL),
	HTML_TAG(COLGROUP),
	HTML_TAG(COMMENT),
	HTML_TAG(CREDIT),
	HTML_TAG(DD),
	HTML_TAG(DEL),
	HTML_TAG(DFN),
	HTML_TAG(DIR),
	HTML_TAG(DIV),
	HTML_TAG(DL),
	HTML_TAG(DLC),
	HTML_TAG(DT),
	HTML_TAG(EM),
	HTML_TAG(EMBED),
	HTML_TAG(FIELDSET),
	HTML_TAG(FIG),
	HTML_TAG(FN),
	HTML_TAG(FONT),
	HTML_TAG(FORM),
	HTML_TAG(FRAME),
	HTML_TAG(FRAMESET),
	HTML_TAG(H1),
	HTML_TAG(H2),
	HTML_TAG(H3),
	HTML_TAG(H4),
	HTML_TAG(H5),
	HTML_TAG(H6),
	HTML_TAG(HEAD),
	HTML_TAG(HR),
	HTML_TAG(HTML),
	HTML_TAG(HY),
	HTML_TAG(I),
	HTML_TAG(IFRAME),
	HTML_TAG(IMG),
	HTML_TAG(INPUT),
	HTML_TAG(INS),
	HTML_TAG(ISINDEX),
	HTML_TAG(KBD),
	HTML_TAG(KEYGEN),
	HTML_TAG(LABEL),
	HTML_TAG(LEGEND),
	HTML_TAG(LH),
	HTML_TAG(LI),
	HTML_TAG(LINK),
	HTML_TAG(LISTING),
	HTML_TAG(MAP),
	HTML_TAG(MARQUEE),
	HTML_TAG(MATH),
	HTML_TAG(MENU),
	HTML_TAG(META),
	HTML_TAG(NEXTID),
	HTML_TAG(NOFRAMES),
	HTML_TAG(NOTE),
	HTML_TAG(OBJECT),
	HTML_TAG(OL),
	HTML_TAG(OPTION),
	HTML_TAG(OVERLAY),
	HTML_TAG(P),
	HTML_TAG(PARAM),
	HTML_TAG(PLAINTEXT),
	HTML_TAG(PRE),
	HTML_TAG(Q),
	HTML_TAG(S),
	HTML_TAG(SAMP),
	HTML_TAG(SCRIPT),
	HTML_TAG(SELECT),
	HTML_TAG(SHY),
	HTML_TAG(SMALL),
	HTML_TAG(SPAN),
	HTML_TAG(SPOT),
	HTML_TAG(STRIKE),
	HTML_TAG(STRONG),
	HTML_TAG(STYLE),
	HTML_TAG(SUB),
	HTML_TAG(SUP),
	HTML_TAG(TAB),
	HTML_TAG(TABLE),
	HTML_TAG(TBODY),
	HTML_TAG(TD),
	HTML_TAG(TEXTAREA),
	HTML_TAG(TEXTFLOW),
	HTML_TAG(TFOOT),
	HTML_TAG(TH),
	HTML_TAG(THEAD),
	HTML_TAG(TITLE),
	HTML_TAG(TR),
	HTML_TAG(TT),
	HTML_TAG(U),
	HTML_TAG(UL),
	HTML_TAG(VAR),
	HTML_TAG(WBR),
	HTML_TAG(XMP),

	//additional tag
	HTML_TAG(TEXT),

  /* old TDXML
    HTML_TAG(A)  ,
    HTML_TAG(AREA)	,
    HTML_TAG(B)		,
    HTML_TAG(BODY)	,
    HTML_TAG(BR)	,
    HTML_TAG(BUTTON)	,
    HTML_TAG(COMMENT),
    HTML_TAG(FONT),
    HTML_TAG(GROUP)	,
    HTML_TAG(HEAD)	,
    HTML_TAG(I),
    HTML_TAG(IFRAME),
    HTML_TAG(IMG),
    HTML_TAG(INPUT),
    HTML_TAG(LABEL),
    HTML_TAG(MAP),
    HTML_TAG(META),
    HTML_TAG(P)	,
    HTML_TAG(SCRIPT),
    HTML_TAG(SCROLL),
    HTML_TAG(STRIKE),
    HTML_TAG(SUB),
    HTML_TAG(SUP),
    HTML_TAG(TEXT),
    HTML_TAG(TEXTAREA),
    HTML_TAG(TITLE),
    HTML_TAG(U)	,
    HTML_TAG(VAR),
    //HTML_TAG(VIDEO),

	//below add by me
	HTML_TAG(TABLE),
	HTML_TAG(TR),
	HTML_TAG(TD),
	HTML_TAG(TH),
	HTML_TAG(TBODY),
	*/

	//{ #t,HTML_##t }
	{ "HTML", HTML_HTML},
};

#ifdef HTML_ATTR
#undef HTML_ATTR
#endif
#define	HTML_ATTR(a)		{ #a,HTML_ATTR_##a }

 BitTag gHTMLAttr[HTML_ATTR_ATTRIBUTES] = {
	HTML_ATTR(ACCESSKEY),
	HTML_ATTR(ALIGN),
	HTML_ATTR(ALPHA),
	HTML_ATTR(ALT),
	HTML_ATTR(BKCOLOR),
	HTML_ATTR(BKSOUND),
	HTML_ATTR(BORDER),
	HTML_ATTR(CHARSET) ,
	HTML_ATTR(CLASS),
	HTML_ATTR(COLOR),
	HTML_ATTR(COLORKEY),
	HTML_ATTR(COLS),
	HTML_ATTR(COORDS),
	HTML_ATTR(DEFER),
	HTML_ATTR(DISABLED),
	HTML_ATTR(END),
	HTML_ATTR(EXTER),
	HTML_ATTR(HREF),
	HTML_ATTR(HREFLANG),
	HTML_ATTR(HTTP_EQUIV),    /* { "HTTP-EQUIV" ) */
	HTML_ATTR(ID),
	HTML_ATTR(ISMAP),
	HTML_ATTR(LANG),
	HTML_ATTR(MARGINHEIGHT),
	HTML_ATTR(MARGINWIDTH),
	HTML_ATTR(NAME),
	HTML_ATTR(NOHREF),
	HTML_ATTR(NORESIZE),
	HTML_ATTR(ONBLUR),
	HTML_ATTR(ONDBCLICK),
	HTML_ATTR(ONFOCUS),
	HTML_ATTR(ONMOUSEDOWN),
	HTML_ATTR(ONMOUSEMOVE),
	HTML_ATTR(ONMOUSEOVER),
	HTML_ATTR(ONMOUSEOUT),
	HTML_ATTR(ONMOUSEUP),
	HTML_ATTR(ONKEYDOWN),
	HTML_ATTR(ONKEYUP),
	HTML_ATTR(PROFILE),
	HTML_ATTR(READONLY),
	HTML_ATTR(ROWS),
	HTML_ATTR(SCHEME),
	HTML_ATTR(SHAPE),
	HTML_ATTR(SCROLLING),
	HTML_ATTR(SIZE),
	HTML_ATTR(SRC),
	HTML_ATTR(SRCDOWN),
	HTML_ATTR(SRCOVER),
	HTML_ATTR(START),
	HTML_ATTR(STYLE),
	HTML_ATTR(TARGET),
	HTML_ATTR(TITLE),
	HTML_ATTR(TYPE),
	HTML_ATTR(USRMAP),
	HTML_ATTR(VALUE),
	HTML_ATTR(VERSION),
	HTML_ATTR(ZORDER),

	// below add by me
	HTML_ATTR(ACTION),
	HTML_ATTR(BACKGROUND),
	HTML_ATTR(BGCOLOR),
	HTML_ATTR(CELLPADDING),
	HTML_ATTR(CELLSPACING),
	HTML_ATTR(HEIGHT),
	HTML_ATTR(LINK),
	HTML_ATTR(TOPMARGIN),
	HTML_ATTR(VALIGN),
	HTML_ATTR(VLINK),
	HTML_ATTR(WIDTH),
};

/*
JSHashTable* gHTMLTagHashTable=NULL,*gHTMLAttrHashTable=NULL;

int BitHTML_InitHashTable()
{
	int i;
	gHTMLTagHashTable=JS_NewHashTable(HTML_ELEMENTS,NULL,NULL,NULL,NULL,NULL);
	gHTMLAttrHashTable=JS_NewHashTable(HTML_ELEMENTS,NULL,NULL,NULL,NULL,NULL);
	if(!gHTMLTagHashTable || !gHTMLAttrHashTable)
		return BitFALSE;
	for(i=0;i<HTML_ELEMENTS;i++)
	{
		JS_HashTableAdd(gHTMLTagHashTable,gHTMLTags[i].name,(void*)gHTMLTags[i].id);
	}
	for(i=0;i<HTML_ATTR_ATTRIBUTES;i++)
	{
		JS_HashTableAdd(gHTMLAttrHashTable,gHTMLAttr[i].name,(void*)gHTMLAttr[i].id);
	}
	return BitTRUE;
}
*/

int BitHTML_GetTagID(char *pName)
{
	int i;
	BitTag *ct;
	if(!pName)
		return 0;
	for (i = 0; i<HTML_ELEMENTS; i++) 
	{
			ct =&gHTMLTags[i];
			if(stricmp(pName,ct->name)==0)
				return i;
	}
	return -1;
}

int BitHTML_GetAtrID(char *pName)
{
	int i;
	BitTag *ct;
	if(!pName)
		return 0;
	for (i = 0; i<HTML_ATTR_ATTRIBUTES; i++) 
	{
			ct =&gHTMLAttr[i];
			if(stricmp(pName,ct->name)==0)
				return i;
	}
	return -1;
}

char *BitHTML_GetTagName(int  id)
{
	if(id>=0 && id< HTML_ELEMENTS)
		return gHTMLTags[id].name;
	return NULL;
}

char * BitHTML_GetAtrName(int id)
{
	if(id>=0 && id< HTML_ATTR_ATTRIBUTES)
		return gHTMLAttr[id].name;
	return NULL;
}

/*
typedef struct BitTagList {
  BitHTML_GET_ID		fGetTagID;
  BitHTML_GET_NAME	fGetTagName;
  BitHTML_GET_ID		fGetAtrID;
  BitHTML_GET_NAME	fGetAtrName;
  int			mEntityCount;
  BitPEntity		pEntity;
  int			nComment;
  char*		pCommentEnd;
  int			nPlainText;
}BitTagList,*BitPTagList;
*/

BitTagList gHTMLTagList={BitHTML_GetTagID,BitHTML_GetTagName,
						BitHTML_GetAtrID,BitHTML_GetAtrName,
						100,gHTMLEntity,
						HTML_COMMENT,"-->",HTML_TEXT};
BitPTagList	Bit_GetHTMLTagList()
{
	return &gHTMLTagList;
}