bithtmldtd.h

一个很好的html网页分析工具

#ifndef __NETBIT_HTMLDTD_H__
#define __NETBIT_HTMLDTD_H__

#define HTML_ENTITIES 67

#include "BitDtd.h"

typedef enum _HTMLElement {
	HTML_A=0,
	HTML_ABBREV,
	HTML_ACRONYM,
	HTML_ADDRESS,
	HTML_APPLET,
	HTML_AREA,
	HTML_AU,
	HTML_AUTHOR,
	HTML_B,
	HTML_BANNER,
	HTML_BASE,
	HTML_BASEFONT,
	HTML_BDO,
	HTML_BGSOUND,
	HTML_BIG,
	HTML_BLINK,
	HTML_BLOCKQUOTE,
	HTML_BODY,
	HTML_BODYTEXT,
	HTML_BQ,
	HTML_BR,
	HTML_BUTTON,
	HTML_CAPTION,
	HTML_CENTER,
	HTML_CITE,
	HTML_CODE,
	HTML_COL,
	HTML_COLGROUP,
	HTML_COMMENT,
	HTML_CREDIT,
	HTML_DD,
	HTML_DEL,
	HTML_DFN,
	HTML_DIR,
	HTML_DIV,
	HTML_DL,
	HTML_DLC,
	HTML_DT,
	HTML_EM,
	HTML_EMBED,
	HTML_FIELDSET,
	HTML_FIG,
	HTML_FN,
	HTML_FONT,
	HTML_FORM,
	HTML_FRAME,
	HTML_FRAMESET,
	HTML_H1,
	HTML_H2,
	HTML_H3,
	HTML_H4,
	HTML_H5,
	HTML_H6,
	HTML_HEAD,
	HTML_HR,
	HTML_HTML,
	HTML_HY,
	HTML_I,
	HTML_IFRAME,
	HTML_IMG,
	HTML_INPUT,
	HTML_INS,
	HTML_ISINDEX,
	HTML_KBD,
	HTML_KEYGEN,
	HTML_LABEL,
	HTML_LEGEND,
	HTML_LH,
	HTML_LI,
	HTML_LINK,
	HTML_LISTING,
	HTML_MAP,
	HTML_MARQUEE,
	HTML_MATH,
	HTML_MENU,
	HTML_META,
	HTML_NEXTID,
	HTML_NOFRAMES,
	HTML_NOTE,
	HTML_OBJECT,
	HTML_OL,
	HTML_OPTION,
	HTML_OVERLAY,
	HTML_P,
	HTML_PARAM,
	HTML_PLAINTEXT,
	HTML_PRE,
	HTML_Q,
	HTML_S,
	HTML_SAMP,
	HTML_SCRIPT,
	HTML_SELECT,
	HTML_SHY,
	HTML_SMALL,
	HTML_SPAN,
	HTML_SPOT,
	HTML_STRIKE,
	HTML_STRONG,
	HTML_STYLE,
	HTML_SUB,
	HTML_SUP,
	HTML_TAB,
	HTML_TABLE,
	HTML_TBODY,
	HTML_TD,
	HTML_TEXTAREA,
	HTML_TEXTFLOW,
	HTML_TFOOT,
	HTML_TH,
	HTML_THEAD,
	HTML_TITLE,
	HTML_TR,
	HTML_TT,
	HTML_U,
	HTML_UL,
	HTML_VAR,
	HTML_WBR,
	HTML_XMP,

	//additional tag
	HTML_TEXT,

	/* old TDXML
	HTML_A = 0,
	HTML_AREA,
	HTML_B,
	HTML_BODY,
	HTML_BR,
	HTML_BUTTON,
	HTML_COMMENT, 
	HTML_FONT,
	HTML_GROUP,
	HTML_HEAD,
	HTML_I,
	HTML_IFRAME,
	HTML_IMG,
	HTML_INPUT,
	HTML_LABEL,
	HTML_MAP,
	HTML_META,
	HTML_P,
	HTML_SCRIPT,
	HTML_SCROLL,
	HTML_STRIKE,
	HTML_SUB,
	HTML_SUP,
	HTML_TEXT,    
	HTML_TEXTAREA,
	HTML_TITLE,
	HTML_U,
	HTML_VAR,
	//HTML_VIDEO,

	//below add by me
	HTML_TABLE,
	HTML_TR,
	HTML_TD,
	HTML_TH,
	HTML_TBODY,

	HTML_CAPTION,
	HTML_CENTER,
	HTML_DIV,
	HTML_FORM,
	HTML_FRAME,
	HTML_FRAMESET,
	HTML_H1,
	HTML_H2,
	HTML_H3,
	HTML_H4,
	HTML_H5,
	HTML_H6,
	HTML_HR,
	HTML_LI,
	HTML_MARQUEE,
	HTML_MATH,
	HTML_MENU,
	HTML_HTML,
*/
	HTML_ELEMENTS		/* This must be the last entry */
} HTMLElement;
/*XUL*/

#define UndefAtr	-1

/* Define an attribute as HTML__ */
#define	HTML_ATTR(a)		HTML_ATTR_##a

enum _HTML_Attributes {
	HTML_ATTR(ACCESSKEY) = 0,
	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),
	
	HTML_ATTR_ATTRIBUTES
};

BitPTagList	Bit_GetHTMLTagList();
int BitHTML_GetTagID(char * pName);
char *BitHTML_GetTagName(int  id);
int BitHTML_GetAtrID(char * pName);
char * BitHTML_GetAtrName(int id);

#endif