htmldtd.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

/*		Our Static DTD for HTML
**		-----------------------
*/

/* Implements:
*/

#include <HTUtils.h>
#include <HTMLDTD.h>
#include <LYLeaks.h>
#include <LYJustify.h>

/*
 *     Character entities like &nbsp now excluded from our DTD tables,
 *     they are mapped to Unicode and handled by chartrans code directly
 *     the similar way the numeric entities like &#123 does.
 *     See  src/chrtrans/entities.h  for real mapping.
 */

/*	Entity Names
**	------------
**
**	This table must be matched exactly with ALL the translation tables
**		(this is an obsolete translation mechanism, probably unused,
**		currently replaced with Unicode chartrans in most cases...)
*/
static CONST char* entities[] = {
  "AElig",	/* capital AE diphthong (ligature) */
  "Aacute",	/* capital A, acute accent */
  "Acirc",	/* capital A, circumflex accent */
  "Agrave",	/* capital A, grave accent */
  "Aring",	/* capital A, ring */
  "Atilde",	/* capital A, tilde */
  "Auml",	/* capital A, dieresis or umlaut mark */
  "Ccedil",	/* capital C, cedilla */
  "Dstrok",	/* capital Eth, Icelandic */
  "ETH",	/* capital Eth, Icelandic */
  "Eacute",	/* capital E, acute accent */
  "Ecirc",	/* capital E, circumflex accent */
  "Egrave",	/* capital E, grave accent */
  "Euml",	/* capital E, dieresis or umlaut mark */
  "Iacute",	/* capital I, acute accent */
  "Icirc",	/* capital I, circumflex accent */
  "Igrave",	/* capital I, grave accent */
  "Iuml",	/* capital I, dieresis or umlaut mark */
  "Ntilde",	/* capital N, tilde */
  "Oacute",	/* capital O, acute accent */
  "Ocirc",	/* capital O, circumflex accent */
  "Ograve",	/* capital O, grave accent */
  "Oslash",	/* capital O, slash */
  "Otilde",	/* capital O, tilde */
  "Ouml",	/* capital O, dieresis or umlaut mark */
  "THORN",	/* capital THORN, Icelandic */
  "Uacute",	/* capital U, acute accent */
  "Ucirc",	/* capital U, circumflex accent */
  "Ugrave",	/* capital U, grave accent */
  "Uuml",	/* capital U, dieresis or umlaut mark */
  "Yacute",	/* capital Y, acute accent */
  "aacute",	/* small a, acute accent */
  "acirc",	/* small a, circumflex accent */
  "acute",	/* spacing acute */
  "aelig",	/* small ae diphthong (ligature) */
  "agrave",	/* small a, grave accent */
  "amp",	/* ampersand */
  "aring",	/* small a, ring */
  "atilde",	/* small a, tilde */
  "auml",	/* small a, dieresis or umlaut mark */
  "brkbar",	/* broken vertical bar */
  "brvbar",	/* broken vertical bar */
  "ccedil",	/* small c, cedilla */
  "cedil",	/* spacing cedilla */
  "cent",	/* cent sign */
  "copy",	/* copyright sign */
  "curren",	/* currency sign */
  "deg",	/* degree sign */
  "die",	/* spacing dieresis */
  "divide",	/* division sign */
  "eacute",	/* small e, acute accent */
  "ecirc",	/* small e, circumflex accent */
  "egrave",	/* small e, grave accent */
  "emdash",	/* dash the width of emsp */
  "emsp",	/* em space - not collapsed */
  "endash",	/* dash the width of ensp */
  "ensp",	/* en space - not collapsed */
  "eth",	/* small eth, Icelandic */
  "euml",	/* small e, dieresis or umlaut mark */
  "frac12",	/* fraction 1/2 */
  "frac14",	/* fraction 1/4 */
  "frac34",	/* fraction 3/4 */
  "gt",		/* greater than */
  "hibar",	/* spacing macron */
  "iacute",	/* small i, acute accent */
  "icirc",	/* small i, circumflex accent */
  "iexcl",	/* inverted exclamation mark */
  "igrave",	/* small i, grave accent */
  "iquest",	/* inverted question mark */
  "iuml",	/* small i, dieresis or umlaut mark */
  "laquo",	/* angle quotation mark, left */
  "lt",		/* less than */
  "macr",	/* spacing macron */
  "mdash",	/* dash the width of emsp */
  "micro",	/* micro sign */
  "middot",	/* middle dot */
  "nbsp",	/* non breaking space */
  "ndash",	/* dash the width of ensp */
  "not",	/* negation sign */
  "ntilde",	/* small n, tilde */
  "oacute",	/* small o, acute accent */
  "ocirc",	/* small o, circumflex accent */
  "ograve",	/* small o, grave accent */
  "ordf",	/* feminine ordinal indicator */
  "ordm",	/* masculine ordinal indicator */
  "oslash",	/* small o, slash */
  "otilde",	/* small o, tilde */
  "ouml",	/* small o, dieresis or umlaut mark */
  "para",	/* paragraph sign */
  "plusmn",	/* plus-or-minus sign */
  "pound",	/* pound sign */
  "quot",	/* quote '"' */
  "raquo",	/* angle quotation mark, right */
  "reg",	/* circled R registered sign */
  "sect",	/* section sign */
  "shy",	/* soft hyphen */
  "sup1",	/* superscript 1 */
  "sup2",	/* superscript 2 */
  "sup3",	/* superscript 3 */
  "szlig",	/* small sharp s, German (sz ligature) */
  "thinsp",	/* thin space (not collapsed) */
  "thorn",	/* small thorn, Icelandic */
  "times",	/* multiplication sign */
  "trade",	/* trade mark sign (U+2122) */
  "uacute",	/* small u, acute accent */
  "ucirc",	/* small u, circumflex accent */
  "ugrave",	/* small u, grave accent */
  "uml",	/* spacing dieresis */
  "uuml",	/* small u, dieresis or umlaut mark */
  "yacute",	/* small y, acute accent */
  "yen",	/* yen sign */
  "yuml",	/* small y, dieresis or umlaut mark */
};


/*		Attribute Lists
**		---------------
**
**	Lists must be in alphabetical order by attribute name
**	The tag elements contain the number of attributes
*/
#ifdef USE_PRETTYSRC
# define N HTMLA_NORMAL
# define i HTMLA_ANAME
# define h HTMLA_HREF
# define c HTMLA_CLASS
# define x HTMLA_AUXCLASS
# define T(t) , t
#else
# define T(t) /*nothing*/
#endif

static attr a_attr[] = {		/* Anchor attributes */
	{ "ACCESSKEY"     T(N) },
	{ "CHARSET"       T(N) },
	{ "CLASS"         T(c) },
	{ "CLEAR"         T(N) },
	{ "COORDS"        T(N) },
	{ "DIR"           T(N) },
	{ "HREF"          T(h) },
	{ "ID"            T(i) },
	{ "ISMAP"         T(N) },
	{ "LANG"          T(N) },
	{ "MD"            T(N) },
	{ "NAME"          T(i) },
	{ "NOTAB"         T(N) },
	{ "ONCLICK"       T(N) },
	{ "ONMOUSEOUT"    T(N) },
	{ "ONMOUSEOVER"   T(N) },
	{ "REL"           T(N) },
	{ "REV"           T(N) },
	{ "SHAPE"         T(N) },
	{ "STYLE"         T(N) },
	{ "TABINDEX"      T(N) },
	{ "TARGET"        T(N) },
	{ "TITLE"         T(N) },
	{ "TYPE"          T(N) },
	{ "URN"           T(N) },
	{ 0               T(N) }	/* Terminate list */
};

static attr address_attr[] = {		/* ADDRESS attributes */
	{ "CLASS"         T(c) },
	{ "CLEAR"         T(N) },
	{ "DIR"           T(N) },
	{ "ID"            T(i) },
	{ "LANG"          T(N) },
	{ "NOWRAP"        T(N) },
	{ "STYLE"         T(N) },
	{ "TITLE"         T(N) },
	{ 0               T(N) }	/* Terminate list */
};

static attr applet_attr[] = {		/* APPLET attributes */
	{ "ALIGN"         T(N) },
	{ "ALT"           T(N) },
	{ "CLASS"         T(c) },
	{ "CLEAR"         T(N) },
	{ "CODE"          T(N) },
	{ "CODEBASE"      T(h) },
	{ "DIR"           T(N) },
	{ "DOWNLOAD"      T(N) },
	{ "HEIGHT"        T(N) },
	{ "HSPACE"        T(N) },
	{ "ID"            T(i) },
	{ "LANG"          T(N) },
	{ "NAME"          T(i) },
	{ "STYLE"         T(N) },
	{ "TITLE"         T(N) },
	{ "VSPACE"        T(N) },
	{ "WIDTH"         T(N) },
	{ 0               T(N) }	/* Terminate list */
};

static attr area_attr[] = {		/* AREA attributes */
	{ "ALT"           T(N) },
	{ "CLASS"         T(c) },
	{ "CLEAR"         T(N) },
	{ "COORDS"        T(N) },
	{ "DIR"           T(N) },
	{ "HREF"          T(h) },
	{ "ID"            T(i) },
	{ "LANG"          T(N) },
	{ "NOHREF"        T(N) },
	{ "NOTAB"         T(N) },
	{ "ONCLICK"       T(N) },
	{ "ONMOUSEOUT"    T(N) },
	{ "ONMOUSEOVER"   T(N) },
	{ "SHAPE"         T(N) },
	{ "STYLE"         T(N) },
	{ "TABINDEX"      T(N) },
	{ "TARGET"        T(N) },
	{ "TITLE"         T(N) },
	{ 0               T(N) }	/* Terminate list */
};

static attr base_attr[] = {		/* BASE attributes */
	{ "HREF"          T(h) },
	{ "TARGET"        T(N) },
	{ "TITLE"         T(N) },
	{ 0               T(N) }	/* Terminate list */
};

static attr bgsound_attr[] = {		/* BGSOUND attributes */
	{ "CLASS"         T(c) },
	{ "CLEAR"         T(N) },
	{ "DIR"           T(N) },
	{ "ID"            T(i) },
	{ "LANG"          T(N) },
	{ "LOOP"          T(N) },
	{ "SRC"           T(h) },
	{ "STYLE"         T(N) },
	{ "TITLE"         T(N) },
	{ 0               T(N) }	/* Terminate list */
};

static attr body_attr[] = {		/* BODY attributes */
	{ "ALINK"         T(N) },
	{ "BACKGROUND"    T(h) },
	{ "BGCOLOR"       T(N) },
	{ "CLASS"         T(c) },
	{ "CLEAR"         T(N) },
	{ "DIR"           T(N) },
	{ "ID"            T(i) },
	{ "LANG"          T(N) },
	{ "LINK"          T(N) },
	{ "ONLOAD"        T(N) },
	{ "ONUNLOAD"      T(N) },
	{ "STYLE"         T(N) },
	{ "TEXT"          T(N) },
	{ "TITLE"         T(N) },
	{ "VLINK"         T(N) },
	{ 0               T(N) } /* Terminate list */
};

static attr bodytext_attr[] = {		/* BODYTEXT attributes */
	{ "CLASS"         T(c) },
	{ "CLEAR"         T(N) },
	{ "DATA"          T(N) },
	{ "DIR"           T(N) },
	{ "ID"            T(i) },
	{ "LANG"          T(N) },
	{ "NAME"          T(N) },
	{ "OBJECT"        T(N) },
	{ "REF"           T(N) },
	{ "STYLE"         T(N) },
	{ "TITLE"         T(N) },
	{ "TYPE"          T(N) },
	{ "VALUE"         T(N) },
	{ "VALUETYPE"     T(N) },
	{ 0               T(N) } /* Terminate list */
};

static attr bq_attr[] = {		/* BQ (BLOCKQUOTE) attributes */
	{ "CITE"          T(h) },
	{ "CLASS"         T(c) },
	{ "CLEAR"         T(N) },
	{ "DIR"           T(N) },
	{ "ID"            T(i) },
	{ "LANG"          T(N) },
	{ "NOWRAP"        T(N) },
	{ "STYLE"         T(N) },
	{ "TITLE"         T(N) },
	{ 0               T(N) }	/* Terminate list */
};

static attr button_attr[] = {		/* BUTTON attributes */
	{ "CLASS"         T(c) },
	{ "CLEAR"         T(N) },
	{ "DIR"           T(N) },
	{ "DISABLED"      T(N) },
	{ "ID"            T(i) },
	{ "LANG"          T(N) },
	{ "NAME"          T(N) },
	{ "ONBLUR"        T(N) },
	{ "ONFOCUS"       T(N) },
	{ "STYLE"         T(N) },
	{ "TABINDEX"      T(N) },
	{ "TITLE"         T(N) },
	{ "TYPE"          T(N) },
	{ "VALUE"         T(N) },
	{ 0               T(N) }	/* Terminate list */