sgml.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

/*			General SGML Parser code		SGML.c
**			========================
**
**	This module implements an HTStream object.  To parse an
**	SGML file, create this object which is a parser.  The object
**	is (currently) created by being passed a DTD structure,
**	and a target HTStructured object at which to throw the parsed stuff.
**
**	 6 Feb 93  Binary searches used. Interface modified.
*/

#include <HTUtils.h>

/* Remove the following to disable the experimental HTML DTD parsing.
   Currently only used in this source file. - kw */

#ifndef NO_EXTENDED_HTMLDTD
#define EXTENDED_HTMLDTD
#endif

#include <SGML.h>
#include <HTMLDTD.h>
#include <HTCJK.h>
#include <UCMap.h>
#include <UCDefs.h>
#include <UCAux.h>

#include <HTChunk.h>

#include <LYCharSets.h>
#include <LYCharVals.h>	 /* S/390 -- gil -- 0635 */
#include <LYGlobalDefs.h>
#include <LYStrings.h>
#include <LYLeaks.h>

#ifdef USE_COLOR_STYLE
# include <LYStyle.h>
#endif
#ifdef USE_PRETTYSRC
# include <LYPrettySrc.h>
#endif

#define INVALID (-1)

#ifdef USE_PRETTYSRC

char* entity_string; /* this is used for printing entity name.
    Unconditionally added since redundant assigments don't hurt much*/

PRIVATE void fake_put_character ARGS2(
		    void*, p GCC_UNUSED,
		    char,  c GCC_UNUSED)
{
}

#define START TRUE
#define STOP FALSE

#define PUTS_TR(x) psrc_convert_string = TRUE; PUTS(x)

#endif

/* my_casecomp() - optimized by the first character, NOT_ASCII ok */
#define my_casecomp(a,b)  ((TOUPPER(*a) == TOUPPER(*b)) ? \
			AS_casecomp(a,b) : \
			(TOASCII(TOUPPER(*a)) - TOASCII(TOUPPER(*b))))

#if ANSI_PREPRO
 /* will use partially inlined version */
#define orig_HTChunkPutUtf8Char HTChunkPutUtf8Char
#undef HTChunkPutUtf8Char

/* ...used for comments and attributes value like href... */
#define HTChunkPutUtf8Char(ch,x) \
    { \
    if ((TOASCII(x) < 128)  && (ch->size < ch->allocated)) \
	ch->data[ch->size++] = (char)x; \
    else \
	orig_HTChunkPutUtf8Char(ch,x); \
    }

#if 0
#define orig_HTChunkPutc HTChunkPutc
#undef HTChunkPutc

#define HTChunkPutc(ch,x) \
    { \
    if (ch->size < ch->allocated) \
	ch->data[ch->size++] = x; \
    else \
	orig_HTChunkPutc(ch,x); \
    }

#undef HTChunkTerminate

#define HTChunkTerminate(ch) \
    HTChunkPutc(ch, (char)0)
#endif /* */
#endif	/* ANSI_PREPRO */

#define PUTS(str) ((*context->actions->put_string)(context->target, str))
#define PUTC(ch)  ((*context->actions->put_character)(context->target, ch))
#define PUTUTF8(code) (UCPutUtf8_charstring((HTStream *)context->target, \
		      (putc_func_t*)(context->actions->put_character), code))

#define OPT 1


/*the following macros are used for pretty source view. */
#define IS_C(attr) (attr.type == HTMLA_CLASS)

PUBLIC HTCJKlang HTCJK = NOCJK;		/* CJK enum value.		*/
PUBLIC BOOL HTPassEightBitRaw = FALSE;	/* Pass 161-172,174-255 raw.	*/
PUBLIC BOOL HTPassEightBitNum = FALSE;	/* Pass ^ numeric entities raw. */
PUBLIC BOOL HTPassHighCtrlRaw = FALSE;	/* Pass 127-160,173,&#127; raw. */
PUBLIC BOOL HTPassHighCtrlNum = FALSE;	/* Pass &#128;-&#159; raw.	*/

/*	The State (context) of the parser
**
**	This is passed with each call to make the parser reentrant
**
*/

#define MAX_ATTRIBUTES 36	/* Max number of attributes per element */


/*		Element Stack
**		-------------
**	This allows us to return down the stack reselecting styles.
**	As we return, attribute values will be garbage in general.
*/
typedef struct _HTElement HTElement;
struct _HTElement {
	HTElement *	next;	/* Previously nested element or 0 */
	HTTag*		tag;	/* The tag at this level  */
};

typedef enum {
    S_text = 0
    ,S_attr
    ,S_attr_gap
    ,S_comment
    ,S_cro
    ,S_doctype
    ,S_dollar
    ,S_dollar_dq
    ,S_dollar_paren
    ,S_dollar_paren_dq
    ,S_dollar_paren_sq
    ,S_dollar_sq
    ,S_dquoted
    ,S_end
    ,S_entity
    ,S_equals
    ,S_ero
    ,S_esc
    ,S_esc_dq
    ,S_esc_sq
    ,S_exclamation
    ,S_in_kanji
    ,S_incro
    ,S_junk_pi
    ,S_junk_tag
    ,S_litteral
    ,S_marked
    ,S_nonascii_text
    ,S_nonascii_text_dq
    ,S_nonascii_text_sq
    ,S_paren
    ,S_paren_dq
    ,S_paren_sq
    ,S_pcdata
    ,S_script
    ,S_sgmlatt
    ,S_sgmlele
    ,S_sgmlent
    ,S_squoted
    ,S_tag
    ,S_tag_gap
    ,S_tagname_slash
    ,S_value
} sgml_state;

/*	Internal Context Data Structure
**	-------------------------------
*/
struct _HTStream {

    CONST HTStreamClass *	isa;		/* inherited from HTStream */

    CONST SGML_dtd		*dtd;
    CONST HTStructuredClass	*actions;	/* target class	 */
    HTStructured		*target;	/* target object */

    HTTag			*current_tag;
    HTTag			*slashedtag;
    CONST HTTag			*unknown_tag;
    BOOL			inSELECT;
    BOOL			no_lynx_specialcodes;
    int				current_attribute_number;
    HTChunk			*string;
    int				leading_spaces;
    int				trailing_spaces;
    HTElement			*element_stack;
    sgml_state			state;
    unsigned char kanji_buf;
#ifdef CALLERDATA
    void *			callerData;
#endif /* CALLERDATA */
    BOOL present[MAX_ATTRIBUTES];	/* Flags: attribute is present? */
    char * value[MAX_ATTRIBUTES];	/* NULL, or strings alloc'd with StrAllocCopy_extra() */

    BOOL			lead_exclamation;
    BOOL			first_dash;
    BOOL			end_comment;
    BOOL			doctype_bracket;
    BOOL			first_bracket;
    BOOL			second_bracket;
    BOOL			isHex;

    HTParentAnchor *		node_anchor;
    LYUCcharset *		inUCI;		/* pointer to anchor UCInfo */
    int				inUCLYhndl;	/* charset we are fed	    */
    LYUCcharset *		outUCI;		/* anchor UCInfo for target */
    int				outUCLYhndl;	/* charset for target	    */
    char			utf_count;
    UCode_t			utf_char;
    char			utf_buf[8];
    char *			utf_buf_p;
    UCTransParams		T;
    int				current_tag_charset; /* charset to pass attributes */

    char *			recover;
    int				recover_index;
    char *			include;
    char *			active_include;
    int				include_index;
    char *			url;
    char *			csi;
    int				csi_index;
#ifdef USE_PRETTYSRC
    BOOL			cur_attr_is_href;
    BOOL			cur_attr_is_name;
    BOOL			seen_nonwhite_in_junk_tag;
#endif
};

#ifndef NO_LYNX_TRACE
PRIVATE char *state_name ARGS1(sgml_state, n)
{
    char *result = "?";
    switch (n) {
    case S_attr:                result = "S_attr";              break;
    case S_attr_gap:            result = "S_attr_gap";          break;
    case S_comment:             result = "S_comment";           break;
    case S_cro:                 result = "S_cro";               break;
    case S_doctype:             result = "S_doctype";           break;
    case S_dollar:              result = "S_dollar";            break;
    case S_dollar_dq:           result = "S_dollar_dq";         break;
    case S_dollar_paren:        result = "S_dollar_paren";      break;
    case S_dollar_paren_dq:     result = "S_dollar_paren_dq";   break;
    case S_dollar_paren_sq:     result = "S_dollar_paren_sq";   break;
    case S_dollar_sq:           result = "S_dollar_sq";         break;
    case S_dquoted:             result = "S_dquoted";           break;
    case S_end:                 result = "S_end";               break;
    case S_entity:              result = "S_entity";            break;
    case S_equals:              result = "S_equals";            break;
    case S_ero:                 result = "S_ero";               break;
    case S_esc:                 result = "S_esc";               break;
    case S_esc_dq:              result = "S_esc_dq";            break;
    case S_esc_sq:              result = "S_esc_sq";            break;
    case S_exclamation:         result = "S_exclamation";       break;
    case S_in_kanji:            result = "S_in_kanji";          break;
    case S_incro:               result = "S_incro";             break;
    case S_junk_pi:             result = "S_junk_pi";           break;
    case S_junk_tag:            result = "S_junk_tag";          break;
    case S_litteral:            result = "S_litteral";          break;
    case S_marked:              result = "S_marked";            break;
    case S_nonascii_text:       result = "S_nonascii_text";     break;
    case S_nonascii_text_dq:    result = "S_nonascii_text_dq";  break;
    case S_nonascii_text_sq:    result = "S_nonascii_text_sq";  break;
    case S_paren:               result = "S_paren";             break;
    case S_paren_dq:            result = "S_paren_dq";          break;
    case S_paren_sq:            result = "S_paren_sq";          break;
    case S_pcdata:              result = "S_pcdata";            break;
    case S_script:              result = "S_script";            break;
    case S_sgmlatt:             result = "S_sgmlatt";           break;
    case S_sgmlele:             result = "S_sgmlele";           break;
    case S_sgmlent:             result = "S_sgmlent";           break;
    case S_squoted:             result = "S_squoted";           break;
    case S_tag:                 result = "S_tag";               break;
    case S_tag_gap:             result = "S_tag_gap";           break;
    case S_tagname_slash:       result = "S_tagname_slash";     break;
    case S_text:                result = "S_text";              break;
    case S_value:               result = "S_value";             break;
    }
    return result;
}
#endif

/* storage for Element Stack */
#define DEPTH 10
static HTElement pool[DEPTH];
static int depth = 0;

PRIVATE HTElement* pool_alloc NOARGS
{
    depth++;
    if (depth > DEPTH)
	return (HTElement*) malloc(sizeof(HTElement));
    return (pool + depth - 1);
}

PRIVATE void pool_free ARGS1(HTElement*, e)
{
    if (depth > DEPTH)
	FREE(e);
    depth--;
    return;
}

#ifdef USE_PRETTYSRC

PRIVATE void HTMLSRC_apply_markup ARGS3(
	    HTStream *,	      context,
	    HTlexeme,	      lexeme,
	    BOOL,	      start)
{
    HT_tagspec* ts = *( ( start ? lexeme_start : lexeme_end ) + lexeme);

    while (ts) {
#ifdef USE_COLOR_STYLE
	if (ts->start) {
	    current_tag_style = ts->style;
	    force_current_tag_style = TRUE;
	    forced_classname = ts->class_name;
	    force_classname = TRUE;
	}
#endif
	CTRACE((tfp,ts->start ? "SRCSTART %d\n" : "SRCSTOP %d\n",(int)lexeme));
	if (ts->start)
	    (*context->actions->start_element)(
		context->target,
		ts->element,
		ts->present,
		(CONST char **)ts->value,
		context->current_tag_charset,
		(char **)&context->include);
	else
	    (*context->actions->end_element)(
		context->target,
		ts->element,
		(char **)&context->include);
	ts = ts->next;
    }
}

#if ANSI_PREPRO
#  define PSRCSTART(x)	HTMLSRC_apply_markup(context,HTL_##x,START)
#  define PSRCSTOP(x)   HTMLSRC_apply_markup(context,HTL_##x,STOP)
#else
#  define PSRCSTART(x)	HTMLSRC_apply_markup(context,HTL_/**/x,START)
#  define PSRCSTOP(x)   HTMLSRC_apply_markup(context,HTL_/**/x,STOP)
#endif

#define attr_is_href context->cur_attr_is_href
#define attr_is_name context->cur_attr_is_name
#endif

PRIVATE void set_chartrans_handling ARGS3(
	HTStream *,		context,
	HTParentAnchor *,	anchor,
	int,			chndl)
{
    if (chndl < 0) {
	/*
	**  Nothing was set for the parser in earlier stages,
	**  so the HTML parser's UCLYhndl should still be its
	**  default. - FM
	*/
	chndl = HTAnchor_getUCLYhndl(anchor, UCT_STAGE_STRUCTURED);
	if (chndl < 0)
	    /*
	    **	That wasn't set either, so seek the HText default. - FM
	    */
	    chndl = HTAnchor_getUCLYhndl(anchor, UCT_STAGE_HTEXT);
	if (chndl < 0)
	    /*
	    **	That wasn't set either, so assume the current display
	    **	character set. - FM
	    */
	    chndl = current_char_set;
	/*
	**  Try to set the HText and HTML stages' chartrans info
	**  with the default lock level (will not be changed if
	**  it was set previously with a higher lock level). - FM
	*/
	HTAnchor_setUCInfoStage(anchor, chndl,
				UCT_STAGE_HTEXT,
				UCT_SETBY_DEFAULT);
	HTAnchor_setUCInfoStage(anchor, chndl,
				UCT_STAGE_STRUCTURED,
				UCT_SETBY_DEFAULT);
	/*
	**  Get the chartrans info for output to the HTML parser. - FM
	*/
	context->outUCI = HTAnchor_getUCInfoStage(anchor,
						  UCT_STAGE_STRUCTURED);
	context->outUCLYhndl = HTAnchor_getUCLYhndl(context->node_anchor,
						    UCT_STAGE_STRUCTURED);
    }
    /*
    **	Set the in->out transformation parameters. - FM
    */
    UCSetTransParams(&context->T,
		     context->inUCLYhndl, context->inUCI,
		     context->outUCLYhndl, context->outUCI);
    /*
    **	This is intended for passing the SGML parser's input
    **	charset as an argument in each call to the HTML
    **	parser's start tag function, but it would be better
    **	to call a Lynx_HTML_parser function to set an element
    **	in its HTStructured object, itself, if this were
    **	needed. - FM
    */
    if (HTCJK != NOCJK) {
	context->current_tag_charset = -1;
    } else if (context->T.transp) {
	context->current_tag_charset = context->inUCLYhndl;
    } else if (context->T.decode_utf8) {
	context->current_tag_charset = context->inUCLYhndl;
    } else if (context->T.do_8bitraw ||
	       context->T.use_raw_char_in) {
	context->current_tag_charset = context->inUCLYhndl;
    } else if (context->T.output_utf8 ||
	       context->T.trans_from_uni) {
	context->current_tag_charset = UCGetLYhndl_byMIME("utf-8");
    } else {
	context->current_tag_charset = LATIN1;
    }
}

PRIVATE void change_chartrans_handling ARGS1(
	HTStream *,		context)
{
    int new_LYhndl = HTAnchor_getUCLYhndl(context->node_anchor,
					  UCT_STAGE_PARSER);
    if (new_LYhndl != context->inUCLYhndl &&
	new_LYhndl >= 0) {
	/*
	 *  Something changed. but ignore if a META wants an unknown charset.
	 */
	LYUCcharset * new_UCI = HTAnchor_getUCInfoStage(context->node_anchor,
							UCT_STAGE_PARSER);
	if (new_UCI) {
	    LYUCcharset * next_UCI = HTAnchor_getUCInfoStage(
				    context->node_anchor, UCT_STAGE_STRUCTURED
							    );
	    int next_LYhndl = HTAnchor_getUCLYhndl(
				    context->node_anchor, UCT_STAGE_STRUCTURED
						  );
	    context->inUCI = new_UCI;
	    context->inUCLYhndl = new_LYhndl;
	    context->outUCI = next_UCI;
	    context->outUCLYhndl = next_LYhndl;
	    set_chartrans_handling(context,
				   context->node_anchor, next_LYhndl);
	}
    }
}

#ifdef USE_COLOR_STYLE