sgml.c

www工具包. 这是W3C官方支持的www支撑库. 其中提供通用目的的客户端的WebAPI: complete HTTP/1.1 (with caching, pipelining, PUT, POS

			** Character reference or Entity
			*/
		    case S_ero:
			if (c == '#')
			    {
				/*   &# is Char Ref Open */ 
				context->state = S_cro;
				break;
			    }
			context->state = S_entity;

			/** FALL THROUGH TO S_entity !! ***/

			/*
			** Handle Entities
			*/
		    case S_entity:
			if (isalnum((int) c))
				HTChunk_putc(string, c);
			else
			    {
				HTChunk_terminate(string);
				handle_entity(context);
				text = b;
				count = 0;
				if (c != ';')
				    {
					--text;
					goto S_text;
				    }
				context->state = S_text;
			    }
			break;

			/*	Character reference
			 */
		    case S_cro:
			if (isalnum((int)c))
				/* accumulate a character NUMBER */
				HTChunk_putc(string, c);
			else
			    {
				int value;
				HTChunk_terminate(string);
				if (sscanf(HTChunk_data(string), "%d", &value)==1)
					PUTC((char)value);
				else
				    {
					PUTB("&#", 2);
					PUTB(HTChunk_data(string), HTChunk_size(string)-1);
				    }
				text = b;
				count = 0;
				if (c != ';')
				    {
					--text;
					goto S_text;
				    }
				context->state = S_text;
			    }
			break;

		    case S_tag:		/* new tag */
		    handle_S_tag:
			if (isalnum((int)c))
				HTChunk_putc(string, c);
			else { /* End of tag name */
			    int i;
			    if (c == '/') {
				if (HTChunk_size(string) > 0)
				    HTTRACE(SGML_TRACE, "`<%s/' found!\n" _ HTChunk_data(string));
				context->state = S_end;
				break;
			    } else if (c == '!') {
				if (HTChunk_size(string) > 0)
				    HTTRACE(SGML_TRACE, " `<%s!' found!\n" _ HTChunk_data(string));
				context->state = S_md;
				break;
			    }
			    HTChunk_terminate(string);
			    context->current_tag  = SGMLFindTag(dtd, HTChunk_data(string));
			    if (context->current_tag == NULL) {
				HTTRACE(SGML_TRACE, "*** Unknown element %s\n" _ HTChunk_data(string));
				(*context->actions->unparsed_begin_element)
				    (context->target, HTChunk_data(string), HTChunk_size(string));
			    } else {
				for (i=0; i<context->current_tag->number_of_attributes; i++) {
				    context->present[i] = NO;
				    context->value[i] = -1;
				}
			    }
			    context->token = 0;
			    HTChunk_clear(string);
			    context->current_attribute_number = INVALID;
			    goto S_tag_gap;
			}
			break;

		    S_tag_gap:
			context->state = S_tag_gap;
		    case S_tag_gap:		/* Expecting attribute or > */
			if (isspace((int) c))
				break;	/* Gap between attributes */

			if (c == '>')
				goto got_element_open;
			else
				goto S_attr;

		    S_attr:
			/*
			** Start collecting the attribute name and collect
			** it in S_attr.
			*/
			context->state = S_attr;
			HTChunk_truncate(string, context->token);
		    case S_attr:
			if (isspace((int) c) || c == '>' || c == '=')
				goto got_attribute_name;
			else
				HTChunk_putc(string, c);
			break;

		    got_attribute_name:
			/*
			** This label is entered when attribute name has been
			** collected. Process it and enter S_attr_gap for
			** potential value or start of the next attribute.
			*/
			HTChunk_terminate(string) ;
			handle_attribute_name
				(context, HTChunk_data(string) + context->token);
			HTChunk_truncate(string, context->token);
			context->state = S_attr_gap;
		    case S_attr_gap:	/* Expecting attribute or = or > */
			if (isspace((int) c))
				break;	/* Gap after attribute */

			if (c == '>')
				goto got_element_open;
			else if (c == '=')
				context->state = S_equals;
			else
				goto S_attr; /* Get next attribute */
			break;

		    case S_equals:	/* After attr = */ 
			if (isspace((int) c))
				break;	/* Before attribute value */

			if (c == '>')
			    {		/* End of tag */
				HTTRACE(SGML_TRACE, "found = but no value\n");
				goto got_element_open;
			    }
			else if (c == '\'')
				context->state = S_squoted;
			else if (c == '"')
				context->state = S_dquoted;
			else
				goto S_value;
			break;

		    S_value:
			context->state = S_value;
			HTChunk_truncate(string, context->token);
		    case S_value:
			if (isspace((int) c) || c == '>')
			    {
				HTChunk_terminate(string);
				handle_attribute_value(context);
				context->token = HTChunk_size(string);
				goto S_tag_gap;
			    }
			else
				HTChunk_putc(string, c);
			break;
		
		    case S_squoted:	/* Quoted attribute value */
			if (c == '\'')
			    {
				HTChunk_terminate(string);
				handle_attribute_value(context);
				context->token = HTChunk_size(string);
				context->state = S_tag_gap;
			    }
			else if (c && c != '\n' && c != '\r')
				HTChunk_putc(string, c);
			break;
	
		    case S_dquoted:	/* Quoted attribute value */
			if (c == '"')
			    {
				HTChunk_terminate(string);
				handle_attribute_value(context);
				context->token = HTChunk_size(string);
				context->state = S_tag_gap;
			    }
			else if (c && c != '\n' && c != '\r')
				HTChunk_putc(string, c);
			break;

		    case S_end:	/* </ */
			if (isalnum((int) c))
				HTChunk_putc(string, c);
			else
			    {		/* End of end tag name */
				HTTag *t;
				char * first;
				HTChunk_terminate(string);
				if ((first=HTChunk_data(string))!=NULL && *first != '\0')
				        t = SGMLFindTag(dtd, HTChunk_data(string));
				else
				    	/* Empty end tag */
					/* Original code popped here one
					   from the stack. If this feature
					   is required, I have to put the
					   stack back... -- msa */
					t = NULL;
				if (!t) {
				    HTTRACE(SGML_TRACE, "Unknown end tag </%s>\n" _ HTChunk_data(string));
				    (*context->actions->unparsed_end_element)
					(context->target, HTChunk_data(string), HTChunk_size(string));
				} else {
				    context->current_tag = NULL;
				    end_element(context, t);
				}
				HTChunk_clear(string);
				context->current_attribute_number = INVALID;
				if (c != '>')
				    {
					if (!isspace((int) c))
					    HTTRACE(SGML_TRACE, "`</%s%c' found!\n" _ HTChunk_data(string) _ c);
					context->state = S_junk_tag;
				    }
				else
				    {
					text = b;
					count = 0;
					context->state = S_text;
				    }
			    }
			break;

		    case S_junk_tag:
			if (c == '>')
			    {
				text = b;
				count = 0;
				context->state = S_text;
			    }
			break;

			/*
			** Scanning (actually skipping) declarations
			*/
		    case S_md:
			if (c == '-')
				context->state = S_com_1;
			else if (c == '"')
				context->state = S_md_dqs;
			else if (c == '\'')
				context->state = S_md_sqs;
			else if (c == '>')
			    {
				text = b;
				count = 0;
				context->state = S_text;
			    }
			break;

		    case S_md_dqs: /* Skip double quoted string */
			if (c == '"')
				context->state = S_md;
			else if (c == '>')
			    {
				text = b;
				count = 0;
				context->state = S_text;
			    }
			break;

		    case S_md_sqs: /* Skip single quoted string */
			if (c == '\'')
				context->state = S_md;
			else if (c == '>')
			    {
				text = b;
				count = 0;
				context->state = S_text;
			    }
			break;

		    case S_com_1: /* Starting a comment? */
			context->state = (c == '-') ? S_com : S_md;
			if (c == '>')
			    {
				text = b;
				count = 0;
				context->state = S_text;
			    }
			break;

		    case S_com: /* ..within comment */
			if (c == '-')
				context->state = S_com_2;
			break;

		    case S_com_2: /* Ending a comment ? */
			context->state = (c == '-') ? S_com_2a : S_com;
			break;
		    
		    case S_com_2a:
			if (c == '>') {
			    text = b;
			    count = 0;
			    context->state = S_text;
			} else if (c == '-') {
			    context->state = S_com_2a;
			} else
			    context->state = S_com;
			break;
		    }
	    }
	if (count > 0)
		PUTB(text, count);
	return HT_OK;
    }


PRIVATE int SGML_string (HTStream * context, const char* s)
    {
	return SGML_write(context, s, (int) strlen(s));
    }


PRIVATE int SGML_character (HTStream * context, char c)
    {
	return SGML_write(context, &c, 1);
    }

/*_______________________________________________________________________
*/

/*	Structured Object Class
**	-----------------------
*/
PRIVATE const HTStreamClass SGMLParser = 
{
    "SGML",
    SGML_flush,
    SGML_free,
    SGML_abort,
    SGML_character, 
    SGML_string,
    SGML_write
}; 

/*	Create SGML Engine
**	------------------
**
** On entry,
**	dtd		represents the DTD, along with
**	actions		is the sink for the data as a set of routines.
**
*/
PUBLIC HTStream *SGML_new(const SGML_dtd * dtd, HTStructured * target)
{
    int i;
    HTStream* context;
    if ((context = (HTStream  *) HT_CALLOC(1, sizeof(HTStream))) == NULL)
	HT_OUTOFMEM("SGML_begin");

    context->isa = &SGMLParser;
    context->string = HTChunk_new(128);	/* Grow by this much */
    context->dtd = dtd;
    context->target = target;
    context->actions = (HTStructuredClass*)(((HTStream*)target)->isa);
    /* Ugh: no OO */
    context->state = S_text;
    for(i=0; i<MAX_ATTRIBUTES; i++)
	context->value[i] = 0;
    return context;
}

PUBLIC HTTag * SGML_findTag (SGML_dtd * dtd, int element_number)
{
    return (dtd && element_number>=0 && element_number<dtd->number_of_tags) ?
	(dtd->tags+element_number) : NULL;
}

PUBLIC char * SGML_findTagName (SGML_dtd * dtd, int element_number)
{
    return (dtd && element_number>=0 && element_number<dtd->number_of_tags) ?
	(dtd->tags+element_number)->name : NULL;
}

PUBLIC SGMLContent SGML_findTagContents (SGML_dtd * dtd, int element_number)
{
    return (dtd && element_number>=0 && element_number<dtd->number_of_tags) ?
	(dtd->tags+element_number)->contents : SGML_ELEMENT;
}

PUBLIC int SGML_findElementNumber (SGML_dtd * dtd, char * name_element)
{
    if (dtd && name_element) {
	int i;
	HTTag *ct;
	for (i = 0; i< dtd->number_of_tags; i++) {
	    ct = &(dtd->tags[i]);
	    if (!strcasecomp(ct->name,name_element))
		return i;
	}
    }
    return -1;
}

PUBLIC char * HTTag_name (HTTag * tag)
{
    return tag ? tag->name : NULL;
}

PUBLIC SGMLContent HTTag_content (HTTag * tag)
{
    return tag ? tag->contents : SGML_EMPTY;
}

PUBLIC int HTTag_attributes (HTTag * tag)
{
    return tag ? tag->number_of_attributes : -1;
}

PUBLIC char * HTTag_attributeName (HTTag * tag, int attribute_number)
{
    return (tag && attribute_number>=0 && attribute_number<tag->number_of_attributes) ?
	(tag->attributes+attribute_number)->name : NULL;
}

PUBLIC char * HTAttr_name (HTAttr * attr)
{
    return attr ? attr->name : NULL;
}