httexgen.c

www工具包

    } else if (me->markup || me->preformatted) {
	*me->write_pointer++ = c;
    } else if (c==' ' || c=='\t') {	              /* Skip space and tabs */
	if (*(me->write_pointer-1) != ' ')
	    *me->write_pointer++ = ' ';
	else
	    return HT_OK;
    } else {
	if (c=='$' || c=='&' || c=='%' || c=='#' ||         /* Special chars */
	    c=='{' || c=='}' || c=='_') {
	    *me->write_pointer++ = '\\';
	    *me->write_pointer++ = c;
	} else if (c=='\\') {			            /* Special names */
	    char *temp = "$\\backslash$";
	    strcpy(me->write_pointer, temp);
	    me->write_pointer += strlen(temp);
	} else if (c=='^') {
	    char *temp = "$\\hat{ }$";
	    strcpy(me->write_pointer, temp);
	    me->write_pointer += strlen(temp);	    
	} else if (c=='~') {
	    char *temp = "$\\tilde{ }$";
	    strcpy(me->write_pointer, temp);
	    me->write_pointer += strlen(temp);
	} else if (c=='|' || c=='<' || c=='>') {       	        /* Math mode */
	    *me->write_pointer++ = '$';
	    *me->write_pointer++ = c;
	    *me->write_pointer++ = '$';
	} else
	    *me->write_pointer++ = c;	      	        /* Char seems normal */
    }

    if (c==' ')						   /* Find delimiter */
	me->line_break = me->write_pointer;
    else if (strchr(WORD_DELIMITERS, c))
	me->line_break = me->write_pointer-1;

    /* Flush buffer out when full */
    if (me->write_pointer >= me->buffer+BUFFER_SIZE-3) {
#ifdef OLD_CODE
	if (me->markup || me->preformatted) {
#endif /* OLD_CODE */
	if (me->preformatted) {
	    *me->write_pointer = '\n';
	    (*me->targetClass.put_block)(me->target,
					 me->buffer,
					 me->write_pointer-me->buffer+1);
	    me->write_pointer = me->buffer;
	} else {		       		          /* Use break-point */
	    char line_break_char = *me->line_break;
	    char *saved = me->line_break;
	    *me->line_break = '\n';
	    (*me->targetClass.put_block)(me->target,
					 me->buffer,
					 me->line_break-me->buffer+1);
	    *me->line_break = line_break_char;
	    {                                           /* move next line in */
		char *p = saved;
		char *q;
		for(q=me->buffer; p<me->write_pointer; )
		    *q++ = *p++;
	    }
	    me->write_pointer = me->buffer + (me->write_pointer-saved);
	}	    
	me->line_break = me->buffer;
    }
    return HT_OK;
}



/*	String handling
**	---------------
*/
PRIVATE int HTTeXGen_put_string (HTStructured * me, const char* s)
{
    while (*s)
	HTTeXGen_put_character(me, *s++);
    return HT_OK;
}


PRIVATE int HTTeXGen_write (HTStructured * me, const char* b, int l)
{
    while (l-- > 0)
	HTTeXGen_put_character(me, *b++);
    return HT_OK;
}


/*	Start Element
**	-------------
**
**     	No attributes are put to the output		Henrik 07/03-94
**	Does no assumptions of WHAT element is started...
*/
PRIVATE void HTTeXGen_start_element (HTStructured * 	me,
				     int		element_number,
				     const BOOL *	present,
				     const char **	value)
{
    me->startup = YES;			        /* Now, let's get down to it */
    if (me->preformatted == YES) {	       /* Don't start markup in here */
	HTTRACE(SGML_TRACE, "LaTeX....... No Markup in verbatim mode\n");
	return;
    }
    if (element_number == HTML_PRE)
	me->preformatted = YES;
    if (element_number == HTML_CITE ||	              /* No \n here, please! */
	element_number == HTML_DT ||
	element_number == HTML_H1 ||
	element_number == HTML_H2 ||
	element_number == HTML_H3 ||
	element_number == HTML_H4 ||
	element_number == HTML_H5 ||
	element_number == HTML_H6 ||
	element_number == HTML_H7 ||
	element_number == HTML_TITLE)
	me->sensitive = YES;
    else if (element_number == HTML_DD)         /* Only way to turn <DT> off */
	me->sensitive = NO;
    me->markup = element_number == HTML_A ? NO : YES;
    HTTeXGen_put_string(me, *TeX_names[element_number]);
    me->markup = NO;
}


/*		End Element
**		-----------
**
**	Ends an markup element			Henrik 07/03-94
**	Does no assumptions of WHAT element is ended...
*/
PRIVATE void HTTeXGen_end_element (HTStructured * me, int element_number)
{
    if (me->preformatted && element_number != HTML_PRE) {
	HTTRACE(SGML_TRACE, "LaTeX....... No markup in verbatim mode\n");
	return;
    }
    me->preformatted = NO;
    me->markup = YES;
    HTTeXGen_put_string(me, *(TeX_names[element_number]+1));
    me->markup = NO;
    if (element_number == HTML_CITE ||	
	element_number == HTML_DL ||
	element_number == HTML_H1 ||
	element_number == HTML_H2 ||
	element_number == HTML_H3 ||
	element_number == HTML_H4 ||
	element_number == HTML_H5 ||
	element_number == HTML_H6 ||
	element_number == HTML_H7 ||
	element_number == HTML_TITLE)
	me->sensitive = NO;
}


/*		Expanding entities
**		------------------
**
*/
PRIVATE void HTTeXGen_put_entity (HTStructured * me, int entity_number)
{
    BOOL mark = me->markup;
    if (*TeX_entities[entity_number] != '&' && /* Theese are converted later */
	*TeX_entities[entity_number] != '<' &&
	*TeX_entities[entity_number] != '>')
	me->markup = YES;
    HTTeXGen_put_string(me, TeX_entities[entity_number]);
    me->markup = mark;
}

PRIVATE int HTTeXGen_unparsedBeginElement (HTStructured * me, const char * b, int l)
{
    return (*me->targetClass.put_block)(me->target, b, l);
}

PRIVATE int HTTeXGen_unparsedEndElement (HTStructured * me, const char * b, int l)
{
    return (*me->targetClass.put_block)(me->target, b, l);
}

PRIVATE int HTTeXGen_unparsedEntity (HTStructured * me, const char * b, int l)
{
    return (*me->targetClass.put_block)(me->target, b, l);
}

/*	Free an HTML object
**	-------------------
**
*/
PRIVATE int HTTeXGen_free (HTStructured * me)
{
    HTTeXGen_flush(me);
    (*me->targetClass.put_string)(me->target, "\n\\end{document}\n");
    HTTeXGen_flush(me);
    (*me->targetClass._free)(me->target);	/* ripple through */
    HT_FREE(me);
    return HT_OK;
}


PRIVATE int HTTeXGen_abort (HTStructured * me, HTList * e)
{
    HTTeXGen_free(me);
    return HT_ERROR;
}


/*	Structured Object Class
**	-----------------------
*/
PRIVATE const HTStructuredClass HTTeXGeneration = /* As opposed to print etc */
{		
	"HTMLToTeX",
	HTTeXGen_flush,
	HTTeXGen_free,
	HTTeXGen_abort,
	HTTeXGen_put_character,        	HTTeXGen_put_string,	HTTeXGen_write,
	HTTeXGen_start_element, 	HTTeXGen_end_element,
	HTTeXGen_put_entity,
	HTTeXGen_unparsedBeginElement,
	HTTeXGen_unparsedEndElement,
	HTTeXGen_unparsedEntity
}; 


/*	HTConverter from HTML to TeX Stream
**	------------------------------------------
**
*/
PUBLIC HTStream* HTMLToTeX (HTRequest *	request,
			    void *	param,
			    HTFormat	input_format,
			    HTFormat	output_format,
			    HTStream *	output_stream)
{
    HTStructured* me;
    if ((me = (HTStructured *) HT_CALLOC(1, sizeof(*me))) == NULL)
        HT_OUTOFMEM("HTMLToTeX");

    me->isa = (HTStructuredClass*) &HTTeXGeneration;
    me->dtd = HTML_dtd();
    me->target = output_stream;
    me->targetClass = *me->target->isa;/* Copy pointers to routines for speed*/
    me->write_pointer = me->buffer;
    me->line_break = 	me->buffer;
    (*me->targetClass.put_string)(me->target,
        "\\documentstyle[11pt]{report}\n\\begin{document}\n");
    return SGML_new(HTML_dtd(), me);
}