[BACK]Return to SGML.html CVS log [TXT] [DIR] Up to [Public] / libwww / Library / src
File:  [Public] / libwww / Library / src / SGML.html
Revision 2.16: download - view: text, annotated - select for diffs
Tue Mar 21 17:48:07 1995 UTC (29 years, 2 months ago) by frystyk
Branches: MAIN
CVS tags: v3/0, WindowsNT, HEAD
Getting Library back on main branch

<HTML>
<HEAD>
<TITLE>SGML parse and stream definition for libwww</TITLE>
</HEAD>
<BODY>

<H1>SGML and Structured streams</H1>

<PRE>
/*
**	(c) COPYRIGHT CERN 1994.
**	Please first read the full copyright statement in the file COPYRIGH.
*/
</PRE>

The SGML parser is a state machine.  It is called for every character
of the input stream. The DTD data structure contains pointers to
functions which are called to implement the actual effect of the text
read. When these functions are called, the attribute structures
pointed to by the DTD are valid, and the function is parsed a pointer
to the curent tag structure, and an "element stack" which represents
the state of nesting within SGML elements. See also the <A
HREF="HTStream.html">the generic Stream definition</A><P>

The following aspects are from Dan Connolly's suggestions: Binary
search, Strcutured object scheme basically, SGML content enum type.<P>

This module is implemented by <A HREF="SGML.c">SGML.c</A>, and it is a
part of the <A
HREF="http://info.cern.ch/hypertext/WWW/Library/User/Guide/Guide.html">
Library of Common Code</A>.

<PRE>
#ifndef SGML_H
#define SGML_H

#include "HTStream.h"
</PRE>

<H2>SGML content types</H2>
<PRE>typedef enum _SGMLContent{
  SGML_EMPTY,    /* no content */
  SGML_LITERAL, /* character data. Recognized exact close tag only.
  		    Old www server compatibility only! Not SGML */
  SGML_CDATA,    /* character data. recognize &lt;/ only */
  SGML_RCDATA,   /* replaceable character data. recognize &lt;/ and &amp;ref; */
  SGML_MIXED,    /* elements and parsed character data. recognize all markup */
  SGML_ELEMENT   /* any data found will be returned as an error*/
  } SGMLContent;


typedef struct {
    char *	name;		/* The (constant) name of the attribute */
    				/* Could put type info in here */
} attr;


/*		A tag structure describes an SGML element.
**		-----------------------------------------
**
**
**	name		is the string which comes after the tag opener "&lt;".
**
**	attributes	points to a zero-terminated array
**			of attribute names.
**
**	litteral	determines how the SGML engine parses the charaters
**			within the element. If set, tag openers are ignored
**			except for that which opens a matching closing tag.
**
*/
typedef struct _tag HTTag;
struct _tag{
    char * 	name;			/* The name of the tag */
    attr * 	attributes;		/* The list of acceptable attributes */
    int		number_of_attributes;	/* Number of possible attributes */
    SGMLContent contents;		/* End only on end tag @@ */		
};

/*		DTD Information
**		---------------
**
** Not the whole DTD, but all this parser usues of it.
*/
typedef struct {
    HTTag *		tags;		/* Must be in strcmp order by name */ 
    int			number_of_tags;
    CONST char **	entity_names;	/* Must be in strcmp order by name */
    int			number_of_entities;
} SGML_dtd;

#define MAX_ATTRIBUTES 20	/* Max number of attributes per element */

/*	SGML context passed to parsers
*/
typedef struct _HTSGMLContext *HTSGMLContext;	/* Hidden */


/*__________________________________________________________________________
*/

</PRE>
<H2>Structured Object definition</H2>A structured object is something
which can reasonably be represented
in SGML.  I'll rephrase that.  A
structured object is am ordered tree-structured
arrangement of data which is representable
as text.The SGML parer outputs to
a Structured object. A Structured
object can output its contents to
another Structured Object. It's a
kind of typed stream. The architecure
is largely Dan Conolly's. Elements
and entities are passed to the sob
by number, implying a knowledge of
the DTD. Knowledge of the SGML syntax
is not here, though.<P>
Superclass: HTStream<P>
The creation methods will vary on
the type of Structured Object.Maybe
the callerData is enough info to
pass along. <P>

<B>NOTE: </B>The <CODE>put_block</CODE> method was <CODE>write</CODE>,
but this upset systems which had macros for <CODE>write()</CODE>. See
<A HREF="HTStream.html">the generic stream definition</A> for valid
return codes.<P>

<PRE>typedef struct _HTStructured HTStructured;

typedef struct _HTStructuredClass{

	char*  name;				/* Just for diagnostics */

	int (*_free) PARAMS((
		HTStructured*	me));

	int (*abort) PARAMS((
		HTStructured*	me,
		HTError		e));
		
	void (*put_character) PARAMS((
		HTStructured*	me,
		char		ch));
				
	void (*put_string) PARAMS((
		HTStructured*	me,
		CONST char *	str));
		
	void (*put_block) PARAMS((
		HTStructured*	me,
		CONST char *	str,
		int		len));
		
	void (*start_element) PARAMS((
		HTStructured*	me,
		int		element_number,
		CONST BOOL*		attribute_present,
		CONST char**		attribute_value));
		
	void (*end_element) PARAMS((
		HTStructured*	me,
		int		element_number));

	void (*put_entity) PARAMS((
		HTStructured*	me,
		int		entity_number));
		
} HTStructuredClass;

</PRE>
<H2>Find a Tag by Name</H2>Returns a pointer to the tag within
the DTD.
<PRE>extern HTTag * SGMLFindTag PARAMS((CONST SGML_dtd* dtd, CONST char * string));


</PRE>
<H2>Find a Attribute by Name</H2>Returns the number of the
atribute or -1 if failure.
<PRE>extern int SGMLFindAttribute PARAMS((HTTag* tag, CONST char * string));


</PRE>
<H2>Create an SGML parser</H2>
<PRE>/*
** On entry,
**	dtd		must point to a DTD structure as defined above
**	callbacks	must point to user routines.
**	callData	is returned in callbacks transparently.
** On exit,
**		The default tag starter has been processed.
*/


extern HTStream* SGML_new PARAMS((
	CONST SGML_dtd * 		dtd,
	HTStructured *		target));

extern CONST HTStreamClass SGMLParser;


#endif	/* SGML_H */










</PRE></BODY>
</HTML>
Webmaster