HighWire and DOM 2 correspondences and notes
date:  June 24,  2002
author:  Dan Ackerman (baldrick@netset.com)

Contents
   1. Purpose
   2. HighWire Internal  Document Structures.
   3. STRUCT S_CONTAINR
   4. STRUCT FRAME_ITEM
   5. STRUCT PARAGRAPH_ITEM
   6. STRUCT WORD_ITEM
   7. STRUCT S_IMAGE  (Note this is posibly in transition)
   8. STRUCT S_TABLE
       8.1 STRUCT S_TABLE_ROW
       8.2 STRUCT S_TABLE_CELL
   9. Implementation NOTES
       9.1 HTMLCollection Interface and attribute ID
       9.2 Matching DOM and HighWire DOM
  10. References

1. Purpose

This Document is meant to describe HighWires internal storage
mechanisms and cover the corespondences between
HighWire's internal storage of a Document and the W3 specifications
for the Document Object Model as found at
http://www.w3.org/TR/2002/CR-DOM-Level-2-HTML-20020605/
Also contained are some implementation notes.

2. HighWire Internal  Document Structures.

There are 4 main internal structures for document storage in 
HighWire.

These are the Container (s_containr), the frame (frame_item),
the paragraph (paragraph_item) and the word (word_item).
They operate exactly in that order of Heirachy.

Containers contain Frames or other containers
Frames contain Paragraphs
Paragraphs contain Words


3. STRUCT S_CONTAINR

The highest level structure is the struct s_containr, it's definition
is located in containr.h.

/*
 * Containr.h -- Tree hirachy to hold frames.
 *
 * A container is a polymorph struct which can either hold one or more children
 * containers or exactly one frame (or is empty).  This way a tree structure of
 * containers  represents  the  hirachy as given by  and  tags
 * and  allowes  every  node (= container)  in the tree to have it's content be
 * replaced at any time.
 *
 * AltF4 - Feb. 17, 2002
 *
 */


4. STRUCT FRAME_ITEM


The next level of the structure is the struct frame_item, it's
definition is located in defs.h.  This is the highest level structure
that contains HTML data beyond the FRAMESET element.
Beyond holding the bodies attributes and some buffers used
in parsing the document, The main contents of the Frame are
the lists for the object (clickable and anchors) and the paragraph
list.  All HTML data is contained in the paragraph list.

  clickable_area list.
    This is a list of all the area's in this frame that respond to user interaction
    via a mouseclick.

  
  named_location list.
    This is a list of all the target area's in the frame.  (anchors)

NOTE:  These correspond in a way to the DOM model lists.  More research
should be done on whether these lists are adequate.  I am uncertain on some
topics at the moment.  (ie should all paragraphs have id's or only those that
explicitly have an id attribute)



5. STRUCT PARAGRAPH_ITEM

The paragraph_item structure is the main organizational structure
in HighWire.  It's definition can be found in defs.h.   Paragraphs
are of many different natures.  At their simplest a Paragraph is
a list of word_item structs.  However a Paragraph can also be
a Table, IMG, Horizontal Ruler, Ordered List or Unorder List.
Basically a PARAGRAPH_ITEM is a block level data container.


6. STRUCT WORD_ITEM

The word_item structure is the most basic element structure
of HighWire.  It contains formatting and style attributes and
one of the following.

  line_brk code - this forces an end to the current line

  struct s_image * - this contains information on an image.
   In this case we would be working with an IMG tag contained
   within a paragraph (ie NOT align LEFT or align RIGHT)

  UWORD *item - This would be the data of this word.  Simplistically
   it can be thought of exactly as it's name implies,  this is a word.
  ex.  "Atari"

7. STRUCT S_IMAGE  (Note this is posibly in transition)

An image structure (struct s_image) can exist in 2 different manners.
It's most common form, is most likely as a link to a word_item struct
inside of a paragraph.  This is where we have an IMG tag without
left or right alignment.  It's other posibility is in a Paragraph_struct
as a word_item, with the paragraph_code being  PAR_IMG.  The effect
then is to allow the side aligned IMG to be grabbed out at display
and moved to the appropriate border.

8. STRUCT S_TABLE

struct s_table definition is found in table.h.
A table is an escaped paragraph with a code of PAR_TABLE.
It is itself a formatting container for the sub container
struct s_table_row.  Which are contained in a link list.

8.1 STRUCT S_TABLE_ROW

A struct s_table_row, is a list of table_cells comprising one
line (or row) of a table.

8.2 STRUCT S_TABLE_CELL

A table_cell is exactly what it is named.  It is one data cell of a table.
It holds formating issues inherited from the table or row, offset information
and a paragraph_item list of contained data.  A table cell structure can 
contain any type of paragraph as discussed in 5. STRUCT PARAGRAPH_ITEM.

9. Implementation NOTES

9.1 HTMLCollection Interface and attribute ID

When the attribute ID is processed we will need to build a list of these objects, 
as well as storing the name.

example... (from html 4.01 specs)

 This is a uniquely named paragraph.
 This is also a uniquely named paragraph.

In this case "myparagraph" could be referenced by index 1 as well the
as "myparagraph".  Additionally "yourpargraph" would have an index of
2.

This will be needed to properly support the HTMLCollection interface.  In addition
an index list of Images, links, forms and anchors would need to be
created and stored in a central location for the document.   These could be
contained as they are now in the frame_item struct, with additions made for
the objects that are currently not listed.   If upon study it is determined that
this is inadequate for our needs then  another good candidate for this would
be additions into the struct s_containr.  Since this is a super frame element.

9.2 Matching W3 DOM and HighWire DOM

It may sound overly simplistic.  But in most cases so far the additions noted
in HTMLCollection Interface and attribute ID (section 9.1).  Will cover the holes
in the DOM mapping.  There are still missing HTML tag supports in HighWire
dealing with FORMS and iFrames as well as others.   There are also some
missing HTML tags dealing with Tables.   These currently would not be available
for DOM mapping.  However we can currently with a litle work map Links and
Anchors to the DOM model.  We will need to look at the EMCAscript definitions
and modify our returns in a method that is in compliance with what it expects
in these situations.

10. References

HighWire Source code can be retrieved from http://highwire.atari-users.net/

Document Object Model (DOM) Level 2 HTML Specification
Version 1.0
W3C Candidate Recommendation 05 June 2002
can be retrieved at http://www.w3.org/TR/DOM-Level-2-HTML/

HTML 4.01 Specification can be retrieved at http://www.w3.org/TR/html401