freebsd-dev/usr.bin/sgmls/instant

#
# Copyright (c) 1994  
# Open Software Foundation, Inc. 
#  
# Permission is hereby granted to use, copy, modify and freely distribute 
# the software in this file and its documentation for any purpose without 
# fee, provided that the above copyright notice appears in all copies and 
# that both the copyright notice and this permission notice appear in 
# supporting documentation.  Further, provided that the name of Open 
# Software Foundation, Inc. ("OSF") not be used in advertising or 
# publicity pertaining to distribution of the software without prior 
# written permission from OSF.  OSF makes no representations about the 
# suitability of this software for any purpose.  It is provided "as is" 
# without express or implied warranty. 
#

	instant - a formatting application for OSF SGML instances
____________________________________________________________________________

Requirements

    ANSI C compiler (gcc is one)

    sgmls 1.1 -- sgml parser from James Clark.  Based on Goldfarb's ARC parser.

    Vanilla unix make

    POSIX C libraries


Files for instant program

    Module		Function
    ------		--------
    browse.c		interactive browser
    general.h		general definitions
    info.c		print information about the instances
    main.c		main entry, arg parsing, instance reading
    tables.c		table-specific formatting routines (TeX and tbl)
    traninit.c		translator initialization (read spec, etc.)
    translate.c		main translator
    translate.h		structure definitions for translation code
    tranvar.c		routines for handling "special variables"
    util.c		general utilities


Also required

    1. Translation spec (transpec) files.  (../transpecs/*.ts)
    2. SDATA mapping files for mapping sdata entities.  (../transpecs/*.sdata)
    3. Character mapping files for mapping characters.  (../transpecs/*.cmap)


Platforms tried on

    OSF1 1.3 (i486)
    Ultrix 4.2 (mips)
    HP-UX 9.01 (hp 9000/700)
    AIX 3.2 (rs6000)
    SunOS [missing strerror()]

____________________________________________________________________________

			General outline of program
			------- ------- -- -------

To summarize in a sentence, instant reads the output of sgmls, builds a tree
of the instnace in memory, then traverses the tree in in-order, processing
the nodes according to a translation spec.

Element tree storage
------- ---- -------

The first thing instant must do is read the ESIS (output of sgmls) from the
specified file or stdin, forming a tree in memory.  (Nothing makes sense
without an instance...)  Each element of the instance is a node in the tree,
stored as a structure called Element_t.  Elements contain content (what
else?), which is a mixture of data (#PCDATA, #CDATA, #RCDATA - all the same
in the ESIS), child elements, and PIs.  Each 'chunk' of content is referred
to by a Content_t structure.  A Content_t contains an enum that can point to
a string (data or PI), another Element_t.  For example, if a <p> element
contains some characters, an <emphasis> element, some more characters,
a <function> element, then some more characters, it has 5 Content_t children
as an array.

Element_t's have pointers to their parents, and a next element in a linked
list (they're stored as a linked list, for cases when you'd want to quickly
travers all the nodes, in no particular order).
For convenience, Element_t's have an array of pointers to it's child
Element_t's.  These are just pointers to the same thing contained in the
Content_t array, without the intervening data or PIs.  This makes it easier
for the program to traverse the elements of the tree (it does not have to
be concerned with skipping data, etc.).  There is an analagous array of
pointers for the data content, an array of (char *)'s.  This makes it easier
to consider the immediate character content of an element.

Attributes are kept as an array of name-value mappings (using the typedef
Mapping_t).  ID attributes are also stored in a separate list of ID value -
element pointer pairs so that it is quick to find an element by ID.

Other information kept about each element (in the Element_t struct) includes
the line number in the EISI where the element occurs, the input filename.
(These depend on sgmls being run with the "-l" option.)  Also stored is
an element's order in its parent's collection of children and an element's
depth in the tree.

Translation specs
----------- -----

A translation spec is read into a linked list in memory.  As the instance
tree is traversed, the transpecs are searched in order for a match.  As a
rule, one should position the less specific transpecs later in the file.
Also, specs for seldom-used element are best placed later in the file, since
it takes cpu cycles to skip over them for each of the more-used elements.

During translation of a particular element, the list of Content_t structures
are processed in order.  If a content 'chunk' is data, it is printed to
the output stream.  If it is an element, the translation routine is called
for that elemen, recursively.  Hence, in-order traversal.

Miscellaneous information displays
------------- ----------- --------

There are several informational display options available. They include:
    -	a list of element usage (-u) -- lists each element in the instance,
	it's attributes, number of children, parent, data content 'nodes'.
    -	statistics about elements (-S) -- lists elements, number of times
	each is used, percent of elements that this is, total char content
	in that element, average number of characters in they element.
    -	show context of each element (-x) -- lists each element and its
	context, looking up the tree (this is the same context that
	would match the Context field of a transpec).
    -	show the hierarchy of the instance (-h) -- show an ascii 'tree' of
	the instance, where elements deeper in the tree are indented more.
	Numbers after the element name in parentheses are the number of
	element content nodes and the number of data content nodes.

Interactive browser
----------- -------

Originally, the interactive browser was intended as a debugging aid for
the code developer.  It was a way to examine a particular node of the
instance tree or process a subtree without being distracted by the rest
of the instance.  Many of the commands test functionality of the query
and search code (such as testing whether a certain element has a given
relationship to the current position in the tree).


____________________________________________________________________________