introduction.tex

Nano的XML解析器

\chapter{Introduction}

This chapter gives a short introduction to \XML{} and \NanoXML{}.

\section{About \XML{}}

The extensible markup language,
\href{http://www.w3c.org/TR/REC-xml}{\XML{}}, is a way to mark up text in
a structured document.

\XML  is a simplification of the complex \ltext{SGML} standard.
\ltext{SGML}, the Standard Generalized Markup Language, is an international
(\ltext{ISO}) standard for marking up text and graphics.
The best known application of \ltext{SGML} is \ltext{HTML}.

Although \ltext{SGML} data is very easy to write, it's very difficult to write a
generic \ltext{SGML} parser.
When designing \XML{} however, the authors removed much of the flexibility
of \ltext{SGML} making it much easier to parse \XML{} documents correctly.

\XML{} data is structured as a tree of \term{entities}.
An entity can be a string of character data or an element which can contain other
entities.
Elements can optionally have a set of attributes.
Attributes are key/value pairs which set some properties of an element.

The following example shows some \XML{} data:

\begin{example}
$<$book$>$
~~$<$chapter id="my chapter"$>$
~~~~$<$title$>$The title$<$/title$>$
~~~~Some text.
~~$<$/chapter$>$
$<$/book$>$
\end{example}

At the root of the tree, you can find the element ``book''.
This element contains one child element: ``chapter''.
The chapter element has one attribute which maps the key ``id'' to
``my chapter''.
The chapter element has two child entities: the element ``title'' and the
character data ``Some text.''.
Finally, the title element has one child, the string ``The title''.

\section{About \NanoXML{}}

In April 2000, \NanoXML{} was first released as a spin-off project of
\ltext{AUIT}, the Abstract User Interface Toolkit.

The intent of NanoXML was to be a small parser which was easy to use.
\ltext{SAX} and \ltext{DOM} are much too complex for what I needed and the
mainstream parsers were either much too big or had a very restrictive license.

\ltext{NanoXML 1} has all the features I needed: it is very small (about 6K),
is reasonably fast for small \XML{} documents, is very easy to use and is
free (\ltext{zlib/libpng} license).
As I never intended to use \NanoXML{} to parse \ltext{DocBook} documents,
there was no support for mixed data or \ltext{DTD} parsing.

\NanoXML{} was released as a \ltext{SourceForge} project and, because of the
very good response from its users, it matured to a small and stable parser.
The final version, release \ltext{1.6.8} was released in May 2001.

Because of its small size, people started to use \NanoXML{} for embedded
systems (\ltext{KVM}, \ltext{J2ME}) and kindly submitted patches to make
\NanoXML{} work in such restricted environment.

\section{\NanoXML{} 2}

In July 2001, \ltext{NanoXML 2} has been released.
Unlike \ltext{NanoXML 1}, speed and \XML{} compliancy were considered to be
very important when the new parser was designed.
\ltext{NanoXML 2} is also very modular: you can easily replace the different
components in the parser to customize it to your needs.
The modularity of \ltext{NanoXML 2} also benefits extensions like \acronym{e.g.} \ltext{SAX} support which can now directly access the parser.
In \ltext{NanoXML 1}, the \ltext{SAX} adapter had to iterate the data structure built by the base product.

Although many features were added to \NanoXML{}, the second release was
still very small.
The full parser with builder fits in a \ltext{JAR} file of about 32K.
This is still very tiny, especially when you compare this with the ``standard'' parsers of more than four times its size.

As there is still need for a tiny parser like \ltext{NanoXML 1}, there is a
special branch of \ltext{NanoXML 2}: \ltext{NanoXML/Lite}. This parser is source compatible with \ltext{NanoXML 1} but features a new parsing algorithm which makes it more than twice as fast as the older version.
It is however more restrictive on the \XML{} data it parses: the older
version allowed some not-wellformed data to be parsed.

There are three branches of \ltext{NanoXML 2}:
\begin{itemize}
  \item[$\bullet$]
    \term{NanoXML/Lite} is the successor of \ltext{NanoXML 1}.
    It features an almost compatible parser which is extremely small.
  \item[$\bullet$]
    \term{NanoXML/Java} is the standard parser.
  \item[$\bullet$]
    \term{NanoXML/SAX} is the \ltext{SAX} adapter for \ltext{NanoXML/Java}.
\end{itemize}

The latest version of \NanoXML{} is \ltext{NanoXML 2.2.1}, which has been
released in April 2002.

\section{\NanoXML{} Extension to the \XML{} System ID}

Because it's convenient to put data files into jar files, we need some way to specify that we want some resource which can be found in the class path.
There is no support for such resources in the \XML{} 1.0 specification.
NanoXML allows you to specify such resources using the
\emph{reference part} of a \ltext{URL}.

This means that if the \ltext{DTD} of the \XML{} data is put in the
resource \filename{/data/foo.dtd}, you can specify such path using the 
following document type declaration:

\begin{example}
$<$!DOCTYPE foo SYSTEM 'file:\#/data/foo.dtd'$>$
\end{example}

It's even possible to specify a resource found in a particular jar, like in the following example:

\begin{example}
$<$!DOCTYPE foo SYSTEM 'http://myserver.com/dtds.jar\#/foo.dtd'$>$
\end{example}