[CZT-Devel] Re: Parser: Integrating Circus.cup into Parser.xml

[Petra M. <pe...@cs...>]

Hi Leo,

Leo Freitas wrote:
> I noticed that there are 14 ant tasks associated to parser generation.
> How many of those I need to be worried about? In other words, apart from 
> the Parser.xml
> file, what other files do I need to change? Perhaps this is in direct 
> connection with the next question.

You need all of them, but you won't need to edit all corresponding xml 
files. I added all required circus tasks to the build.xml file and 
commited it to CVS, so you don't have to worry about that.

> 1) LaTeX Markup
> 
> In Circus, there are environments for declaring new circus paragraphs as 
> \begin{circus}...\end{circus},

The CZT parser for LaTeX markup consists of a LaTex to Unicode converter 
  followed by a Unicode scanner and parser. That's because the Z grammar 
given in the ISO Standard is defined for Unicode only.
That means, you need to think about how a circus specification is 
supposed to look like in Unicode.

Since Unicode characters are difficult to type (using an editor that 
doesn't support Unicode), all Unicode charcters used in Z (or OZ, TCOZ) 
got a name. These are defined in <CZT_HOME>/devtools/zchar.xml (or 
<CZT_HOME>/devtools/ozchar.xml, CZT_HOME>/devtools/tcozchar.xml).

These XML files are used to generate the corresponding classes
net.sourceforge.czt.z.util.{ZChar,ZString}
net.sourceforge.czt.oz.util.{OzChar,OzString}
net.sourceforge.czt.tcoz.util.{TcozChar,TcozString}

So I guess we need to write a circuschar.xml file where Circus specific 
Unicode characters are listed. Tim might be able to help with this ... 
he has done that for Object Z.

To define the \begin{circus} ... \end{circus} paragraphs, we need to 
modify the LaTeX to Unicode converter. It's in
parser/templates/Latex2Unicode.xml.

Next, the Unicode scanner needs to know about Circus. The Unicode 
scanner consists of the context-free scanner followed by other kinds of 
scanners. The context-free scanner is in
parser/templates/ContextFreeScanner.xml.



Just to get you started, I added all the stuff to the scanner and parser 
so that
----------------------------------------
\begin{zed}
   \vdash?~true
\end{zed}

\begin{circus}
   x : \nat
\end{circus}
----------------------------------------

can now be parsed (with the Circus parser) and the resulting XML looks like

<Spec Version="1.3" xmlns="http://czt.sourceforge.net/zml">
     <ZSect>
         <Name>Specification</Name>
         <Parent>
             <Word>standard_toolkit</Word>
         </Parent>
         <LatexMarkupPara/>
         <ConjPara>
             <TruePred/>
         </ConjPara>
         <ns1:ChannelPara xmlns:ns1="http://czt.sourceforge.net/circus">
             <VarDecl>
                 <DeclName>
                     <Word>x</Word>
                 </DeclName>
                 <RefExpr Mixfix="false">
                     <RefName>
                         <Word><E2><84><95></Word>
                     </RefName>
                 </RefExpr>
             </VarDecl>
         </ns1:ChannelPara>
     </ZSect>
</Spec>

You can try that by calling "ant circus2zml" in the parser directory.

Hope that gives you a bit of a start,
Petra