On Dec 10, 2:02 am, Hannes Wallnoefer <[EMAIL PROTECTED]> wrote:
> On Dec 9, 10:14 pm, [EMAIL PROTECTED] wrote:
>
>
>
> > Hi!
>
> > I need to be able to parse one 'token' at a time of a JavaScript
> > source (fx. a java.lang.String) and build an AST tree out of it.
>
> > The source string will not contain the complete JavaScript source when
> > I start parsing it but it should always contain valid JavaScript from
> > start to end of string at the time I parse.
>
> > An example:
>
> > I might start by having the following:
>
> > String javascript =
> > "var x = 0;\n" +
> > "for (int i = 0; i < 5; i++) {";
>
> > which is missing the ending "}" to have a valid for-loop.... but never
> > the less this will be the case when I start building the AST tree, so
> > I need to put each 'token' into the AST tree and keep checking for an
> > end-of-string event.
>
> > When I have parsed the above example I will add some more javascript,
> > for example:
>
> > javascript = javascript +
> > " x = x + i;\n" +
> > "}\n" +
> > "x\n";
>
> > and I should then be able to continue parsing from where I left off
> > before and add additional tokens to the AST tree.
>
> I think the closest thing Rhino supports is the
> Context.stringIsCompilableUnit() method. This tries to compile a
> string and returns false if an error occured and the string was fully
> read, in other words if the error could be fixed by adding more code
> to the source string.
>
> The Rhino shell uses this for expressions spanning over multiple
> lines.
>
> Of course this doesn't give you access to the partial parse tree. I'm
> not so familiar with the new parsing code, but maybe it would be
> possible to attach the parse data to the exception in some way when an
> error occurs.
>
> Hannes
>
> > I need to always be able to know if I'm inside a loop, a nested loop n
> > times, a function or what-ever, and I think that AST is the way to
> > go..... or can I keep this 'depth' by just reading tokens?
>
> > Anyone have an example of how to do this?
> > Also please do not hesitate to ask further questions if you didn't
> > understand the above ;-)
The new parser code includes support for IDE mode, which is supposed
to be better at computing error bounds. Look for
CompilerEvirons.setIdeMode.
--N
_______________________________________________
dev-tech-js-engine-rhino mailing list
[email protected]
https://lists.mozilla.org/listinfo/dev-tech-js-engine-rhino
