Hi,
I'm trying to understand the transform functionality of Parslet. For
this reason I parsed the string 'a + b + c '.
Transforming the parse tree of this string results in the following
transformation output:
PARSING:a + b + c
ORIGINAL TREE:
{:result=>
[{:term=>{:factor=>{:ident=>"a"@1}}},
{:plus=>"+ "@3, :term=>{:factor=>{:ident=>"b"@5}}},
{:plus=>"+ "@7, :term=>{:factor=>{:ident=>"c"@9}}}]}
TRANSFORMED BY RULE: Terminals
{:result=>
[{:term=>"a"@1},
{:plus=>"+ "@3, :term=>"b"@5},
{:plus=>"+ "@7, :term=>"c"@9}]}
TRANSFORMED BY RULE: Test1
{:result=>
[{:left=>"a"@1, :plus=>"+ "@3, :right=>"b"@5},
{:plus=>"+ "@7, :term=>"c"@9}]}
TRANSFORMED BY RULE: Test2
{:result=>[{:left=>"t3", :plus=>"+ "@7, :term=>"c"@9}]}
In this example, I used the following rules:
@t.rule(:result=> [{:term=>simple(:a)}, {:plus=>simple(:b),
:term=>simple(:c)},
{:plus=>simple(:d), :term=>simple(:e)} ] ){
{:result=> [{:left=>a, :plus=>b, :right=>c},{:plus=>d, :term=>e } ] }
}
and
@t.rule(:result=>[{:left=>simple(:a), :plus=>simple(:b),
:right=>simple(:c)},
{:plus=>simple(:d), :term=>simple(:e) } ]) {
{:result=>[{:left=>newtmp, :plus=>d, :term=>e } ]}
}
To make these rules, I used the intermediate results of the parse tree.
The problem is when I altered this string 'a + b + c' into 'a + b + c + d'
the above rules don't give the required transformation output; see next
example:
PARSING:a + b + c + d
ORIGINAL TREE:
{:result=>
[{:term=>{:factor=>{:ident=>"a"@1}}},
{:plus=>"+ "@3, :term=>{:factor=>{:ident=>"b"@5}}},
{:plus=>"+ "@7, :term=>{:factor=>{:ident=>"c"@9}}},
{:plus=>"+ "@11, :term=>{:factor=>{:ident=>"d"@13}}}]}
TRANSFORMED BY RULE: Test2
{:result=>
[{:term=>"a"@1},
{:plus=>"+ "@3, :term=>"b"@5},
{:plus=>"+ "@7, :term=>"c"@9},
{:plus=>"+ "@11, :term=>"d"@13}]}
For the new input string 'a + b + c + d' it seems to be that I should
have to make new rules for every
new (random) input string of a given grammar. I hardly can't believe
that this is the way I should have to work with Parslet!
My question is how to define transformation rules in a smart way? My
guess is that I'm not using the right matching features
like, for example, sequence or subtree?
Many thanks in advance,
Thiel
