Question

I’ve been trying to create a parser using simpleparse. I’ve defined the grammar like this:

<w> := [ \n]* statement_list := statement,(w,statement)? statement := 'MOVE',w,word,w,'TO',w,(word,w)+ word := [A-Za-z],[A-Za-z0-9]*,([-]+,[A-Za-z0-9]+)* 

Now if I try to parse a string

MOVE ABC-DEF TO ABC MOVE DDD TO XXX 

The second statement gets interpreted as parameters of the first one… This sucks and is obviously not what I want. I have been able to get this working using pyparsing like this:

word = Word(alphas,alphanums+'-') statement = 'MOVE'+word+'TO'+word statement_list = OneOrMore(statement.setResultsName('statement',True)) 

Is there any way to get this working in simpleparse as well?

EDIT: clarification below

I am not trying to achieve a line-based grammar. What I would like to see being parsed is:

Simple case

MOVE AA TO BB 

More comlex case

MOVE AA TO BB             CC DD             EE FF 

Several of the above statments

MOVE AA TO BB             CC MOVE CC TO EE MOVE EE TO FF             GG             HH IIJJK 

Answer 1

The grammar is currently ambiguous. On paper you cannot parse if ‘MOVE A TO B MOVE C TO D’ is two statements, or one statement with particular badly named destinations.

You have two answers. You may like neither.

1. You explicitly make your WORD not match any reserved word. That is, you specifically disallow matching MOVE or TO. This is equivalent to saying ‘MOVE is not a valid parameter name’. This makes ‘MOVE TL TO TM TN TO’ an error.

2. You modify your grammar so that you can tell where the statement ends. You could add commas ‘MOVE AA TO BB, CC MOVE TM TO TN, TO, TP’. You could add semi-colons or blank lines at the end of statements. You could require that MOVE be the least indented, like Python.