Lark: How to rule/terminal priority works

[solartes]

Hello,

I have the following grammar:

    Lark(r"""
    documento: PALABRA+ monto PALABRA+
    monto: SIGNO ENTERO 
    SIGNO: "$" INT    
    ENTERO.5: INT+    
    PALABRA.1: /[^\s]+/
    """, start='documento', parser="earley", lexer="dynamic", ambiguity="resolve")

The priority does not work in the last grammar, the curious is the priority works in the following code

Lark(r"""
documento:PALABRA+ acciondemandante PALABRA+
    acciondemandante: PALABRASDEMANDA informacionentidad+
     PALABRASDEMANDA: "demandante"i                
  informacionentidad: nombres distancia identificacion
             nombres: NOMBRE+ 
              NOMBRE: NOMBRENORMAL+
        NOMBRENORMAL: (UCASE_LETTER LCASE_LETTER+)
           distancia: CONECTOR~0..3
            CONECTOR: LCASE_LETTER+
    identificacion.5: tipo INT+
                tipo: (LCASE_LETTER | UCASE_LETTER) (LCASE_LETTER | UCASE_LETTER)
           PALABRA.1: /.+/

    """, start='documento', parser="earley", lexer="dynamic", ambiguity="resolve")

[erezsh]

Hi,
The current Earley implementation supports rule priority, but not terminal priority.

LALR supports terminal priority but not rule priority.

There are reasons! But it's also something I hope to correct in the future.

[solartes]

I already tried with rule priority but still does not work

json_parser3 = Lark(r"""
    documento: palabra+ monto palabra+
    monto.5: "$" INT+      
    palabra.1: /[^\s]+/

    %import common.INT             
    %import common.WS
    %ignore WS
    %ignore "." | "," | ";" | ":" | "-" | "(" | ")" | "\"" | "_" | "#" | "”"
    """, start='documento', parser="earley", lexer="dynamic", ambiguity="resolve")

and I tried

`   json_parser3 = Lark(r"""
    documento: PALABRA+ monto PALABRA+
    monto.5: "$" INT+      
    PALABRA.1: /[^\s]+/

    %import common.INT             
    %import common.WS
    %ignore WS
    %ignore "." | "," | ";" | ":" | "-" | "(" | ")" | "\"" | "_" | "#" | "”"
    """, start='documento', parser="earley", lexer="dynamic", ambiguity="resolve")`

I got this
Tree(monto, [Token(INT, '63')]), Token(PALABRA, '000.000.00')
and I wanna this
Tree(monto, [Token(INT, '63')], Token(INT, '000'), Token(INT, '000'), Token(INT, '00')])])

[erezsh]

You're right, it should work and it doesn't.

Meanwhile, try this

x = Lark(r"""
    documento: palabra? monto palabra?
    monto: "$" INT+
    palabra.0: /[^\s]+/

    %import common.INT
    %import common.WS
    %ignore WS
    %ignore "." | "," | ";" | ":" | "-" | "(" | ")" | "\"" | "_" | "#" | "”"
    """, start='documento')     

[solartes]

It does not works for me because the text that I send is:
...bla bla Limítese la medida en la suma de $63.000.000.00 Sírvase proceder de conformidad y consignar los dineros retenidos a Órdenes de este Juzgado para el proceso de referencia bla bla..
As you can see the $63.000.000.00 is among the text, so I need to use "palabra+"

[erezsh]

I think plain old regular expressions will solve your problem very easily.