Use Tatsu to get generic AST and create IR

tatsu-example.py

#!/usr/bin/env python3

import tatsu
import json
from pprint import pprint

GRAMMAR = """@@grammar::ZQUERY

start = expression $ ;

expression
    =
    | or_expr
    | term
    ;

or_expr
    =
    left:expression op:'|' right:term
    ;

term
    =
    | and_expr
    | factor
    ;

and_expr
    = left:term op:and_op right:factor
    ;

and_op
    = op:'&'
    | op:'!'
    ;

not_expr
    = left:term op:'!' right:factor
    ;

factor
    =
    | '(' @:expression ')'
    | z_field
    ;

z_field
    = field:literal ':' text:literal
    ;

literal
    = word:/"(\s+|\w+)*"/
    | word:/\w+/
    ;
"""

from collections import namedtuple

Field = namedtuple("Field", ["field","text"])
OpNode = namedtuple("OpNode", ["op", "left", "right"])

class Semantics(object):
    def literal(self, ast):
        return ast.word

    def z_field(self, ast):
        return Field(ast.field, ast.text)

    def and_expr(self, ast):
        return OpNode(ast.op, ast.left, ast.right)

    def and_op(self, ast):
        return ast.op

    def or_expr(self, ast):
        return OpNode(ast.op, ast.left, ast.right)

def go():
    parser = tatsu.compile(GRAMMAR)
    #line = input("> ")
    line = "subject:x & (title:urgent | title:whatever) | to:gkt"

    # Semantics-free AST - parser can inject semantics (ZettelGeist shows how to do this)
    ast = parser.parse(line)
    print(ast)

    ir = parser.parse(line, semantics=Semantics()) 
    print(ir)

    # Dump JSON just to get pretty-prineted "tree" representation
    #print(json.dumps(ast, indent=3))

if __name__ == '__main__':
    go()

Output looks like this:

{'left': {'left': {'field': {'word': 'subject'}, 'text': {'word': 'x'}}, 'op': {'op': '&'}, 'right': {'left': {'field': {'word': 'title'}, 'text': {'word': 'urgent'}}, 'op': '|', 'right': {'field': {'word': 'title'}, 'text': {'word': 'whatever'}}}}, 'op': '|', 'right': {'field': {'word': 'to'}, 'text': {'word': 'gkt'}}}
OpNode(op='|', left=OpNode(op='&', left=Field(field='subject', text='x'), right=OpNode(op='|', left=Field(field='title', text='urgent'), right=Field(field='title', text='whatever'))), right=Field(field='to', text='gkt'))