F5 Config Parser

f5_config_parser.py

import lark
import re


f5_config_grammar = r"""
    // define rules
    ?start: array
    ?value: object
          | value_set
          | string
          | ESCAPED_STRING -> quoted_string
    
    array  : [command_block ("\n" command_block)*]
    // Command block = command + method + arguments (command token + key-value pair)
    command_block: command " " pair
    object : "{\n" [pair ("\n" pair)*] "\n}"
    value_set : "{ " [(value " ")*] "}"
    pair   : string " " value
    command: COMMAND_TOKEN
    string : STRING_TOKEN

    // define tokens (NOTE: longer tokens should be defined first, i.e. "auth token" before "auth")
    COMMAND_TOKEN: "auth token" | "auth" | "sys"
    STRING_TOKEN : /([a-zA-Z0-9\-_.=$\/\:])+/
    
    %import common.ESCAPED_STRING
"""


class F5ConfigTransformer(lark.Transformer):
    
    def pair(self, values):
        assert len(values) == 2
        k, v = values
        return {k: v}
    
    def object(self, values):
        merged = {}
        for value in values:
            merged = {**merged, **value}
        return merged
    
    @lark.v_args(inline=True)
    def quoted_string(self, value):
        return value[1:-1]

    array = list
    command_block = pair
    command = lark.v_args(inline=True)(str)
    string = lark.v_args(inline=True)(str)


parser = lark.Lark(f5_config_grammar, parser="lalr", transformer=F5ConfigTransformer())

# TODO: figure out how to ignore tabs directly in parser
# clean_config_text = config_text.replace("    ", "")
# parser.parse(clean_config_text)