Concrete syntax parser for the rule DSL. The parsed grammar is defined as follows:

ident = [a-zA-Z] [a-zA-Z0-9_-]*
str = '"' <any character with " and \ escaped> '"'
regex = '/' <any character with / and \ escaped> '/'

file = ruleset*

ruleset = 'ruleset' 'for' filename '{' rule ('or' rule)* '}'

filename = 'agency'
          | 'stops'
          | 'routes'
          | 'trips'
          | 'stop_times'
          | 'calendar'
          | 'calendar_dates'
          | 'fare_attributes'
          | 'fare_rules'
          | 'shapes'
          | 'frequencies'
          | 'transfers'
          | 'pathways'
          | 'levels'
          | 'feed_info'
          | 'translations'
          | 'attributions'

rule = 'rule' ident '{' 'when' matcher 'then' action '}'

matcher = '(' matcher ')'
        | '!' factor
        | 'any'
        | variable 'exists'
        | value valueMatcher
        | matcher 'and' matcher
        | matcher 'or' matcher

valueMatcher = '==' value
              | 'in' '[' value* ']'
              | 'matches' regex

action = 'delete'
        | transformation ('then' transformation)*
        | log

transformation = 'set' 'row' '[' str ']' 'to' value
                | 'search' regex 'in' 'row' '[' str ']' 'and' 'replace' 'by' value

log = 'debug' '(' expr ')'
    | 'info' '(' expr ')'
    | 'warning' '(' expr ')'
    | 'error' '(' expr ')'

expr = ident '(' (expr (',' expr)*)? ')'
      | value
      | expr + expr

value = str
      | variable

variable = 'row' '[' value ']'
          | 'ctx' '[' value ']' ('[' value ']')*

For matchers, operators have natural precendence: ! has precedence over and which as precedence over or. This means that !m1 and m2 or m3 is equivalent to ((!m1) and m2) or m3. Combinators are left associative.