Add an optimized version of str(…) | str(…) | …

This commit is contained in:
Claire 2023-09-01 11:21:47 +02:00
parent 32eb0e7744
commit e5caa462c1

View file

@ -1,11 +1,47 @@
# frozen_string_literal: true
class SearchQueryParser < Parslet::Parser
# Efficiently matches disjoint strings
class StrList < Parslet::Atoms::Base
attr_reader :strings
def initialize(strings)
super()
@strings = strings
@pattern = Regexp.union(strings)
@min_length = strings.map(&:length).min
end
def error_msgs
@error_msgs ||= {
premature: 'Premature end of input',
failed: "Expected any of #{strings.inspect}, but got ",
}
end
def try(source, context, _consume_all)
match = source.match(@pattern)
return succ(source.consume(match)) unless match.nil?
# Input ending early:
return context.err(self, source, error_msgs[:premature]) if source.chars_left < @min_length
# Expected something, but got something else instead:
error_pos = source.pos
context.err_at(self, source, [error_msgs[:failed], source.consume(@len)], error_pos)
end
def to_s_inner(_prec)
"[#{strings.map { |str| "'#{str}'" }.join(',')}]"
end
end
rule(:term) { match('[^\s]').repeat(1).as(:term) }
rule(:colon) { str(':') }
rule(:space) { match('\s').repeat(1) }
rule(:operator) { (str('+') | str('-')).as(:operator) }
rule(:prefix_operator) { str('has') | str('is') | str('language') | str('from') | str('before') | str('after') | str('during') | str('in') }
rule(:prefix_operator) { StrList.new(%w(has is language from before after during in)) }
rule(:prefix) { prefix_operator.as(:prefix_operator) >> colon }
rule(:phrase) do
(str('"') >> match('[^"]').repeat.as(:phrase) >> str('"')) |