Magic in Fitnext

the core algorithm of generating meals for fitness

Transcript

1. Magic in Fitnext — “engine, start”
3. ? ? ? ? ? Where is the magic?
4. Ingredient Rules User’s program and diagnosis Food constraints
5. Food Constraints
6. • “detox = yes” • “cycle_index is odd” • “moment is breakfast” • “day_index included_in 1,3” Raw materials
7. Steps 1. Manufacture (mapping, translate) 2. Dynamic evaluation
8. Manufacture #<Condition id: 6, attr: "detox", matcher: nil, value: "true"> #<Condition id: 147, attr: "cycle_index", matcher: "odd", value: ""> operator = (matcher.nil? || matcher.eql?('=') || ['odd', 'even'].include? (matcher)) ? '==' : matcher key = case attr when 'detox': 'has_detox_readiness' when 'objective': 'objective_id' when 'cycle_index' ['odd', 'even'].include?(matcher) ? 'cycle_parity' : 'cycle_number' else attr end val = ['odd', 'even'].include?(matcher) ? "\'#{matcher}'" : value
9. #<Condition id: 6, attr: "detox", matcher: nil, value: "true"> "#{key} #{operator} #{val}" "has_detox_readiness == true" #<Condition id: 147, attr: "cycle_index", matcher: "odd", value: ""> "cycle_parity == 'odd'"
10. The object object = attr.eql?('sport_sessions_per_week') ? diagnosis : diagnosis.program #<Program id: 4, user_id: 26, objective_id: 5, completed: nil, started: nil, has_detox_readiness: true,…> # program.rb def cycle_number user.program_histories.count + 1 end def cycle_parity (cycle_number % 2) == 0 ? 'even' : 'odd' end
11. Dynamic evaluation object.send(:eval, "#{key} #{operator} #{val}") — Code is data, data is code It is important for one to think of code as data to understand what eval is all about. The idea is that the code itself can be manifested as a primitive data type which the language can then make sense of. program.has_detox_readiness == true program.cycle_parity == 'odd' Kernel.eval()
12. eval() can be evil eval(“program.has_detox_readiness == User.delete_all”) Check and validate the user input!
13. constraint 1 constraint 2 constraint 3 constraint 4
14. Ingredient rules
17. # Diagnosis.rb include InferenceEngine::Inputs diagnosis.respond_to(“season_january”) == true
18. module InferenceEngine module Rule def evaluate(input_object) @evaluation_result = false if input_object.respond_to?(self.attribute_name) @evaluation_result = (input_object.send(self.attribute_name) == (self.value == 1)) else false end end end end # ingredient.rb include InferenceEngine::Inference # ingredient_rule.rb include InferenceEngine::Rule
19. module InferenceEngine module Inference def infere!(input_object) @excluded = false @priority_level = 0 @evaluated_rules = [] rules.each do |rule| @evaluated_rules << rule if rule.evaluate(input_object) case rule.human_rule_type when :exclude @excluded = true return when :prioritize @priority_level += 1 when :penalize @priority_level -= 1 end end end end
20. constraint 3
21. Thanks!