Module: Legion::LLM::Discovery::RuleGenerator

Extended by:
Legion::Logging::Helper
Defined in:
lib/legion/llm/discovery/rule_generator.rb

Constant Summary collapse

EMBEDDING_PATTERNS =
%w[
  embed mxbai-embed nomic-embed bge- snowflake-arctic-embed
  text-embedding titan-embed
].freeze
DISCOVERABLE_PROVIDERS =
%i[ollama mlx vllm].freeze
TIER_MAP =
{
  ollama:    :local,
  mlx:       :local,
  vllm:      :fleet,
  openai:    :cloud,
  bedrock:   :cloud,
  azure:     :cloud,
  gemini:    :cloud,
  anthropic: :frontier
}.freeze
DEFAULT_TIER_PRIORITY =
%i[local direct fleet openai_compat cloud frontier].freeze

Class Method Summary collapse

Class Method Details

.build_rule(provider, instance, model_data, capability, tier, priority) ⇒ Object



105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
# File 'lib/legion/llm/discovery/rule_generator.rb', line 105

def build_rule(provider, instance, model_data, capability, tier, priority)
  model_name = model_data.is_a?(Hash) ? (model_data[:name] || model_data['name']).to_s : model_data.to_s
  target = {
    provider:           provider.to_sym,
    instance:           instance.to_sym,
    model:              model_name,
    tier:               tier,
    model_capabilities: extract_capabilities(model_data),
    context_length:     extract_field(model_data, :context_length),
    parameter_count:    extract_field(model_data, :parameter_count)
  }.compact
  {
    name:     "auto:#{provider}/#{instance}:#{model_name}:#{capability}",
    when:     { capability: capability },
    then:     target,
    priority: priority
  }
end

.embedding_model?(model_data) ⇒ Boolean

Returns:

  • (Boolean)


65
66
67
68
69
70
71
72
73
74
75
# File 'lib/legion/llm/discovery/rule_generator.rb', line 65

def embedding_model?(model_data)
  if model_data.is_a?(Hash)
    caps = model_data[:capabilities] || model_data['capabilities']
    return caps.any? { |c| c.to_s == 'embedding' } if caps.is_a?(Array) && caps.any?
  end

  # Fall back to name pattern matching when no capability data
  name = model_data.is_a?(Hash) ? (model_data[:name] || model_data['name']).to_s : model_data.to_s
  name = name.downcase
  EMBEDDING_PATTERNS.any? { |pat| name.include?(pat) }
end

.extension_providersObject



159
160
161
162
163
164
165
166
167
# File 'lib/legion/llm/discovery/rule_generator.rb', line 159

def extension_providers
  ext = Legion::Settings[:extensions]
  return ext[:llm] if ext.is_a?(Hash) && ext[:llm].is_a?(Hash)

  {}
rescue StandardError => e
  handle_exception(e, level: :debug, handled: true, operation: 'rule_generator.extension_providers')
  {}
end

.extract_capabilities(model_data) ⇒ Object



124
125
126
127
128
129
130
131
# File 'lib/legion/llm/discovery/rule_generator.rb', line 124

def extract_capabilities(model_data)
  return nil unless model_data.is_a?(Hash)

  caps = model_data[:capabilities] || model_data['capabilities']
  return caps if caps.is_a?(Array) && caps.any?

  nil
end

.extract_field(model_data, field) ⇒ Object



133
134
135
136
137
# File 'lib/legion/llm/discovery/rule_generator.rb', line 133

def extract_field(model_data, field)
  return nil unless model_data.is_a?(Hash)

  model_data[field] || model_data[field.to_s]
end

.generate(discovered_instances) ⇒ Object



33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
# File 'lib/legion/llm/discovery/rule_generator.rb', line 33

def generate(discovered_instances)
  rules = []
  discovered_instances.each do |provider, instances|
    next unless DISCOVERABLE_PROVIDERS.include?(provider.to_sym)
    next unless instances.is_a?(Hash)

    tier = TIER_MAP[provider.to_sym] || :local
    order = 0
    instances.each do |instance_id, data|
      models = data.is_a?(Hash) ? Array(data[:models]) : []

      models.each do |model|
        model_data = model.is_a?(Hash) ? model : { name: model.to_s }
        model_name = (model_data[:name] || model_data['name']).to_s
        next if model_name.empty?

        model_tier = extract_field(model_data, :tier)&.to_sym ||
                     extract_field(model_data, 'tier')&.to_sym ||
                     tier
        capability = embedding_model?(model_data) ? :embed : :chat
        priority = tier_weight(model_tier) - order
        rules << build_rule(provider, instance_id, model_data, capability, model_tier, priority)
        rules << build_rule(provider, instance_id, model_data, :stream, model_tier, priority) if capability == :chat
        order += 1
      end
    end
  end

  rules += generate_configured_provider_rules
  rules.sort_by { |r| -r[:priority] }
end

.generate_configured_provider_rulesObject



77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
# File 'lib/legion/llm/discovery/rule_generator.rb', line 77

def generate_configured_provider_rules
  rules = []
  providers_config = extension_providers
  return rules unless providers_config.is_a?(Hash)

  providers_config.each do |provider_name, config|
    next unless config.is_a?(Hash)
    next if config[:enabled] == false
    next if DISCOVERABLE_PROVIDERS.include?(provider_name.to_sym)

    tier = TIER_MAP[provider_name.to_sym]
    next unless tier

    default_model = config[:default_model]
    next unless default_model

    model_data = { name: default_model }
    priority = tier_weight(tier)
    rules << build_rule(provider_name, :default, model_data, :chat, tier, priority)
    rules << build_rule(provider_name, :default, model_data, :stream, tier, priority)
  end

  rules
rescue StandardError => e
  handle_exception(e, level: :warn, handled: true, operation: 'llm.discovery.rule_generator.configured_providers')
  []
end

.tier_priorityObject



147
148
149
150
151
152
153
154
155
156
157
# File 'lib/legion/llm/discovery/rule_generator.rb', line 147

def tier_priority
  configured = Legion::LLM::Settings.value(:routing, :tier_priority, default: DEFAULT_TIER_PRIORITY)
  normalized = Array(configured).filter_map do |tier|
    tier.to_sym if tier.respond_to?(:to_sym)
  end
  normalized = DEFAULT_TIER_PRIORITY if normalized.empty?
  (normalized + DEFAULT_TIER_PRIORITY).uniq
rescue StandardError => e
  handle_exception(e, level: :warn, handled: true, operation: 'rule_generator.tier_priority')
  DEFAULT_TIER_PRIORITY
end

.tier_weight(tier) ⇒ Object



139
140
141
142
143
144
145
# File 'lib/legion/llm/discovery/rule_generator.rb', line 139

def tier_weight(tier)
  tier_sym = tier.respond_to?(:to_sym) ? tier.to_sym : tier
  index = tier_priority.index(tier_sym)
  return 0 unless index

  (tier_priority.length - index) * 100
end