llm_cost_tracker 0.5.0 → 0.5.2

data/lib/tasks/llm_cost_tracker.rake

@@ -1,5 +1,7 @@
 # frozen_string_literal: true
 
+require "fileutils"
+
 # rubocop:disable Metrics/BlockLength
 namespace :llm_cost_tracker do
   desc "Check LLM Cost Tracker setup"
@@ -26,56 +28,49 @@ namespace :llm_cost_tracker do
 
   namespace :prices do
     desc(
-      "Sync the configured pricing file from LiteLLM/OpenRouter JSON sources. " \
-      "Use PREVIEW=1 to preview, STRICT=1 to fail on provider errors, " \
-      "or OUTPUT=path/to/file.json."
+      "Refresh the configured pricing file from the maintained LLM Cost Tracker price snapshot. " \
+      "Use PREVIEW=1 to preview, URL=... to override the source, or OUTPUT=path/to/file.json."
     )
-    task :sync do
+    task :refresh do
       Rake::Task["environment"].invoke if Rake::Task.task_defined?("environment")
       require_relative "../llm_cost_tracker"
 
-      output_path = price_sync_output_path
-      strict = ENV["STRICT"] == "1" || ARGV.include?("--strict")
-      result = LlmCostTracker::PriceSync.sync(
+      output_path = price_refresh_output_path
+      source_url = LlmCostTracker::PriceSync.configured_remote_url
+      preview = ENV["PREVIEW"] == "1"
+      result = LlmCostTracker::PriceSync.refresh(
         path: output_path,
-        preview: ENV["PREVIEW"] == "1",
-        strict: strict
+        url: source_url,
+        preview: preview
       )
 
-      action = if ENV["PREVIEW"] == "1"
+      action = if preview
                  "previewed"
                elsif result.written
-                 "updated"
+                 "refreshed"
                else
                  "kept"
                end
 
       puts "llm_cost_tracker: #{action} pricing file #{result.path}"
-      print_source_usage(result.sources_used)
+      puts "  source: #{result.source_url}"
+      puts "  version: #{result.source_version.inspect}" if result.source_version
       print_changes(result.changes)
-      print_discrepancies(result.discrepancies)
-      print_issues("validator rejected", result.rejected)
-      print_issues("validator flagged", result.flagged)
-      print_models("orphaned models (no JSON source match)", result.orphaned_models)
-      print_failures(result.failed_sources, heading: "source failures (kept existing values)")
     end
 
-    desc "Compare the current pricing snapshot with LiteLLM/OpenRouter JSON sources and exit non-zero on drift."
+    desc "Compare the current pricing file with the maintained LLM Cost Tracker price snapshot."
     task :check do
       Rake::Task["environment"].invoke if Rake::Task.task_defined?("environment")
       require_relative "../llm_cost_tracker"
 
-      output_path = price_sync_output_path
-      result = LlmCostTracker::PriceSync.check(path: output_path)
+      output_path = price_refresh_output_path
+      source_url = LlmCostTracker::PriceSync.configured_remote_url
+      result = LlmCostTracker::PriceSync.check(path: output_path, url: source_url)
 
       puts "llm_cost_tracker: checked pricing file #{result.path}"
-      print_source_usage(result.sources_used)
+      puts "  source: #{result.source_url}"
+      puts "  version: #{result.source_version.inspect}" if result.source_version
       print_changes(result.changes)
-      print_discrepancies(result.discrepancies)
-      print_issues("validator rejected", result.rejected)
-      print_issues("validator flagged", result.flagged)
-      print_models("orphaned models (no JSON source match)", result.orphaned_models)
-      print_failures(result.failed_sources, heading: "source failures")
       puts "  pricing is up to date" if result.up_to_date
       abort("llm_cost_tracker: pricing check failed") unless result.up_to_date
     end
@@ -83,16 +78,6 @@ namespace :llm_cost_tracker do
 end
 # rubocop:enable Metrics/BlockLength
 
-def print_source_usage(sources_used)
-  return if sources_used.empty?
-
-  puts "  sources used:"
-  sources_used.each do |source, usage|
-    version = usage.source_version ? ", version=#{usage.source_version.inspect}" : ""
-    puts "    - #{source} (#{usage.prices_count} prices#{version})"
-  end
-end
-
 def print_changes(changes)
   puts "  changed models: #{changes.size}"
   return if changes.empty?
@@ -105,47 +90,8 @@ def print_changes(changes)
   end
 end
 
-def print_discrepancies(discrepancies)
-  return if discrepancies.empty?
-
-  puts "  source discrepancies: #{discrepancies.size}"
-  discrepancies.each do |issue|
-    formatted = issue.values.map { |source, value| "#{source}=#{value.inspect}" }.join(", ")
-    puts "    - #{issue.model} #{issue.field}: #{formatted}"
-  end
-end
-
-def print_issues(heading, issues)
-  return if issues.empty?
-
-  puts "  #{heading}: #{issues.size}"
-  issues.each do |issue|
-    puts "    - #{issue.model}: #{issue.reason}"
-  end
-end
-
-def print_models(heading, models)
-  return if models.empty?
-
-  puts "  #{heading}: #{models.size}"
-  models.each { |model| puts "    - #{model}" }
-end
-
-def print_failures(failed_sources, heading:)
-  return if failed_sources.empty?
-
-  puts "  #{heading}: #{failed_sources.size}"
-  failed_sources.each do |source, message|
-    puts "    - #{source}: #{message}"
-  end
-end
-
-def price_sync_output_path
+def price_refresh_output_path
   path = LlmCostTracker::PriceSync.configured_output_path
-  return path if path
-
-  abort(
-    "llm_cost_tracker: configure prices_file, run bin/rails generate llm_cost_tracker:prices, " \
-    "or set OUTPUT=config/llm_cost_tracker_prices.yml"
-  )
+  FileUtils.mkdir_p(File.dirname(path))
+  path
 end

metadata

@@ -1,14 +1,14 @@
 --- !ruby/object:Gem::Specification
 name: llm_cost_tracker
 version: !ruby/object:Gem::Version
-  version: 0.5.0
+  version: 0.5.2
 platform: ruby
 authors:
 - Sergii Khomenko
 autorequire:
 bindir: bin
 cert_chain: []
-date: 2026-04-25 00:00:00.000000000 Z
+date: 2026-04-27 00:00:00.000000000 Z
 dependencies:
 - !ruby/object:Gem::Dependency
   name: activesupport
@@ -84,6 +84,20 @@ dependencies:
     - - "<"
       - !ruby/object:Gem::Version
         version: '9.0'
+- !ruby/object:Gem::Dependency
+  name: nokogiri
+  requirement: !ruby/object:Gem::Requirement
+    requirements:
+    - - "~>"
+      - !ruby/object:Gem::Version
+        version: '1.16'
+  type: :development
+  prerelease: false
+  version_requirements: !ruby/object:Gem::Requirement
+    requirements:
+    - - "~>"
+      - !ruby/object:Gem::Version
+        version: '1.16'
 - !ruby/object:Gem::Dependency
   name: railties
   requirement: !ruby/object:Gem::Requirement
@@ -208,10 +222,10 @@ dependencies:
     - - "~>"
       - !ruby/object:Gem::Version
         version: '3.0'
-description: Tracks token usage, latency, and estimated costs for OpenAI, Anthropic,
-  Google Gemini, OpenRouter, DeepSeek, and OpenAI-compatible APIs. Works through Faraday
-  middleware or explicit track/track_stream helpers, with ActiveRecord storage, tag-based
-  attribution, price sync tasks, and budget guardrails.
+description: Tracks token usage, latency, and estimated costs for RubyLLM, OpenAI,
+  Anthropic, Google Gemini, OpenRouter, DeepSeek, and OpenAI-compatible APIs. Works
+  through Faraday middleware or explicit track/track_stream helpers, with ActiveRecord
+  storage, tag-based attribution, price sync tasks, and budget guardrails.
 email:
 - sergey@mm.st
 executables: []
@@ -241,6 +255,7 @@ files:
 - app/helpers/llm_cost_tracker/pagination_helper.rb
 - app/services/llm_cost_tracker/dashboard/data_quality.rb
 - app/services/llm_cost_tracker/dashboard/data_quality_aggregate.rb
+- app/services/llm_cost_tracker/dashboard/date_range.rb
 - app/services/llm_cost_tracker/dashboard/filter.rb
 - app/services/llm_cost_tracker/dashboard/overview_stats.rb
 - app/services/llm_cost_tracker/dashboard/provider_breakdown.rb
@@ -303,6 +318,7 @@ files:
 - lib/llm_cost_tracker/integrations/object_reader.rb
 - lib/llm_cost_tracker/integrations/openai.rb
 - lib/llm_cost_tracker/integrations/registry.rb
+- lib/llm_cost_tracker/integrations/ruby_llm.rb
 - lib/llm_cost_tracker/llm_api_call.rb
 - lib/llm_cost_tracker/logging.rb
 - lib/llm_cost_tracker/middleware/faraday.rb
@@ -322,17 +338,9 @@ files:
 - lib/llm_cost_tracker/price_registry.rb
 - lib/llm_cost_tracker/price_sync.rb
 - lib/llm_cost_tracker/price_sync/fetcher.rb
-- lib/llm_cost_tracker/price_sync/merger.rb
-- lib/llm_cost_tracker/price_sync/model_catalog.rb
-- lib/llm_cost_tracker/price_sync/raw_price.rb
-- lib/llm_cost_tracker/price_sync/refresh_plan_builder.rb
+- lib/llm_cost_tracker/price_sync/registry_diff.rb
 - lib/llm_cost_tracker/price_sync/registry_loader.rb
 - lib/llm_cost_tracker/price_sync/registry_writer.rb
-- lib/llm_cost_tracker/price_sync/source.rb
-- lib/llm_cost_tracker/price_sync/source_result.rb
-- lib/llm_cost_tracker/price_sync/sources/litellm.rb
-- lib/llm_cost_tracker/price_sync/sources/open_router.rb
-- lib/llm_cost_tracker/price_sync/validator.rb
 - lib/llm_cost_tracker/prices.json
 - lib/llm_cost_tracker/pricing.rb
 - lib/llm_cost_tracker/railtie.rb
@@ -343,11 +351,14 @@ files:
 - lib/llm_cost_tracker/retention.rb
 - lib/llm_cost_tracker/storage/active_record_rollups.rb
 - lib/llm_cost_tracker/storage/active_record_store.rb
+- lib/llm_cost_tracker/storage/dispatcher.rb
+- lib/llm_cost_tracker/stream_capture.rb
 - lib/llm_cost_tracker/stream_collector.rb
 - lib/llm_cost_tracker/tag_accessors.rb
 - lib/llm_cost_tracker/tag_context.rb
 - lib/llm_cost_tracker/tag_key.rb
 - lib/llm_cost_tracker/tag_query.rb
+- lib/llm_cost_tracker/tag_sanitizer.rb
 - lib/llm_cost_tracker/tags_column.rb
 - lib/llm_cost_tracker/tracker.rb
 - lib/llm_cost_tracker/unknown_pricing.rb

data/lib/llm_cost_tracker/price_sync/merger.rb

@@ -1,72 +0,0 @@
-# frozen_string_literal: true
-
-module LlmCostTracker
-  module PriceSync
-    class Merger
-      Discrepancy = Data.define(:model, :field, :values)
-
-      PRIORITY_ORDER = %i[litellm openrouter].freeze
-      SUPPLEMENTAL_FIELDS = %i[cache_read_input cache_write_input].freeze
-
-      def merge(results_by_source)
-        prices = collect_prices(results_by_source)
-        discrepancies = []
-
-        merged = prices.group_by(&:model).sort.to_h.transform_values do |candidates|
-          sorted = sort_candidates(candidates)
-          discrepancies.concat(detect_discrepancies(sorted))
-          fill_missing_fields(sorted.first, sorted.drop(1))
-        end
-
-        [merged, discrepancies]
-      end
-
-      private
-
-      def collect_prices(results_by_source)
-        results_by_source.flat_map do |source_name, result|
-          result.prices.map do |price|
-            price.with(source: source_name)
-          end
-        end
-      end
-
-      def sort_candidates(candidates)
-        candidates.sort_by do |price|
-          PRIORITY_ORDER.index(price.source.to_sym) || PRIORITY_ORDER.length
-        end
-      end
-
-      def fill_missing_fields(primary, fallbacks)
-        SUPPLEMENTAL_FIELDS.reduce(primary) do |current, field|
-          next current if current.public_send(field)
-
-          fallback = fallbacks.find { |candidate| candidate.public_send(field) }
-          fallback ? current.with(field => fallback.public_send(field)) : current
-        end
-      end
-
-      def detect_discrepancies(candidates)
-        return [] if candidates.length < 2
-
-        RawPrice::PRICE_FIELDS.filter_map do |field|
-          values = candidates.each_with_object({}) do |price, collected|
-            value = price.public_send(field)
-            collected[price.source] = value unless value.nil?
-          end
-          next if values.size < 2
-          next unless discrepant?(values.values)
-
-          Discrepancy.new(model: candidates.first.model, field: field, values: values)
-        end
-      end
-
-      def discrepant?(values)
-        min, max = values.minmax
-        return max != min if min.to_f.zero?
-
-        ((max - min).abs / min.to_f) >= 0.05
-      end
-    end
-  end
-end

data/lib/llm_cost_tracker/price_sync/model_catalog.rb

@@ -1,77 +0,0 @@
-# frozen_string_literal: true
-
-module LlmCostTracker
-  module PriceSync
-    class ModelCatalog
-      OPENROUTER_PROVIDER_PREFIXES = {
-        openai: %w[openai],
-        anthropic: %w[anthropic],
-        gemini: %w[google]
-      }.freeze
-      LITELLM_PROVIDER_PREFIXES = {
-        openai: [nil, "openai"],
-        anthropic: [nil, "anthropic"],
-        gemini: [nil, "gemini"]
-      }.freeze
-      ALIASES = {
-        "gpt-4o-2024-05-13" => "gpt-4o"
-      }.freeze
-
-      class << self
-        def resolve_from_litellm(our_model, payload)
-          litellm_candidates(our_model).find { |candidate| payload.key?(candidate) }
-        end
-
-        def resolve_from_openrouter(our_model, index)
-          openrouter_candidates(our_model).find { |candidate| index.key?(candidate) }
-        end
-
-        def guess_provider(our_model)
-          case our_model.to_s
-          when /\A(?:gpt-|o1|o3|o4|chatgpt|text-embedding)/
-            :openai
-          when /\Aclaude-/
-            :anthropic
-          when /\Agemini-/
-            :gemini
-          end
-        end
-
-        private
-
-        def litellm_candidates(our_model)
-          provider = guess_provider(our_model)
-          prefixes = LITELLM_PROVIDER_PREFIXES.fetch(provider, [nil])
-
-          model_variants(our_model).flat_map do |variant|
-            prefixes.map { |prefix| prefix ? "#{prefix}/#{variant}" : variant }
-          end.uniq
-        end
-
-        def openrouter_candidates(our_model)
-          provider = guess_provider(our_model)
-          prefixes = OPENROUTER_PROVIDER_PREFIXES.fetch(provider, [])
-
-          model_variants(our_model).flat_map do |variant|
-            prefixes.map { |prefix| "#{prefix}/#{variant}" }
-          end.uniq
-        end
-
-        def model_variants(our_model)
-          model = our_model.to_s
-          canonical = ALIASES.fetch(model, model)
-
-          [model, canonical].flat_map do |variant|
-            [variant, anthropic_version_variant(variant)]
-          end.compact.uniq
-        end
-
-        def anthropic_version_variant(model)
-          return nil unless guess_provider(model) == :anthropic
-
-          model.gsub(/(?<=\d)-(?=\d)/, ".")
-        end
-      end
-    end
-  end
-end

data/lib/llm_cost_tracker/price_sync/raw_price.rb

@@ -1,33 +0,0 @@
-# frozen_string_literal: true
-
-module LlmCostTracker
-  module PriceSync
-    RawPrice = Data.define(
-      :model,
-      :provider,
-      :input,
-      :output,
-      :cache_read_input,
-      :cache_write_input,
-      :source,
-      :source_version,
-      :fetched_at
-    )
-
-    class RawPrice
-      PRICE_FIELDS = %w[input output cache_read_input cache_write_input].freeze
-
-      def to_registry_entry(today:)
-        {
-          "input" => input,
-          "output" => output,
-          "cache_read_input" => cache_read_input,
-          "cache_write_input" => cache_write_input,
-          "_source" => source.to_s,
-          "_source_version" => source_version,
-          "_fetched_at" => fetched_at || today.iso8601
-        }.compact
-      end
-    end
-  end
-end

data/lib/llm_cost_tracker/price_sync/refresh_plan_builder.rb

@@ -1,164 +0,0 @@
-# frozen_string_literal: true
-
-module LlmCostTracker
-  module PriceSync
-    class RefreshPlanBuilder
-      def initialize(sources:, loader: RegistryLoader.new)
-        @sources = sources
-        @loader = loader
-      end
-
-      def call(path:, seed_path:, fetcher:, today:)
-        path = path.to_s
-        registry = loader.call(path: path, seed_path: seed_path)
-        current_models = registry.fetch("models", {})
-        source_results, failed_sources = fetch_all(current_models, fetcher)
-        merged, discrepancies = Merger.new.merge(source_results)
-        validated = Validator.new.validate_batch(merged, existing_registry: current_models)
-        updated_models = apply_changes(current_models, validated.accepted, today)
-
-        PriceSync::RefreshPlan.new(
-          path: path,
-          registry: registry,
-          updated_registry: registry.merge(
-            "metadata" => updated_metadata(
-              registry["metadata"],
-              today,
-              refresh_succeeded: source_results.any? { |_source, result| result.prices.any? },
-              source_results: source_results
-            ),
-            "models" => updated_models
-          ),
-          accepted: validated.accepted,
-          changes: price_changes(current_models, updated_models),
-          orphaned_models: compute_orphaned(current_models, merged.keys, source_results),
-          failed_sources: failed_sources,
-          discrepancies: discrepancies,
-          rejected: validated.rejected,
-          flagged: validated.flagged,
-          sources_used: source_usage(source_results),
-          source_results: source_results
-        )
-      end
-
-      private
-
-      attr_reader :sources, :loader
-
-      def fetch_all(current_models, fetcher)
-        results = {}
-        failures = {}
-
-        sources.each do |source|
-          results[source.name.to_sym] = source.fetch(current_models: current_models, fetcher: fetcher)
-        rescue Error => e
-          failures[source.name.to_sym] = e.message
-        end
-
-        [results, failures]
-      end
-
-      def apply_changes(current_models, accepted, today)
-        merged = seed_models(current_models)
-
-        accepted.each do |model, price|
-          next if manual_model?(merged[model])
-
-          merged[model] = registry_entry_for(merged[model], price, today)
-        end
-
-        merged.sort.to_h
-      end
-
-      def compute_orphaned(current_models, merged_models, source_results)
-        return [] if source_results.empty?
-
-        seed_models(current_models).keys.reject do |model|
-          manual_model?(current_models[model]) || merged_models.include?(model)
-        end.sort
-      end
-
-      def seed_models(current_models)
-        normalize_models(current_models).transform_values do |entry|
-          next entry if entry.key?("_source")
-
-          entry.merge("_source" => "seed")
-        end
-      end
-
-      def normalize_models(models)
-        normalize_hash(models).each_with_object({}) do |(model, entry), normalized|
-          normalized[model.to_s] = normalize_hash(entry)
-        end
-      end
-
-      def normalize_hash(hash)
-        return {} if hash.nil?
-        raise ArgumentError, "price sync entries must be hashes" unless hash.is_a?(Hash)
-
-        hash.each_with_object({}) do |(key, value), normalized|
-          normalized[key.to_s] = value
-        end
-      end
-
-      def manual_model?(entry)
-        normalize_hash(entry)["_source"] == "manual"
-      end
-
-      def registry_entry_for(existing_entry, price, today)
-        normalize_hash(existing_entry)
-          .except(*PriceRegistry::PRICE_KEYS)
-          .merge(price.to_registry_entry(today: today))
-      end
-
-      def updated_metadata(existing, today, refresh_succeeded:, source_results:)
-        metadata = normalize_hash(existing)
-        metadata["currency"] ||= "USD"
-        metadata["unit"] ||= "1M tokens"
-        return metadata unless refresh_succeeded
-
-        metadata["updated_at"] = today.iso8601
-        metadata["source_urls"] = source_urls(source_results)
-        metadata
-      end
-
-      def source_usage(source_results)
-        source_results.transform_values do |result|
-          PriceSync::SourceUsage.new(prices_count: result.prices.size, source_version: result.source_version)
-        end
-      end
-
-      def price_changes(current_models, updated_models)
-        current_models = normalize_models(current_models)
-        updated_models = normalize_models(updated_models)
-
-        (current_models.keys | updated_models.keys).sort.each_with_object({}) do |model, changes|
-          fields = price_field_changes(current_models[model], updated_models[model])
-          changes[model] = fields if fields.any?
-        end
-      end
-
-      def price_field_changes(current_entry, updated_entry)
-        current_price = comparable_price(current_entry)
-        updated_price = comparable_price(updated_entry)
-
-        (current_price.keys | updated_price.keys).sort.each_with_object({}) do |field, changes|
-          from = current_price[field]
-          to = updated_price[field]
-          next if from == to
-
-          changes[field] = { "from" => from, "to" => to }
-        end
-      end
-
-      def comparable_price(entry)
-        normalize_hash(entry).slice(*PriceRegistry::PRICE_KEYS)
-      end
-
-      def source_urls(source_results)
-        names = source_results.keys.map(&:to_sym)
-        sources.select { |source| names.include?(source.name.to_sym) }.map(&:url)
-      end
-    end
-  end
-end

data/lib/llm_cost_tracker/price_sync/source.rb

@@ -1,29 +0,0 @@
-# frozen_string_literal: true
-
-module LlmCostTracker
-  module PriceSync
-    class Source
-      def fetch(current_models:, fetcher:)
-        raise NotImplementedError
-      end
-
-      def name
-        self.class.name.split("::").last.downcase.to_sym
-      end
-
-      def priority
-        100
-      end
-
-      def url
-        raise NotImplementedError
-      end
-
-      private
-
-      def response_version(response)
-        response.source_version
-      end
-    end
-  end
-end

data/lib/llm_cost_tracker/price_sync/source_result.rb

@@ -1,7 +0,0 @@
-# frozen_string_literal: true
-
-module LlmCostTracker
-  module PriceSync
-    SourceResult = Data.define(:prices, :missing_models, :source_version)
-  end
-end