feat/fix model download

.gitignore

@@ -6,3 +6,4 @@
 .rspec_status
 Gemfile.lock
 /onnx/onnx/
+.cache

README.md

@@ -19,11 +19,10 @@ require "gliner"
 
 Gliner.configure do |config|
   config.threshold = 0.2
-  # If unset, auto! downloads the default model to .cache/
+  # By default, the gem downloads the default model to .cache/
   # Or set a local path explicitly:
   # config.model = "/path/to/gliner2-multi-v1"
   config.variant = :fp16
-  config.auto!
 end
 
 text = "Apple CEO Tim Cook announced iPhone 15 in Cupertino yesterday."

bin/console

@@ -1,4 +1,5 @@
 #!/usr/bin/env ruby
+
 # frozen_string_literal: true
 
 begin
@@ -7,53 +8,20 @@ rescue LoadError
 end
 
 require "gliner"
-require "fileutils"
-require "httpx"
 require "irb"
 
-DEFAULT_REPO_ID = "cuerbot/gliner2-multi-v1"
-DEFAULT_MODEL_FILE = "model_fp16.onnx"
-DEFAULT_MODEL_SUBDIR = "onnx"
-
-def ensure_model_dir!(repo_id:, model_file:, model_subdir:)
-  dir = File.expand_path("../tmp/models/#{repo_id.tr('/', '__')}", __dir__)
-  FileUtils.mkdir_p(dir)
-
-  base = "https://huggingface.co/#{repo_id}/resolve/main"
-  base = "#{base}/#{model_subdir}" unless model_subdir.nil? || model_subdir.empty?
-  files = ["tokenizer.json", "config.json", model_file]
-
-  files.each do |file|
-    dest = File.join(dir, file)
-    next if File.exist?(dest) && File.size?(dest)
-    download("#{base}/#{file}", dest)
-  end
-
-  dir
-end
-
-def download(url, dest)
-  response = HTTPX.get(url)
-  raise "Download failed: #{url} (status: #{response.status})" unless response.status.between?(200, 299)
-
-  File.binwrite(dest, response.body.to_s)
-end
-
 model_dir = ARGV[0] || ENV["GLINER_MODEL_DIR"]
-repo_id = ENV["GLINER_REPO_ID"] || DEFAULT_REPO_ID
-model_file = ENV["GLINER_MODEL_FILE"] || DEFAULT_MODEL_FILE
-model_subdir = ENV["GLINER_MODEL_SUBDIR"] || DEFAULT_MODEL_SUBDIR
+model_file = ENV["GLINER_MODEL_FILE"]
 
 if model_dir && !model_dir.empty?
-  $gliner_model = Gliner.load(model_dir, file: model_file)
+  $gliner_model = model_file ? Gliner.load(model_dir, file: model_file) : Gliner.load(model_dir)
 else
   begin
-    require "fileutils"
-    model_dir = ensure_model_dir!(repo_id: repo_id, model_file: model_file, model_subdir: model_subdir)
-    $gliner_model = Gliner.load(model_dir, file: model_file)
+    Gliner.configure { |config| config.auto = true }
+    $gliner_model = Gliner.model
   rescue => e
     warn "No model loaded (auto-download failed: #{e.class}: #{e.message})"
-    warn "Set GLINER_MODEL_DIR to a local model dir, or set GLINER_REPO_ID/GLINER_MODEL_FILE for auto-download."
+    warn "Set GLINER_MODEL_DIR or configure Gliner.config.model to a local model dir."
   end
 end
 
@@ -79,6 +47,5 @@ puts "- helper: gliner_classify(text, tasks)"
 puts "- helper: gliner_extract_json(text, structures)"
 puts "- model variable: $gliner_model"
 puts "- model dir: #{model_dir.inspect}"
-puts "- auto-download env: GLINER_REPO_ID=#{repo_id.inspect} GLINER_MODEL_FILE=#{model_file.inspect}" unless $gliner_model
 
 IRB.start(__FILE__)

gliner.gemspec

@@ -21,6 +21,7 @@ Gem::Specification.new do |spec|
   spec.add_dependency 'tokenizers', '~> 0.6'
 
   spec.add_development_dependency 'rake', '~> 13.0'
+  spec.add_development_dependency 'irb', '~> 1.16.0'
   spec.add_development_dependency 'rspec', '~> 3.13'
   spec.add_development_dependency 'rubocop', '~> 1.50'
 

lib/gliner.rb

@@ -6,6 +6,7 @@
 require 'gliner/configuration'
 require 'gliner/model'
 require 'gliner/runners/prepared_task'
+require 'gliner/runners/inspectable'
 require 'gliner/runners/entity_runner'
 require 'gliner/runners/structured_runner'
 require 'gliner/runners/classification_runner'
@@ -68,7 +69,10 @@ def load(dir, file: nil)
     end
 
     def model
-      @model ||= model_from_config || model_from_env
+      @model ||= begin
+        apply_model_source!
+        model_from_config || model_from_env
+      end
     end
 
     def [](config)
@@ -98,7 +102,7 @@ def model_from_config
     end
 
     def model_from_env
-      dir = ENV.fetch('GLINER_MODEL_DIR', nil)
+      dir = env_model_dir
       return if dir.nil?
 
       file = ENV['GLINER_MODEL_FILE'] || model_file_for_variant(config.variant)
@@ -130,7 +134,8 @@ def apply_model_source!
       return unless config.auto?
 
       source = config.model
-      return unless source.nil? || source.empty?
+      return unless source.nil?
+      return if env_model_dir
 
       config.model = download_default_model
     end
@@ -143,15 +148,32 @@ def download_default_model
       FileUtils.mkdir_p(dir)
 
       files = ['tokenizer.json', 'config.json', model_file]
-      client = HTTPX.plugin(:follow_redirects)
+      client = HTTPX.plugin(:follow_redirects).with(max_redirects: 5)
 
       files.each do |file|
-        response = client.get("#{DEFAULT_MODEL_BASE}/#{file}")
-        raise Error, "Download failed: #{file}" if response.error?
+        dest = File.join(dir, file)
+        next if File.exist?(dest) && File.size?(dest)
+        download_file!(client, "#{DEFAULT_MODEL_BASE}/#{file}", dest)
+      end
+
+      dir
+    end
 
-        File.binwrite(File.join(dir, file), response.body.to_s)
+    def download_file!(client, url, dest)
+      response = client.get(url)
+      status = response.status
+
+      unless status && status.between?(200, 299)
+        raise Error, "Download failed: #{url} (status: #{status || 'unknown'})"
       end
 
+      File.binwrite(dest, response.body.to_s)
+    end
+
+    def env_model_dir
+      dir = ENV.fetch('GLINER_MODEL_DIR', nil)
+      return nil if dir.nil? || dir.empty?
+
       dir
     end
 

lib/gliner/configuration.rb

@@ -11,7 +11,7 @@ def initialize
       @threshold = DEFAULT_THRESHOLD
       @model = nil
       @variant = :fp16
-      @auto = false
+      @auto = true
     end
 
     def variant=(value)

lib/gliner/runners/classification_runner.rb

@@ -3,6 +3,8 @@
 module Gliner
   module Runners
     class ClassificationRunner
+      include Inspectable
+
       def self.[](tasks)
         new(Gliner.model!, tasks)
       end
@@ -21,6 +23,11 @@ def [](text, **options)
       end
 
       alias call []
+
+      private
+
+      def inspect_label = 'Classification'
+      def inspect_items = @tasks.keys
     end
   end
 end

lib/gliner/runners/entity_runner.rb

@@ -3,8 +3,12 @@
 module Gliner
   module Runners
     class EntityRunner
+      include Inspectable
+
       def initialize(model, config)
         parsed = model.entity_task.parse_config(config)
+
+        @labels = parsed[:labels]
         @task = PreparedTask.new(model.entity_task, parsed)
       end
 
@@ -14,6 +18,11 @@ def [](text, **options)
       end
 
       alias call []
+
+      private
+
+      def inspect_label = 'Entity'
+      def inspect_items = @labels
     end
   end
 end

lib/gliner/runners/inspectable.rb

@@ -0,0 +1,13 @@
+# frozen_string_literal: true
+
+module Gliner
+  module Runners
+    module Inspectable
+      def inspect
+        items = Array(inspect_items).map(&:to_s)
+
+        "#<Gliner(#{inspect_label}) input=#{items.inspect}>"
+      end
+    end
+  end
+end

lib/gliner/runners/structured_runner.rb

@@ -3,6 +3,8 @@
 module Gliner
   module Runners
     class StructuredRunner
+      include Inspectable
+
       def initialize(model, config)
         @tasks = build_tasks(model, config)
       end
@@ -17,6 +19,9 @@ def [](text, **options)
 
       private
 
+      def inspect_label = 'Structure'
+      def inspect_items = @tasks.keys
+
       def build_tasks(model, config)
         raise Error, 'structures must be a Hash' unless config.is_a?(Hash)
 

spec/gliner/configuration_spec.rb

@@ -72,5 +72,35 @@
         expect(Gliner).not_to have_received(:download_default_model)
       end
     end
+
+    it 'auto! does not download when model is explicitly set' do
+      allow(Gliner).to receive(:download_default_model)
+
+      Gliner.configure do |config|
+        config.model = ''
+        config.auto!
+      end
+
+      expect(Gliner.config.model).to eq('')
+      expect(Gliner).not_to have_received(:download_default_model)
+    end
+
+    it 'auto! respects GLINER_MODEL_DIR when set' do
+      allow(Gliner).to receive(:download_default_model)
+
+      previous = ENV['GLINER_MODEL_DIR']
+      ENV['GLINER_MODEL_DIR'] = '/tmp/env-model'
+
+      begin
+        Gliner.configure do |config|
+          config.auto!
+        end
+
+        expect(Gliner.config.model).to be_nil
+        expect(Gliner).not_to have_received(:download_default_model)
+      ensure
+        ENV['GLINER_MODEL_DIR'] = previous
+      end
+    end
   end
 end