Release/v5.0.0 alpha.1

.github/workflows/_test-cli.yml

@@ -0,0 +1,47 @@
+name: CLI Test
+
+on:
+  workflow_call:
+  workflow_dispatch:
+
+env:
+  MINDEE_API_KEY: ${{ secrets.MINDEE_API_KEY_SE_TESTS }}
+  MINDEE_V2_API_KEY: ${{ secrets.MINDEE_V2_SE_TESTS_API_KEY }}
+  MINDEE_V2_SE_TESTS_FINDOC_MODEL_ID: ${{ secrets.MINDEE_V2_SE_TESTS_FINDOC_MODEL_ID }}
+
+jobs:
+  test:
+    name: Run Tests
+    timeout-minutes: 30
+    strategy:
+      max-parallel: 4
+      matrix:
+        os_config:
+          - os: "ubuntu-24.04"
+            rid: "linux-x64"
+          - os: "macos-latest"
+            rid: "osx-x64"
+          - os: "windows-latest"
+            rid: "win-x64"
+        ruby:
+          - "3.0"
+          - "4.0"
+    runs-on: ${{ matrix.os_config.os }}
+    steps:
+      - uses: actions/checkout@v4
+        with:
+          submodules: recursive
+
+      - name: set up Ruby ${{ matrix.ruby }}
+        uses: ruby/setup-ruby@v1
+        with:
+          ruby-version: ${{ matrix.ruby }}
+          bundler-cache: true
+
+      - name: Tests V2 CLI
+        run: |
+          ./spec/test_v2_cli.sh ./spec/data/file_types/pdf/blank_1.pdf ${{ matrix.os_config.rid }} 
+
+      - name: Tests V1 CLI
+        run: |
+          ./spec/test_v1_cli.sh ./spec/data/file_types/pdf/blank_1.pdf ${{ matrix.os_config.rid }}

.github/workflows/_test-integrations.yml

@@ -7,6 +7,19 @@ on:
   workflow_call:
   workflow_dispatch:
 
+env:
+  MINDEE_API_KEY: ${{ secrets.MINDEE_API_KEY_SE_TESTS }}
+  WORKFLOW_ID: ${{ secrets.WORKFLOW_ID_SE_TESTS }}
+  MINDEE_V2_API_KEY: ${{ secrets.MINDEE_V2_SE_TESTS_API_KEY }}
+  MINDEE_V2_FINDOC_MODEL_ID: ${{ secrets.MINDEE_V2_SE_TESTS_FINDOC_MODEL_ID }}
+  MINDEE_V2_SE_TESTS_BLANK_PDF_URL: ${{ secrets.MINDEE_V2_SE_TESTS_BLANK_PDF_URL }}
+  MINDEE_V2_SE_TESTS_FINDOC_MODEL_ID: ${{ secrets.MINDEE_V2_SE_TESTS_FINDOC_MODEL_ID }}
+  MINDEE_V2_SE_TESTS_CLASSIFICATION_MODEL_ID: ${{ secrets.MINDEE_V2_SE_TESTS_CLASSIFICATION_MODEL_ID }}
+  MINDEE_V2_SE_TESTS_CROP_MODEL_ID: ${{ secrets.MINDEE_V2_SE_TESTS_CROP_MODEL_ID }}
+  MINDEE_V2_SE_TESTS_OCR_MODEL_ID: ${{ secrets.MINDEE_V2_SE_TESTS_OCR_MODEL_ID }}
+  MINDEE_V2_SE_TESTS_SPLIT_MODEL_ID: ${{ secrets.MINDEE_V2_SE_TESTS_SPLIT_MODEL_ID }}
+  MINDEE_LOG_LEVEL: DEBUG
+
 jobs:
   integration-tests:
     name: Run Integration Tests
@@ -50,17 +63,6 @@ jobs:
           sudo sed -i "s/$SRC/$RPL/" /etc/ImageMagick-6/policy.xml
 
       - name: Run Rspec for integration tests
-        env:
-          MINDEE_API_KEY: ${{ secrets.MINDEE_API_KEY_SE_TESTS }}
-          WORKFLOW_ID: ${{ secrets.WORKFLOW_ID_SE_TESTS }}
-          MINDEE_V2_API_KEY: ${{ secrets.MINDEE_V2_SE_TESTS_API_KEY }}
-          MINDEE_V2_FINDOC_MODEL_ID: ${{ secrets.MINDEE_V2_SE_TESTS_FINDOC_MODEL_ID }}
-          MINDEE_V2_SE_TESTS_BLANK_PDF_URL: ${{ secrets.MINDEE_V2_SE_TESTS_BLANK_PDF_URL }}
-          MINDEE_V2_SE_TESTS_CLASSIFICATION_MODEL_ID: ${{ secrets.MINDEE_V2_SE_TESTS_CLASSIFICATION_MODEL_ID }}
-          MINDEE_V2_SE_TESTS_CROP_MODEL_ID: ${{ secrets.MINDEE_V2_SE_TESTS_CROP_MODEL_ID }}
-          MINDEE_V2_SE_TESTS_OCR_MODEL_ID: ${{ secrets.MINDEE_V2_SE_TESTS_OCR_MODEL_ID }}
-          MINDEE_V2_SE_TESTS_SPLIT_MODEL_ID: ${{ secrets.MINDEE_V2_SE_TESTS_SPLIT_MODEL_ID }}
-          MINDEE_LOG_LEVEL: DEBUG
         run: |
           bundle exec rake integration
 

.github/workflows/_test-code-samples.yml → .github/workflows/_test-smoke.yml

@@ -1,4 +1,4 @@
-name: Test Code Samples
+name: Smoke Test
 
 on:
   workflow_call:
@@ -9,12 +9,13 @@ env:
   MINDEE_ACCOUNT_SE_TESTS: ${{ secrets.MINDEE_ACCOUNT_SE_TESTS }}
   MINDEE_API_KEY_SE_TESTS: ${{ secrets.MINDEE_API_KEY_SE_TESTS }}
   MINDEE_V2_API_KEY: ${{ secrets.MINDEE_V2_SE_TESTS_API_KEY }}
-  MINDEE_V2_SE_TESTS_FINDOC_MODEL_ID: ${{ secrets.MINDEE_V2_SE_TESTS_FINDOC_MODEL_ID }}
-  MINDEE_V2_SE_TESTS_FAILURE_WEBHOOK_ID: ${{ secrets.MINDEE_V2_SE_TESTS_FAILURE_WEBHOOK_ID }}
-  MINDEE_V2_SE_TESTS_CLASSIFICATION_MODEL_ID: ${{ secrets.MINDEE_V2_SE_TESTS_CLASSIFICATION_MODEL_ID }}
-  MINDEE_V2_SE_TESTS_CROP_MODEL_ID: ${{ secrets.MINDEE_V2_SE_TESTS_CROP_MODEL_ID }}
-  MINDEE_V2_SE_TESTS_OCR_MODEL_ID: ${{ secrets.MINDEE_V2_SE_TESTS_OCR_MODEL_ID }}
-  MINDEE_V2_SE_TESTS_SPLIT_MODEL_ID: ${{ secrets.MINDEE_V2_SE_TESTS_SPLIT_MODEL_ID }}
+  MINDEE_V2_FINDOC_MODEL_ID: ${{ secrets.MINDEE_V2_SE_TESTS_FINDOC_MODEL_ID }}
+  MINDEE_V2_SE_TESTS_BLANK_PDF_URL: ${{ secrets.MINDEE_V2_SE_TESTS_BLANK_PDF_URL }}
+  MINDEE_V2_CLASSIFICATION_MODEL_ID: ${{ secrets.MINDEE_V2_SE_TESTS_CLASSIFICATION_MODEL_ID }}
+  MINDEE_V2_CROP_MODEL_ID: ${{ secrets.MINDEE_V2_SE_TESTS_CROP_MODEL_ID }}
+  MINDEE_V2_FAILURE_WEBHOOK_ID: ${{ secrets.MINDEE_V2_SE_TESTS_FAILURE_WEBHOOK_ID }}
+  MINDEE_V2_OCR_MODEL_ID: ${{ secrets.MINDEE_V2_SE_TESTS_OCR_MODEL_ID }}
+  MINDEE_V2_SPLIT_MODEL_ID: ${{ secrets.MINDEE_V2_SE_TESTS_SPLIT_MODEL_ID }}
 
 jobs:
   test:
@@ -40,14 +41,15 @@ jobs:
           ruby-version: ${{ matrix.ruby }}
           bundler-cache: true
 
-      - name: Tests V1 code samples
+      - name: Tests V2 code samples
         env:
           MINDEE_LOG_LEVEL: DEBUG
         run: |
-          ./spec/test_code_samples_v1.sh
+          ./spec/test_v2_code_samples.sh
 
-      - name: Tests V2 code samples
+      - name: Tests V1 code samples
         env:
           MINDEE_LOG_LEVEL: DEBUG
         run: |
-          ./spec/test_code_samples_v2.sh
+          ./spec/test_v1_code_samples.sh ${{ secrets.MINDEE_ACCOUNT_SE_TESTS }} ${{ secrets.MINDEE_ENDPOINT_SE_TESTS }}
+

.github/workflows/cron.yml

@@ -5,6 +5,9 @@ on:
     - cron:  '0 0 * * *'
 
 jobs:
-  test_code_samples:
-    uses: mindee/mindee-api-ruby/.github/workflows/_test-code-samples.yml@main
+  test-smoke:
+    uses: mindee/mindee-api-ruby/.github/workflows/_test-smoke.yml@main
+    secrets: inherit
+  test-cli:
+    uses: mindee/mindee-api-ruby/.github/workflows/_test-cli.yml@main
     secrets: inherit

.github/workflows/publish-release.yml

@@ -5,13 +5,13 @@ on:
     types: [ published ]
 
 jobs:
-  publish_docs:
+  publish-docs:
     uses: mindee/mindee-api-ruby/.github/workflows/_publish-docs.yml@main
     secrets: inherit
-  publish_guide:
+  publish-guide:
     uses: mindee/mindee-api-ruby/.github/workflows/_publish-guide.yml@main
-    needs: publish_docs
+    needs: publish-docs
     secrets: inherit
-  publish_code:
+  publish-code:
     uses: mindee/mindee-api-ruby/.github/workflows/_publish-code.yml@main
     secrets: inherit

.github/workflows/pull-request.yml

@@ -8,17 +8,21 @@ permissions:
   pull-requests: read
 
 jobs:
-  static_analysis:
+  static-analysis:
     uses: ./.github/workflows/_static-analysis.yml
-  test_units:
+  test-units:
     uses: ./.github/workflows/_test-units.yml
-    needs: static_analysis
+    needs: static-analysis
     secrets: inherit
-  test_integrations:
+  test-integrations:
     uses: ./.github/workflows/_test-integrations.yml
-    needs: test_units
+    needs: test-units
     secrets: inherit
-  test_code_samples:
-    uses: ./.github/workflows/_test-code-samples.yml
-    needs: test_units
+  test-smoke:
+    uses: ./.github/workflows/_test-smoke.yml
+    needs: test-units
+    secrets: inherit
+  test-cli:
+    uses: ./.github/workflows/_test-cli.yml
+    needs: test-units
     secrets: inherit

.github/workflows/push-main-branch.yml

@@ -6,15 +6,15 @@ on:
       - main
 
 jobs:
-  static_analysis:
+  static-analysis:
     uses: mindee/mindee-api-ruby/.github/workflows/_static-analysis.yml@main
-  test_units:
+  test-units:
     uses: mindee/mindee-api-ruby/.github/workflows/_test-units.yml@main
-    needs: static_analysis
+    needs: static-analysis
     secrets: inherit
   tag:
     uses: mindee/client-lib-actions/.github/workflows/tag-version.yml@main
-    needs: test_units
+    needs: test-units
   release:
     uses: mindee/client-lib-actions/.github/workflows/create-release.yml@main
     needs: tag

.rubocop.yml

@@ -37,7 +37,8 @@ Metrics/MethodLength:
 
 Metrics/ClassLength:
   Max: 200
-
+Metrics/ModuleLength:
+  Max: 200
 Metrics/ParameterLists:
   Max: 8
 

bin/mindee.rb

@@ -3,127 +3,28 @@
 
 require 'bundler/setup'
 require 'optparse'
-require 'mindee'
-
-require_relative 'cli_products'
-
-options = {}
-
-# Initializes universal-specific options
-# @param cli_parser [OptionParser]
-def custom_subcommand(cli_parser, options)
-  cli_parser.on('-v [VERSION]', '--version [VERSION]', 'Model version for the API') do |v|
-    options[:endpoint_version] = v
-  end
-  cli_parser.on('-a ACCOUNT_NAME', '--account ACCOUNT_NAME', 'API account name for the endpoint') do |v|
-    options[:account_name] = v
-  end
-end
-
-product_parser = {}
-PRODUCTS.each do |doc_key, doc_value|
-  product_parser[doc_key] = OptionParser.new do |opts|
-    opts.on('-w', '--all-words', 'Include words in response') do |v|
-      options[:all_words] = v
-    end
-    opts.on('-c', '--cut-pages', "Cut document pages") do |v|
-      options[:cut_pages] = v
-    end
-    opts.on('-k [KEY]', '--key [KEY]', 'API key for the endpoint') do |v|
-      options[:api_key] = v
-    end
-    opts.on('-f', '--full', "Print the full data, including pages") do |v|
-      options[:print_full] = true
-    end
-    opts.on('-F', '--fix-pdf', "Attempts to fix broken PDF files before sending them to the server.") do |v|
-      options[:repair_pdf] = true
-    end
-    if doc_key != 'universal'
-      opts.banner = "#{doc_value[:description]}. \nUsage: \nmindee.rb universal [options] endpoint_name file\nor\nmindee.rb universal [options] endpoint_name file"
-      custom_subcommand(opts, options)
-    end
-    if doc_value[:async]
-      if doc_value[:sync]
-        opts.on("-A", "--async", "Call asynchronously") do |v|
-          options[:parse_async] = v
-        end
-      end
-    end
+require_relative 'v1/parser'
+require_relative 'v2/parser'
+
+def setup_main_parser
+  v1_parser = MindeeCLI::V1Parser.new(ARGV)
+  v2_parser = MindeeCLI::V2Parser.new(ARGV)
+  main_parser = OptionParser.new do |opts|
+    opts.banner = "Usage: mindee [command]"
+    opts.separator "Commands:"
+    opts.separator "  v1    Use Version 1 of the Mindee API"
+    opts.separator "  v2    Use Version 2 of the Mindee API"
   end
-end
-
-global_parser = OptionParser.new do |opts|
-  opts.banner = "Usage: mindee.rb product [options] file"
-  opts.separator "Available products:"
-  opts.separator "  #{PRODUCTS.keys.join("\n  ")}"
-end
-
-command = ARGV.shift
-unless PRODUCTS.include?(command)
-  abort(global_parser.help)
-end
-doc_class = PRODUCTS[command][:doc_class]
-product_parser[command].parse!
+  main_command = ARGV.shift
 
-if command == 'universal'
-  if ARGV.length < 2
-    $stderr.puts "The '#{command}' command requires both ENDPOINT_NAME and file arguments."
-    abort(product_parser[command].help)
-  end
-  endpoint_name = ARGV[0]
-  options[:file_path] = ARGV[1]
-else
-  if ARGV.length < 1
-    $stderr.puts "file missing"
-    abort(product_parser[command].help)
-  end
-  endpoint_name = nil
-  options[:file_path] = ARGV[0]
-end
-
-mindee_client = Mindee::Client.new(api_key: options[:api_key])
-if options[:file_path].start_with?("https://")
-  input_source = mindee_client.source_from_url(options[:file_path])
-else
-  input_source = mindee_client.source_from_path(options[:file_path], repair_pdf: options[:repair_pdf])
-end
-
-if command == 'universal'
-  custom_endpoint = mindee_client.create_endpoint(
-    endpoint_name: endpoint_name,
-    account_name: options[:account_name],
-    version: options[:endpoint_version].nil? ? "1" : options[:endpoint_version]
-  )
-else
-  custom_endpoint = nil
-end
-
-if options[:cut_pages].nil? || !options[:cut_pages].is_a?(Integer) || options[:cut_pages] < 0
-  page_options = nil
-else
-  page_options = Mindee::PageOptions.new(params: {
-    page_indexes: (0..options[:cut_pages].to_i).to_a,
-    operation: :KEEP_ONLY,
-    on_min_pages: 0,
-  })
-end
-
-if options[:parse_async].nil?
-  if !PRODUCTS[command][:sync]
-    options[:parse_async] = true
+  case main_command
+  when 'v1'
+    v1_parser.execute
+  when 'v2'
+    v2_parser.execute
   else
-    options[:parse_async] = false
+    abort(main_parser.help)
   end
 end
-result = mindee_client.parse(
-  input_source,
-  doc_class,
-  options: { endpoint: custom_endpoint,
-             options: Mindee::ParseOptions.new(params: { page_options: page_options }), enqueue: options[:parse_async] }
-)
 
-if options[:print_full]
-  puts result.document
-else
-  puts result.document.inference.prediction
-end
+setup_main_parser