✨ add support for remote resource fetching

bin/mindee.rb

@@ -97,12 +97,6 @@
     sync: false,
     async: true,
   },
-  "fr-carte-vitale" => {
-    description: "FR Carte Vitale",
-    doc_class: Mindee::Product::FR::CarteVitale::CarteVitaleV1,
-    sync: true,
-    async: false,
-  },
   "fr-id-card" => {
     description: "FR ID Card",
     doc_class: Mindee::Product::FR::IdCard::IdCardV2,

lib/mindee/input/sources/url_input_source.rb

@@ -1,5 +1,9 @@
 # frozen_string_literal: true
 
+require 'net/http'
+require 'uri'
+require 'fileutils'
+
 module Mindee
   module Input
     module Source
@@ -13,6 +17,110 @@ def initialize(url)
 
           @url = url
         end
+
+        # Downloads the file from the URL and saves it to the specified path.
+        #
+        # @param path [String] Path to save the file to.
+        # @param filename [String, nil] Optional name to give to the file.
+        # @param username [String, nil] Optional username for authentication.
+        # @param password [String, nil] Optional password for authentication.
+        # @param token [String, nil] Optional token for JWT-based authentication.
+        # @param max_redirects [Integer] Maximum amount of redirects to follow.
+        # @return [String] The full path of the saved file.
+        def save_to_file(path, filename: nil, username: nil, password: nil, token: nil, max_redirects: 3)
+          response_body = fetch_file_content(username: username, password: password, token: token,
+                                             max_redirects: max_redirects)
+
+          filename = fill_filename(filename)
+
+          full_path = File.join(path.chomp('/'), filename)
+          File.write(full_path, response_body)
+
+          full_path
+        end
+
+        # Downloads the file from the url, and returns a BytesInputSource wrapper object for it.
+        #
+        # @param filename [String, nil] Optional name to give to the file.
+        # @param username [String, nil] Optional username for authentication.
+        # @param password [String, nil] Optional password for authentication.
+        # @param token [String, nil] Optional token for JWT-based authentication.
+        # @param max_redirects [Integer] Maximum amount of redirects to follow.
+        # @return [BytesInputSource] The full path of the saved file.
+        def as_local_input_source(filename: nil, username: nil, password: nil, token: nil, max_redirects: 3)
+          filename = fill_filename(filename)
+          response_body = fetch_file_content(username: username, password: password, token: token,
+                                             max_redirects: max_redirects)
+          bytes = StringIO.new(response_body)
+
+          BytesInputSource.new(bytes.read, filename)
+        end
+
+        # Fetches the file content from the URL.
+        #
+        # @param username [String, nil] Optional username for authentication.
+        # @param password [String, nil] Optional password for authentication.
+        # @param token [String, nil] Optional token for JWT-based authentication.
+        # @param max_redirects [Integer] Maximum amount of redirects to follow.
+        # @return [String] The downloaded file content.
+        def fetch_file_content(username: nil, password: nil, token: nil, max_redirects: 3)
+          uri = URI.parse(@url)
+          request = Net::HTTP::Get.new(uri)
+
+          request['Authorization'] = "Bearer #{token}" if token
+          request.basic_auth(username, password) if username && password
+
+          response = make_request(uri, request, max_redirects)
+          if response.code.to_i > 299
+            raise "Failed to download file: HTTP status code #{response.code}"
+          elsif response.code.to_i < 200
+            raise "Failed to download file: Invalid response code #{response.code}."
+          end
+
+          response.body
+        end
+
+        private
+
+        def extract_filename_from_url(uri)
+          filename = File.basename(uri.path)
+          filename.empty? ? '' : filename
+        end
+
+        def fill_filename(filename)
+          filename ||= extract_filename_from_url(URI.parse(@url))
+          if filename.empty? || File.extname(filename).empty?
+            filename = generate_file_name(extension: get_file_extension(filename))
+          end
+          filename
+        end
+
+        def make_request(uri, request, max_redirects)
+          Net::HTTP.start(uri.hostname, uri.port, use_ssl: true) do |http|
+            response = http.request(request)
+            if response.is_a?(Net::HTTPRedirection) && max_redirects.positive?
+              location = response['location']
+              raise 'No location in redirection header.' if location.nil?
+
+              new_uri = URI.parse(location)
+              request = Net::HTTP::Get.new(new_uri)
+              make_request(new_uri, request, max_redirects - 1)
+            else
+              response
+            end
+          end
+        end
+
+        def get_file_extension(filename)
+          ext = File.extname(filename)
+          ext.empty? ? nil : ext.downcase
+        end
+
+        def generate_file_name(extension: nil)
+          extension ||= '.tmp'
+          random_string = Array.new(8) { rand(36).to_s(36) }.join
+          "mindee_temp_#{Time.now.strftime('%Y-%m-%d_%H-%M-%S')}_#{random_string}#{extension}"
+        end
       end
     end
   end

lib/mindee/product.rb

@@ -13,7 +13,6 @@
 require_relative 'product/fr/bank_account_details/bank_account_details_v2'
 require_relative 'product/fr/bank_statement/bank_statement_v1'
 require_relative 'product/fr/carte_grise/carte_grise_v1'
-require_relative 'product/fr/carte_vitale/carte_vitale_v1'
 require_relative 'product/fr/id_card/id_card_v1'
 require_relative 'product/fr/id_card/id_card_v2'
 require_relative 'product/fr/energy_bill/energy_bill_v1'

spec/input/sources_spec.rb

@@ -50,18 +50,6 @@
     end
   end
 
-  context 'A remote url file' do
-    it 'should not send an invalid URL' do
-      expect do
-        Mindee::Input::Source::UrlInputSource.new('http://invalid-url')
-      end.to raise_error 'URL must be HTTPS'
-    end
-    it 'should send a valid URL' do
-      input = Mindee::Input::Source::UrlInputSource.new('https://platform.mindee.com')
-      expect(input.url).to eq('https://platform.mindee.com')
-    end
-  end
-
   context 'A broken fixable PDF' do
     mindee_client = Mindee::Client.new(api_key: 'invalid-api-key')
     it 'Should not raise a mime error' do

spec/input/url_input_source_integration.rb

@@ -0,0 +1,18 @@
+# frozen_string_literal: true
+
+require 'rspec'
+require 'mindee'
+
+describe Mindee::Input::Source::UrlInputSource do
+  it 'retrieves response from a remote file' do
+    api_key = ENV.fetch('MINDEE_API_KEY', nil)
+    client = Mindee::Client.new(api_key: api_key)
+    remote_input = Mindee::Input::Source::UrlInputSource.new('https://github.com/mindee/client-lib-test-data/blob/main/products/invoice_splitter/invoice_5p.pdf?raw=true')
+
+    local_input = remote_input.as_local_input_source
+    expect(local_input.filename).to eq('invoice_5p.pdf')
+
+    result = client.parse(local_input, Mindee::Product::Invoice::InvoiceV4)
+    expect(result.document.n_pages).to eq(5)
+  end
+end