adobe_pdfservices_ruby 0.1.2 → 0.1.4

checksums.yaml

@@ -1,7 +1,7 @@
 ---
 SHA256:
-  metadata.gz: 7874ed45d46a4c10b5a44574d45b49946707cebb0c05447416bcedfbc777e090
-  data.tar.gz: 2a920b6f69c0abd16245d1f08486387993a5232c2e6bd5f005112dadec28b88d
+  metadata.gz: 6147a7ab0e87ee51554ec2136b1f21ff3e2d4630760a6ea07432744d8a145699
+  data.tar.gz: 7f0f7c7f48470007c8078531598ca63dc3969628ea44a12b612d221c07703833
 SHA512:
-  metadata.gz: 384c4c8555dd64925398fdd6d0c5e5b46d8333893c84684a100ab84c54793f58f754dd515c9e8d426f9a46450b178eea2f4d440c93dde3f5699de9f9d2ac0092
-  data.tar.gz: a7e0e3c4cf47c017abdb1fa6bc0d61daa83e438f2ce8eb68eeecc8bf87f7de0dca670983e24f97853158019917f4299c925953c64080dccb9a6d477b310224a5
+  metadata.gz: 32a89f111dd8b13884f7dd854c59e6e74e8dde5d77adc6035cb2af98527ddccc9cd97d9bd854a7972e33217628fa9b37bbc64bac9b8f30174f660d4bd06f38af
+  data.tar.gz: '028ecf12bb31bf37a3f48137a9d5ad55d18fb23b8026f121773cc36788032e9587e7fc3b75ccbd696cf2895693dd7e8c5eb32e44a63c6e3fcef5ac2562ba7a48'

data/CHANGELOG.md

@@ -22,3 +22,16 @@
 - Allow client to be initialized with no secret_key if there's an access_token provided (useful for development)
 - Asset delete request is properly formed
 - Use MimeMagic to determine content-type of files
+
+## [0.1.3] - 2024-01-27
+
+- Fix incorrect URLs in README
+- Fix incorrect URLs in gemspec
+- Add support for document generation via spike
+- Fix passed blocks not being called for most operations
+- Add support for html_to_pdf operation via spike
+- Add support for OCR operation via spike
+
+## [0.1.4] - 2024-01-29
+
+- Remove require for multipart parser

data/README.md

@@ -1,6 +1,6 @@
 # PDF Services for Ruby
 
-### Originally forked from [Adobe Document Services PDF Tools SDK for Ruby](https://github.com/arpc/adobe_pdfservices_ruby)
+### Originally forked from [Adobe Document Services PDF Tools SDK for Ruby](https://github.com/arpc/pdfservices-ruby-sdk)
 
 This gem provides a Ruby wrapper for the [Adobe PDF Services API](https://developer.adobe.com/document-services/docs/overview/). It allows you to perform various PDF operations, such as extracting content from PDFs, OCR, HTML to PDF, and document generation.
 
@@ -58,13 +58,13 @@ To install this gem onto your local machine, run `bundle exec rake install`. To
 Work is in progress on getting the gem to support all of the operations available in the [Adobe PDF Services API](https://developer.adobe.com/document-services/docs/overview/). Below is a list of the operations and their current support status:
 
 - ✅ Extract PDF
-- ❗ OCR
-  - ❗ EXPERIMENTAL: Internal
+- ✅ OCR
+  - ✅ Internal
   - ❗ EXPERIMENTAL: External
-- ❗ HTML to PDF
-  - ❗ EXPERIMENTAL: Internal
+- ✅ HTML to PDF
+  - ✅ Internal
   - ❗ EXPERIMENTAL: External
-- ❗ Document Generation:
+- ✅ Document Generation:
   - ✅ Internal
   - ❗ EXPERIMENTAL: External
 

data/lib/adobe_pdfservices_ruby.rb

@@ -1,7 +1,6 @@
 # frozen_string_literal: true
 
 require 'json'
-require 'multipart_parser/reader'
 require 'yaml'
 require 'faraday'
 require 'mimemagic'
@@ -25,9 +24,13 @@ require_relative 'pdfservices/operations/document_generation/external'
 
 # OCR
 require_relative 'pdfservices/operations/ocr'
+require_relative 'pdfservices/operations/ocr/internal'
+require_relative 'pdfservices/operations/ocr/external'
 
 # HTML to PDF
 require_relative 'pdfservices/operations/html_to_pdf'
+require_relative 'pdfservices/operations/html_to_pdf/internal'
+require_relative 'pdfservices/operations/html_to_pdf/external'
 
 # Extract PDF
 require_relative 'pdfservices/operations/extract_pdf'

data/lib/pdfservices/asset.rb

@@ -5,14 +5,17 @@ module PdfServices
     attr_reader :id
 
     def initialize(api, id = nil)
+      # MimeMagic can't detect docx files, and will return `application/zip` so we need to add it manually
+      MimeMagic.add('application/vnd.openxmlformats-officedocument.wordprocessingml.document',
+                    magic: [[0..2000, 'word/']])
       raise ArgumentError, 'Api is nil' unless api
 
       @api = api
       @id = id
     end
 
-    def upload(file, media_type: 'application/pdf')
-      url = presigned_url(media_type:)
+    def upload(file)
+      url = presigned_url(file:)
       upload_uri = url['uploadUri']
       asset_id = url['assetID']
 
@@ -39,7 +42,7 @@ module PdfServices
       @api.get(download_uri)
     end
 
-    def delete
+    def delete_asset
       raise AssetError, 'Asset ID is nil' unless @id
 
       @api.delete("#{ASSETS_ENDPOINT}/#{@id}")
@@ -54,9 +57,10 @@ module PdfServices
       }
     end
 
-    def presigned_url(operation = :upload, media_type: 'application/pdf')
+    def presigned_url(operation = :upload, file: nil)
       case operation
       when :upload
+        media_type = file ? MimeMagic.by_magic(file).type : 'application/pdf'
         response = @api.post(ASSETS_ENDPOINT, body: { mediaType: media_type },
                                               headers: { 'Content-Type' => 'application/json' })
       when :download

data/lib/pdfservices/operations/base.rb

@@ -26,6 +26,15 @@ module PdfServices
         handle_polling_result(url, json_response, original_asset, &block)
       end
 
+      def handle_response(response, asset, &block)
+        unless response.status == 201
+          raise "Unexpected response status from operation endpoint: #{response.status}, #{response.body}"
+        end
+
+        document_url = response.headers['location']
+        poll_document_result document_url, asset, &block
+      end
+
       private
 
       def handle_polling_result(url, json_response, original_asset, &block)
@@ -64,16 +73,16 @@ module PdfServices
       end
 
       def handle_polling_done(_json_response, original_asset)
-        original_asset.delete
+        original_asset.delete_asset
       end
 
       def handle_polling_failed(json_response, original_asset)
-        original_asset.delete
+        original_asset.delete_asset
         raise PollingError, "Document extraction failed: #{json_response['error']}"
       end
 
       def handle_polling_unexpected_status(json_response, original_asset)
-        original_asset.delete
+        original_asset.delete_asset
         raise PollingError, "Unexpected status: #{json_response['status']}"
       end
     end

data/lib/pdfservices/operations/document_generation/internal.rb

@@ -5,7 +5,7 @@ module PdfServices
     class Internal < Operation
       INTERNAL_OPTIONS = %i[output_format json_data_for_merge fragments notifiers].freeze
 
-      def execute(template_path, options = {})
+      def execute(template_path, options = {}, &block)
         validate_options(options)
         asset = upload_asset(template_path)
 
@@ -13,7 +13,7 @@ module PdfServices
                              body: request_body(asset.id, options),
                              headers: request_headers)
 
-        handle_response(response, asset.id)
+        handle_response(response, asset, &block)
       end
 
       private
@@ -64,7 +64,7 @@ module PdfServices
         raise ArgumentError, "Invalid options: #{invalid_keys}" unless invalid_keys.empty?
       end
 
-      def handle_polling_done(json_response, original_asset)
+      def handle_polling_done(json_response, original_asset, &block)
         asset_id = json_response['asset']['assetID']
         file = Asset.new(@api).download(asset_id).body
         super

data/lib/pdfservices/operations/document_generation.rb

@@ -9,15 +9,6 @@ module PdfServices
         { 'Content-Type' => 'application/json' }
       end
 
-      def handle_response(response, asset_id)
-        unless response.status == 201
-          raise "Unexpected response status from document merge endpoint: #{response.status}, asset_id: #{asset_id}"
-        end
-
-        document_url = response.headers['location']
-        poll_document_result document_url, asset_id
-      end
-
       def internal_class
         Internal
       end

data/lib/pdfservices/operations/extract_pdf.rb

@@ -8,7 +8,7 @@ module PdfServices
       TABLE_OUTPUT_FORMATS = %w[csv xlsx].freeze
       RENDITIONS_EXTRACTS = %w[tables figures].freeze
 
-      def execute(source_pdf_path = nil, options = {})
+      def execute(source_pdf_path = nil, options = {}, &block)
         validate_options(options)
         @download_zip = options.delete(:download_zip) || false
         asset = upload_asset(source_pdf_path)
@@ -16,7 +16,7 @@ module PdfServices
         response = @api.post(OPERATION_ENDPOINT,
                              body: extract_pdf_request_body(asset.id, options),
                              headers: extract_pdf_request_headers)
-        handle_extract_pdf_response(response, asset)
+        handle_response(response, asset, &block)
       end
 
       private
@@ -36,13 +36,6 @@ module PdfServices
         { 'Content-Type' => 'application/json' }
       end
 
-      def handle_extract_pdf_response(response, asset)
-        raise OperationError, "Extract PDF operation failed: #{response.body}" unless response.status == 201
-
-        polling_url = response.headers['location']
-        poll_document_result polling_url, asset
-      end
-
       def handle_polling_done(json_response, original_asset)
         file_key = @download_zip ? 'resource' : 'content'
         asset_id = json_response[file_key]['assetID']

data/lib/pdfservices/operations/html_to_pdf/internal.rb

@@ -6,15 +6,18 @@ module PdfServices
       INTERNAL_OPTIONS = %i[input_url json include_header_footer page_layout notifiers].freeze
       PAGE_LAYOUT_OPTIONS = %i[page_width page_height].freeze
 
-      def execute(html_file_path, options = {})
-        validate_options(options)
-        asset = upload_asset(html_file_path)
+      def execute(html_file_path, options = {}, &block)
+        validate_options(options, html_file_path)
+
+        asset = upload_asset(html_file_path) unless options[:input_url]
+
+        asset_id = asset.id if asset
 
         response = @api.post(OPERATION_ENDPOINT,
-                             body: request_body(asset.id, options),
+                             body: request_body(asset_id, options),
                              headers: request_headers)
 
-        handle_response(response, asset.id)
+        handle_response(response, asset, &block)
       end
 
       private
@@ -24,13 +27,13 @@ module PdfServices
         Asset.new(@api).download(asset_id).body
       end
 
-      def request_body(asset_id, options)
-        body = {
-          assetID: asset_id,
-          inputUrl: options.fetch(:input_url, ''),
-          pageLayout: camelize_keys(options.fetch(:page_layout, {})),
-          json: transform_json(options.fetch(:json, ''))
-        }
+      def request_body(asset_id, options) # rubocop:disable Metrics/AbcSize
+        body = {}
+        body[:includeHeaderFooter] = options[:include_header_footer] if options[:include_header_footer]
+        body[:pageLayout] = options[:page_layout] if options[:page_layout]
+        body[:json] = transform_json(options[:json]) if options[:json]
+        body[:assetID] = asset_id if asset_id
+        body[:inputUrl] = options[:input_url] if options[:input_url]
         body[:notifiers] = options[:notifiers] if options[:notifiers]
         body
       end
@@ -39,22 +42,19 @@ module PdfServices
         json.is_a?(String) ? json : json.to_json
       end
 
-      def validate_options(options)
+      def validate_options(options, source = nil)
         raise ArgumentError, 'Invalid options' unless options.is_a?(Hash)
 
         options.each_key do |key|
           raise ArgumentError, "Invalid option: #{key}" unless INTERNAL_OPTIONS.include?(key)
         end
 
-        validate_required_keys(options)
+        validate_source(source, options)
         validate_page_layout_options(options[:page_layout]) if options[:page_layout]
       end
 
-      def validate_required_keys(options)
-        required_keys = INTERNAL_OPTIONS - %i[page_layout notifiers]
-        required_keys.each do |key|
-          raise ArgumentError, "Missing required option: #{key}" unless options.key?(key)
-        end
+      def validate_source(source, options)
+        raise OperationError, "Cannot specify both 'input_url' and a HTML file" if options[:input_url] && source
       end
 
       def validate_page_layout_options(options)

data/lib/pdfservices/operations/html_to_pdf.rb

@@ -9,15 +9,6 @@ module PdfServices
         { 'Content-Type' => 'application/json' }
       end
 
-      def handle_response(response, asset_id)
-        unless response.status == 201
-          raise "Unexpected response status from document merge endpoint: #{response.status}, asset_id: #{asset_id}"
-        end
-
-        document_url = response.headers['location']
-        poll_document_result document_url, asset_id
-      end
-
       def internal_class
         Internal
       end

data/lib/pdfservices/operations/internal_external_operation.rb

@@ -1,9 +1,9 @@
 module PdfServices
   module InternalExternalOperation
     class Operation < Base::Operation
-      def execute(source_file_path, options = {})
+      def execute(source_file_path, options = {}, &block)
         operation_class = switch_on_type(options)
-        operation_class.new(@api).execute(source_file_path, options)
+        operation_class.new(@api).execute(source_file_path, options, &block)
       end
 
       private

data/lib/pdfservices/operations/ocr/internal.rb

@@ -5,21 +5,21 @@ module PdfServices
     class Internal < Operation
       INTERNAL_OPTIONS = %i[ocr_lang ocr_type notifiers].freeze
 
-      def execute(html_file_path, options = {})
+      def execute(source_pdf, options = {}, &block)
         validate_options(options)
-        asset = upload_asset(html_file_path)
+        asset = upload_asset(source_pdf)
 
         response = @api.post(OPERATION_ENDPOINT,
-                             body: request_body(asset.id, options),
-                             headers: request_headers)
+                             body: request_body(asset.id, options), headers: { 'Content-Type' => 'application/json' })
 
-        handle_response(response, asset.id)
+        handle_response(response, asset, &block)
       end
 
       private
 
-      def handle_polling_done(json_response, _original_asset_id)
+      def handle_polling_done(json_response, _original_asset)
         asset_id = json_response['asset']['assetID']
+        super
         Asset.new(@api).download(asset_id).body
       end
 

data/lib/pdfservices/operations/ocr.rb

@@ -3,7 +3,7 @@
 module PdfServices
   module Ocr
     class Operation < InternalExternalOperation::Operation
-      OCR_ENDPOINT = 'https://pdf-services-ue1.adobe.io/operation/ocr'
+      OPERATION_ENDPOINT = "#{BASE_ENDPOINT}ocr".freeze
       OCR_LANGS = %w[
         da-DK lt-LT sl-SI el-GR ru-RU en-US zh-HK hu-HU et-EE
         pt-BR uk-UA nb-NO pl-PL lv-LV fi-FI ja-JP es-ES bg-BG
@@ -16,6 +16,14 @@ module PdfServices
 
       private
 
+      def internal_class
+        Internal
+      end
+
+      def external_class
+        External
+      end
+
       def validate_ocr_lang_option(ocr_lang)
         raise ArgumentError, "Invalid ocr_lang option: #{ocr_lang}" unless OCR_LANGS.include?(ocr_lang)
       end

data/lib/pdfservices/version.rb

@@ -1,5 +1,5 @@
 # frozen_string_literal: true
 
 module PdfServices
-  VERSION = '0.1.2'
+  VERSION = '0.1.4'
 end

metadata

@@ -1,7 +1,7 @@
 --- !ruby/object:Gem::Specification
 name: adobe_pdfservices_ruby
 version: !ruby/object:Gem::Version
-  version: 0.1.2
+  version: 0.1.4
 platform: ruby
 authors:
 - Jimmy Bosse
@@ -9,7 +9,7 @@ authors:
 autorequire:
 bindir: bin
 cert_chain: []
-date: 2024-01-25 00:00:00.000000000 Z
+date: 2024-01-29 00:00:00.000000000 Z
 dependencies:
 - !ruby/object:Gem::Dependency
   name: faraday