Merge pull request #7 from ARPC/document-merge-new-api

Document merge new api

Gemfile.lock

@@ -1,7 +1,7 @@
 PATH
   remote: .
   specs:
-    pdfservices (0.1.2)
+    pdfservices (0.1.3)
       http (~> 5.1)
       json (~> 2.6)
       jwt (>= 1.0, < 3.0)
@@ -31,7 +31,7 @@ GEM
       domain_name (~> 0.5)
     http-form_data (2.3.0)
     json (2.6.2)
-    jwt (2.6.0)
+    jwt (2.7.0)
     llhttp-ffi (0.4.0)
       ffi-compiler (~> 1.0)
       rake (~> 13.0)

lib/pdfservices/document_merge/operation.rb

@@ -8,8 +8,9 @@
 module PdfServices
   module DocumentMerge
     class Operation
-      ENDPOINT = "https://cpf-ue1.adobe.io/ops/:create?respondWith=%7B%22reltype%22%3A%20%22http%3A%2F%2Fns.adobe.com%2Frel%2Fprimary%22%7D"
-      DOCUMENT_GENERATION_ASSET_ID = "urn:aaid:cpf:Service-52d5db6097ed436ebb96f13a4c7bf8fb"
+      PRESIGNED_URL_ENDPOINT = "https://pdf-services.adobe.io/assets"
+      OPERATION_ENDPOINT = "https://pdf-services.adobe.io/operation/documentgeneration"
+      ASSETS_ENDPOINT = "https://pdf-services.adobe.io/assets"
 
       def initialize(credentials = nil, template_path = nil, json_data_for_merge = nil, output_format = nil)
         @credentials = credentials
@@ -18,85 +19,92 @@ def initialize(credentials = nil, template_path = nil, json_data_for_merge = nil
         @output_format = output_format
       end
 
-      def execute
-        form = {
-          contentAnalyzerRequests: build_content_analyzer_requests.to_json,
-          InputFile0: build_input_file
-        }
-        response = api.post(ENDPOINT, form: form)
-        if response.status == 202
-          document_url = response.headers["Location"]
-          poll_document_result(document_url)
+      def get_presigned_url
+        response = api.post(PRESIGNED_URL_ENDPOINT, json: {mediaType: "application/vnd.openxmlformats-officedocument.wordprocessingml.document"})
+        if response.status == 200
+          JSON.parse(response.body.to_s)
         else
-          Result.new(nil, "Unexpected response status: #{response.status}")
+          Result.new(nil, "Unexpected response status from get presigned url: #{response.status}")
         end
       end
 
-      private
+      def upload_asset(template_path)
+        presigned_url = get_presigned_url
+        upload_uri = presigned_url["uploadUri"]
+        asset_id = presigned_url["assetID"]
+        aws = HTTP.headers({"Content-Type": "application/vnd.openxmlformats-officedocument.wordprocessingml.document"})
+        response = aws.put(upload_uri, body: File.open(template_path))
+        if response.status == 200
+          asset_id
+        else
+          Result.new(nil, "Unexpected response status from asset upload: #{response.status}")
+        end
+      end
 
-      def build_content_analyzer_requests
-        {
-          "cpf:engine": {
-            "repo:assetId": DOCUMENT_GENERATION_ASSET_ID
-          },
-          "cpf:inputs": {
-            documentIn: {
-              "dc:format": "application/vnd.openxmlformats-officedocument.wordprocessingml.document",
-              "cpf:location": "InputFile0"
-            },
-            params: {
-              "cpf:inline": {
-                outputFormat: @output_format,
-                jsonDataForMerge: @json_data_for_merge
-              }
-            }
-          },
-          "cpf:outputs": {
-            documentOut: {
-              "dc:format": mime_type(@output_format),
-              "cpf:location": "multipartLabel"
-            }
-          }
-        }
+      def delete_the_asset(asset_id)
+        api.delete("#{ASSETS_ENDPOINT}/#{asset_id}")
       end
 
-      def build_headers
+      def execute
+        asset_id = upload_asset(@template_path)
+        response = api.post(OPERATION_ENDPOINT, json: {
+          assetID: asset_id,
+          outputFormat: @output_format,
+          jsonDataForMerge: @json_data_for_merge
+        })
+        if response.status == 201
+          document_url = response.headers["location"]
+          poll_document_result(document_url, asset_id)
+        else
+          Result.new(nil, "Unexpected response status from document merge endpoint: #{response.status}\nasset_id: #{asset_id}")
+        end
+      end
+
+      private
+
+      def api_headers
         {
           Authorization: "Bearer #{JwtProvider.get_jwt(@credentials)}",
           "x-api-key": @credentials.client_id,
-          Prefer: "respond-async,wait=0"
+          "Content-Type": "application/json"
         }
       end
 
-      def build_input_file
-        HTTP::FormData::File.new(@template_path)
-      end
-
       def api
-        @api ||= HTTP.headers(build_headers)
+        @api ||= HTTP.headers(api_headers)
       end
 
-      def poll_document_result(url)
+      def poll_document_result(url, original_asset_id)
         sleep(1)
-        document_response = api.get(url)
-        case document_response.content_type.mime_type
-        when "application/json"
-          poll_document_result(url)
-        when "multipart/mixed"
-          Result.from_multipart_response(document_response)
-        else
-          Result.new(nil, "Unexpected response content type: #{document_response.content_type.mime_type}; status: #{document_response.status}")
-        end
-      end
-
-      def mime_type(format)
-        case format
-        when :docx
-          "application/vnd.openxmlformats-officedocument.wordprocessingml.document"
-        when :pdf
-          "application/pdf"
+        response = api.get(url)
+        if response.status == 200
+          json_response = JSON.parse(response.body.to_s)
+          document_merge_asset_id = json_response&.[]("asset")&.[]("assetID")
+          case json_response["status"]
+          when "in progress"
+            poll_document_result(url, original_asset_id)
+          when "done"
+            # download_the_asset
+            response = HTTP.get(json_response["asset"]["downloadUri"])
+            # delete the assets
+            delete_the_asset(original_asset_id) if !original_asset_id.nil?
+            delete_the_asset(document_merge_asset_id) if !document_merge_asset_id.nil?
+            # return the result
+            Result.new(response.body, nil)
+          when "failed"
+            # delete the original asset
+            message = json_response["error"]["message"]
+            delete_the_asset(original_asset_id) if !original_asset_id.nil?
+            Result.new(nil, message)
+          else
+            # delete the original asset
+            delete_the_asset(original_asset_id) if original_asset_id.present?
+            Result.new(nil, "Unexpected status from polling: #{json_response["status"]}")
+          end
         else
-          "text/plain"
+          # delete the original asset
+          delete_the_asset(original_asset_id) if original_asset_id.present?
+          Result.new(nil, "Unexpected response status from polling: #{json_response["status"]}")
         end
       end
     end

lib/pdfservices/version.rb

@@ -1,5 +1,5 @@
 # frozen_string_literal: true
 
 module PdfServices
-  VERSION = "0.1.2"
+  VERSION = "0.1.3"
 end

test/fixtures/merge_request_done.json

@@ -0,0 +1,7 @@
+{
+  "status": "done",
+  "asset": {
+    "assetID": "merged:asset-id",
+    "downloadUri": "https://documentmerge.file.url"
+  }
+}

test/fixtures/merge_request_in_progress.json

@@ -1,25 +1,3 @@
 {
-  "cpf:status": {
-    "completed": false,
-    "type": "",
-    "title": "In Progress",
-    "status": 202
-  },
-  "cpf:engine": {
-    "repo:assetId": "urn:aaid:cpf:Service-52d5db6097ed436ebb96f13a4c7bf8fb"
-  },
-  "cpf:inputs": {
-    "documentIn": {
-      "dc:format": "application/vnd.openxmlformats-officedocument.wordprocessingml.document",
-      "cpf:location": "InputFile0"
-    },
-    "params": {
-      "cpf:inline": {
-        "outputFormat": "pdf",
-        "jsonDataForMerge": {
-          "message": "World"
-        }
-      }
-    }
-  }
+  "status": "in progress"
 }

test/fixtures/ocr_done.json

@@ -1,3 +1,7 @@
 {
-  "status": "in progress"
+  "status": "done",
+  "asset": {
+    "assetID": "ocr'd:asset-id",
+    "downloadUri": "https://ocr.file.url"
+  }
 }

test/fixtures/ocr_in_progress.json

@@ -1,7 +1,3 @@
 {
-  "status": "done",
-  "asset": {
-    "assetID": "ocr'd:asset-id",
-    "downloadUri": "https://ocr.file.url"
-  }
+  "status": "in progress"
 }

test/integration_spikes/document_merge.rb

@@ -0,0 +1,21 @@
+# frozen_string_literal: true
+
+require "lib/pdfservices"
+
+credentials = ::PdfServices::CredentialsBuilder.new
+  .with_client_id(ENV["PDF_SERVICES_CLIENT_ID"])
+  .with_client_secret(ENV["PDF_SERVICES_CLIENT_SECRET"])
+  .with_organization_id(ENV["PDF_SERVICES_ORGANIZATION_ID"])
+  .with_account_id(ENV["PDF_SERVICES_ACCOUNT_ID"])
+  .with_private_key(ENV["PDF_SERVICES_PRIVATE_KEY"])
+  .build
+
+json = {message: "World"}
+word_file = File.join(Dir.pwd, "test", "fixtures", "files", "sample_template.docx")
+operation = ::PdfServices::DocumentMerge::Operation.new(credentials, word_file, json, "pdf")
+
+result = operation.execute
+
+puts(result.error)
+
+result.save_as_file("tmp/document_merge_result.pdf")

test/pdf_services_sdk/test_document_merge.rb

@@ -31,27 +31,65 @@ def test_it_works
   private
 
   def stub_valid_response_sequence
+    # get JWT for requests
     stub_request(:post, "https://ims-na1.adobelogin.com/ims/exchange/jwt/")
       .to_return(status: 200, body: json_fixture("valid_jwt_response"))
 
-    stub_request(:post, "https://cpf-ue1.adobe.io/ops/:create?respondWith=%7B%22reltype%22:%20%22http://ns.adobe.com/rel/primary%22%7D")
+    # get a presigned url to upload the source pdf
+    stub_request(:post, "https://pdf-services.adobe.io/assets")
+      .with(headers: secured_headers)
       .to_return(
-        status: 202,
-        headers: {"Location" => "https://cpf-ue1.adobe.io/ops/id/some-document-token"}.merge(json_headers),
-        body: json_fixture("merge_request_in_progress")
+        status: 200,
+        headers: json_headers,
+        body: json_fixture("presigned_upload_url_response")
       )
 
-    stub_request(:get, "https://cpf-ue1.adobe.io/ops/id/some-document-token")
-      .to_return(status: 202, headers: json_headers, body: json_fixture("merge_request_in_progress"))
-      .to_return(status: 202, headers: json_headers, body: json_fixture("merge_request_in_progress"))
-      .to_return(status: 200, headers: multipart_headers, body: multipart_fixture("merge_request_response"))
+    # upload the source template
+    stub_request(:put, "https://a.presigned.url").to_return(status: 200)
+
+    # request the Document Merge operation
+    stub_request(:post, "https://pdf-services.adobe.io/operation/documentgeneration")
+      .with(headers: secured_headers)
+      .to_return(
+        status: 201,
+        headers: {"location" => "https://some.polling.url"}.merge(json_headers)
+      )
+
+    # poll for the result
+    stub_request(:get, "https://some.polling.url")
+      .with(headers: secured_headers)
+      .to_return(status: 200, headers: json_headers, body: json_fixture("merge_request_in_progress"))
+      .to_return(status: 200, headers: json_headers, body: json_fixture("merge_request_in_progress"))
+      .to_return(status: 200, headers: json_headers, body: json_fixture("merge_request_done"))
+
+    # download the merged pdf
+    stub_request(:get, "https://documentmerge.file.url")
+      .to_return(status: 200, headers: pdf_headers, body: file_fixture("merge_done.pdf"))
+
+    # delete the original asset
+    stub_request(:delete, "https://pdf-services.adobe.io/assets/urn:a-real-long-asset-asset-id")
+      .with(headers: secured_headers)
+      .to_return(status: 200, body: "", headers: {})
+
+    # delete the merged asset
+    stub_request(:delete, "https://pdf-services.adobe.io/assets/merged:asset-id")
+      .with(headers: secured_headers)
+      .to_return(status: 200, body: "", headers: {})
+  end
+
+  def secured_headers
+    {
+      Authorization: "Bearer fake1.fake2.fake3",
+      "Content-Type": "application/json",
+      "X-Api-Key": "123someclientid"
+    }
   end
 
   def json_headers
     {"Content-Type" => "application/json;charset=UTF-8"}
   end
 
-  def multipart_headers
-    {"Content-Type" => "multipart/mixed; boundary=Boundary_962026_575353369_1661875317362;charset=UTF-8"}
+  def pdf_headers
+    {"Content-Type" => "application/pdf"}
   end
 end