cnvrg 1.11.2 → 1.11.8

checksums.yaml

@@ -1,7 +1,7 @@
 ---
 SHA256:
-  metadata.gz: 70906f4870c9aa2b3457673212893d09e6b5f22e15a5dc4c3b0cc2115e74d8d9
-  data.tar.gz: ff9de0565cf2c81102c68a21154af1f97d843e65d51fc8eb69f3a490b64bb9a3
+  metadata.gz: 75a73acbac75b972943707353f376c6c6cbda9fabc9c29b4a4dc5a454e9f531c
+  data.tar.gz: fd8b9ae0ea9c0ddeef86c6f25a8679457a66a6fd0dfdb4f06f8d7fa5cdf8e63b
 SHA512:
-  metadata.gz: d839249d813d8004c9d00553f81273e890d7beb3bfab4bb65b64453f459bd6e4c366c59b35cee2773716e2c26f6f5b30223937837e088fe964e98a60b242c6e3
-  data.tar.gz: f7cac4b24a2f7fc275e28dd03b2fad4b7374dd52c648dd84f36652d6e7bd61f24e834aa89b1f826ac39e7680aa7cbfc6ca1564bc3611fdfe53dfd650368d7319
+  metadata.gz: c034b4c2bab31d420d8343d1007824baf78ada4ca20f245848d12ed61f37941ebe2d6a9e707d6d5f80f989b519623b640d941197a1a9ff3b8f5035c0142e6d2e
+  data.tar.gz: aff33e7fecc6419102cc02082116efd341b53f395f8362258015369839d87b68131bced0125b083ef3afc6bd7d233d3ad29fde9e549b07e1d74f285f5748a592

data/lib/cnvrg/cli.rb

@@ -896,10 +896,12 @@ module Cnvrg
             return
           end
 
+
           if only_tree
-            success = Dataset.clone_tree(commit: commit, dataset_home: dataset_home)
-            return if success
+            Dataset.clone_tree(commit: commit, dataset_home: dataset_home)
+            return
           end
+
           commit = response["result"]["commit"]
           files_count = response["result"]["file_count"]
           files = @files.get_clone_chunk(commit: commit)
@@ -911,6 +913,7 @@ module Cnvrg
                                            :total => files_count,
                                            :autofinish => true)
 
+          Dataset.clone_tree(commit: commit, dataset_home: dataset_home, progressbar: progressbar)
 
           while files['keys'].length > 0
             Cnvrg::Logger.log_info("download multiple files, #{downloaded_files.size} files downloaded")
@@ -969,6 +972,7 @@ module Cnvrg
 
         if Dataset.blank_clone(owner, dataset_name, dataset_slug)
           dataset = Dataset.new(dataset_home)
+          downloader = dataset.get_storage_client
           log_message("Cloning #{dataset_name}", Thor::Shell::Color::BLUE)
           parallel_options = {
               :progress => {
@@ -991,15 +995,20 @@ module Cnvrg
               relative_path_dir = relative_path_dir.join("/")
               abs_path = dataset_home + "/" + relative_path_dir
               abs_path = dataset_home if flatten
+              fullpath = abs_path + "/" + file_name
 
               begin
-                FileUtils.mkdir_p(abs_path) unless File.exist? (abs_path + "/" + file_name)
+                FileUtils.mkdir_p(abs_path) unless File.exist? (fullpath)
               rescue
                 log_message("Could not create directory: #{abs_path}", Thor::Shell::Color::RED)
                 exit(1)
               end
               begin
-                File.write "#{abs_path}/#{file_name}", open(f["url"]).read unless File.exist? (abs_path + "/" + file_name)
+                unless File.exist?(fullpath)
+                  downloader.safe_operation("#{abs_path}/#{file_name}") do
+                    File.open(fullpath, "w") { |file| file.write open(f["url"]).read }
+                  end
+                end
               rescue => e
                 log_message("Could not download file: #{f["fullpath"]}", Thor::Shell::Color::RED)
                 exit(1)
@@ -1905,7 +1914,7 @@ module Cnvrg
               log_message("Remove dir in order to clone #{project_name}", Thor::Shell::Color::RED)
               exit(1)
             end
-  
+
           end
           clone_resp = Project.clone_dir(slug, owner, project_name,git)
           project_home = Dir.pwd + "/" + project_name

data/lib/cnvrg/datafiles.rb

@@ -1338,7 +1338,7 @@ module Cnvrg
             #   Cnvrg::Logger.log_info("Trying to download #{local_path} but its already exists, skipping..")
             #   next
             # end
-            resp = @downloader.download(storage_path, local_path)
+            resp = @downloader.safe_download(storage_path, local_path)
             Cnvrg::Logger.log_info("Download #{local_path} success resp: #{resp}")
           rescue => e
             Cnvrg::Logger.log_error(e)

data/lib/cnvrg/dataset.rb

@@ -471,23 +471,38 @@ module Cnvrg
       response.to_json
     end
 
-    def self.clone_tree(commit: 'latest', dataset_home: nil)
+    def self.clone_tree(commit: 'latest', dataset_home: nil, progressbar: nil)
       @dataset = Cnvrg::Dataset.new(dataset_home)
       @files = Cnvrg::Datafiles.new(@dataset.owner, @dataset.slug, dataset: @dataset)
       trees = @files.get_trees(commit: commit)
       return false if trees.nil?
-      pb = ProgressBar.create(:title => "Download Progress",
-                              :progress_mark => '=',
-                              :format => "%b>>%i| %p%% %t",
-                              :starting_at => 0,
-                              :total => trees.size,
-                              :autofinish => true)
+
+      if progressbar
+        pb = progressbar
+        pb.total += trees.size
+      else
+        pb = ProgressBar.create(
+          :title => "Download Progress",
+          :progress_mark => '=',
+          :format => "%b>>%i| %p%% %t",
+          :starting_at => 0,
+          :total => trees.size,
+          :autofinish => true
+        )
+      end
+
       trees.each do |tree|
         pb.progress += 1
         @files.download_dir(dataset_home, tree)
       end
-      pb.finish
-      @dataset.write_success
+
+      unless progressbar
+        # if progessbar sent it means that its not only tree so we dont want to finish the progress bar
+        # and we dont want to write success
+
+        pb.finish
+        @dataset.write_success
+      end
       true
     end
 

data/lib/cnvrg/downloader/client.rb

@@ -30,10 +30,14 @@ module Cnvrg
         file.gsub(prefix, '').gsub(/^\/*/, '')
       end
 
-      def download(storage_path, local_path)
+      def download(storage_path, local_path, decrypt: true)
         ### need to be implemented..
       end
 
+      def safe_download(storage_path, local_path, decrypt: true)
+        safe_operation(local_path) { self.download(storage_path, local_path, decrypt: decrypt) }
+      end
+
       def upload(storage_path, local_path)
         ### need to be implemented..
       end
@@ -51,17 +55,34 @@ module Cnvrg
       end
 
       def safe_upload(storage_path, local_path)
+        safe_operation(local_path) { self.upload(storage_path, local_path) }
+      end
+
+      def self.factory(params)
+        params = params.as_json
+        case params["storage"]
+        when 's3', 'minio'
+          return Cnvrg::Downloader::Clients::S3Client.new(sts_path: params["path_sts"], access_key: params["sts_a"], secret: params["sts_s"], session_token: params["sts_st"], region: params["region"], bucket: params["bucket"], encryption: params["encryption"], endpoint: params["endpoint"], storage: params["storage"])
+        when 'azure'
+          azure_params = params.symbolize_keys.slice(*[:storage_account_name, :storage_access_key, :container, :sts])
+          return Cnvrg::Downloader::Clients::AzureClient.new(**azure_params)
+        when 'gcp'
+          return Cnvrg::Downloader::Clients::GcpClient.new(project_id: params["project_id"], credentials: params["credentials"], bucket_name: params["bucket_name"], sts: params["sts"])
+        end
+      end
+
+      def safe_operation(local_path)
         n = 1
         error = nil
         while n <= RETRIES
           begin
-            self.upload(storage_path, local_path)
+            yield
             error = nil
             break
           rescue => e
             backoff_time_seconds = backoff_time(n)
 
-            message = "Got error: #{e.class.name} with message: #{e.message} while uploading a single file: #{local_path}, retry: #{n} of: #{RETRIES}"
+            message = "Got error: #{e.class.name} with message: #{e.message} while uploading / downloading a single file: #{local_path}, retry: #{n} of: #{RETRIES}"
             if n < RETRIES
               message += ", next retry in: #{backoff_time_seconds} seconds"
             else
@@ -79,19 +100,6 @@ module Cnvrg
         true
       end
 
-      def self.factory(params)
-        params = params.as_json
-        case params["storage"]
-        when 's3', 'minio'
-          return Cnvrg::Downloader::Clients::S3Client.new(sts_path: params["path_sts"], access_key: params["sts_a"], secret: params["sts_s"], session_token: params["sts_st"], region: params["region"], bucket: params["bucket"], encryption: params["encryption"], endpoint: params["endpoint"], storage: params["storage"])
-        when 'azure'
-          azure_params = params.symbolize_keys.slice(*[:storage_account_name, :storage_access_key, :container, :sts])
-          return Cnvrg::Downloader::Clients::AzureClient.new(**azure_params)
-        when 'gcp'
-          return Cnvrg::Downloader::Clients::GcpClient.new(project_id: params["project_id"], credentials: params["credentials"], bucket_name: params["bucket_name"], sts: params["sts"])
-        end
-      end
-
       private
 
       def random_number_milliseconds

data/lib/cnvrg/downloader/clients/gcp_client.rb

@@ -31,7 +31,7 @@ module Cnvrg
           @tempfile = t
         end
 
-        def download(storage_path, local_path)
+        def download(storage_path, local_path, decrypt: true)
           prepare_download(local_path)
           file = @bucket.file(decrypt(storage_path))
           file.download local_path

data/lib/cnvrg/files.rb

@@ -789,7 +789,7 @@ module Cnvrg
               end
               local_path = project_home+"/"+file_path
               storage_path = f["path"]
-              @client.download(storage_path, local_path)
+              @client.safe_download(storage_path, local_path)
               progress.progress += 1 if progress.present?
               download_succ_count += 1
             rescue => e
@@ -962,7 +962,7 @@ module Cnvrg
 
     def download_file(file_path: '', key: '', iv: '', bucket: '', path: '', client: nil)
       local_path = @project_home+"/"+file_path
-      @client.download(path, local_path)
+      @client.safe_download(path, local_path)
     end
 
     def delete(file)

data/lib/cnvrg/helpers.rb

@@ -3,7 +3,7 @@ module Cnvrg
 
     extend self
     def parallel_threads
-      5
+      15
     end
 
     def self.parallel_options

data/lib/cnvrg/storage.rb

@@ -34,7 +34,7 @@ module Cnvrg
         @element.get_clone_chunk(commit: commit, chunk_size: params[:limit], offset: params[:offset])
       end
       action = Proc.new do |storage, local|
-        @client.download(storage, local)
+        @client.safe_download(storage, local)
       end
 
       @stats = @element.get_stats

data/lib/cnvrg/version.rb

@@ -1,3 +1,3 @@
 module Cnvrg
-  VERSION = '1.11.2'
+  VERSION = '1.11.8'
 end

metadata

@@ -1,7 +1,7 @@
 --- !ruby/object:Gem::Specification
 name: cnvrg
 version: !ruby/object:Gem::Version
-  version: 1.11.2
+  version: 1.11.8
 platform: ruby
 authors:
 - Yochay Ettun
@@ -10,7 +10,7 @@ authors:
 autorequire:
 bindir: bin
 cert_chain: []
-date: 2020-11-06 00:00:00.000000000 Z
+date: 2020-11-19 00:00:00.000000000 Z
 dependencies:
 - !ruby/object:Gem::Dependency
   name: bundler