cnvrg 0.5.6 → 0.5.8

data/lib/cnvrg/data.rb

@@ -27,15 +27,16 @@ module Cnvrg
     method_option :force, :type => :boolean, :aliases => ["-f","--force"], :default => false
     method_option :sync, :type => :boolean, :aliases => ["-s","--sync"], :default => false
     method_option :tags, :type => :boolean, :aliases => ["--tags"], :desc => "upload file tags", :default => false
-
+    method_option :chunk_size, :type => :numeric, :aliases => ["--chunk"], :desc => "upload file tags", :default => 1000
     def upload
       cli = Cnvrg::CLI.new()
       verbose = options["verbose"]
       sync = options["sync"]
       force = options["force"]
       new_branch = options["new_branch"]
+      chunk_size = options["chunk_size"]
       tags = options["tags"]
-      cli.upload_data_new(new_branch,  verbose,sync,force, tags)
+      cli.upload_data_new(new_branch,  verbose,sync,force, tags, chunk_size)
     end
     desc 'sync', 'sync_data_new'
     method_option :new_branch, :type => :boolean, :aliases => ["-nb"], :desc => "create new branch of commits"
@@ -45,6 +46,8 @@ module Cnvrg
     method_option :all_files, :type => :boolean, :aliases => ["--all"], :desc => "download specified commit", :default => false
     method_option :tags, :type => :boolean, :aliases => ["--tags"], :desc => "upload file tags", :default => false
     method_option :parallel, :type => :numeric, :aliases => ["-p", "--parallel"], :desc => "uparallel upload at the same time", :default => 15
+    method_option :chunk_size, :type => :numeric, :aliases => ["--chunk_size"], :desc => "chunk size to communicate with the server", :default => 1000
+    method_option :init, :type => :boolean, :aliases => ["--initial"], :desc => "chunk size to communicate with the server", :default => false
 
     def sync_data_new()
       cli = Cnvrg::CLI.new()
@@ -55,9 +58,9 @@ module Cnvrg
       all_files = options["all_files"]
       tags = options["tags"]
       parallel=options["parallel"]
-
-
-      cli.sync_data_new(new_branch, force, verbose,commit,all_files, tags,parallel)
+      chunk_size = options["chunk_size"]
+      init = options["init"]
+      cli.sync_data_new(new_branch, force, verbose,commit,all_files, tags,parallel, chunk_size, init)
     end
     desc 'data download', 'pull data'
     method_option :new_branch, :type => :boolean, :aliases => ["-nb"], :desc => "create new branch of commits", :default => false
@@ -73,7 +76,6 @@ module Cnvrg
       new_branch = options["new_branch"]
       commit = options["commit"]
       all_files = options["all_files"]
-
       cli.download_data_new(verbose,sync,new_branch, commit,all_files)
 
     end
@@ -87,9 +89,7 @@ module Cnvrg
       only_tree =options[:only_tree]
       commit =options[:commit]
       query =options[:query]
-
-        cli.clone_data(dataset_url, only_tree=only_tree,commit=commit, query=query)
-
+      cli.clone_data(dataset_url, only_tree=only_tree,commit=commit, query=query)
     end
 
     desc 'data clone_query', 'clone query datset'

data/lib/cnvrg/datafiles.rb

@@ -20,6 +20,34 @@ module Cnvrg
       @base_resource = "users/#{owner}/datasets/#{dataset_slug}/"
     end
 
+    def upload_multiple_files(commit_sha1, tree, threads: ParallelThreads, force: false, new_branch: false)
+      random_file_name = (0...8).map { (65 + rand(26)).chr }.join #needed to create a file for post_file..
+      # each file have: {sha1: sha1, file_name: file_name, file_size: file_size, content_type: content_type, absolute_path, relative_path} 
+
+      #this call should also implement compare_idx... 
+      upload_resp = Cnvrg::API.request(@base_resource + "upload_files", 'POST', {commit_sha1: commit_sha1, tree: tree, force: force, is_branch: new_branch})
+      return Cnvrg::Result.new(false, "Failed to upload files") unless Cnvrg::CLI.is_response_success(upload_resp, false)
+      results = upload_resp['result'].with_indifferent_access
+      props = Cnvrg::Helpers.get_s3_props(results)
+      client = props[:client]
+      bucket = props[:bucket]
+      upload_options = props[:upload_options]
+      s3_bucket = Aws::S3::Resource.new(client: client).bucket(bucket)
+      files = results['files']
+      Parallel.map((files.keys), {in_threads: threads}) do |k|
+        o = tree[k].merge(files[k])
+        upload_single_file(o, s3_bucket, upload_options)
+      end
+      return files.keys.length
+    end
+
+    def upload_single_file(file, s3_bucket, upload_options={})
+      file = file.with_indifferent_access
+      resp = s3_bucket.
+        object(file[:path]).
+        upload_file(file[:absolute_path], upload_options)
+    end
+
     def upload_file(absolute_path, relative_path, commit_sha1)
       file_name = File.basename relative_path
       file_size = File.size(absolute_path).to_f
@@ -62,6 +90,8 @@ module Cnvrg
         upload_resp = Cnvrg::API.request(@base_resource + "upload_file", 'POST_FILE', {absolute_path: absolute_path, relative_path: relative_path,
                                                                                        commit_sha1: commit_sha1, file_name: file_name,
                                                                                        file_size: file_size, file_content_type: content_type, sha1: sha1})
+        puts upload_resp
+
       end
 
       if Cnvrg::CLI.is_response_success(upload_resp, false)
@@ -69,7 +99,7 @@ module Cnvrg
 
         return s3_res
       end
-      end
+    end
     def upload_tar_file(absolute_path, relative_path, commit_sha1)
       begin
         file_name = File.basename relative_path
@@ -653,12 +683,12 @@ module Cnvrg
     def delete_commit_files_local(deleted)
       begin
         FileUtils.rm_rf(deleted) unless (deleted.nil? or deleted.empty?)
-        return true
+        return Cnvrg::Result.new(true, '')
       rescue => e
-        return false
+        return Cnvrg::Result.new(false, '')
       end
 
-      return true
+      return Cnvrg::Result.new(true, '')
 
     end
 
@@ -716,9 +746,9 @@ module Cnvrg
 
     end
 
-    def end_commit(commit_sha1,force)
+    def end_commit(commit_sha1,force, is_success)
       begin
-        response = Cnvrg::API.request("#{base_resource}/commit/end", 'POST', {commit_sha1: commit_sha1,force:force})
+        response = Cnvrg::API.request("#{base_resource}/commit/end", 'POST', {commit_sha1: commit_sha1,force:force, success:is_success})
         Cnvrg::CLI.is_response_success(response, true)
         return response
       rescue => e
@@ -736,5 +766,120 @@ module Cnvrg
       response = Cnvrg::API.request("#{base_resource}/commit/rollback", 'POST', {commit_sha1: commit_sha1})
       Cnvrg::CLI.is_response_success(response, false)
     end
+
+    def clone_in_chunks(commit: 'latest', chunk_size: 1000)
+      begin
+
+      end
+    end
+
+    def get_clone_chunk(latest_id: nil, chunk_size: 1000, commit: 'latest')
+      response = Cnvrg::API.request("#{@base_resource}/clone_chunk", 'POST',{commit: commit, chunk_size: chunk_size, latest_id: latest_id})
+      return nil unless Cnvrg::CLI.is_response_success(response, false)
+      response['result']['files']
+    end
+
+    def download_files_in_chunks(files, chunk_size: 1000, conflict: false, commit: 'latest')
+      begin
+        ttl_files = 0
+        files.each_slice(chunk_size).each do |files|
+          ttl_files += download_files_chunk(files, conflict: conflict)
+        end
+        return Cnvrg::Result.new(true, "Download Completed")
+      rescue Exception => e
+        return Cnvrg::Result.new(false, "Can`t download files")
+      end
+    end
+
+    def download_files_chunk(files, conflict: false)
+      (1..5).each do |i|
+        response = Cnvrg::API.request("users/#{@owner}/datasets/#{@dataset_slug}/download_multi", 'POST', {files: files})
+        next unless Cnvrg::CLI.is_response_success(response, false) #trying to api request 5 times.
+        files = response['files']
+        data_home = "#{Dir.pwd}/#{response['name']}"
+        res = download_multiple_files_s3(files, data_home, conflict: conflict)
+        next unless res.is_success? #try again..
+        return files['keys'].length
+      end
+    end
+
+    def download_multiple_chunks(commit, chunk_size=1000)
+      last_chunk_size = chunk_size
+      q = { commit: commit, chunk_size: chunk_size}
+      overall = 0
+      while last_chunk_size > 0
+        response = Cnvrg::API.request("users/#{@owner}/datasets/#{@dataset_slug}/clone", 'POST', q)
+        if Cnvrg::CLI.is_response_success(response, false) 
+          files = response['files']
+          data_home = "#{Dir.pwd}/#{response['name']}"
+          last_chunk_size = files['keys'].length
+          break if last_chunk_size == 0
+          res = download_multiple_files_s3(files, data_home)
+          overall += last_chunk_size
+          q[:latest] = files['latest']
+        else
+          last_chunk_size = 0
+        end
+      end
+      Cnvrg::Result.new(true, "Cloned #{overall} files!")
+    end
+
+    def generate_parallel_idx
+      
+    end
+
+    def download_multiple_files_s3(files, project_home, conflict: false)
+      begin
+        props = Cnvrg::Helpers.get_s3_props(files)
+        client = props[:client]
+        iv = props[:iv]
+        key = props[:key]
+        bucket = props[:bucket]
+        download_succ_count = 0
+        parallel_options = {
+            in_threads: Cnvrg::Helpers.parallel_threads,
+            isolation: true
+        }
+        Parallel.map(files["keys"], parallel_options) do |f|
+
+          file_path = f["name"]
+          if file_path.end_with? "/"
+            # dir
+            if download_dir(file_path, file_path)
+              download_succ_count += 1
+            else
+              return Cnvrg::Result.new(false,"Could not create directory: #{file_path}")
+              raise Parallel::Kill
+            end
+          else
+            # blob
+            begin
+            file_key =  Cnvrg::Helpers.decrypt(key,iv, f["path"])
+            resp = false
+            file_path = "#{file_path}.conflict" if conflict
+            File.open(project_home+"/"+file_path, 'w+') do |file|
+              resp = client.get_object({bucket:bucket,
+                                        key:file_key}, target: file)
+            end
+            if resp
+              download_succ_count +=1
+            else
+              return Cnvrg::Result.new(false,"Could not create file: #{file_path}")
+            end
+            rescue => e
+              puts(e)
+              puts(e.backtrace)
+              return Cnvrg::Result.new(false,"Could not create file: #{file_path}", e.message, e.backtrace)
+              raise Parallel::Kill
+            end
+          end
+        end
+        if download_succ_count == files["keys"].size
+          return Cnvrg::Result.new(true,"Done.\nDownloaded #{download_succ_count} files")
+        end
+        rescue => e
+          return Cnvrg::Result.new(false,"Could not download some files", e.message, e.backtrace)
+        end
+      end
   end
 end

data/lib/cnvrg/dataset.rb

@@ -292,15 +292,47 @@ module Cnvrg
       url = Cnvrg::Helpers.remote_url
       "#{url}/#{self.owner}/datasets/#{self.slug}"
     end
+    def generate_chunked_idx(list_files=[], threads: IDXParallelThreads)
+      tree = {}
+      Parallel.map(list_files, in_threads: IDXParallelThreads) do |file|
+        label = file.gsub(self.local_path + "/", "")
+        if File.directory? file
+          tree[label+"/"] = nil
+        else
+          sha1 =  OpenSSL::Digest::SHA1.file(file).hexdigest
+          file_name = File.basename file
+          file_size = File.size(file).to_f
+          mime_type = MimeMagic.by_path(file)
+          content_type = !(mime_type.nil? or mime_type.text?) ? mime_type.type : "text/plain"
+          relative_path = file.gsub(/^#{@local_path + "/"}/, "")
+          tree[label] = {sha1: sha1, file_name: file_name, file_size: file_size, content_type: content_type, absolute_path: file, relative_path: relative_path}
+        end
+      end
+      return tree
+    end
+
+    def list_all_files
+      list = Dir.glob("#{self.local_path}/**/*", File::FNM_DOTMATCH).reject { |x| (x =~ /\/\.{1,2}$/) or (x =~ /^#{self.local_path}\/\.cnvrg\/*/) or (x =~/^#{self.local_path}\/\.cnvrgignore.conflict*/) and not (x =~/^#{self.local_path}\/\.cnvrgignore/) }
+      list_ignore = self.get_ignore_list()
+      return list.select{|file| !list_ignore.include? file}
+    end
+
+    def write_idx(tree=nil, commit=nil)
+      if tree.blank?
+        tree = self.generate_idx[:tree]
+        tree = tree.map{|k,v| (v.present?)? [k, {sha1: v[:sha1], commit_time: Time.now}] : [k,v]}.to_h
+      end
+      idx = {tree: tree, commit: commit}
+      File.open("#{self.local_path}/.cnvrg/idx.yml", 'w') { |f| f.write idx.to_yaml }
+    end
+
     def generate_idx(show_progress=false)
       if File.exists? "#{self.local_path}/.cnvrg/idx.yml"
         old_idx = YAML.load_file("#{self.local_path}/.cnvrg/idx.yml")
       else
         old_idx = nil
       end
-
       tree_idx = Hash.new(0)
-
       list = Dir.glob("#{self.local_path}/**/*", File::FNM_DOTMATCH).reject { |x| (x =~ /\/\.{1,2}$/) or (x =~ /^#{self.local_path}\/\.cnvrg\/*/) or (x =~/^#{self.local_path}\/\.cnvrgignore.conflict*/) and not (x =~/^#{self.local_path}\/\.cnvrgignore/) }
       list_ignore = self.get_ignore_list()
       if show_progress
@@ -348,10 +380,8 @@ module Cnvrg
         idx = {commit: old_idx.to_h[:commit], tree: tree_idx, next_commit:old_idx[:next_commit] }
       else
         idx = {commit: old_idx.to_h[:commit], tree: tree_idx}
-
       end
       idx_yaml = idx.to_yaml
-
       File.open("#{self.local_path}/.cnvrg/idx.yml", 'w') { |f| f.write idx_yaml }
       return idx
     end
@@ -361,9 +391,8 @@ module Cnvrg
       return response
     end
 
-    def downlowd_updated_data(current_commit)
-
-      response = Cnvrg::API.request("users/#{self.owner}/datasets/#{self.slug}/downlowd_updated_data", 'POST', {current_commit: current_commit})
+    def download_updated_data(current_commit)
+      response = Cnvrg::API.request("users/#{self.owner}/datasets/#{self.slug}/download_updated_data", 'POST', {current_commit: current_commit})
       CLI.is_response_success(response,false)
       return response
     end

data/lib/cnvrg/files.rb

@@ -455,58 +455,13 @@ module Cnvrg
 
         end
       end
-    def download_multiple_files_s3(files, project_home)
+    def download_multpile_files_s3(files, project_home)
       begin
-
-          sts_path = files["path_sts"]
-          retries = 0
-          success= false
-          while !success and retries < 20
-            begin
-              if !Helpers.is_verify_ssl
-                body = open(sts_path, {ssl_verify_mode: OpenSSL::SSL::VERIFY_NONE}).read
-              else
-                body = open(sts_path).read
-              end
-              success = true
-            rescue => e
-              retries +=1
-              sleep(5)
-
-            end
-          end
-          if !success
-            return Cnvrg::Result.new(false,"couldn't download some files", "error in sts", "" )
-          end
-          split = body.split("\n")
-          key = split[0]
-          iv = split[1]
-
-          access =  Cnvrg::Helpers.decrypt(key, iv, files["sts_a"])
-
-          secret =  Cnvrg::Helpers.decrypt(key,iv, files["sts_s"])
-
-          session =  Cnvrg::Helpers.decrypt(key,iv, files["sts_st"])
-          region =  Cnvrg::Helpers.decrypt(key,iv, files["region"])
-
-          bucket =  Cnvrg::Helpers.decrypt(key,iv, files["bucket"])
-          is_s3 = files["is_s3"]
-          if is_s3 or is_s3.nil?
-            client = Aws::S3::Client.new(
-                :access_key_id =>access,
-                :secret_access_key => secret,
-                :session_token => session,
-                :region => region,
-                :http_open_timeout => 60, :retry_limit => 20)
-          else
-            endpoint = Cnvrg::Helpers.decrypt(key,iv, files["endpoint_url"])
-            client = Aws::S3::Client.new(
-                :access_key_id =>access,
-                :secret_access_key => secret,
-                :region => region,
-                :endpoint=> endpoint,:force_path_style=> true,:ssl_verify_peer=>false,
-                :http_open_timeout => 60, :retry_limit => 20)
-          end
+          props = Cnvrg::Helpers.get_s3_props(files)
+          client = props[:client]
+          iv = props[:iv]
+          key = props[:key]
+          bucket = props[:bucket]
           download_succ_count = 0
           parallel_options = {
               :progress => {
@@ -539,11 +494,12 @@ module Cnvrg
               File.open(project_home+"/"+file_path, 'w+') do |file|
                 resp = client.get_object({bucket:bucket,
                                           key:file_key}, target: file)
+                #TODO update idx here!
               end
               if resp
                 download_succ_count +=1
               else
-                return Cnvrg::Result(false,"Could not create file: #{file_path}")
+                return Cnvrg::Result.new(false,"Could not create file: #{file_path}")
               end
 
 
@@ -697,10 +653,9 @@ module Cnvrg
       return true
 
     end
-      def start_commit(new_branch,force:false, exp_start_commit:nil)
-
+      def start_commit(new_branch,force:false, exp_start_commit:nil, job_slug: nil, job_type: nil)
         response = Cnvrg::API.request("#{base_resource}/commit/start", 'POST', {project_slug: @project_slug, new_branch: new_branch,force:force,
-                                                                                username: @owner, exp_start_commit:exp_start_commit})
+                                                                                username: @owner, exp_start_commit:exp_start_commit, job_slug: job_slug, job_type: job_type})
         Cnvrg::CLI.is_response_success(response,false)
         return response
       end

data/lib/cnvrg/helpers.rb

@@ -219,5 +219,61 @@ parameters:
     def get_mem(pid)
     end
 
+    def get_s3_props(files) #will return client and decryptor
+      sts_path = files["path_sts"]
+      retries = 0
+      success= false
+      while !success and retries < 20
+        begin
+          if !Helpers.is_verify_ssl
+            body = open(sts_path, {ssl_verify_mode: OpenSSL::SSL::VERIFY_NONE}).read
+          else
+            body = open(sts_path).read
+          end
+          success = true
+        rescue => e
+          retries +=1
+          sleep(5)
+
+        end
+      end
+      if !success
+        return Cnvrg::Result.new(false,"couldn't download some files", "error in sts", "" )
+      end
+      split = body.split("\n")
+      key = split[0]
+      iv = split[1]
+
+      access =  Cnvrg::Helpers.decrypt(key, iv, files["sts_a"])
+
+      secret =  Cnvrg::Helpers.decrypt(key,iv, files["sts_s"])
+
+      session =  Cnvrg::Helpers.decrypt(key,iv, files["sts_st"])
+      region =  Cnvrg::Helpers.decrypt(key,iv, files["region"])
+
+      bucket =  Cnvrg::Helpers.decrypt(key,iv, files["bucket"])
+      is_s3 = files["is_s3"]
+      if is_s3 or is_s3.nil?
+        client = Aws::S3::Client.new(
+            :access_key_id =>access,
+            :secret_access_key => secret,
+            :session_token => session,
+            :region => region,
+            :http_open_timeout => 60, :retry_limit => 20)
+        use_accelerate_endpoint = true
+      else
+        endpoint = Cnvrg::Helpers.decrypt(key,iv, files["endpoint_url"])
+        client = Aws::S3::Client.new(
+            :access_key_id =>access,
+            :secret_access_key => secret,
+            :region => region,
+            :endpoint=> endpoint,:force_path_style=> true,:ssl_verify_peer=>false,
+            :http_open_timeout => 60, :retry_limit => 20)
+        use_accelerate_endpoint = false
+      end
+      upload_options = {:use_accelerate_endpoint => use_accelerate_endpoint,:server_side_encryption => 'AES256'}
+      return {client: client, key: key, iv: iv, bucket: bucket, upload_options: upload_options}
+    end
+
   end
 end