ebulk: new features and fixes

- minor fixes in storage inputs
- retry on failed connections
- resume split operation for download
- control on first ingestion / partial ingestion
- report messages when storage ingestion finishes
- better control on exception and messages

ebulk

@@ -151,13 +151,19 @@ function updateConfigFile {
     	echo
 	OPTION=""
     	if [ "$ADVANCED" = true ] ; then
-    		echo "[INFO] If you want to edit the configuration file by yourself (advanced) please type CONFIG, otherwise press enter to continue."
+		echo "[INFO] If you want to edit the configuration file by yourself (advanced) please type YES, otherwise press enter to continue."
+		echo -n "** Open configuration file? ('YES' to confirm): "
 		read -e OPTION
-    		if [ "$OPTION" = "CONFIG" ] ; then
+		if [ "$OPTION" = "YES" ] ; then
         		CUSTOM=true
     		fi
     	fi
-	if [ "$OPTION" != "CONFIG" ] ; then
+	if [ "$OPTION" != "YES" ] ; then
+		if [ "$ADVANCED" = true ] ; then
+			echo
+			echo "** Please enter the information needed to configure the storage **"
+			echo
+		fi
 	    	$PARAMETER_FUNCTION
 	    	if [ "$ADVANCED" = false ] ; then
 			echo "** If you need a more advanced storage configuration, you can run the tool with the parameter --advanced **"
@@ -231,7 +237,7 @@ function runProcess {
     if [ ! -d $LOG_DIR ]; then
         mkdir $LOG_DIR 2>/dev/null
     fi
-    if [ -z "$RESUME_STORAGE_INGESTION" ]; then
+    if [ "$RESTART_STORAGE_INGESTION" ]; then
 	rm -f ${DIFF_FILE} 2>/dev/null
     fi
     $embulk run -I $TOOL_PATH/embulk-wendelin-dataset-tool/lib $FILE $DIFF_COMMAND 2> "$LOG_DIR/error.log" || {
@@ -322,9 +328,7 @@ function checkStoragePlugin {
     if [ "$STORAGE_GEM" != "" ] ; then
       echo -n "[INFO] Checking if '$STORAGE' plugin is installed... "
       if [ "$CUSTOM" = false ] ; then
-        sleep 1
         echo -e "${GREEN}OK${NC}"
-	sleep 1
       else
 	$embulk gem list 2>/dev/null | grep -q "$STORAGE_GEM" 2>/dev/null
 	if [ $? == 0 ]; then
@@ -386,12 +390,15 @@ function askFTPparameters {
         echo -e "${ORANGE}[ERROR] Empty host.${NC}"
 	exit
     fi
+    FTP_HOST="${FTP_HOST/ftp:\/\//$empty}"
+    if [[ $FTP_HOST == *"/"* ]]; then
+        echo -e "${ORANGE}[ERROR] Please, enter only the ftp host, without '/' or path. Path will be requested after.${NC}"
+	exit
+    fi
     echo "Path prefix:"
     echo "* (e.g. /mydata/sample/dir/) *"
     read -e FTP_PATH
-    if [ "$FTP_PATH" = "" ] ; then
-        FTP_PATH="/"
-    fi
+    FTP_PATH="/$FTP_PATH"
     echo "FTP User:"
     echo "* you can leave this input empty and anonymous authentication will be used  *"
     read -e FTP_USER
@@ -493,7 +500,7 @@ while [ "$1" != "" ]; do
 					;;
 	-a | --advanced )		ADVANCED=true
 					;;
-	-rs | --resume )		RESUME_STORAGE_INGESTION=true
+	-rs | --restart )		RESTART_STORAGE_INGESTION=true
 					;;
 	-dc | --discard-changes )	DISCARD_CHANGES=true
 					;;

ebulk-data/embulk-wendelin-dataset-tool/lib/embulk/dataset_utils.rb

@@ -9,9 +9,12 @@ class DatasetUtils
   DATASET_TEMP_REPORT_FILE = ".temp-dataset-task-report"
   DATASET_COMPLETED_FILE = ".dataset-completed"
   RESUME_OPERATION_FILE = ".resume-operation"
-  INITIAL_INGESTION_FILE = ".initial-ingestion"
+  PARTIAL_INGESTION_FILE = ".partial-ingestion"
   STAGED_FILE = ".staged"
   DISCARD_CHANGES_FILE = ".discard-changes"
+  SPLIT_FILE = ".split-operation"
+  SPLIT_CONTROL_FILE = ".control-split-operation"
+  FIRST_INGESTION_FILE = ".first-ingestion"
 
   RUN_DONE = "done"
   RUN_ERROR = "error"
@@ -19,6 +22,7 @@ class DatasetUtils
   DELETE = "DELETE"
   RENAME = "RENAME"
   INGESTION = "ingestion"
+  DOWNLOAD = "download"
   ADD = "add"
   REMOVE = "remove"
   STATUS_NEW = "new"
@@ -45,15 +49,18 @@ class DatasetUtils
   DATE_FORMAT = "%Y-%m-%d-%H-%M-%S"
 
   def initialize(data_set_directory)
-    @data_set_directory = data_set_directory
+    @data_set_directory = appendSlashTo(data_set_directory)
     @logger = LogManager.instance()
     @task_report_file = @data_set_directory + DATASET_REPORT_FILE
     @temp_report_file = @data_set_directory + DATASET_TEMP_REPORT_FILE
     @completed_file = @data_set_directory + DATASET_COMPLETED_FILE
     @resume_operation_file = @data_set_directory + RESUME_OPERATION_FILE
-    @initial_ingestion_file = @data_set_directory + INITIAL_INGESTION_FILE
+    @partial_ingestion_file = @data_set_directory + PARTIAL_INGESTION_FILE
+    @first_ingestion_file = @data_set_directory + FIRST_INGESTION_FILE
     @staged_file = @data_set_directory + STAGED_FILE
     @discard_changes_file = @data_set_directory + DISCARD_CHANGES_FILE
+    @split_file = @data_set_directory + SPLIT_FILE
+    @split_control_file = @data_set_directory + SPLIT_CONTROL_FILE
   end
 
   def getLocalPaths(paths)
@@ -207,7 +214,7 @@ class DatasetUtils
     end
   end
 
-  def showTaskErrors(failed_tasks)
+  def showTaskErrors(failed_tasks, storage=FALSE)
     puts
     @logger.error("The following files could not be processed. Please check the details in the log file: " + @logger.getLogPath(), print=TRUE)
     if failed_tasks.length > 15
@@ -217,10 +224,77 @@ class DatasetUtils
     else
       @logger.error(failed_tasks, print=TRUE)
     end
+    if storage
+      puts
+      @logger.info("You can retry the operation for those files by running your command again with parameter --restart .", print=TRUE)
+    else
       @logger.info("You can retry the operation for those files.", print=TRUE)
+    end
     puts
   end
 
+  def splitOperationControlFileExist(reference)
+    file_reference = reference.gsub('/', '__')
+    return File.exist?(@split_control_file + "__" + file_reference)
+  end
+
+  def createSplitOperationControlFile(reference)
+    file_reference = reference.gsub('/', '__')
+    File.open(@split_control_file + "__" + file_reference, 'w') {}
+  end
+
+  def deleteSplitOperationControlFile(reference=nil)
+    if reference.nil?
+      Dir.glob("#{@split_control_file}*").each { |file| File.delete(file) }
+    else
+      file_reference = reference.gsub('/', '__')
+      File.delete(@split_control_file + "__" + file_reference) if File.exist?(@split_control_file + "__" + file_reference)
+    end
+  end
+
+  def deleteSplitOperationFile(reference=nil)
+    if reference.nil?
+      Dir.glob("#{@split_file}*").each { |file| File.delete(file) }
+    else
+      file_reference = reference.gsub('/', '__')
+      File.delete(@split_file + "__" + file_reference) if File.exist?(@split_file + "__" + file_reference)
+    end
+  end
+
+  def saveSplitOperation(operation, reference, eof, hash, chunk_size)
+    file_reference = reference.gsub('/', '__')
+    record = [operation, reference, eof, hash, Integer(chunk_size)].join(RECORD_SEPARATOR)
+    File.open(@split_file + "__" + file_reference, 'w') { |file| file.puts(record) }
+    deleteSplitOperationControlFile(reference)
+  end
+
+  def splitOperationFileExist(reference)
+    file_reference = reference.gsub('/', '__')
+    return File.exist?(@split_file + "__" + file_reference)
+  end
+
+  def getLastSplitOperation(operation, reference, hash, chunk_size)
+    file_reference = reference.gsub('/', '__')
+    return 0 if not File.exist?(@split_file + "__" + file_reference)
+    record = File.open(@split_file + "__" + file_reference).read.chomp.split(RECORD_SEPARATOR)
+    if record[0] == operation && record[1] == reference && record[3] == hash && record[4] == Integer(chunk_size).to_s && record[2] != EOF
+      # discard if user interrupted (ctrl+c) the operation
+      if splitOperationControlFileExist(file_reference)
+	@logger.warn("Previous split operation attempt for file #{reference} was interrupt by user (aborted tool execution), it will be restarted.", print=TRUE)
+	deleteSplitOperationFile(file_reference)
+	deleteSplitOperationControlFile(file_reference)
+	return 0
+      end
+      createSplitOperationControlFile(file_reference)
+      return record[2].to_i
+    end
+    return 0
+  rescue Exception => e
+    @logger.error("An error occurred in getLastSplitOperation method:" + e.to_s)
+    @logger.error(e.backtrace)
+    return 0
+  end
+
   def deleteDiscardChangesFile()
     File.delete(@discard_changes_file) if File.exist?(@discard_changes_file)
   end
@@ -255,16 +329,28 @@ class DatasetUtils
     end
   end
 
-  def deleteInitialIngestionFile()
-    File.delete(@initial_ingestion_file) if File.exist?(@initial_ingestion_file)
+  def deleteFirstIngestionFile()
+    File.delete(@first_ingestion_file) if File.exist?(@first_ingestion_file)
+  end
+
+  def createFirstIngestionFile()
+    File.open(@first_ingestion_file, 'w') {}
+  end
+
+  def firstIngestionFileExist()
+    return File.exist?(@first_ingestion_file)
+  end
+
+  def deletePartialIngestionFile()
+    File.delete(@partial_ingestion_file) if File.exist?(@partial_ingestion_file)
   end
 
-  def createInitialIngestionFile()
-    File.open(@initial_ingestion_file, 'w') {}
+  def createPartialIngestionFile()
+    File.open(@partial_ingestion_file, 'w') {}
   end
 
-  def initialIngestionFileExist()
-    return File.exist?(@initial_ingestion_file)
+  def partialIngestionFileExist()
+    return File.exist?(@partial_ingestion_file)
   end
 
   def stagedFileExist()

ebulk-data/embulk-wendelin-dataset-tool/lib/embulk/input/fif.rb

@@ -22,7 +22,7 @@ module Embulk
 	       ]
 
       def self.status(task, push=FALSE)
-	partial_ingestion = @dataset_utils.initialIngestionFileExist()
+	partial_ingestion = @dataset_utils.partialIngestionFileExist()
 	staged_changes, untracked_changes = @dataset_utils.getLocalChanges(task['paths'], task['data_set'], staged=TRUE, partial_ingestion=partial_ingestion)
 	staged = (not staged_changes.empty?)
 	task['paths'] = staged ? staged_changes : untracked_changes
@@ -96,7 +96,7 @@ module Embulk
 	  @status = @status == "" ? FALSE : @status
 	  @dataset_utils.deleteDiscardChangesFile()
 	  if @status
-	    if not @dataset_utils.initialIngestionFileExist()
+	    if not @dataset_utils.partialIngestionFileExist()
 	      if not @dataset_utils.reportFileExist()
 		puts
 		@logger.error("The dataset directory does not contain a valid dataset.", print=TRUE)
@@ -123,11 +123,12 @@ module Embulk
 	    @logger.abortExecution()
 	  end
 	  task['data_streams'] = data_stream_dict["result"]
-
+	  first_ingestion = task['data_streams'].length == 0
 	  if not @dataset_utils.reportFileExist()
-	    @dataset_utils.createInitialIngestionFile()
+	    @dataset_utils.createPartialIngestionFile()
+	    @dataset_utils.createFirstIngestionFile() if first_ingestion
 	  else
-	    if not @dataset_utils.initialIngestionFileExist()
+	    if not @dataset_utils.partialIngestionFileExist()
 	      @logger.info("Checking local dataset...", print=TRUE)
 	      if not @dataset_utils.reportUpToDate(data_stream_dict, @data_set)
 		puts
@@ -188,12 +189,17 @@ module Embulk
 	task_reports = yield(task, columns, count)
 	next_config_diff = task_reports.map{|hash| hash[DatasetUtils::RUN_DONE]}.flatten.compact
 	@dataset_utils.showTaskReport(next_config_diff)
-	element_output = @dataset_utils.initialIngestionFileExist() ? "new file" : "change"
+	element_output = @dataset_utils.partialIngestionFileExist() ? "new file" : "change"
 	@logger.info("#{next_config_diff.length} #{element_output}(s) ingested.", print=TRUE)
 	if(next_config_diff.length == count)
 	  @logger.info("Dataset successfully ingested.", print=TRUE)
 	  @wendelin.increaseDatasetVersion(@data_set)
 	  @dataset_utils.deleteStagedFile()
+	  if @dataset_utils.firstIngestionFileExist()
+	    @dataset_utils.createCompletedFile()
+	    @dataset_utils.deletePartialIngestionFile()
+	    @dataset_utils.deleteFirstIngestionFile()
+	  end
 	else
 	  failed_tasks = task_reports.map{|hash| hash[DatasetUtils::RUN_ERROR] || hash[DatasetUtils::RUN_ABORTED] }.flatten.compact
 	  @dataset_utils.showTaskErrors(failed_tasks)
@@ -221,6 +227,7 @@ module Embulk
 	  hash = file_dict["hash"]
 	  delete = hash == DatasetUtils::DELETE
 	  rename = file_dict["status"] == DatasetUtils::STATUS_RENAMED
+	  split = FALSE
 	  if size == "" and hash == "" #new file
 	    size = File.size(path)
 	    hash = @dataset_utils.getHash(path)
@@ -229,10 +236,17 @@ module Embulk
 	  filename, extension, reference = @dataset_utils.getPathInfo(path, @dataset)
 	  operation = rename ? DatasetUtils::RENAME : DatasetUtils::INGESTION
 	  @dataset_utils.saveCurrentOperation(operation, reference, new_reference)
-	  each_chunk(path, filename, extension, size, hash, schema[1..-1].map{|elm| elm.name}, @chunk_size, delete, new_reference) do |entry|
+	  resume_split = @dataset_utils.splitOperationFileExist(reference) ? @dataset_utils.getLastSplitOperation(operation, reference, hash, @chunk_size) : 0
+	  each_chunk(path, filename, extension, size, hash, schema[1..-1].map{|elm| elm.name}, @chunk_size, delete, new_reference, resume_split) do |entry|
+	    @dataset_utils.createSplitOperationControlFile(reference) if split
 	    @page_builder.add(entry)
+	    if ! delete && ! rename && entry[5] != ""
+	      split = TRUE
+	      @dataset_utils.saveSplitOperation(operation, reference, entry[5], hash, @chunk_size)
+	    end
 	  end
 	  @page_builder.finish
+	  @dataset_utils.deleteSplitOperationFile(reference) if split
 	rescue java.lang.OutOfMemoryError
 	  @logger.logOutOfMemoryError(path)
 	  return_value = DatasetUtils::RUN_ABORTED
@@ -244,6 +258,7 @@ module Embulk
 	else
 	  return_value = DatasetUtils::RUN_DONE
 	end
+	@dataset_utils.deleteSplitOperationControlFile(reference) if return_value != DatasetUtils::RUN_DONE
 	if return_value == DatasetUtils::RUN_DONE
 	  if delete
 	    if @dataset_utils.reportFileExist()
@@ -261,7 +276,7 @@ module Embulk
 
       private
 
-      def each_chunk(path, filename, extension, size, hash, fields, chunk_size=DatasetUtils::CHUNK_SIZE, delete=FALSE, new_reference=FALSE)
+      def each_chunk(path, filename, extension, size, hash, fields, chunk_size=DatasetUtils::CHUNK_SIZE, delete=FALSE, new_reference=FALSE, resume_split=0)
 	if delete
 	  File.delete(path) if File.exist?(path)
 	  values = [@supplier, @dataset, filename, extension, "", DatasetUtils::DELETE, "", ""]
@@ -284,13 +299,19 @@ module Embulk
 		end
 		break
 	      end
+	      if resume_split > 0
+		@logger.info("Resuming interrupted split ingestion", print=TRUE)
+		first = FALSE
+		file_object.seek(chunk_size*resume_split+resume_split) # ignore ingested part
+		next_byte = file_object.read(1)
+		data = next_byte
+		npart = resume_split
+		resume_split = 0
+	      end
 	      data += file_object.read(chunk_size)
 	      next_byte = file_object.read(1)
 	      if not next_byte
-		  eof = DatasetUtils::EOF
-		  if first # this means that the whole file will be ingested at once (not split)
-		    eof = ""
-		  end
+		eof = first ? "" : DatasetUtils::EOF
 	      else
 		npart += 1
 		eof = npart.to_s.rjust(3, "0")

ebulk-data/embulk-wendelin-dataset-tool/lib/embulk/input/wendelin.rb

@@ -68,6 +68,8 @@ module Embulk
 	      @logger.info("Your downloaded dataset is already up to date.", print=TRUE)
 	    end
 	  when DOWNLOAD
+	    @dataset_utils.deleteSplitOperationControlFile()
+	    @dataset_utils.deleteSplitOperationFile()
 	    @logger.info("Checking remote files and posible local conflicts...", print=TRUE)
 	    self.warnConflicts(task['data_streams'], task['data_set'])
 	    @dataset_utils.deleteCompletedFile()
@@ -112,7 +114,7 @@ module Embulk
 	  task['data_set_directory'] = @dataset_utils.appendSlashTo(@output_path)
 	  @data_set_directory = task['data_set_directory']
 	  @dataset_utils = DatasetUtils.new(@data_set_directory)
-	  if @dataset_utils.reportFileExist() && @dataset_utils.completedFileExist() && @dataset_utils.discardChangesFileExist() && ! @dataset_utils.initialIngestionFileExist()
+	  if @dataset_utils.reportFileExist() && @dataset_utils.completedFileExist() && @dataset_utils.discardChangesFileExist() && ! @dataset_utils.partialIngestionFileExist()
 	    task['discard_changes'] = @dataset_utils.discardChangesFileExist()
 	    local_changes = @dataset_utils.getRemoteFileListForDiscardLocalChanges([], @data_set, check_changes=TRUE)
 	    if local_changes.empty?
@@ -159,7 +161,7 @@ module Embulk
 		puts
 		self.askUserForAction(task, action=UPDATE)
 	      end
-	    elsif not @dataset_utils.initialIngestionFileExist()
+	    elsif not @dataset_utils.partialIngestionFileExist()
 	      puts
 	      @logger.info("There was a previous attempt to download this dataset but it did not finish successfully.", print=TRUE)
 	      @logger.info("What do you want to do?", print=TRUE)
@@ -195,7 +197,7 @@ module Embulk
 	    end
 	    @dataset_utils.createReportFile()
 	  end
-	  @dataset_utils.deleteInitialIngestionFile()
+	  @dataset_utils.deletePartialIngestionFile()
 	  @dataset_utils.deleteDiscardChangesFile()
 	  columns = [
 	    Column.new(0, "reference", :string),
@@ -254,34 +256,42 @@ module Embulk
       def run
 	data_stream = task['data_streams'][@index]
 	id = data_stream["id"]
-	ref = data_stream["reference"]
+	reference = data_stream["reference"]
 	size = data_stream["size"]
 	hash = data_stream["hash"]
 	renamed = data_stream["status"] == DatasetUtils::STATUS_RENAMED
 	deleted = hash.to_s == DatasetUtils::DELETE
 	begin
 	  if deleted
-	    entry = [ref, "", @data_set, DatasetUtils::DELETE, renamed]
+	    entry = [reference, "", @data_set, DatasetUtils::DELETE, renamed]
 	    page_builder.add(entry)
 	  elsif renamed
 	    new_reference = data_stream["new_reference"]
-	    entry = [ref, new_reference, @data_set, TRUE, renamed]
+	    entry = [reference, new_reference, @data_set, TRUE, renamed]
   	    page_builder.add(entry)
 	  else
 	    @logger.info("Discarding local change on '#{data_stream["path"]}'", print=TRUE) if task['discard_changes']
-	    @logger.info("Getting content from remote #{ref}", print=TRUE)
-	    n_chunk = 0
-	    @wendelin.eachDataStreamContentChunk(id, @chunk_size) do |chunk|
+	    @logger.info("Getting content from remote #{reference}", print=TRUE)
+	    @logger.info("Downloading...", print=TRUE)
+	    resume_split = @dataset_utils.splitOperationFileExist(reference) ? @dataset_utils.getLastSplitOperation(DatasetUtils::DOWNLOAD, reference, hash, @chunk_size) : 0
+	    n_chunk = resume_split == 0 ? 0 : resume_split+1
+	    split = n_chunk > 0
+	    @logger.info("Resuming interrupted split download...", print=TRUE) if split
+	    @wendelin.eachDataStreamContentChunk(id, @chunk_size, n_chunk) do |chunk|
 	      content = chunk.nil? || chunk.empty? ? "" : Base64.encode64(chunk)
 	      begin_of_file = n_chunk == 0
-	      entry = [ref, content, @data_set, begin_of_file, renamed]
+	      split = n_chunk > 0
+	      @dataset_utils.createSplitOperationControlFile(reference) if split
+	      entry = [reference, content, @data_set, begin_of_file, renamed]
 	      page_builder.add(entry)
+	      @dataset_utils.saveSplitOperation(DatasetUtils::DOWNLOAD, reference, n_chunk, hash, @chunk_size) if split
 	      n_chunk += 1
 	    end
 	  end
 	  page_builder.finish
+	  @dataset_utils.deleteSplitOperationFile(reference) if split
 	rescue java.lang.OutOfMemoryError
-	  @logger.logOutOfMemoryError(ref)
+	  @logger.logOutOfMemoryError(reference)
 	  return_value = DatasetUtils::RUN_ABORTED
 	rescue Exception => e  
 	  @logger.error(e.to_s, print=TRUE)
@@ -291,14 +301,15 @@ module Embulk
 	else
 	  return_value = DatasetUtils::RUN_DONE
 	end
+	@dataset_utils.deleteSplitOperationControlFile(reference) if return_value != DatasetUtils::RUN_DONE
 	if return_value == DatasetUtils::RUN_DONE
 	  if deleted
-	    @dataset_utils.deleteFromReport(ref, return_value)
+	    @dataset_utils.deleteFromReport(reference, return_value)
 	  else
-	    @dataset_utils.addToReport(ref, return_value, size, hash, task['data_set'], new_reference)
+	    @dataset_utils.addToReport(reference, return_value, size, hash, task['data_set'], new_reference)
 	  end
 	end
-	return {return_value => ref}
+	return {return_value => reference}
       end
     end
   end

ebulk-data/embulk-wendelin-dataset-tool/lib/embulk/output/wendelin.rb

@@ -18,18 +18,31 @@ module Embulk
 	  "data_set" => config.param("data_set", :string, default: nil),
 	  "erp5_base_url" => config.param("erp5_base_url", :string, default: nil)
 	}
+	storage_ingestion = ! task["type_input"]
 	task_reports = yield(task)
 	next_config_diff = {}
 	@logger = LogManager.instance()
+	@dataset_utils = DatasetUtils.new(Dir.pwd)
 	if task_reports.length > 0
 	  @logger.info("Your ingested files will be available in the site in a few minutes. Thank for your patience.", print=TRUE)
-	  # if ingestion was done from a storage different than filesystem, increase dataset version
-	  if not task["type_input"] and task["data_set"] and task["erp5_base_url"]
+	  if storage_ingestion
+	    done_tasks = task_reports.map{|hash| hash[DatasetUtils::RUN_DONE]}.flatten.compact
+	    @dataset_utils.showTaskReport(done_tasks)
+	    @logger.info("#{done_tasks.length} new file(s) ingested.", print=TRUE)
+	    if(done_tasks.length == count)
+	      @logger.info("Dataset successfully ingested.", print=TRUE)
 	      @wendelin = WendelinClient.new(task["erp5_base_url"], task["user"], task["password"])
 	      @wendelin.increaseDatasetVersion(task["data_set"])
+	    else
+	      failed_tasks = task_reports.map{|hash| hash[DatasetUtils::RUN_ERROR] || hash[DatasetUtils::RUN_ABORTED] }.flatten.compact
+	      @dataset_utils.showTaskErrors(failed_tasks, storage=TRUE)
+	    end
 	  end
 	else
-	  @logger.info("No new files where processed for ingestion.", print=TRUE)
+	  @logger.info("No new files found for ingestion.", print=TRUE)
+	  if storage_ingestion
+	    @logger.info("You can restart the ingestion from this storage by running your command again with parameter --restart .", print=TRUE)
+	  end
 	end
 	return next_config_diff
       end
@@ -43,11 +56,9 @@ module Embulk
 	@wendelin = WendelinClient.new(@erp5_url, @user, @password)
       end
 
-      def close
-      end
-
       def add(page)
 	page.each do |record|
+	  @return_value = DatasetUtils::RUN_ERROR
 	  supplier = (record[0].nil? || record[0].empty?) ? "default" : record[0]
 	  dataset = (record[1].nil? || record[1].empty?) ? "default" : record[1]
 	  filename = record[2]
@@ -58,23 +69,28 @@ module Embulk
 	  hash = record[7]
 	  begin
 	    if eof == DatasetUtils::DELETE
-	      reference = [dataset, filename, extension].join(DatasetUtils::REFERENCE_SEPARATOR)
-	      @wendelin.delete(reference)
+	      @reference = [dataset, filename, extension].join(DatasetUtils::REFERENCE_SEPARATOR)
+	      @wendelin.delete(@reference)
 	    elsif eof == DatasetUtils::RENAME
-	      reference = [dataset, filename, extension].join(DatasetUtils::REFERENCE_SEPARATOR)
-	      @wendelin.rename(reference, record[4].to_s)
+	      @reference = [dataset, filename, extension].join(DatasetUtils::REFERENCE_SEPARATOR)
+	      @wendelin.rename(@reference, record[4].to_s)
 	    else
-	      reference = [supplier, dataset, filename, extension, eof, size, hash].join(DatasetUtils::REFERENCE_SEPARATOR)
+	      @reference = [supplier, dataset, filename, extension, eof, size, hash].join(DatasetUtils::REFERENCE_SEPARATOR)
 	      split = eof != ""
-	      if not @wendelin.ingest(reference, data_chunk, split)
-		raise "could not ingest"
+	      ingestion_response = @wendelin.ingest(@reference, data_chunk, split)
+	      if not ingestion_response["success"]
+		raise ingestion_response["message"]
 	      end
 	    end
 	  rescue Exception => e
-	    raise e
 	    @logger.error(e.backtrace)
+	    raise e
+	  end
+	  @return_value = DatasetUtils::RUN_DONE
 	end
       end
+
+      def close
       end
 
       def finish
@@ -84,7 +100,7 @@ module Embulk
       end
 
       def commit
-	task_report = {}
+	task_report = {@return_value => @reference}
 	return task_report
       end
     end

ebulk-data/embulk-wendelin-dataset-tool/lib/embulk/parser/binary.rb

 require_relative '../filelogger'
 require_relative '../dataset_utils'
+require_relative '../wendelin_client'
 
 class Index
   include Singleton
@@ -53,8 +54,10 @@ module Embulk
       def run(file_input)
 	@index = Index.instance().get()
 	@logger = LogManager.instance()
+	split = FALSE
 	while file = file_input.next_file
 	  begin
+	    @dataset_utils = DatasetUtils.new(Dir.pwd)
 	    metadata_file = Dir.pwd + METADATA_FILE_NAME
 	    metadata = File.open(metadata_file) {|f| f.readline} if File.exist?(metadata_file)
 	    File.delete(metadata_file) if File.exist?(metadata_file)
@@ -72,43 +75,65 @@ module Embulk
 	      filename = "file_from_#{task['input_plugin']}_#{task['date']}"
 	      extension = @index.to_s.rjust(3, "0")
 	    end
-	    each_chunk(file, filename.chomp, extension.chomp, task['chunk_size']) do |record|
+	    reference = [task['data_set'], filename, extension.chomp].join(DatasetUtils::REFERENCE_SEPARATOR)
+	    resume_split = @dataset_utils.splitOperationFileExist(reference) ? @dataset_utils.getLastSplitOperation(DatasetUtils::INGESTION, reference, "", task['chunk_size']) : 0
+	    each_chunk(file, filename.chomp, extension.chomp, task['chunk_size'], resume_split) do |record|
+	      @dataset_utils.createSplitOperationControlFile(reference) if split
 	      @page_builder.add(record)
+	      if record[5] != ""
+		split = TRUE
+		@dataset_utils.saveSplitOperation(DatasetUtils::INGESTION, reference, record[5], "", task['chunk_size'])
+	      end
 	    end
 	    @page_builder.finish
 	    Index.instance().increase()
+	    @dataset_utils.deleteSplitOperationFile(reference) if split
 	  rescue java.lang.OutOfMemoryError
-	    @logger.logOutOfMemoryError(path)
+	    @logger.logOutOfMemoryError(reference)
+	    @dataset_utils.deleteSplitOperationControlFile(reference)
 	    @logger.abortExecution()
 	  rescue Exception => e
 	    @logger.error("An error occurred during file ingestion: " + e.to_s, print=TRUE)
 	    @logger.error(e.backtrace)
 	    puts "[INFO] For more detailed information, please refer to the log file: " + @logger.getLogPath()
+	    @dataset_utils.deleteSplitOperationControlFile(reference)
+	    @logger.abortExecution(error=FALSE) if (e.to_s.include? '401' or e.to_s.include? 'Unauthorized')
+	    raise e if not e.to_s.include? WendelinClient::HTTP_REFERENCE_EXIST
 	  end
 	end
       end
 
       private
-      def each_chunk(file, filename, extension, chunk_size=DatasetUtils::CHUNK_SIZE)
+      def each_chunk(stream, filename, extension, chunk_size=DatasetUtils::CHUNK_SIZE, resume_split=0)
 	npart = 0
-	next_byte = file.read(1)
+	next_byte = stream.read(1)
 	first = TRUE
 	while true
 	    data = next_byte
 	    if not next_byte
 		if first
-		  # this means this is an empty file
+		  # this means this is an empty stream
 		  values = [task['supplier'], task['data_set'], filename, extension, "", "", "", ""]
 		  yield(values)
 		end
 		break
 	    end
-	    data += file.read(chunk_size)
-	    next_byte = file.read(1)
+	    if resume_split > 0
+	      @logger.warn("Large files resuming could take some minutes", print=TRUE) if resume_split > 10
+	      @logger.info("Resuming interrupted split ingestion...", print=TRUE)
+	      first = FALSE
+	      discardIngestedPart(stream, resume_split, chunk_size)
+	      next_byte = stream.read(1)
+	      data = next_byte
+	      npart = resume_split
+	      resume_split = 0
+	    end
+	    data += stream.read(chunk_size)
+	    next_byte = stream.read(1)
 	    if not next_byte
 		eof = DatasetUtils::EOF
 		if first
-		  # this means that the whole file will be ingested at once (not split)
+		  # this means that the whole stream will be ingested at once (not split)
 		  eof = ""
 		end
 	    else
@@ -122,6 +147,16 @@ module Embulk
 	end
       end
 
+      def discardIngestedPart(stream, resume_split, chunk_size)
+	read = 0
+	while read < resume_split
+	  stream.read(chunk_size)
+	  read += 1
+	  sleep 1
+	end
+	stream.read(resume_split-1)
+      end
+
     end
   end
 end

ebulk-data/embulk-wendelin-dataset-tool/lib/embulk/wendelin_client.rb

@@ -7,6 +7,14 @@ require_relative 'filelogger'
 # class representing a Wendelin client
 class WendelinClient
 
+  HTTP_MESSAGE_401 = "Unauthorized access. Please check your user credentials and try again."
+  HTTP_MESSAGE_5XX = "Internal Server Error: if the error persists, please contact the administrator."
+  HTTP_MESSAGE_OTHER = "Sorry, an error ocurred. If the error persists, please contact the administrator."
+  HTTP_MESSAGE_NOT_2XX = "HTTP-NOT-OK"
+  HTTP_MESSAGE_EXCEPTION = "HTTP-ERROR"
+  HTTP_MEMORY_ERROR = "MEMORY-ERROR"
+  HTTP_REFERENCE_EXIST = "REFERENCE-EXIST"
+
   def initialize(erp5_url, user, password)
     @erp5_url = erp5_url
     @user = user
@@ -44,7 +52,7 @@ class WendelinClient
     checkReferenceChars(reference)
     uri = URI(URI.escape("#{@erp5_url}/ERP5Site_invalidateIngestionObjects?reference=#{reference}"))
     res = handleRequest(uri)
-    if res == FALSE
+    if res["success"] == FALSE
       @logger.abortExecution()
     end
     @logger.info("Remote file successfully ingested.", print=TRUE)
@@ -56,7 +64,7 @@ class WendelinClient
     checkReferenceChars(new_reference)
     uri = URI(URI.escape("#{@erp5_url}/ERP5Site_renameIngestion?reference=#{reference}&new_reference=#{new_reference}"))
     res = handleRequest(uri)
-    if res == FALSE
+    if res["success"] == FALSE
       @logger.abortExecution()
     end
     @logger.info("Remote file successfully renamed.", print=TRUE)
@@ -67,8 +75,8 @@ class WendelinClient
       @logger.warn("Could not increase data set version because dataset reference is empty.")
     else
       @logger.info("Increasing dataset version")
-      uri = URI(URI.escape("#{@erp5_url}/ERP5Site_increaseDatasetVersion?reference=#{reference}"))
       begin
+	uri = URI(URI.escape("#{@erp5_url}/ERP5Site_increaseDatasetVersion?reference=#{reference}"))
 	res = open(uri, http_basic_authentication: [@user, @password]).read
       rescue Exception => e
 	@logger.error("An error occurred while increasing dataset version: " + e.to_s)
@@ -87,46 +95,77 @@ class WendelinClient
 	@logger.info("There is another ingestion already done for the pair dataset-filename. Reference "\
 	      + reference, print=TRUE)
 	@logger.info("Rename your file or download the full dataset to make local changes.", print=TRUE)
-	return FALSE
+	return {"success"=>FALSE, "message"=>HTTP_REFERENCE_EXIST}
       end
-      checkReferenceChars(reference)
       uri = URI(URI.escape("#{@erp5_url}/ingest?reference=#{reference}"))
+      n_retry = 0
       response = handleRequest(uri, reference, data_chunk)
-      if response == FALSE
-	return FALSE
+      while response["success"] == FALSE && n_retry < 10
+	if response["message"] != HTTP_MESSAGE_EXCEPTION && response["message"] != HTTP_MESSAGE_NOT_2XX
+	  return response
+	else
+	  n_retry += 1
+	  @logger.info("Retrying #{n_retry}/10...", print=TRUE)
+	  sleep 30
+	  response = handleRequest(uri, reference, data_chunk)
+	end
+      end
+      if response["success"] == FALSE
+	return response
       end
       @logger.info("Record successfully ingested.", print=TRUE)
       @last_ingestion = Time.new
-      return TRUE
+      return {"success"=>TRUE, "message"=>"success"}
   end
 
-  def eachDataStreamContentChunk(id, chunk_size)
-    uri = URI(URI.escape("#{@erp5_url}#{id}/getData"))
-    @logger.info("Downloading...", print=TRUE)
+  def eachDataStreamContentChunk(id, chunk_size, n_chunk=0)
+    n_part = n_chunk
+    done = FALSE
     first = TRUE
+    while not done
+      start_offset = n_part*chunk_size
+      end_offset = n_part*chunk_size+chunk_size
+      uri = URI(URI.escape("#{@erp5_url}getDataStreamChunk?id=#{id}&start_offset=#{start_offset}&end_offset=#{end_offset}"))
+      success = FALSE
+      n_retry = 0
+      while ! success && n_retry < 10
+	begin
 	  res = open(uri, http_basic_authentication: [@user, @password])  {
 	    |content|
-	while true
-	  chunk = content.read(chunk_size)
-	  if chunk.nil?
+	      chunk = content.read()
+	      if chunk.nil? || chunk.empty?
 		if first
 		  yield chunk
 		end
 	        @logger.info("Done", print=TRUE)
-	    break
-	  end
+	        done = TRUE
+	      else
 		first = FALSE
+	        n_part += 1
 	        yield chunk
 	      end
 	  }
+	  success = TRUE
+	rescue Exception => e
+	  exception = e
+	  @logger.error("Error downloading data: " + e.to_s, print=TRUE)
+	  @logger.error(e.backtrace)
+	  n_retry += 1
+	  @logger.info("Retrying #{n_retry}/10...", print=TRUE)
+	  sleep 30
+	end
+      end
+      raise exception if not success
+    end
   end
 
   def getDataStreams(data_set_reference)
     uri = URI(URI.escape("#{@erp5_url}getDataStreamList?data_set_reference=#{data_set_reference}"))
-    str = handleRequest(uri)
-    if str == FALSE
+    response = handleRequest(uri)
+    if response["success"] == FALSE
       @logger.abortExecution()
     end
+    str = response["message"]
     if not str.nil?
 	str.gsub!(/(\,)(\S)/, "\\1 \\2")
     	return YAML::load(str)
@@ -145,11 +184,14 @@ class WendelinClient
 	  req.set_form_data('data_chunk' => data_chunk)
 	rescue java.lang.OutOfMemoryError
 	  @logger.logOutOfMemoryError(reference)
-	  return FALSE
+	  return {"success"=>FALSE, "message"=>HTTP_MEMORY_ERROR}
+	rescue Exception => e
+	  @logger.error("Error setting form data: " + e.to_s, print=TRUE)
+	  @logger.error(e.backtrace)
+	  return {"success"=>FALSE, "message"=>HTTP_MESSAGE_EXCEPTION}
 	end
 	@logger.info("Sending record:'#{reference}'...", print=TRUE)
       end
-
       begin
 	  res = Net::HTTP.start(uri.hostname, uri.port,
 		  :use_ssl      => (uri.scheme == 'https'),
@@ -161,22 +203,25 @@ class WendelinClient
       rescue Exception => e
 	  @logger.error("HTTP ERROR: " + e.to_s, print=TRUE)
 	  @logger.error(e.backtrace)
-	  return FALSE
+	  return {"success"=>FALSE, "message"=>HTTP_MESSAGE_EXCEPTION}
       else
 	  if res.kind_of?(Net::HTTPSuccess) # res.code is 2XX
 	      @logger.info("Done")
-	      return res.body
+	      return {"success"=>TRUE, "message"=>res.body}
 	  else
 	      @logger.error("HTTP FAIL -  code: #{res.code}", print=TRUE)
 	      if res.code == '500' or res.code == '502' or res.code == '503'
-		@logger.error("Internal Server Error: if the error persists, please contact the administrator.", print=TRUE)
+		@logger.error(HTTP_MESSAGE_5XX, print=TRUE)
 	      elsif res.code == '401'
-		@logger.error("Unauthorized access. Please check your user credentials and try again.", print=TRUE)
+		@logger.error(HTTP_MESSAGE_401, print=TRUE)
+		@logger.abortExecution()
+	      elsif res.code == '400'
+		@logger.error(HTTP_MESSAGE_400, print=TRUE)
 		@logger.abortExecution()
 	      else
-		@logger.error("Sorry, an error ocurred. If the error persists, please contact the administrator.", print=TRUE)
+		@logger.error(HTTP_MESSAGE_OTHER, print=TRUE)
 	      end
-	      return FALSE
+	      return {"success"=>FALSE, "message"=>HTTP_MESSAGE_NOT_2XX}
 	  end
       end 
   end