lib/sysmodb/extractor.rb in simple-spreadsheet-extractor-0.16.1 vs lib/sysmodb/extractor.rb in simple-spreadsheet-extractor-0.17.0

@@ -1,71 +1,64 @@
-require 'open4'
+require 'terrapin'
 
 module SysMODB
-  #Exception that is thrown when a problem occurs during the extraction
+  # Exception that is thrown when a problem occurs during the extraction
   class SpreadsheetExtractionException < Exception; end
 
-  #handles the delegation to java, and executes the extraction passing the
-  #input file through STDIN, and reading the results through STDOUT.
+  # handles the delegation to java
   class Extractor
-    JAR_VERSION="0.16.0".freeze
+    JAR_VERSION = '0.16.0'.freeze
     DEFAULT_PATH = File.dirname(__FILE__) + "/../../jars/simple-spreadsheet-extractor-#{JAR_VERSION}.jar"
-    BUFFER_SIZE=250000 # 1/4 a megabyte
 
     def initialize(memory_allocation)
       @memory_allocation = memory_allocation
-      if is_windows?
-        raise Exception.new("Windows is not currently supported")
-      end
+      raise Exception, 'Windows is not currently supported' if is_windows?
     end
 
+    # spreadsheet_data can be an IO like object or the path to a file
     def spreadsheet_to_xml(spreadsheet_data)
-      read_with_open4 spreadsheet_data,"xml"
+      spreadsheet_to_requested_format(spreadsheet_data, 'xml')
     end
 
-    def spreadsheet_to_csv(spreadsheet_data,sheet=1,trim=false)
-      read_with_open4 spreadsheet_data,"csv",sheet,trim
+    # spreadsheet_data can be an IO like object or the path to a file
+    def spreadsheet_to_csv(spreadsheet_data, sheet = 1, trim = false)
+      spreadsheet_to_requested_format(spreadsheet_data, 'csv', sheet, trim)
     end
 
     private
 
-    def spreadsheet_extractor_command(format="xml",sheet=nil,trim=false)
+    def spreadsheet_to_requested_format(spreadsheet_data, format, sheet = nil, trim = nil)
+      if spreadsheet_data.is_a?(IO) || spreadsheet_data.is_a?(StringIO)
+        Tempfile.create('spreadsheet-extraction') do |f|
+          f.write(spreadsheet_data.read)
+          f.flush
+          execute_command_line f.path, format, sheet, trim
+        end
+      elsif spreadsheet_data.is_a?(String)
+        execute_command_line spreadsheet_data, format, sheet, trim
+      end
+    end
+
+    def spreadsheet_extractor_command(filepath, format = 'xml', sheet = nil, trim = false)
       command = "java -Xmx#{@memory_allocation} -jar #{(defined? SPREADSHEET_EXTRACTOR_JAR_PATH) ? SPREADSHEET_EXTRACTOR_JAR_PATH : DEFAULT_PATH}"
-      command +=  " -o #{format}"
+      command += " -o #{format}"
       command += " -s #{sheet}" if sheet
-      command += " -t" if trim
+      command += ' -t' if trim
+      command += " < #{filepath}"
       command
     end
 
     def is_windows?
       !(RUBY_PLATFORM =~ /mswin32/ || RUBY_PLATFORM =~ /mingw32/).nil?
     end
 
-    def read_with_open4(spreadsheet_data,format="xml",sheet=nil,trim=false)
-      output = ""
-      err_message = ""
-      command = spreadsheet_extractor_command format,sheet,trim
-      status = Open4.popen4(command) do |_pid, stdin, stdout, stderr|
-        while ((line = spreadsheet_data.gets(BUFFER_SIZE)) != nil) do
-          stdin << line
-        end
-        stdin.close
-
-        while ((line = stdout.gets(BUFFER_SIZE)) != nil) do
-          output << line
-        end
-        stdout.close
-
-        until ((line=stderr.gets((BUFFER_SIZE))).nil?) do
-          err_message << line
-        end
-        stderr.close
+    def execute_command_line(filepath, format = 'xml', sheet = nil, trim = false)
+      command = spreadsheet_extractor_command filepath, format, sheet, trim
+      begin
+        Terrapin::CommandLine.new(command).run.strip
+      rescue Terrapin::ExitStatusError, Terrapin::CommandNotFoundError => e
+        raise SpreadsheetExtractionException, e.message
       end
-
-      if status.to_i != 0
-        raise SpreadsheetExtractionException.new(err_message)
-      end
-
-      output.strip
     end
+
   end
 end