helpers.rb in eco-helpers-2.5.1

- old
+ new

@@ -10,48 +10,38 @@
         end
       end
 
       module InstanceMethods
         include Eco::Language::AuxiliarLogger
+        include Eco::Data::Files::Encoding
 
         # It offers a resilient way to read content from a file
         # @tolerance [Integer] the number of allowed encoding errors.
         # @return [String] the content of the file
-        def get_file_content(file, encoding, tolerance: 5)
-          unless self.class.file_exists?(file)
-            logger.error("File does not exist: #{file}")
-            exit(1)
-          end
-          encoding ||= self.class.encoding(file)
-          encoding   = (encoding == "bom") ? "#{encoding}|utf-8": encoding
-          unless !encoding || encoding == 'utf-8'
-            msg = "File encoding: '#{encoding}'"
-            logger.debug(msg)
-            puts msg
-          end
+        def get_file_content(file, encoding: nil, tolerance: 5)
           read_with_tolerance(file, encoding: encoding, tolerance: tolerance)
         end
 
         def read_with_tolerance(file, encoding:, tolerance: 5)
-          if content = File.read(file, encoding: encoding)
-            content = content.encode("utf-8") unless encoding.include?('utf-8')
-            content.scrub do |bytes|
-              replacement = '<' + bytes.unpack('H*')[0] + '>'
-              if tolerance <= 0
-                logger.error("There were more than 5 encoding errors in the file '#{file}'.")
-                return content
-              else
-                tolerance -= 1
-                logger.error("Encoding problem in file '#{file}': '#{replacement}'.")
-                replacement
-              end
+          return nil unless content = get_file_content_with_encoding(file, encoding: encoding)
+          content.scrub do |bytes|
+            replacement = '<' + bytes.unpack('H*')[0] + '>'
+            if tolerance <= 0
+              logger.error("There were more than 5 encoding errors in the file '#{file}'.")
+              return content
+            else
+              tolerance -= 1
+              logger.error("Encoding problem in file '#{file}': '#{replacement}'.")
+              replacement
             end
           end
         end
       end
 
       module ClassMethods
+        include Eco::Data::Files::Encoding
+
         def create_directory(path, includes_file: false)
           Directory.create(path, includes_file: includes_file)
         end
 
         def split(path)
@@ -102,22 +92,10 @@
         def file_empty?(path)
           return true if !File.file?(path)
           File.zero?(path)
         end
 
-        def has_bom?(path)
-          return false if !path || file_empty?(path)
-          File.open(path, "rb") do |f|
-            bytes = f.read(3)
-            return bytes.unpack("C*") == [239, 187, 191]
-          end
-        end
-
-        def encoding(path)
-          has_bom?(path) ? "bom" : "utf-8"
-        end
-
         def script_subfolder
           basename = File.basename($0, File.extname($0))
           path = File.dirname($0)
           File.join(path, basename)
         end
@@ -135,10 +113,10 @@
           end.sort
         end
 
         def csv_files(folder = ".", regexp: nil, older_than: nil)
           folder_files(folder, "*.csv", regexp: regexp, older_than: older_than).sort
-        end        
+        end
       end
 
       class << self
         include Files::ClassMethods
       end