emailparser 0.1.6 → 0.1.7

checksums.yaml

@@ -1,7 +1,7 @@
 ---
 SHA1:
-  metadata.gz: 9b4fd8d1224deeaddd1cfc9a658529ad6fba1a63
-  data.tar.gz: 6d2b1dedff2e0164dfd1c0c51f1050d32d124f2e
+  metadata.gz: 9453c8fe2c99ad7fbf91ece0ca61d294bfa2d0b0
+  data.tar.gz: cc5075498b9b81f18bff729c792e70514d8d78d1
 SHA512:
-  metadata.gz: f14f1a15c13634747010860ccb6f9302bdc5fba5007d85d97eae752b6d1a8721655c406c20bee5e8842f2723c7d95a64ac9e2fca5f767b52d4a8dda309476ce2
-  data.tar.gz: 3454b07cdac6301e045f26d4761293f7ccc56f64103a657bbed253d2359850859c24e2a00b2e31e48c088096cc3716c3dd824db389d30798f930dce79379da1f
+  metadata.gz: 4b11c724d3603c5f5775650d5ca21d2d8cb052f7e7068d24f1aa538e268a7db3cca02ec2684b6dde40f0798c3ca0cd901045c4117cffd94e57695570fdda8bf5
+  data.tar.gz: 60cad1df4447f65f44cad337bff57894ef0e4427f65e357dc0a18a4fd70b57e76db5f29882b38730d50db17f26e749604c2d48126a853ff70a0ce4bc9e714078

data/lib/emailparser.rb

@@ -129,54 +129,66 @@ class Emailparser
 
 	# Voodoo to fix nasty encoded strings
 	def fix_encode(text)
-		if text.is_a?(String)
-			return text.unpack('C*').pack('U*')
-		elsif text.is_a?(Array)
-			fixed = []
-			text.each do | item |
-				fixed.push(item.unpack('C*').pack('U*'))
+		begin
+			if text.is_a?(String)
+				return text.unpack('C*').pack('U*')
+			elsif text.is_a?(Array)
+				fixed = []
+				text.each do | item |
+					fixed.push(item.unpack('C*').pack('U*'))
+				end
+				return fixed
+			else
+				return text
 			end
-			return fixed
-		else
-			return text
+		rescue
+			puts " X Fixing encoding failed"
+			return "Unreadable Text"
 		end
 	end
 
 	def make_attachment_folder(attachments, source_hash)
 		if (!attachments.empty?)
-			puts "Creating sub-directory: " + source_hash
+			puts " - Creating directory: " + source_hash
 			attachments_dir = @attachment_dir + source_hash
       		Dir.mkdir(attachments_dir) if !Dir.exist?(attachments_dir)
 		end
 	end
 
 	def save_attachment(attachment, message_id, filename)
-		puts " - found attachment " + filename + "\n"
+		puts " - " + filename + "\n"
 		begin
 			File.open(@attachment_dir + message_id + "/" + filename, "w+b", 0644) do |f|
 				f.write attachment.body.decoded 
 			end
 		rescue => e
-			puts "Unable to save data for #{filename} because #{e.message}"
+			puts " X Unable to save data for #{filename} because #{e.message}"
 		end
 	end
 
 	# Accepts a message
 	def parse_message
 
-		puts "Loading email: " + @message + "\n"
+		puts "-" * 70
+		puts "Loading email: " + @message
 		source_hash = Digest::SHA256.hexdigest(File.read(@message))
-		puts "SHA256 Hash: " + source_hash
-
 		email = Mail.read(@message)
 
 		# Defaults
 		source_file = @message.split("/").last
-		
-		if email.message_id.nil?
-			message_id = ""
-		else
-			message_id = fix_encode(email.message_id)
+		message_id = ""
+		subject = "No Subject"
+		body_plain	= ""
+		body_html	= ""
+		attachments = []
+
+		# Message-ID
+		begin
+			if !email.message_id.nil?
+				message_id = fix_encode(email.message_id)
+			end
+		rescue
+			puts " X - Getting Message-ID failed"
 		end
 
 		# Date
@@ -206,24 +218,23 @@ class Emailparser
 			recipients = email_to.concat(email_cc)
 			addresses = recipients + email_from
 		rescue
-			puts "oops something failed here..."
+			puts " X Some addresses failed"
 			# binding.pry
 		end
 
 		# Subject
-		if email.subject
-			subject = fix_encode(email.subject)
-		else 
-			subject = "No Subject"
+		begin
+			if email.subject
+				subject = fix_encode(email.subject)
+			end
+		rescue
+			puts " X Subject breaks encoding"
+  			subject = "Unreadable Subject"
 		end
-		
-		body_plain	= ""
-		body_html	= ""
-		attachments = []
 
-		# Check for Multipart
+		# Body - Check for Multipart
 		if email.multipart?
-			puts " - is multipart\n"
+			puts " - Multipart message"
 			if email.text_part 
 				body_plain = fix_encode(email.text_part.body.decoded)
 			end
@@ -231,11 +242,12 @@ class Emailparser
 				body_html = fix_encode(email.html_part.body.decoded)
 			end
 		else
-			puts " - is single part\n"
+			puts " - Single part message"
 			if !email.content_type.nil? and email.content_type.start_with?('text/html')
 				body_html = fix_encode(email.body.decoded)
+			else
+				body_plain = fix_encode(email.body.decoded)
 			end
-			body_plain = fix_encode(email.body.decoded)
 		end
 
 		# Handle Attachments
@@ -244,28 +256,27 @@ class Emailparser
 			attachment_save = false
 			filename = fix_encode(attachment.filename)
 			mime_type, remaining = attachment.content_type.split(';', 2)
-			puts " - Attachment mime: " + mime_type
 			# Check Allowed Mime Types
 			if (@allowed_documents.include? mime_type)
-				puts " - Attachment is: document"
+				puts " - Attachment: Document"
 				attachment_save = true
 			elsif (@allowed_spreadsheets.include? mime_type)
-				puts " - Attachment is: spreadsheet"
+				puts " - Attachment: Spreadsheet"
 				attachment_save = true
 			elsif (@allowed_audio.include? mime_type)
-				puts " - Attachment is: audio"
+				puts " - Attachment: Audio"
 				attachment_save = true
 			elsif (@allowed_contacts.include? mime_type)
-				puts " - Attachment is: contact"
+				puts " - Attachment: Contact"
 				attachment_save = true
 			elsif (@allowed_images.include? mime_type)
-				puts " - Attachment is: image"
+				puts " - Attachment: Image"
 				attachment_save = true
 			elsif (@allowed_slideshows.include? mime_type)
-				puts " - Attachment is: slideshow"
+				puts " - Attachment: Slideshow"
 				attachment_save = true
 			elsif (@allowed_videos.include? mime_type)
-				puts " - Attachment is: video"
+				puts " - Attachment: Video"
 				attachment_save = true
 			end
 
@@ -275,7 +286,7 @@ class Emailparser
 				save_attachment(attachment, source_hash, filename)
 			end
 		end
-
+	
 		# Structure Data
 		email_data = {
 			source_file: source_file,

metadata

@@ -1,14 +1,14 @@
 --- !ruby/object:Gem::Specification
 name: emailparser
 version: !ruby/object:Gem::Version
-  version: 0.1.6
+  version: 0.1.7
 platform: ruby
 authors:
 - Brennan Novak
 autorequire: 
 bindir: bin
 cert_chain: []
-date: 2017-01-10 00:00:00.000000000 Z
+date: 2017-01-12 00:00:00.000000000 Z
 dependencies: []
 description: Parses a signle email file to JSON with attachments
 email: hi@brennannovak.com