emailparser 0.1.7 → 0.1.8

checksums.yaml

@@ -1,7 +1,7 @@
 ---
 SHA1:
-  metadata.gz: 9453c8fe2c99ad7fbf91ece0ca61d294bfa2d0b0
-  data.tar.gz: cc5075498b9b81f18bff729c792e70514d8d78d1
+  metadata.gz: f54662f39ecfbd6bf875b3eefc92a427f7a0958e
+  data.tar.gz: fb42fed00d2a7eed6fed4dd4fb102d55e36a7800
 SHA512:
-  metadata.gz: 4b11c724d3603c5f5775650d5ca21d2d8cb052f7e7068d24f1aa538e268a7db3cca02ec2684b6dde40f0798c3ca0cd901045c4117cffd94e57695570fdda8bf5
-  data.tar.gz: 60cad1df4447f65f44cad337bff57894ef0e4427f65e357dc0a18a4fd70b57e76db5f29882b38730d50db17f26e749604c2d48126a853ff70a0ce4bc9e714078
+  metadata.gz: 8869b528a7ee753bac34b8bd685e3f4c96a7b511189de6a9c5b6460b1baa1cb656ed2cb4843486fa3347206c6f59b55112dc713881c2b3c9d740ac4aba08032e
+  data.tar.gz: 868153660b3bf993634a6719311abddae9bcd6ea42215ba41ff2e1f3d2b93fa4af3438a60fea2eaf4240343bdaee15eb0080e72145be7e259454eed983f2aa92

data/lib/emailparser.rb

@@ -2,11 +2,12 @@ require 'pry'
 require 'json'
 require 'mail'
 require 'digest'
+require 'pathname'
 
-class Emailparser
+class EmailParser
 
-	def initialize(message, out_dir, attachment_dir)
-		@message = message
+	def initialize(path, out_dir, attachment_dir)
+		@path = path
 		@attachment_dir = out_dir + "/" + attachment_dir
 		@allowed_documents = [
 			'application/x-mobipocket-ebook',
@@ -129,66 +130,61 @@ class Emailparser
 
 	# Voodoo to fix nasty encoded strings
 	def fix_encode(text)
-		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
+		if text.is_a?(String)
+      		text_out = text.to_s.encode('UTF-8', {
+                                           :invalid => :replace,
+                                           :undef   => :replace,
+                                           :replace => '?'
+                                         })
+			return text_out
+		elsif text.is_a?(Array)
+			fixed = []
+			text.each do | item |
+				item_fixed = item.to_s.encode('UTF-8', {
+                                           :invalid => :replace,
+                                           :undef   => :replace,
+                                           :replace => '?'
+                                         })
+				fixed.push(item_fixed)
 			end
-		rescue
-			puts " X Fixing encoding failed"
-			return "Unreadable Text"
+			return fixed
+		else
+			return text
 		end
 	end
 
 	def make_attachment_folder(attachments, source_hash)
 		if (!attachments.empty?)
-			puts " - Creating directory: " + source_hash
+			puts "Creating sub-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 " - " + filename + "\n"
+		puts " - found attachment " + filename + "\n"
 		begin
 			File.open(@attachment_dir + message_id + "/" + filename, "w+b", 0644) do |f|
 				f.write attachment.body.decoded 
 			end
 		rescue => e
-			puts " X Unable to save data for #{filename} because #{e.message}"
+			puts "Unable to save data for #{filename} because #{e.message}"
 		end
 	end
 
 	# Accepts a message
 	def parse_message
-
-		puts "-" * 70
-		puts "Loading email: " + @message
-		source_hash = Digest::SHA256.hexdigest(File.read(@message))
-		email = Mail.read(@message)
+		puts "Loading email: " + @path + "\n"
+		email_file = File.read(@path).unpack('C*').pack('U*')
+		source_file = File.basename(@path)
+		source_hash = Digest::SHA256.hexdigest(email_file)
+		email = Mail.new(email_file)
 
 		# Defaults
-		source_file = @message.split("/").last
-		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"
+		if email.message_id.nil?
+			message_id = ""
+		else
+			message_id = fix_encode(email.message_id)
 		end
 
 		# Date
@@ -218,23 +214,24 @@ class Emailparser
 			recipients = email_to.concat(email_cc)
 			addresses = recipients + email_from
 		rescue
-			puts " X Some addresses failed"
+			puts "oops something failed here..."
 			# binding.pry
 		end
 
 		# Subject
-		begin
-			if email.subject
-				subject = fix_encode(email.subject)
-			end
-		rescue
-			puts " X Subject breaks encoding"
-  			subject = "Unreadable Subject"
+		if email.subject
+			subject = fix_encode(email.subject)
+		else 
+			subject = "No Subject"
 		end
 
-		# Body - Check for Multipart
+		body_plain	= ""
+		body_html	= ""
+		attachments = []
+
+		# Check for Multipart
 		if email.multipart?
-			puts " - Multipart message"
+			puts " - is multipart\n"
 			if email.text_part 
 				body_plain = fix_encode(email.text_part.body.decoded)
 			end
@@ -242,12 +239,11 @@ class Emailparser
 				body_html = fix_encode(email.html_part.body.decoded)
 			end
 		else
-			puts " - Single part message"
+			puts " - is single part\n"
 			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
@@ -256,27 +252,28 @@ 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: Document"
+				puts " - Attachment is: document"
 				attachment_save = true
 			elsif (@allowed_spreadsheets.include? mime_type)
-				puts " - Attachment: Spreadsheet"
+				puts " - Attachment is: spreadsheet"
 				attachment_save = true
 			elsif (@allowed_audio.include? mime_type)
-				puts " - Attachment: Audio"
+				puts " - Attachment is: audio"
 				attachment_save = true
 			elsif (@allowed_contacts.include? mime_type)
-				puts " - Attachment: Contact"
+				puts " - Attachment is: contact"
 				attachment_save = true
 			elsif (@allowed_images.include? mime_type)
-				puts " - Attachment: Image"
+				puts " - Attachment is: image"
 				attachment_save = true
 			elsif (@allowed_slideshows.include? mime_type)
-				puts " - Attachment: Slideshow"
+				puts " - Attachment is: slideshow"
 				attachment_save = true
 			elsif (@allowed_videos.include? mime_type)
-				puts " - Attachment: Video"
+				puts " - Attachment is: video"
 				attachment_save = true
 			end
 
@@ -286,7 +283,7 @@ class Emailparser
 				save_attachment(attachment, source_hash, filename)
 			end
 		end
-	
+
 		# Structure Data
 		email_data = {
 			source_file: source_file,

metadata

@@ -1,7 +1,7 @@
 --- !ruby/object:Gem::Specification
 name: emailparser
 version: !ruby/object:Gem::Version
-  version: 0.1.7
+  version: 0.1.8
 platform: ruby
 authors:
 - Brennan Novak
@@ -37,7 +37,7 @@ required_rubygems_version: !ruby/object:Gem::Requirement
       version: '0'
 requirements: []
 rubyforge_project: 
-rubygems_version: 2.5.1
+rubygems_version: 2.6.8
 signing_key: 
 specification_version: 4
 summary: Parses a single email file to JSON with attachments