sisimai 4.25.16-java → 5.0.2-java

data/lib/sisimai/data.rb

@@ -1,411 +0,0 @@
-module Sisimai
-  # Sisimai::Data generate parsed data from Sisimai::Message object.
-  class Data
-    # Imported from p5-Sisimail/lib/Sisimai/Data.pm
-    require 'sisimai/address'
-    require 'sisimai/reason'
-    require 'sisimai/string'
-    require 'sisimai/rhost'
-    require 'sisimai/time'
-    require 'sisimai/datetime'
-    require 'sisimai/smtp/error'
-
-    @@rwaccessors = [
-      :catch,           # [?] Results generated by hook method
-      :token,           # [String] Message token/MD5 Hex digest value
-      :lhost,           # [String] local host name/Local MTA
-      :rhost,           # [String] Remote host name/Remote MTA
-      :alias,           # [String] Alias of the recipient address
-      :listid,          # [String] List-Id header of each ML
-      :reason,          # [String] Bounce reason
-      :action,          # [String] The value of Action: header
-      :origin,          # [String] Email path as a data source
-      :subject,         # [String] UTF-8 Subject text
-      :timestamp,       # [Sisimai::Time] Date: header in the original message
-      :addresser,       # [Sisimai::Address] From address
-      :recipient,       # [Sisimai::Address] Recipient address which bounced
-      :messageid,       # [String] Message-Id: header
-      :replycode,       # [String] SMTP Reply Code
-      :smtpagent,       # [String] Module(Engine) name
-      :softbounce,      # [Integer] 1 = Soft bounce, 0 = Hard bounce, -1 = ?
-      :smtpcommand,     # [String] The last SMTP command
-      :destination,     # [String] The domain part of the "recipinet"
-      :senderdomain,    # [String] The domain part of the "addresser"
-      :feedbacktype,    # [String] Feedback Type
-      :diagnosticcode,  # [String] Diagnostic-Code: Header
-      :diagnostictype,  # [String] The 1st part of Diagnostic-Code: Header
-      :deliverystatus,  # [String] Delivery Status(DSN)
-      :timezoneoffset,  # [Integer] Time zone offset(seconds)
-    ]
-    attr_accessor(*@@rwaccessors)
-
-    RetryIndex = Sisimai::Reason.retry
-    RFC822Head = Sisimai::RFC5322.HEADERFIELDS(:all)
-
-    # Constructor of Sisimai::Data
-    # @param    [Hash] argvs    Data
-    # @return   [Sisimai::Data] Structured email data
-    def initialize(argvs)
-      # Create email address object
-      as = Sisimai::Address.make(argvs['addresser'])
-      ar = Sisimai::Address.make(address: argvs['recipient'])
-
-      return nil unless as.is_a? Sisimai::Address
-      return nil unless ar.is_a? Sisimai::Address
-      return nil if as.void
-      return nil if ar.void
-
-      @addresser = as
-      @recipient = ar
-      @senderdomain = as.host
-      @destination  = ar.host
-      @alias = argvs['alias'] || ''
-      @token = Sisimai::String.token(as.address, ar.address, argvs['timestamp'])
-      @timestamp = Sisimai::Time.parse(::Time.at(argvs['timestamp']).to_s)
-      @timezoneoffset = argvs['timezoneoffset'] || '+0000'
-      @lhost          = argvs['lhost']          || ''
-      @rhost          = argvs['rhost']          || ''
-      @catch          = argvs['catch']          || nil
-      @reason         = argvs['reason']         || ''
-      @listid         = argvs['listid']         || ''
-      @subject        = argvs['subject']        || ''
-      @messageid      = argvs['messageid']      || ''
-      @smtpagent      = argvs['smtpagent']      || ''
-      @diagnosticcode = argvs['diagnosticcode'] || ''
-      @diagnostictype = argvs['diagnostictype'] || ''
-      @deliverystatus = argvs['deliverystatus'] || ''
-      @smtpcommand    = argvs['smtpcommand']    || ''
-      @feedbacktype   = argvs['feedbacktype']   || ''
-      @action         = argvs['action']         || ''
-      @origin         = argvs['origin']         || ''
-      @replycode      = argvs['replycode']      || ''
-      @replycode      = Sisimai::SMTP::Reply.find(argvs['diagnosticcode']).to_s if @replycode.empty?
-      @softbounce     = argvs['softbounce']     || ''
-    end
-
-    # Another constructor of Sisimai::Data
-    # @param          [Sisimai::Message] data Data Object
-    # @param          [Hash] argvs            Parser options
-    # @options argvs  [Boolean] delivered     true: Including "delivered" reason
-    # @return         [Array, Nil]            List of Sisimai::Data or Nil if the
-    #                                         argument is not Sisimai::Message object
-    def self.make(data: nil, **argvs)
-      return nil unless data
-      return nil unless data.is_a? Sisimai::Message
-
-      messageobj = data
-      rfc822data = messageobj.rfc822
-      objectlist = []
-      delivered1 = argvs[:delivered] || false
-
-      return nil unless messageobj.ds
-      return nil unless messageobj.rfc822
-
-      eachobject = messageobj.ds.dup
-      while e = eachobject.shift do
-        # Create parameters for new() constructor.
-        p = {
-          'catch'          => messageobj.catch  || nil,
-          'lhost'          => e['lhost']        || '',
-          'rhost'          => e['rhost']        || '',
-          'alias'          => e['alias']        || '',
-          'action'         => e['action']       || '',
-          'reason'         => e['reason']       || '',
-          'replycode'      => e['replycode']    || '',
-          'smtpagent'      => e['agent']        || '',
-          'recipient'      => e['recipient']    || '',
-          'softbounce'     => e['softbounce']   || '',
-          'smtpcommand'    => e['command']      || '',
-          'feedbacktype'   => e['feedbacktype'] || '',
-          'diagnosticcode' => e['diagnosis']    || '',
-          'diagnostictype' => e['spec']         || '',
-          'deliverystatus' => e['status']       || '',
-        }
-        unless delivered1
-          # Skip if the value of "deliverystatus" begins with "2." such as 2.1.5
-          next if p['deliverystatus'].start_with?('2.')
-        end
-
-        # EMAIL_ADDRESS:
-        # Detect email address from message/rfc822 part
-        RFC822Head[:addresser].each do |f|
-          # Check each header in message/rfc822 part
-          next unless rfc822data[f]
-          next if rfc822data[f].empty?
-
-          j = Sisimai::Address.find(rfc822data[f]) || []
-          next if j.empty?
-          p['addresser'] = j[0]
-          break
-        end
-
-        unless p['addresser']
-          # Fallback: Get the sender address from the header of the bounced
-          # email if the address is not set at loop above.
-          j = Sisimai::Address.find(messageobj.header['to']) || []
-          p['addresser'] = j[0] unless j.empty?
-        end
-        next unless p['addresser']
-        next unless p['recipient']
-
-        # TIMESTAMP:
-        # Convert from a time stamp or a date string to a machine time.
-        datestring = nil
-        zoneoffset = 0
-        datevalues = []
-        datevalues << e['date'] unless e['date'].to_s.empty?
-
-        # Date information did not exist in message/delivery-status part,...
-        RFC822Head[:date].each do |f|
-          # Get the value of Date header or other date related header.
-          next unless rfc822data[f]
-          datevalues << rfc822data[f]
-        end
-
-        # Set "date" getting from the value of "Date" in the bounce message
-        datevalues << messageobj.header['date'] if datevalues.size < 2
-
-        while v = datevalues.shift do
-          # Parse each date value in the array
-          datestring = Sisimai::DateTime.parse(v)
-          break if datestring
-        end
-
-        if datestring && cv = datestring.match(/\A(.+)[ ]+([-+]\d{4})\z/)
-          # Get the value of timezone offset from datestring
-          # Wed, 26 Feb 2014 06:05:48 -0500
-          datestring = cv[1]
-          zoneoffset = Sisimai::DateTime.tz2second(cv[2])
-          p['timezoneoffset'] = cv[2]
-        end
-
-        begin
-          # Convert from the date string to an object then calculate time
-          # zone offset.
-          t = Sisimai::Time.strptime(datestring, '%a, %d %b %Y %T')
-          p['timestamp'] = (t.to_time.to_i - zoneoffset) || nil
-        rescue
-          warn ' ***warning: Failed to strptime ' << datestring.to_s
-        end
-        next unless p['timestamp']
-
-        # OTHER_TEXT_HEADERS:
-        recvheader = data.header['received'] || []
-        unless recvheader.empty?
-          # Get localhost and remote host name from Received header.
-          %w[lhost rhost].each { |a| e[a] ||= '' }
-          e['lhost'] = Sisimai::RFC5322.received(recvheader[0]).shift if e['lhost'].empty?
-          e['rhost'] = Sisimai::RFC5322.received(recvheader[-1]).pop  if e['rhost'].empty?
-        end
-
-        # Remove square brackets and curly brackets from the host variable
-        %w[rhost lhost].each do |v|
-          p[v].delete!('[]()')    # Remove square brackets and curly brackets from the host variable
-          p[v].sub!(/\A.+=/, '')  # Remove string before "="
-          p[v].chomp!("\r") if p[v].end_with?("\r") # Remove CR at the end of the value
-
-          # Check space character in each value and get the first element
-          p[v] = p[v].split(' ', 2).shift if p[v].include?(' ')
-          p[v].chomp!('.') if p[v].end_with?('.')   # Remove "." at the end of the value
-        end
-
-        # Subject: header of the original message
-        p['subject'] = rfc822data['subject'] || ''
-        p['subject'].scrub!('?')
-        p['subject'].chomp!("\r") if p['subject'].end_with?("\r")
-
-        # The value of "List-Id" header
-        p['listid'] = rfc822data['list-id'] || ''
-        unless p['listid'].empty?
-          # Get the value of List-Id header like "List name <list-id@example.org>"
-          if cv = p['listid'].match(/\A.*([<].+[>]).*\z/) then p['listid'] = cv[1] end
-          p['listid'].delete!('<>')
-          p['listid'].chomp!("\r") if p['listid'].end_with?("\r")
-          p['listid'] = '' if p['listid'].include?(' ')
-        end
-
-        # The value of "Message-Id" header
-        p['messageid'] = rfc822data['message-id'] || ''
-        unless p['messageid'].empty?
-          # Leave only string inside of angle brackets(<>)
-          if cv = p['messageid'].match(/\A([^ ]+)[ ].*/) then p['messageid'] = cv[1] end
-          if cv = p['messageid'].match(/[<]([^ ]+?)[>]/) then p['messageid'] = cv[1] end
-        end
-
-        # CHECK_DELIVERY_STATUS_VALUE:
-        # Cleanup the value of "Diagnostic-Code:" header
-        unless p['diagnosticcode'].empty?
-          # Count the number of D.S.N. and SMTP Reply Code
-          vs = Sisimai::SMTP::Status.find(p['diagnosticcode'])
-          vr = Sisimai::SMTP::Reply.find(p['diagnosticcode'])
-          vm = 0
-          re = nil
-
-          if vs
-            # How many times does the D.S.N. appeared
-            vm += p['diagnosticcode'].scan(/\b#{vs}\b/).size
-            p['deliverystatus'] = vs if vs =~ /\A[45][.][1-9][.][1-9]\z/
-          end
-
-          if vr
-            # How many times does the SMTP reply code appeared
-            vm += p['diagnosticcode'].scan(/\b#{vr}\b/).size
-            p['replycode'] ||= vr
-          end
-
-          if vm > 2
-            # Build regular expression for removing string like '550-5.1.1'
-            # from the value of "diagnosticcode"
-            re0 = %r/;?[ ]#{vr}[- ](?:#{vs})?/
-            re1 = %r/;?[ ][45]\d\d[- ](?:#{vs})?/
-            re2 = %r/;?[ ]#{vr}[- ](?:[45][.]\d[.]\d+)?/
-
-            # 550-5.7.1 [192.0.2.222] Our system has detected that this message is
-            # 550-5.7.1 likely unsolicited mail. To reduce the amount of spam sent to Gmail,
-            # 550-5.7.1 this message has been blocked. Please visit
-            # 550 5.7.1 https://support.google.com/mail/answer/188131 for more information.
-            p['diagnosticcode'] = p['diagnosticcode'].gsub(re0, ' ')
-            p['diagnosticcode'] = p['diagnosticcode'].gsub(re1, ' ')
-            p['diagnosticcode'] = p['diagnosticcode'].gsub(re2, ' ')
-            p['diagnosticcode'] = Sisimai::String.sweep(p['diagnosticcode'].sub(%r|<html>.+</html>|i, ''))
-          end
-        end
-
-        if Sisimai::String.is_8bit(p['diagnosticcode'])
-          # To avoid incompatible character encodings: ASCII-8BIT and UTF-8 (Encoding::CompatibilityError
-          p['diagnosticcode'] = p['diagnosticcode'].force_encoding('UTF-8').scrub('?')
-        end
-
-        p['diagnostictype'] ||= 'X-UNIX' if p['reason'] == 'mailererror'
-        p['diagnostictype'] ||= 'SMTP' unless %w[feedback vacation].include?(p['reason'])
-
-        # Check the value of SMTP command
-        p['smtpcommand'] = '' unless %w[EHLO HELO MAIL RCPT DATA QUIT].include?(p['smtpcommand'])
-        p['origin'] = argvs[:origin]  # Set the path to the original email
-
-        if p['action'].empty?
-          # Check the value of "action"
-          if p['reason'] == 'expired'
-            # Action: delayed
-            p['action'] = 'delayed'
-          elsif p['deliverystatus'].start_with?('5', '4')
-            # Action: failed
-            p['action'] = 'failed'
-          end
-        end
-
-        o = Sisimai::Data.new(p)
-        next unless o.recipient
-
-        if o.reason.empty? || RetryIndex[o.reason]
-          # Decide the reason of email bounce
-          r = ''; r = Sisimai::Rhost.get(o) if Sisimai::Rhost.match(o.rhost)
-          if r.empty?
-            # Failed to detect a bounce reason by the value of "rhost"
-            r = Sisimai::Rhost.get(o, o.destination) if Sisimai::Rhost.match(o.destination)
-            r = Sisimai::Reason.get(o) if r.nil? || r.empty?
-            r = 'undefined' if r.empty?
-          end
-          o.reason = r
-        end
-
-        if %w[delivered feedback vacation].include?(o.reason)
-          # The value of reason is "vacation" or "feedback"
-          o.softbounce = -1
-          o.replycode = '' unless o.reason == 'delivered'
-        else
-          # Bounce message which reason is "feedback" or "vacation" does
-          # not have the value of "deliverystatus".
-          softorhard = nil
-
-          if o.softbounce.to_s.empty?
-            # The value is not set yet
-            textasargv   = (p['deliverystatus'] + ' ' + p['diagnosticcode']).lstrip
-            softorhard   = Sisimai::SMTP::Error.soft_or_hard(o.reason, textasargv) || ''
-            o.softbounce = if softorhard.size > 0
-                             # Returned value is "soft" or "hard"
-                             (softorhard == 'soft') ? 1 : 0
-                           else
-                             # Returned value is an empty string
-                             -1
-                           end
-          end
-
-          if o.deliverystatus.empty?
-            # Set pseudo status code
-            textasargv = (o.replycode + ' ' + p['diagnosticcode']).lstrip
-            getchecked = Sisimai::SMTP::Error.is_permanent(textasargv)
-            tmpfailure = getchecked.nil? ? false : (getchecked ? false : true)
-
-            if pseudocode = Sisimai::SMTP::Status.code(o.reason, tmpfailure)
-              # Set the value of "deliverystatus" and "softbounce"
-              o.deliverystatus = pseudocode
-
-              if o.softbounce < 0
-                # set the value of "softbounce" again when the value is -1
-                softorhard = Sisimai::SMTP::Error.soft_or_hard(o.reason, pseudocode)
-
-                o.softbounce = if softorhard.size > 0
-                                 # Returned value is "soft" or "hard"
-                                 softorhard == 'soft' ? 1 : 0
-                               else
-                                 # Returned value is an empty string
-                                 -1
-                               end
-              end
-            end
-          end
-
-          unless o.replycode.empty?
-            # Check both of the first digit of "deliverystatus" and "replycode"
-            o.replycode = '' unless o.replycode[0, 1] == o.deliverystatus[0, 1]
-          end
-
-        end
-        objectlist << o
-
-      end
-      return objectlist
-    end
-
-    # Convert from object to hash reference
-    # @return   [Hash] Data in Hash reference
-    def damn
-      data = {}
-      @@rwaccessors.each do |e|
-        next if %w[addresser recipient timestamp].include?(e.to_s)
-        data[e.to_s] = self.send(e) || ''
-      end
-      data['addresser'] = self.addresser.address
-      data['recipient'] = self.recipient.address
-      data['timestamp'] = self.timestamp.to_time.to_i
-      return data
-    end
-    alias :to_hash :damn
-
-    # Data dumper
-    # @param    [String] type   Data format: json, yaml
-    # @return   [String, Nil]   Dumped data or nil if the value of the first
-    #                           argument is neither "json" nor "yaml"
-    def dump(type = 'json')
-      return nil unless %w[json yaml].include?(type)
-      referclass = 'Sisimai::Data::' << type.upcase
-
-      begin
-        require referclass.downcase.gsub('::', '/')
-      rescue
-        warn '***warning: Failed to load' << referclass
-      end
-
-      dumpeddata = Module.const_get(referclass).dump(self)
-      return dumpeddata
-    end
-
-    # JSON handler
-    # @return   [String]            JSON string converted from Sisimai::Data
-    def to_json(*)
-      return self.dump('json')
-    end
-
-  end
-end