Source Code for Module emma.interface.email.message

  1  """ 
  2  email message support 
  3   
  4  @copyright: (c) 2011 hackmeeting U{http://sindominio.net/hackmeeting} 
  5  @author: Ruben Pollan 
  6  @organization: hackmeeting U{http://sindominio.net/hackmeeting} 
  7  @contact: meskio@sindominio.net 
  8  @license: 
  9    This program is free software; you can redistribute it and/or 
 10    modify it under the terms of the Do What The Fuck You Want To 
 11    Public License, Version 2, as published by Sam Hocevar. See 
 12    U{http://sam.zoy.org/projects/COPYING.WTFPL} for more details. 
 13  """ 
 14   
 15  import pyzmail 
 16  import re 
 17  from email.utils import parsedate 
 18  from time import mktime 
 19  from datetime import datetime 
 20   
 21  from emma.interface import message 
 22   
 23   
 24 -class Message(message.Message): 
 25      """ 
 26      email message 
 27   
 28      All the headers apears on the dictionary as msg[hdr_name]. The attachments 
 29      are a list on the msg['attachments']. 
 30      """ 
 31 -    def __init__(self, list_lines): 
 32          message.Message.__init__(self, tpe='email') 
 33   
 34          msgStr = '\n'.join(list_lines) 
 35          msg = pyzmail.PyzMessage.factory(msgStr) 
 36   
 37          d = msg_to_dict(msg) 
 38          self.update(d) 
 39          timestamp = mktime(parsedate(self['Date'])) 
 40          self['Date'] = datetime.fromtimestamp(timestamp) 
 41          self['Commands'] = self.commands() 
 42          self['Tags'] = self.tags() 
 43   
 44 -    def commands(self): 
 45          """ 
 46          get a list of the commands inside the email 
 47   
 48          Everithing with the form __cmd params__ will be reconice as command 
 49   
 50          @returns: [(cmd, params)] 
 51          """ 
 52          if 'Commands' in self: 
 53              return self['Commands'] 
 54   
 55          text = self['Body'] 
 56          commands = [] 
 57          cmd = ["", ""] 
 58          isCmd = False 
 59          isArg = False 
 60          isComment = False 
 61          for i in range(len(text)): 
 62              # don't parse the commented lines 
 63              # ignore everyline starting with '>' 
 64              if text[i] == '>': 
 65                  j = i-1 
 66                  while text[j] in (' ', '\t'): 
 67                      j -= 1 
 68                  if text[j] == '\n': 
 69                      isComment = True 
 70              elif text[i] == '\n': 
 71                  isComment = False 
 72              if isComment: 
 73                  if isArg: 
 74                      cmd[1] += text[i] 
 75                  continue 
 76   
 77              if text[i-1] != '\\' and text[i:i+2] == '._' and (isCmd or isArg): 
 78                  isArg = False 
 79                  commands.append(cmd) 
 80                  cmd = ["", ""] 
 81              elif isCmd: 
 82                  if text[i] == ' ': 
 83                      isArg = True 
 84                      isCmd = False 
 85                  else: 
 86                      cmd[0] += text[i] 
 87              elif isArg: 
 88                  if text[i:i+3] in ('\\._', '\\_.'): 
 89                      pass 
 90                  else: 
 91                      cmd[1] += text[i] 
 92              elif text[i-1] != '\\' and text[i-1:i+1] == '_.': 
 93                      isCmd = True 
 94   
 95          return commands 
 96   
 97 -    def tags(self): 
 98          """ 
 99          get a list of the tags from the subject 
100   
101          The text between [ and ] is consider as a tag 
102   
103          @returns: [tag] 
104          """ 
105          tagexp = re.compile(r"\[([^\]]*)\]") 
106          subject = self['Subject'] 
107          return tagexp.findall(subject) 
108   
109   
110 -def msg_to_dict(msg): 
111      """ 
112      Convert a PyZmail message to a dictionary 
113   
114      @type msg: PyzMessage 
115      @param msg: email to convert 
116      @returns: {'Header': 'content'} 
117      """ 
118      # FIXME: any repeated header will be ignored 
119      # Usually it is only 'Received' header 
120      d = {} 
121   
122      if msg.text_part: 
123          body = msg.text_part.get_payload() 
124          charset = msg.text_part.charset 
125      else: 
126          body = msg.get_payload() 
127          charset = msg.get_charset() 
128      if charset: 
129          charset = charset.lower() 
130          i = charset.find('iso') 
131          u = charset.find('utf') 
132          if i > 0: 
133              charset = charset[i:] 
134          elif u > 0: 
135              charset = charset[u:] 
136          # Some old emails say it's ascii or unkown but in reality is not 
137          # not use any charset not iso or utf 
138          elif i != 0 and u != 0: 
139              charset = None 
140   
141      for header in msg.keys(): 
142          value = msg.get_decoded_header(header) 
143          value, _ = pyzmail.decode_text(value, charset, None) 
144          value = value.encode('UTF-8') 
145          header = header.replace('.', ',')    # mongoDB don't likes '.' on keys 
146          d[header] = value 
147   
148      attach = [] 
149      if type(body) == str: 
150          body, _ = pyzmail.decode_text(body, charset, None) 
151          body = body.encode('UTF-8') 
152      # On attachments of emails sometimes it end up with a list of email.message 
153      elif type(body) == list: 
154          for part in body: 
155              zmail = pyzmail.PyzMessage(part) 
156              a = msg_to_dict(zmail) 
157              attach.append(a) 
158          body = attach[0]['Body'] 
159      d['Body'] = body 
160   
161      if len(msg.mailparts) > 1: 
162          for mailpart in msg.mailparts: 
163              zmail = pyzmail.PyzMessage(mailpart.part) 
164              a = msg_to_dict(zmail) 
165              attach.append(a) 
166   
167      if attach: 
168          d['Attachments'] = attach 
169   
170      return d 
171