PageRenderTime 147ms CodeModel.GetById 11ms RepoModel.GetById 0ms app.codeStats 0ms

/mail_signature.py

https://gitlab.com/yaojian/RenjuAI
Python | 324 lines | 282 code | 20 blank | 22 comment | 1 complexity | 9c99b4f2cbe5adcdc5e1020eeeabcbba MD5 | raw file
    

  1. #!/usr/bin/python
    2. 

  2. #  -*- coding: utf-8 -*-
    3. 

  3. # author: yaojian-xy <yaojian-xy@360.cn>
    4. 

  4. import re
    5. 

  5. from sklearn.externals.joblib import Parallel, delayed
    6. 

  6. 

  7. from mail_global import MailGlobal, TimeoutException, time_limit
    8. 

  8. 

  9. 

  10. class MailSignatureParser:
    11. 

  11.     """
    12. 

  12.         extract mail signature, including: person names, address, phone, fax, zipcode and so on
    13. 

  13.     """
    14. 

  14.     CHINESE_SURNAME = u"李王张刘陈杨赵黄周吴徐孙胡朱高林何郭马罗梁宋郑谢韩唐冯于董萧程曹袁邓许傅沈曾彭吕\
    15. 

  15. 苏卢蒋蔡贾丁魏薛叶阎余潘杜戴夏钟汪田任姜范方石姚谭廖邹熊金陆郝孔白崔康毛邱秦江史\
    16. 

  16. 顾侯邵孟龙万段雷钱汤尹黎易常武乔贺赖龚文"
    17. 

    18. 

  18.     KEY_EMAIL, KEY_PHONE, KEY_ADDRESS, KEY_ZIP_CODE, KEY_COMPANY = 'email', 'phone', 'address', 'zipcode', 'company'
    19. 

  19.     KEY_WEBSITE, KEY_PERSON, KEY_OTHER = 'website', 'person', 'other'
    20. 

  20.     ENTITY_MAP = {KEY_ADDRESS: u"总\s*部|地\s*址|addr?",
    21. 

  21.                   KEY_PHONE: u"电\s*话|手\s*机|手机号码|Tel|座机|Mobile",
    22. 

  22.                   KEY_EMAIL: u"E-?mail",
    23. 

  23.                   KEY_ZIP_CODE: u"邮\s*编|PC\.",
    24. 

  24.                   KEY_PERSON: u"联系人|Contact",
    25. 

  25.                   KEY_WEBSITE: u"网\s*站|website",
    26. 

  26.                   KEY_OTHER: u"Q\s*Q|飞\s*信",
    27. 

  27.                 }
    28. 

  28. 

  29.     # ENTITY_MAP = dict((_v, k) for _v in v.split('|') for k, v in ENTITY_MAP.items())
    30. 

  30.     for k, v in ENTITY_MAP.items():
    31. 

  31.         ENTITY_MAP[k] = re.compile(u'^[ >\*]*(%s)[：:]' % v, re.IGNORECASE)
    32. 

  32. 

  33.     # ENTITY_REGEX = re.compile(u'(%s)[：:]' % '|'.join(ENTITY_MAP.keys()), re.IGNORECASE)
    34. 

  34.     # ENTITY_REGEX = re.compile(u'^[ >\*]*(%s)[：:]' % '|'.join(ENTITY_MAP.keys()), re.IGNORECASE)
    35. 

  35. 

  36.     REG_EMAIL = MailGlobal.email_reg_2
    37. 

  37.     #  ex:  http://www.hylandslaw.com  or  www.hylandslaw.com
    38. 

  38.     REG_URL = re.compile(u"(?:http[s]?://|www\.){1}(?:[a-zA-Z]|[0-9]|[$-_@.&+]|[!*\(\),]|(?:%[0-9a-fA-F][0-9a-fA-F]))+")
    39. 

  39. 

  40.     # ex: ["000-000-0000", "000 000 0000", "+86 10 6588 8825", "+86-21-60410082"]
    41. 

  41.     REG_PHONE = re.compile(
    42. 

  42.         u"(\d{3}[-\.\s]??\d{3}[-\.\s]??\d{4}|\(\d{3}\)\s*\d{3}[-\.\s]??\d{4}|\d{3}[-\.\s]??\d{4}|\d{2}[-\s]?\d{2}[-\s]?\d{4}[-\s]?\d{4})")
    43. 

  43. 

  44.     REG_COMPANY = re.compile(u"[:： ]*([\u4e00-\u9fff|\u0020-\u00ff]{2,}公司)")
    45. 

  45. 

  46.     REG_ADDRESS = re.compile(u"(?:.+省)?(?:.+市)?.+区(?:.+路)?(?:.+号)?")
    47. 

  47. 

  48.     REG_ZIP_CODE = re.compile(u"(?:邮\s*编|PC\.)\s*[：:]?\s*\d{6}")
    49. 

  49. 

  50.     #     REG_PERSON = re.compile(u"(?:(联系人|Contact|)\s*[：:]?\s*[%s][\u4e00-\u9fa5]{1,2})\
    51. 

  51.     # |(?:^[%s][\u4e00-\u9fa5]{1,2}$)" % (CHINESE_SURNAME, CHINESE_SURNAME))
    52. 

    53. 

  53.     REG_PERSON = re.compile(u"(^|[\s：:]+)[%s]([\u4e00-\u9fa5]{1,2}|\s{1,2}[\u4e00-\u9fa5])($|\s+)" % CHINESE_SURNAME)
    54. 

  54. 

  55.     # REG_PERSON = re.compile(u"((?:联系人|Contact)\s*[：:][%s][\u4e00-\u9fa5]{1,2}|\
    56. 

  56.     #                           (?:联系人|Contact)\s*[：:][a-zA-Z]+\s[a-zA-Z]+|\
    57. 

  57.     #                           ^[%s][\u4e00-\u9fa5]{1,2}$|^[a-zA-Z]+\s[a-zA-Z]+$)"
    58. 

  58.     #                         % (CHINESE_SURNAME, CHINESE_SURNAME))
    59. 

  59. 

  60.     REGEX_MAP = {KEY_EMAIL: REG_EMAIL,
    61. 

  61.                  KEY_WEBSITE: REG_URL,
    62. 

  62.                  KEY_PHONE: REG_PHONE,
    63. 

  63.                  KEY_COMPANY: REG_COMPANY,
    64. 

  64.                  KEY_ZIP_CODE: REG_ZIP_CODE,
    65. 

  65.                  KEY_ADDRESS: REG_ADDRESS,
    66. 

  66.                  KEY_PERSON: REG_PERSON,
    67. 

  67.                  }
    68. 

  68. 

  69.     SEPARATE_LINE_REGEX = re.compile(u"^\s*(?:-|_|—){4,}\s*$", re.MULTILINE)
    70. 

  70.     SERIES_EMPTY_LINE_REGEX = re.compile(u"(?:\n *){2,}")
    71. 

  71. 

  72.     REPLACE_REGEX = re.compile('&nbsp;|\"')
    73. 

  73.     REPLY_REGEX = re.compile(u'^\s*[>\*]{,3}\s*\*?(发件人|寄件者|From)[：:]{1}.+', re.MULTILINE)
    74. 

  74.     REPLY_SEPARATE_REGEX = re.compile(u"^\s*(?:-|—){4,}\s*(?:原始邮件|Original Message)\s*(?:-|—){4,}\s*$",
    75. 

  75.                                       re.MULTILINE | re.IGNORECASE)
    76. 

  76. 

  77.     IS_SIGNATURE_MAIL = True
    78. 

  78. 

  79.     def __init__(self):
    80. 

  80.         pass
    81. 

  81. 

  82.     @staticmethod
    83. 

  83.     def preprocess(mail_body):
    84. 

  84.         """
    85. 

  85.             preprocess
    86. 

  86.         :param mail_body:
    87. 

  87.         :return: mail body lines
    88. 

  88.         """
    89. 

  89.         body_stream = re.sub('&nbsp;|\"', '', mail_body)
    90. 

  90.         body_stream = body_stream.replace('\r\n', '\n')
    91. 

  91.         body_stream = MailSignatureParser.SERIES_EMPTY_LINE_REGEX.sub('\n\n', body_stream)
    92. 

  92.         mail_body_lines = body_stream.split('\n')
    93. 

  93.         return mail_body_lines
    94. 

  94. 

  95.     @staticmethod
    96. 

  96.     def remove_reply_pos(mail_body):
    97. 

  97.         match_span = MailSignatureParser.REPLY_SEPARATE_REGEX.search(mail_body)
    98. 

  98.         if match_span is not None:
    99. 

  99.             return mail_body[: match_span.span()[0]]
    100. 

  100.         match_span = MailSignatureParser.REPLY_REGEX.search(mail_body)
    101. 

  101.         if match_span is not None:
    102. 

  102.             return mail_body[: match_span.span()[0]]
    103. 

  103.         return mail_body
    104. 

  104. 

  105.     @staticmethod
    106. 

  106.     def check_sign_zone(mail_body_lines, max_matches=3, max_lines=12):
    107. 

  107.         """
    108. 

  108.             verify whether specail zone is a signature zone.
    109. 

  109.         :param mail_body_lines:
    110. 

  110.         :return:
    111. 

  111.         """
    112. 

  112.         match_list = set()
    113. 

  113.         for idx in xrange(min(len(mail_body_lines), max_lines)):
    114. 

  114.             line = mail_body_lines[idx].strip()
    115. 

  115.             # print "line:", idx, ": ", line
    116. 

  116.             if line.find(u"。") > 0 or line.find(u'，') > 0:  # when line is a param, skip search...
    117. 

  117.                 continue
    118. 

  118.             for _key, _reg in MailSignatureParser.ENTITY_MAP.items():  # loop all entity regex
    119. 

  119.                 fits = _reg.findall(line)
    120. 

  120.                 if len(fits) > 0 and len(fits[0]) > 0:  # when match standard format signature info
    121. 

  121.                     # map(lambda x: MailGlobal.logger.debug("check signature zone, key=%s" % x), fits)
    122. 

  122.                     match_list.add(_key)
    123. 

  123.                     break
    124. 

  124.                 else:  # when not matched
    125. 

  125.                     flag = False
    126. 

  126.                     for key, reg in MailSignatureParser.REGEX_MAP.items():
    127. 

  127.                         if reg.search(line) is not None:
    128. 

  128.                             match_pos = reg.search(line).span()
    129. 

  129.                             MailGlobal.logger.debug("found signature info, key=%s, value=%s" %
    130. 

  130.                                                     (key, line[match_pos[0]: match_pos[1]].encode('utf8')))
    131. 

  131.                             match_list.add(key)
    132. 

  132.                             flag = True
    133. 

  133.                             break
    134. 

  134.                     if flag:  # break current line
    135. 

  135.                         break
    136. 

  136.             if len(match_list) >= max_matches:
    137. 

  137.                 return True
    138. 

  138.         return False
    139. 

  139. 

  140.     # @staticmethod
    141. 

  141.     # def find_sign_zone_v2(mail, max_lines=10, max_matches=3, max_empty_line=3):
    142. 

  142.     #     # check reply mail
    143. 

  143.     #     # mail_body = MailSignatureParser.remove_reply_pos(mail['body'])
    144. 

  144.     #     mail_body = mail['body']
    145. 

  145.     #     # preprocess, split by lines
    146. 

  146.     #     mail_body_lines = MailSignatureParser.preprocess(mail_body)
    147. 

  147.     #     increment_empty_lines = 0
    148. 

  148.     #     sign_zone_start = len(mail_body_lines) - 1
    149. 

  149.     #     for line_no in xrange(len(mail_body_lines) - 1, -1, -1):
    150. 

  150.     #         line = mail_body_lines[line_no].strip()
    151. 

  151.     #         sign_zone_end = line_no
    152. 

  152.     #         sep_match = MailSignatureParser.SINGLE_SEPARATE_LINE_REGEX.search()
    153. 

  153. 

  154. 

  155.     @staticmethod
    156. 

  156.     def find_sign_zone(mail_body, max_lines=12, max_matches=3, max_empty_line=2):
    157. 

  157.         """
    158. 

  158.         :param mail:
    159. 

  159.         :param max_lines:
    160. 

  160.         :param max_matches:  max matched units number
    161. 

  161.         :return:
    162. 

  162.         """
    163. 

  163.         # check reply mail
    164. 

  164.         mail_body = MailSignatureParser.remove_reply_pos(mail_body)
    165. 

  165.         mail_body = mail_body.rstrip("\n\r ")
    166. 

  166.         # found separate line
    167. 

  167.         sep_match = MailSignatureParser.SEPARATE_LINE_REGEX.search(mail_body)
    168. 

  168.         sep_empty_lines = MailSignatureParser.SERIES_EMPTY_LINE_REGEX.search(mail_body)
    169. 

  169.         if sep_match is None and sep_empty_lines is None:
    170. 

  170.             return mail_body, not MailSignatureParser.IS_SIGNATURE_MAIL
    171. 

  171.         if sep_match is not None:
    172. 

  172.             sep_pos = sep_match.span()[0]
    173. 

  173.             if MailSignatureParser.check_sign_zone(mail_body[sep_pos:].split("\n")):  # verify signature zone
    174. 

  174.                 return mail_body[: sep_pos], MailSignatureParser.IS_SIGNATURE_MAIL
    175. 

  175.         # preprocess, split by lines
    176. 

  176.         mail_body_lines = MailSignatureParser.preprocess(mail_body)
    177. 

  177.         step = min(max_lines, len(mail_body_lines))
    178. 

  178.         st, ed = len(mail_body_lines) - step, len(mail_body_lines) - 1
    179. 

  179.         match_list = set()
    180. 

  180.         last_match_pos = len(mail_body_lines)
    181. 

  181.         flag_sign_zone = False
    182. 

  182.         accum_empty_lines = 0
    183. 

  183.         for idx in xrange(ed, st - 1, -1):
    184. 

  184.             line = mail_body_lines[idx].strip()
    185. 

  185.             # print "line:", idx, ": ", line
    186. 

  186.             if line.find(u"。") > 0 or line.count(u'，') > 1:  # when line is a param, skip search...
    187. 

  187.                 continue
    188. 

  188.             if line == "":
    189. 

  189.                 if flag_sign_zone:  # accumulate emtpy lines
    190. 

  190.                     accum_empty_lines += 1
    191. 

  191.                     if accum_empty_lines >= max_empty_line:  # when continuous empty lines reach to threshold value
    192. 

  192.                         last_match_pos = idx
    193. 

  193.                         break
    194. 

  194.                 continue
    195. 

  195.             accum_empty_lines = 0
    196. 

  196.             for _key, _reg in MailSignatureParser.ENTITY_MAP.items():
    197. 

  197.                 fits = _reg.findall(line)
    198. 

  198.                 if len(fits) > 0 and len(fits[0]) > 0:  # when match standard format signature info
    199. 

  199.                     last_match_pos = idx
    200. 

  200.                     # map(lambda x: MailGlobal.logger.debug("found signature info, key=%s" % x), fits)
    201. 

  201.                     match_list.add(_key)
    202. 

  202.                     flag_sign_zone = True
    203. 

  203.                     break
    204. 

  204.                 else:  # when not matched
    205. 

  205.                     for key, reg in MailSignatureParser.REGEX_MAP.items():
    206. 

  206.                         if reg.search(line) is not None:
    207. 

  207.                             last_match_pos = idx
    208. 

  208.                             match_pos = reg.search(line).span()
    209. 

  209.                             MailGlobal.logger.debug("found signature info, key=%s, value=%s" %
    210. 

  210.                                                    (key, line[match_pos[0]: match_pos[1]].encode('utf8')))
    211. 

  211.                             match_list.add(key)
    212. 

  212.                             flag_sign_zone = True
    213. 

  213.         if len(match_list) >= max_matches:
    214. 

  214.             return u'\n'.join(mail_body_lines[: last_match_pos]), MailSignatureParser.IS_SIGNATURE_MAIL
    215. 

  215.         # not found signature
    216. 

  216.         return mail_body, not MailSignatureParser.IS_SIGNATURE_MAIL
    217. 

  217. 

  218. 

  219. # def parallel_handle_signature(mail_sets, n_jobs=MailGlobal.etc_map['n_jobs']):
    220. 

  220. #     result = Parallel(n_jobs=n_jobs, verbose=11)(
    221. 

  221. #         delayed(single_handle_signature)(index, mail['body']) for index, mail in mail_sets.df.iterrows())
    222. 

  222. #     for item in result:
    223. 

  223. #         index, mail_body = item
    224. 

  224. #         mail_sets.df.loc[index, 'body'] = mail_body
    225. 

  225. #     return mail_sets
    226. 

  226. 

  227. def parallel_handle_signature(mails, n_jobs=MailGlobal.etc_map['n_jobs']):
    228. 

  228.     verbose = 1 if MailGlobal.DEBUG_MODEL else 0
    229. 

  229.     result = Parallel(n_jobs=n_jobs, verbose=verbose)(
    230. 

  230.         delayed(single_handle_signature)(i, mail['body']) for i, mail in enumerate(mails))
    231. 

  231.     for item in result:
    232. 

  232.         index, mail_body = item
    233. 

  233.         mails[index]['body'] = mail_body
    234. 

  234.     return mails
    235. 

  235. 

  236. 

  237. def single_handle_signature(index, mail_body, time_it=10):
    238. 

  238.     try:
    239. 

  239.         with time_limit(time_it):
    240. 

  240.             mail_body, _ = MailSignatureParser.find_sign_zone(mail_body)
    241. 

  241.     except TimeoutException:
    242. 

  242.         MailGlobal.logger.warn("signature exec time out, index=%s, text=%s" % (index, mail_body.encode('utf8')))
    243. 

  243.     return index, mail_body
    244. 

  244. 

  245. 

  246. def metric_signature(mail_file):
    247. 

  247.     import random
    248. 

  248.     from mail_io import load_mail_file
    249. 

  249.     mail_datas = load_mail_file(mail_file, raw_type="dict")
    250. 

  250.     accuracy_ratio = 0.0
    251. 

  251.     # import pdb
    252. 

  252.     random.shuffle(mail_datas)
    253. 

  253.     for mail in mail_datas:
    254. 

  254.         # print "\n" * 3, "^" * 40
    255. 

  255.         # print mail['body']
    256. 

  256.         # pdb.set_trace()
    257. 

  257.         mail_body, is_sign = MailSignatureParser.find_sign_zone(mail['body'])
    258. 

  258.         # print "is sign? : ", is_sign
    259. 

  259.         # if not is_sign:
    260. 

  260.         #     mail_body, is_sign = MailSignatureParser.find_sign_zone(mail['body'])
    261. 

  261.         # print "\n++++++++++\n"
    262. 

  262.         # print mail_body
    263. 

  263.         accuracy_ratio += is_sign
    264. 

  264.     accuracy_ratio /= len(mail_datas)
    265. 

  265.     MailGlobal.logger.info("signature mail metric of accuracy is : %.8f" % accuracy_ratio)
    266. 

  266. 

  267. 

  268. def check_signature(mail_file):
    269. 

  269.     import random
    270. 

  270.     from mail_entity import MailSet
    271. 

  271.     mail_sets = MailSet()
    272. 

  272.     mail_sets.load(mail_file, file_format='pkl')
    273. 

  273.     import pdb
    274. 

  274.     # for _, mail in mail_sets.df.iterrows():
    275. 

  275.     while True:
    276. 

  276.         indics = random.randint(0, mail_sets.length() - 1)
    277. 

  277.         mail = mail_sets.df.iloc[indics]
    278. 

  278.         mail_body, is_sign = MailSignatureParser.find_sign_zone(mail['body'])
    279. 

  279.         if is_sign:
    280. 

  280.             print "\n" * 3, "^" * 40
    281. 

  281.             print mail['body']
    282. 

  282.             pdb.set_trace()
    283. 

  283.             print "\n++++++++++\n"
    284. 

  284.             print mail_body
    285. 

  285. 

  286. 

  287. def sample_sign_mail(mail_files, output):
    288. 

  288.     import random
    289. 

  289.     import json
    290. 

  290.     from mail_io import load_mail_files
    291. 

  291. 

  292.     mail_datas = load_mail_files(mail_files)
    293. 

  293.     with open(output, 'a+') as sign_writer:
    294. 

  294.         while True:
    295. 

  295.             mail = random.choice(mail_datas)
    296. 

  296.             if len(mail['body'].split('\n')) < 8:
    297. 

  297.                 continue
    298. 

  298.             print "\n" * 3, "^" * 50
    299. 

  299.             print mail['body']
    300. 

  300.             while True:
    301. 

  301.                 is_sign = raw_input("is sign email ? [0, 1]: ")
    302. 

  302.                 if is_sign == "":
    303. 

  303.                     is_sign = "0"
    304. 

  304.                 if is_sign == '0' or is_sign == '1':
    305. 

  305.                     break
    306. 

  306.             if is_sign == "1":
    307. 

  307.                 json_mail = mail.m
    308. 

  308.                 json_mail['is_sign'] = int(is_sign)
    309. 

  309.                 sign_writer.write("%s\n" % json.dumps(json_mail))
    310. 

  310. 

  311. 

  312. if __name__ == "__main__":
    313. 

  313.     # sample_sign_mail(["./data/2015-12-11"], "./data/test_sign_mail.json")
    314. 

  314.     # metric_signature("./data/test_sign_mail.json")
    315. 

  315. 

  316.     # check_signature("./data/2015-12-18_token.pkl")
    317. 

  317.     from mail_preprocess import parse_mail_body
    318. 

  318.     MailGlobal.logger._debug = True
    319. 

  319.     mail_body = u"\r\n\r\n\r\n<div class=\"\">\r\n<p class=\"\" style=\"mso-margin-top-alt:auto;mso-margin-bottom-alt:auto\"><span lang=\"EN-US\" style=\"font-family:&quot;Arial&quot;,&quot;sans-serif&quot;;color:black\">Hi</span><span style=\"font-family:宋体;color:black\">，赵丽超，</span><span lang=\"EN-US\"><o:p></o:p></span></p>\r\n<p class=\"\" style=\"mso-margin-top-alt:auto;mso-margin-bottom-alt:auto\"><span style=\"font-family:宋体;color:black\">请在</span><span lang=\"EN-US\" style=\"font-family:&quot;Arial&quot;,&quot;sans-serif&quot;;color:black\"><a target=\"_blank\" href=\"http://url.qmail.com/cgi-bin/safejmp?action=check_link&amp;url=http%3A%2F%2Fikang.com%2F\" target=\"_blank\">http://iKang.com</a></span><span style=\"font-family:宋体;color:black\">或</span><span style=\"font-family:&quot;Arial&quot;,&quot;sans-serif&quot;;color:black\">\r\n<span lang=\"EN-US\"><a target=\"_blank\" href=\"http://url.qmail.com/cgi-bin/safejmp?action=check_link&amp;url=http%3A%2F%2Fm.ikang.com%2F\" target=\"_blank\">http://m.iKang.com</a></span></span><span style=\"font-family:宋体;color:black\">使用以下卡号和密码进行预约体检（无需付费，体检报告会由医院直接送往公司无须另外跟进），可在元旦期间进行体检：</span><span lang=\"EN-US\"><o:p></o:p></span></p>\r\n<p class=\"\" style=\"mso-margin-top-alt:auto;mso-margin-bottom-alt:auto\"><span style=\"font-family:宋体;color:black\">卡号：</span><span lang=\"EN-US\" style=\"font-size:10.0pt;font-family:&quot;Arial&quot;,&quot;sans-serif&quot;;color:black\">1111000022977623</span><span lang=\"EN-US\"><o:p></o:p></span></p>\r\n<p class=\"\" style=\"mso-margin-top-alt:auto;mso-margin-bottom-alt:auto\"><span style=\"font-family:宋体;color:black\">密码：</span><span lang=\"EN-US\" style=\"font-size:10.0pt;font-family:&quot;Arial&quot;,&quot;sans-serif&quot;;color:black\">195555</span><span lang=\"EN-US\"><o:p></o:p></span></p>\r\n<p class=\"\" style=\"mso-margin-top-alt:auto;mso-margin-bottom-alt:auto\"><span style=\"font-family:宋体;color:black\">如有疑问，随时沟通。收到请回复，谢谢！</span><span lang=\"EN-US\"><o:p></o:p></span></p>\r\n<p class=\"\"><span lang=\"EN-US\" style=\"color:#1F497D\"><o:p>&nbsp;</o:p></span></p>\r\n<p class=\"\"><span lang=\"EN-US\" style=\"color:#1F497D\"><o:p>&nbsp;</o:p></span></p>\r\n<p class=\"\"><span style=\"font-size:12.0pt;font-family:&quot;微软雅黑&quot;,&quot;sans-serif&quot;;color:#5F497A\">柏雪宇</span><span lang=\"EN-US\" style=\"font-size:12.0pt;font-family:&quot;微软雅黑&quot;,&quot;sans-serif&quot;;color:#5F497A\">&nbsp;<o:p></o:p></span></p>\r\n<p class=\"\"><span style=\"font-size:12.0pt;font-family:&quot;微软雅黑&quot;,&quot;sans-serif&quot;;color:#5F497A\">人力资源部</span><span lang=\"EN-US\" style=\"font-size:12.0pt;font-family:&quot;微软雅黑&quot;,&quot;sans-serif&quot;;color:#5F497A\"><o:p></o:p></span></p>\r\n<p class=\"\"><span lang=\"EN-US\" style=\"font-family:宋体;color:#505050\"><o:p>&nbsp;</o:p></span></p>\r\n<p class=\"\"><b><span style=\"font-size:14.0pt;font-family:&quot;微软雅黑&quot;,&quot;sans-serif&quot;;color:#387D39\">奇虎<span lang=\"EN-US\">360<o:p></o:p></span></span></b></p>\r\n<p class=\"\"><b><span style=\"font-size:12.0pt;font-family:华文细黑;color:#505050\">手机：</span></b><span lang=\"EN-US\" style=\"font-family:华文细黑;color:#505050\">13260819821<o:p></o:p></span></p>\r\n<p class=\"\"><b><span style=\"font-size:12.0pt;font-family:华文细黑;color:#505050\">电话：</span></b><span lang=\"EN-US\" style=\"font-family:华文细黑;color:#505050\">010-5244-8212<o:p></o:p></span></p>\r\n<p class=\"\"><b><span style=\"font-size:12.0pt;font-family:华文细黑;color:#505050\">邮件：</span></b><u><span lang=\"EN-US\" style=\"font-family:华文细黑;color:blue\">baixueyu@360.cn</span></u><span lang=\"EN-US\" style=\"font-family:华文细黑;color:#505050\"><o:p></o:p></span></p>\r\n<p class=\"\"><b><span style=\"font-size:12.0pt;font-family:华文细黑;color:#505050\">地址：</span></b><span style=\"font-family:华文细黑;color:#505050\">北京市朝阳区酒仙桥路<span lang=\"EN-US\">6</span>号院（电子城</span><span style=\"font-family:华文细黑;color:#505050\">&#8226;</span><span style=\"font-family:华文细黑;color:#505050\">国际电子总部）</span><span lang=\"EN-US\" style=\"font-family:华文细黑;color:#505050\">2</span><span style=\"font-family:华文细黑;color:#505050\">号楼</span><span lang=\"EN-US\" style=\"font-family:华文细黑;color:#505050\"><o:p></o:p></span></p>\r\n<p class=\"\"><span lang=\"EN-US\" style=\"color:#1F497D\"><o:p>&nbsp;</o:p></span></p>\r\n<p class=\"\" align=\"left\" style=\"text-align:left\"><b><span style=\"font-size:10.0pt;font-family:宋体\">发件人<span lang=\"EN-US\">:</span></span></b><span lang=\"EN-US\" style=\"font-size:10.0pt;font-family:宋体\"> hengyiponiu@sina.com [mailto:hengyiponiu@sina.com]\r\n<br>\r\n</span><b><span style=\"font-size:10.0pt;font-family:宋体\">发送时间<span lang=\"EN-US\">:</span></span></b><span lang=\"EN-US\" style=\"font-size:10.0pt;font-family:宋体\"> 2015</span><span style=\"font-size:10.0pt;font-family:宋体\">年<span lang=\"EN-US\">12</span>月<span lang=\"EN-US\">29</span>日<span lang=\"EN-US\">\r\n 11:49<br>\r\n</span><b>收件人<span lang=\"EN-US\">:</span></b><span lang=\"EN-US\"> </span>柏雪宇<span lang=\"EN-US\"><br>\r\n</span><b>主题<span lang=\"EN-US\">:</span></b><span lang=\"EN-US\"> </span>回复：答复<span lang=\"EN-US\">:\r\n</span>回复：欢迎加入<span lang=\"EN-US\">360<o:p></o:p></span></span></p>\r\n<p class=\"\" align=\"left\" style=\"text-align:left\"><span lang=\"EN-US\"><o:p>&nbsp;</o:p></span></p>\r\n<p class=\"\"><span lang=\"EN-US\"><o:p>&nbsp;</o:p></span></p>\r\n</div>\r\n\r\n\r\n\r\n"
    320. 

  320.     mail_body = parse_mail_body(mail_body)
    321. 

  321.     print mail_body
    322. 

  322.     mail_body, is_sign = MailSignatureParser.find_sign_zone(mail_body)
    323. 

  323.     print is_sign
    324. 

  324.     print mail_body
    325. 

  325.