def extract_basicinfo(text):
basic_info = {}
basic_info["ori_text_" + str(random.randint(0, 1000))] = text
lines = text.split('\n')
if lines[0].startswith(u"自我评价"):
basic_info["self_remark"] = "\n".join(lines[1:])
return basic_info
for line_pre, line in izip([""] + lines, lines):
## name update
m_update = re.search(u"更新时间：\s*(?P<up>\d{4}-\d{2}-\d{2})", line)
if m_update:
basic_info["updated_at"] = m_update.group("up") + " 00:00:00"
if len(line.split("|")) >= 3:
if len(StringUtils.get_words(line_pre.strip())) in [2, 3, 4]:
basic_info["name"] = line_pre.strip()
## email phone
email = match_basic.match_email(line)
phone = match_basic.match_phone(line)
if email: basic_info["contact_email"] = email
if phone: basic_info["contact_phone"] = phone
## gender age birth
if "U" != match_basic.match_gender(line) and (u"岁" in line
or u"经验" in line):
basic_info["gender"] = match_basic.match_gender(line)
mage = re.search(u"\d+岁", line)
if mage: basic_info["age"] = int(mage.group()[:-1])
mbirth = re.search(u"\d{4}年\s?\d{1,2}月\s?\d{1,2}日", line)
if mbirth and mage:
basic_info["birth"] = mbirth.group().replace(" ", "")
## others
mtel = re.search(u"家庭电话：(.+)", line)
if mtel: basic_info["contact_tel"] = mtel.group(1).strip()
mexp = re.search(u"(\d+)年工作经验", line)
if mexp: basic_info["work_experience"] = int(mexp.group(1))
mmary = re.search(u"婚姻状况：(.+)", line)
if mmary:
basic_info["marital"] = match_basic.match_marital(
mmary.group(1).strip())
maccount = re.search(u"(户口/国籍|户\s*口)(：|:)(?P<acc>.+)", line)
if maccount:
basic_info["account_str"] = maccount.group("acc").strip()
basic_info["account"] = match_region.match_region(
maccount.group("acc").strip())
maddress = re.search(u"(居住地|现居住)(:|：)\s*(?P<s>.+?)(\t|\||$)", line)
if maddress:
basic_info["address_str"] = maddress.group("s").strip()
basic_info["address"] = match_region.match_region(
maddress.group("s").strip())
return basic_info
def extract_basicinfo(text):
basic_info = {}
basic_info["ori_text_" + str(random.randint(0, 1000))] = text
lines = text.split('\n')
for line_pre, line in izip([""] + lines, lines):
## name update
if re.search(u"(.+)\s*流程状态.+标签.+", line) or re.search(
u"(.+)\s*标签.+", line) or re.search(u"(.+)\s*ID:\d+", line):
if u"流程状态" in line:
basic_info["name"] = re.search(u"(.+)\s*流程状态",
line).group(1).strip()
elif u"标签" in line:
basic_info["name"] = re.search(u"(.+)\s*标签",
line).group(1).strip()
else:
basic_info["name"] = re.search(u"(.+)\s*ID",
line).group(1).strip()
if line_pre.strip().startswith(u"更新时间："):
m = re.search("\d{4}-\d{2}-\d{2}", line_pre)
if m: basic_info["updated_at"] = m.group() + " 00:00:00"
if "name" not in basic_info and re.search(u"^ID(:|：)\d{4,}", line):
if len(StringUtils.get_words(line_pre)) in [2, 3, 4]:
basic_info["name"] = line_pre
if "name" not in basic_info and re.search(u"^\d{11}", line):
if len(StringUtils.get_words(line_pre)) in [2, 3, 4]:
basic_info["name"] = line_pre
## email phone
email = match_basic.match_email(line)
phone = match_basic.match_phone(line)
if email: basic_info["contact_email"] = email
if phone: basic_info["contact_phone"] = phone
## gender age birth
if "U" != match_basic.match_gender(line) and (u"岁" in line
or u"经验" in line):
basic_info["gender"] = match_basic.match_gender(line)
mage = re.search(u"\d+岁", line)
if mage: basic_info["age"] = int(mage.group()[:-1])
mbirth = re.search(u"\d{4}年\s?\d{1,2}月\s?\d{1,2}日", line)
if mbirth and mage:
basic_info["birth"] = mbirth.group().replace(" ", "")
## others
mtel = re.search(u"家庭电话：(.+)", line)
if mtel: basic_info["contact_tel"] = mtel.group(1).strip()
mexp = re.search(u"(\d+)年工作经验", line)
if mexp: basic_info["work_experience"] = int(mexp.group(1))
mmary = re.search(u"婚姻状况：(.+)", line)
if mmary:
basic_info["marital"] = match_basic.match_marital(
mmary.group(1).strip())
maccount = re.search(u"户口/国籍：(.+)", line)
if maccount:
basic_info["account_str"] = maccount.group(1).strip()
basic_info["account"] = match_region.match_region(
maccount.group(1).strip())
maddress = re.search(u"现居住?(.+?)(\||$)", line)
if maddress:
basic_info["address_str"] = maddress.group(1).strip()
basic_info["address"] = match_region.match_region(
maddress.group(1).strip())
return basic_info
def extract_basicinfo(text):
basic_info = {}
basic_info["ori_text_" + str(random.randint(0, 1000))] = text
lines = text.split('\n')
if lines[0].startswith(u"自我评价"):
basic_info["self_remark"] = "\n".join(lines[1:])
return basic_info
for line_pre, line in izip([""] + lines, lines):
## update
mupdate = re.search(u"简历更新时间(:|：)\s*(?P<t>\d{4}\.\d{2}\.\d{2})",
line_pre + line)
if mupdate:
basic_info["updated_at"] = mupdate.group("t").replace(
".", "-") + " 00:00:00"
## name
mname_id = re.search(u"^ID：", line_pre)
mname_ph = re.search(u"^手机：", line)
if mname_id and ("name" not in basic_info or not basic_info["name"]):
basic_info["name"] = line.strip() if len(
StringUtils.get_words(line.strip())) in [2, 3, 4] else ""
if mname_ph and ("name" not in basic_info or not basic_info["name"]):
basic_info["name"] = line_pre.strip() if len(
StringUtils.get_words(line_pre.strip())) in [2, 3, 4] else ""
## email phone
email = match_basic.match_email(line)
phone = match_basic.match_phone(line)
m_ph = re.search(u"^手机：(\d+)",
re.sub(u"(\(|（).+(\)|）)", "", line).replace(" ", ""))
if email: basic_info["contact_email"] = email
if m_ph: basic_info["contact_phone"] = m_ph.group(1)
elif phone and "contact_phone" not in basic_info:
basic_info["contact_phone"] = phone
## gender age birth
if "U" != match_basic.match_gender(line) and (u"岁" in line
or u"经验" in line):
basic_info["gender"] = match_basic.match_gender(line)
mage = re.search(u"\d+岁", line)
if mage: basic_info["age"] = int(mage.group()[:-1])
mbirth = re.search(u"(?P<y>\d{4})年\s*(?P<m>\d{1,2})月", line)
if mbirth and mage:
month = mbirth.group('m')
if len(mbirth.group("m")) == 1:
month = '0' + month
basic_info["birth"] = mbirth.group('y') + u'年' + month + u'月'
## others
mtel = re.search(u"家庭电话：(.+)", line)
if mtel: basic_info["contact_tel"] = mtel.group(1).strip()
mexp = re.search(u"(\d+)年工作经验", line)
if mexp: basic_info["work_experience"] = int(mexp.group(1))
mmary = re.search(u"(已|未)婚|Single|Married", line)
if mmary:
basic_info["marital"] = match_basic.match_marital(
mmary.group(0).strip())
maccount = re.search(u"户口：(.+?)(\||$)", line)
if maccount:
basic_info["account_str"] = maccount.group().strip()
basic_info["account"] = match_region.match_region(
maccount.group(1).strip())
maddress = re.search(u"现居住地：(.+?)(\||$)", line)
if maddress:
basic_info["address_str"] = maddress.group().strip()
basic_info["address"] = match_region.match_region(
maddress.group(1).strip())
basic_info["address_province"] = "10" #提醒获取
id_card = re.search(u"[0-9]{17}[xX0-9]", line)
if id_card:
basic_info["card"] = id_card.group(0)
overseas = re.search(u"有海外工作|Overseas Work", line)
if overseas:
basic_info["overseas"] = "Y"
return basic_info