def get_discipline(major_phrase,
prefer_index,
discipline_phrase_dic,
debug=False):
prefer_major = major_phrase[prefer_index]
prefer_major = SegmentHelper.normalize(
TextHelper.unicode_to_ascii(prefer_major))
prefer_major = TextHelper.get_dict_pattern(prefer_major,
discipline_phrase_dic)
if len(prefer_major) == 0:
if debug:
print("prefer major can not found match phrase in dict: %s" %
major_phrase[prefer_index])
prefer_major = ' '.join(major_phrase)
prefer_major = SegmentHelper.normalize(
TextHelper.unicode_to_ascii(prefer_major))
prefer_major = TextHelper.get_dict_pattern(prefer_major,
discipline_phrase_dic)
if len(prefer_major) == 0:
if debug:
print("Can not found major words: %s" % str(major_phrase))
return None
max_length = max([len(key) for key in prefer_major.keys()])
for major in prefer_major:
if len(major) == max_length:
return major
def get_highest_education(profile, education_phrase_dic,
discipline_phrase_dic):
education_dic = {}
for i in range(len(profile['education'])):
education = SegmentHelper.normalize(
TextHelper.unicode_to_ascii(profile['education'][i]))
education_dic[i] = TextHelper.get_dict_pattern(
education, education_phrase_dic)
education_dic = {
e_dic.keys()[0]: index
for index, e_dic in education_dic.items() if len(e_dic) > 0
}
if 'Doctor' in education_dic:
return ['Doctor'], [
ProfileHelper.get_discipline(profile['major'],
education_dic['Doctor'],
discipline_phrase_dic)
]
elif 'Master' in education_dic:
return ['Master'], [
ProfileHelper.get_discipline(profile['major'],
education_dic['Master'],
discipline_phrase_dic)
]
elif 'Bachelor' in education_dic:
return ['Bachelor'], [
ProfileHelper.get_discipline(profile['major'],
education_dic['Bachelor'],
discipline_phrase_dic)
]
else:
return [], []
def _get_education_words(self, education_dict):
education_phrase_dict = TextHelper.get_dict_pattern(
self.raw_position, education_dict)
if len(education_phrase_dict) == 0:
default_education_requirement = "Bachelor"
self.new_words_list.append(default_education_requirement)
return {default_education_requirement: 1}
else:
return education_phrase_dict
def get_skills(profile, skills_dic, debug=False):
skill_phrases = ' '.join(profile['skills'])
skill_phrases = SegmentHelper.normalize(
TextHelper.unicode_to_ascii(skill_phrases))
if debug:
print("right after normalize: %s" % skill_phrases)
skill_phrases_dict = TextHelper.get_dict_pattern(
skill_phrases, skills_dic)
if len(skill_phrases_dict) == 0:
# print ("can not found skills in %s" % str(skills))
return []
else:
return skill_phrases_dict.keys()
def _get_responsibility_words(self, education_dict):
return TextHelper.get_dict_pattern(self.raw_position, education_dict)
def _get_discipline_words(self, discipline_dict):
discipline_phrase_dict = TextHelper.get_dict_pattern(
self.raw_position, discipline_dict)
return discipline_phrase_dict
def _get_skill_words(self, skill_dict):
return TextHelper.get_dict_pattern(self.raw_position, skill_dict)