def clean_authors(authors):
cleaned_authors = []
authors = authors.lower()
# get rid of commas where there are suffixes, like Jr. or III
authors = authors.replace(", jr.", " jr.")
authors = authors.replace(", iii", " iii")
authors = authors.replace(", ph.d", "")
# special cases
authors = authors.replace("organizer:", "")
authors = authors.replace("roel m,", "roel m.")
if authors == 'kozue miyashiro, etsuko harada, t.':
author_list = ['kozue miyashiro', 'etsuko harada, t.']
else:
author_list = authors.split(",")
for author in author_list:
author = HumanName(author.lower())
if author.first == '' or author.last == '':
raise ValueError("invalid author name: {}".format(author))
author.capitalize()
author.string_format = u"{last}, {title} {first} {middle}, {suffix}"
cleaned_authors.append(unicode(author))
return cleaned_authors
def extractFirstName(name, order):
'''Split on dots'''
name = ' '.join(name.split('.'))
'''Split on - '''
name = ' '.join(name.split('-'))
'''Replace numbers by whitespace'''
oldname = name
name = re.sub("\d+", "", name)
if not len(name):
name = re.sub("\d+", "_", oldname)
oldname = name
'''Replace ? by whitespace'''
name = re.sub("\?", "", name)
if not len(name):
name = re.sub("\?", "_", oldname)
name = ' '.join(name.split('_'))
'''Use the Python name parser'''
try:
firstName = getFirstNameFromHumanName(HumanName(name), order)
except:
firstName = getFirstNameFromSplitName(name.split(), order)
'''If fail, use heuristics'''
if firstName.strip() == name.strip():
'''firstName('Ben Voigt') = 'Ben Voigt'!!!'''
if len(name.split()) == 2:
firstName = getFirstNameFromSplitName(name.split(), order)
else:
'''Try CamelCase'''
uncamel = ' '.join(splitCamelCase(name).split('_'))
if uncamel != name:
try:
firstName = HumanName(uncamel).first
if len(firstName.split()) == 2:
firstName = getFirstNameFromSplitName(firstName.split(), order)
except:
firstName = getFirstNameFromSplitName(uncamel.split(), order)
if firstName == 'Mc':
firstName = ''
if len(firstName) == 1:
firstName = ''
return firstName.lower()
def names_compare(name1, name2):
"""
Takes string arguments with human names and returns indication of match between names.
:param name1: String argument with name
:param name2: String argument with name
:return: "exact" for definite matches, "last" for only last name matches, False for non-matches
"""
if not isinstance(name1, str) or not isinstance(name2, str):
raise TypeError(
"CosineCalc.names_compare must receive both string arguments.")
name1 = HumanName(name1.lower())
name2 = HumanName(name2.lower())
# Check for exact matches
last_names_match = name1.last == name2.last
first_names_match = name1.first == name2.first
# Check for short names
order = len(name1.first) < len(name2.first)
if order:
nick_name = name1.first in name2.first
else:
nick_name = name2.first in name1.first
if last_names_match and first_names_match:
result = "exact"
elif last_names_match and nick_name:
result = "exact"
elif last_names_match:
result = "last"
else:
result = False
return result
def extractFirstName(name, order):
'''Split on dots'''
name = ' '.join(name.split('.'))
'''Split on - '''
name = ' '.join(name.split('-'))
'''Replace numbers by whitespace'''
oldname = name
name = re.sub("\d+", "", name)
if not len(name):
name = re.sub("\d+", "_", oldname)
oldname = name
'''Replace ? by whitespace'''
name = re.sub("\?", "", name)
if not len(name):
name = re.sub("\?", "_", oldname)
name = ' '.join(name.split('_'))
'''Use the Python name parser'''
try:
firstName = getFirstNameFromHumanName(HumanName(name), order)
except:
firstName = getFirstNameFromSplitName(name.split(), order)
'''If fail, use heuristics'''
if firstName.strip() == name.strip():
'''firstName('Ben Voigt') = 'Ben Voigt'!!!'''
if len(name.split()) == 2:
firstName = getFirstNameFromSplitName(name.split(), order)
else:
'''Try CamelCase'''
uncamel = ' '.join(splitCamelCase(name).split('_'))
if uncamel != name:
try:
firstName = HumanName(uncamel).first
if len(firstName.split()) == 2:
firstName = getFirstNameFromSplitName(
firstName.split(), order)
except:
firstName = getFirstNameFromSplitName(
uncamel.split(), order)
if firstName == 'Mc':
firstName = ''
if len(firstName) == 1:
firstName = ''
return firstName.lower()
def initContact(contactId: str):
assert contactId not in activeContacts
activeContacts.add(contactId)
contact = dir.getContact(contactId)
contactsToEmails[contactId] = contact['email']
name = " ".join(
filter(None, [
contact.get('title_before_name'),
contact.get('first_name'),
contact.get('last_name'),
contact.get('title_after_name')
]))
#name = name.translate(str.maketrans('', '', string.punctuation))
name = name.translate(str.maketrans('', '', '@'))
name = HumanName(name.lower().strip())
name.capitalize()
if re.search('^(\w\.)+$', name.first):
name.first = name.first.upper()
contactsToNames[contactId] = name.__str__()
