def expand_by_link():
# 임시로 마지막 어미의 키워드 부착: 집에서 따라하지 마세요 :P
for key in groups.keys():
for klass in groups[key]:
for rule in klass['rules']:
rule.append(key)
refgroups = groups.copy()
def find_rules_to_attach(last):
rules = []
for key in refgroups.keys():
g = refgroups[key]
for k in g:
if (('-' in k['after'] or last in k['after']) and
(not ('notafter' in k) or not (last in k['notafter']))):
for r in k['rules']:
if re.match('.*' + ENC(r[1]) + '$',
ENC(last[:-1])):
rules.append(r)
return rules
def expand_class(klass):
while True:
rules_to_expand = [r for r in klass['rules'] if r[0][-1] == '-']
if not rules_to_expand:
return
new_rules = [r for r in klass['rules'] if r[0][-1] != '-']
for r in rules_to_expand:
last = r[-1]
attaches = find_rules_to_attach(last)
for a in attaches:
if a[2]:
striplen = len(ENC(a[2]))
else:
striplen = 0
new_suffix = DEC(ENC(r[0])[:-1-striplen] + a[0][1:])
new_rules.append([new_suffix] + r[1:3] + a[3:])
klass['rules'] = new_rules
for key in groups.keys():
if key[-1] != '-':
continue
for klass in groups[key]:
expand_class(klass)
# 임시로 부착했던 마지막 어미 키워드 삭제
for key in groups.keys():
for klass in groups[key]:
for rule in klass['rules']:
del rule[-1]
def expand_by_link():
# 임시로 마지막 어미의 키워드 부착: 집에서 따라하지 마세요 :P
for key in groups.keys():
for klass in groups[key]:
for rule in klass['rules']:
rule.append(key)
refgroups = groups.copy()
def find_rules_to_attach(last):
rules = []
for key in refgroups.keys():
g = refgroups[key]
for k in g:
if (('-' in k['after'] or last in k['after']) and
(not k.has_key('notafter') or not last in k['notafter'])):
for r in k['rules']:
if re.match(NFD(u'.*' + r[1] + '$'),
NFD(last[:-1].decode('utf-8'))):
rules.append(r)
return rules
def expand_class(klass):
while True:
rules_to_expand = [r for r in klass['rules'] if r[0][-1] == '-']
if not rules_to_expand:
return
new_rules = [r for r in klass['rules'] if r[0][-1] != '-']
for r in rules_to_expand:
last = r[-1]
attaches = find_rules_to_attach(last)
for a in attaches:
if a[2]:
striplen = len(NFD(a[2]))
else:
striplen = 0
new_suffix = NFC(NFD(r[0])[:-1-striplen] + a[0][1:])
new_rules.append([new_suffix] + r[1:3] + a[3:])
klass['rules'] = new_rules
for key in groups.keys():
if key[-1] != '-':
continue
for klass in groups[key]:
expand_class(klass)
# 임시로 부착했던 마지막 어미 키워드 삭제
for key in groups.keys():
for klass in groups[key]:
for rule in klass['rules']:
del rule[-1]
def expand_by_cond():
for key in groups.keys():
for klass in groups[key]:
new_rules = []
for rule in klass['rules']:
if isinstance(rule[1], list):
for c in rule[1]:
new_rules.append([rule[0], c] + rule[2:])
else:
new_rules.append(rule)
klass['rules'] = new_rules
def expand_by_cond():
for key in groups.keys():
for klass in groups[key]:
new_rules = []
for rule in klass['rules']:
if isinstance(rule[1], list) :
for c in rule[1]:
new_rules.append([rule[0], c] + rule[2:])
else:
new_rules.append(rule)
klass['rules'] = new_rules
def clean_up_cond():
for key in groups.keys():
for klass in groups[key]:
for c in ['after', 'notafter', 'cond', 'notcond']:
if c not in klass:
continue
new = set()
for item in klass[c]:
if item == '#용언':
new.add('#동사')
new.add('#형용사')
else:
new.add(item)
klass[c] = sorted(list(new))
return groups
def clean_up_cond():
for key in groups.keys():
for klass in groups[key]:
for c in ['after', 'notafter', 'cond', 'notcond']:
if not klass.has_key(c):
continue
new = set()
for item in klass[c]:
if item == '#용언':
new.add('#동사')
new.add('#형용사')
else:
new.add(item)
klass[c] = sorted(list(new))
return groups
expand_class(klass)
# 임시로 부착했던 마지막 어미 키워드 삭제
for key in groups.keys():
for klass in groups[key]:
for rule in klass['rules']:
del rule[-1]
expand_by_cond()
clean_up_cond()
expand_by_link()
# 연결이 끝나면 그룹끼리 구분할 필요가 없다.
klasses = []
for key in groups.keys():
for r in groups[key]:
r['name'] = key
klasses += groups[key]
# 선어말어미 연결 정보도 필요 없다.
for klass in klasses:
for condname in ['after', 'notafter']:
try:
klass[condname] = [c for c in klass[condname] if c[0] != '-']
except:
pass
# 같은 조건의 클래스를 머지한다.
def eq_klass_cond(a, b):
for klass in groups[key]:
expand_class(klass)
# 임시로 부착했던 마지막 어미 키워드 삭제
for key in groups.keys():
for klass in groups[key]:
for rule in klass['rules']:
del rule[-1]
expand_by_cond()
clean_up_cond()
expand_by_link()
# 연결이 끝나면 그룹끼리 구분할 필요가 없다.
klasses = []
for key in groups.keys():
klasses += groups[key]
# 선어말어미 연결 정보도 필요 없다.
for klass in klasses:
for condname in ['after', 'notafter']:
try:
klass[condname] = [c for c in klass[condname] if c[0] != '-']
except:
pass
# 같은 조건의 클래스를 머지한다.
def eq_klass_cond(a, b):
for condname in ['after', 'notafter', 'cond', 'notcond']:
if a.has_key(condname) and b.has_key(condname):
if a[condname] != b[condname]:
