def __init__(self, spec, bnst_id=0, newstyle=False):
self._mrph_list = MList()
self._tag_list = TList()
self.parent_id = -1
self.parent = None
self.children = []
self.dpndtype = ''
self.fstring = ''
self._pstring = ''
self.bnst_id = bnst_id
spec = spec.strip()
if spec == '*':
pass
elif newstyle:
items = spec.split(u"\t")
self.parent_id = int(items[2])
self.dpndtype = items[3]
self.fstring = items[17]
self.repname = items[6]
elif re.match(r'\* (-?\d+)([DPIA])(.*)$', spec):
match = re.match(r'\* (-?\d+)([DPIA])(.*)$', spec)
self.parent_id = int(match.group(1))
self.dpndtype = match.group(2)
self.fstring = match.group(3).strip()
else:
sys.stderr.write("Illegal bunsetsu spec: %s\n" % spec)
quit(1)
# Extract 正規化代表表記
if not newstyle:
self.repname = ''
match = re.search(r"<正規化代表表記:([^\"\s]+?)>", self.fstring)
if match:
self.repname = match.group(1)
def __init__(self, spec, bnst_id=0, juman_format=JUMAN_FORMAT.DEFAULT):
self._mrph_list = MList()
self._tag_list = TList()
self.midasi = ''
self.parent_id = -1
self.parent = None
self.children = []
self.prev = None
self.next = None
self.dpndtype = ''
self.fstring = ''
self._pstring = ''
self.bnst_id = bnst_id
spec = spec.strip()
if spec == '*':
pass
elif juman_format != JUMAN_FORMAT.DEFAULT: # TODO
items = spec.split("\t")
self.parent_id = int(items[2])
self.dpndtype = items[3]
self.fstring = items[17]
self.repname = items[6]
elif re.match(r'\* (-?\d+)([DPIA])(.*)$', spec):
match = re.match(r'\* (-?\d+)([DPIA])(.*)$', spec)
self.parent_id = int(match.group(1))
self.dpndtype = match.group(2)
self.fstring = match.group(3).strip()
else:
raise Exception("Illegal bunsetsu spec: %s" % spec)
self.features = Features(self.fstring)
# Extract 正規化代表表記
if juman_format == JUMAN_FORMAT.DEFAULT:
self.repname = ''
self.normalized_repname = ''
self.head_repname = ''
self.head_prime_repname = ''
normalized_repname = self.features.get("正規化代表表記")
if normalized_repname:
self.repname = normalized_repname
self.normalized_repname = normalized_repname
head_repname = self.features.get("主辞代表表記")
if head_repname:
self.head_repname = head_repname
head_prime_repname = self.features.get("主辞’代表表記")
if head_prime_repname:
self.head_prime_repname = head_prime_repname
def __init__(self, spec, bnst_id=0, juman_format=JUMAN_FORMAT.DEFAULT):
self._mrph_list = MList()
self._tag_list = TList()
self.midasi = ''
self.parent_id = -1
self.parent = None
self.children = []
self.dpndtype = ''
self.fstring = ''
self._pstring = ''
self.bnst_id = bnst_id
spec = spec.strip()
if spec == '*':
pass
elif juman_format != JUMAN_FORMAT.DEFAULT: # TODO
items = spec.split("\t")
self.parent_id = int(items[2])
self.dpndtype = items[3]
self.fstring = items[17]
self.repname = items[6]
elif re.match(r'\* (-?\d+)([DPIA])(.*)$', spec):
match = re.match(r'\* (-?\d+)([DPIA])(.*)$', spec)
self.parent_id = int(match.group(1))
self.dpndtype = match.group(2)
self.fstring = match.group(3).strip()
else:
raise Exception("Illegal bunsetsu spec: %s" % spec)
self.features = Features(self.fstring)
# Extract 正規化代表表記
if juman_format == JUMAN_FORMAT.DEFAULT:
self.repname = ''
self.normalized_repname = ''
self.head_repname = ''
self.head_prime_repname = ''
normalized_repname = self.features.get("正規化代表表記")
if normalized_repname:
self.repname = normalized_repname
self.normalized_repname = normalized_repname
head_repname = self.features.get("主辞代表表記")
if head_repname:
self.head_repname = head_repname
head_prime_repname = self.features.get("主辞’代表表記")
if head_prime_repname:
self.head_prime_repname = head_prime_repname
def draw_tag_tree(self, fh=None):
""" タグ列の依存関係を木構造として表現して出力する． """
tlist = TList()
for tag in self.tag_list():
tlist.push_tag(tag)
tlist.draw_tree(fh=fh)
class Bunsetsu(object):
"""
KNP による係り受け解析の単位である文節の各種情報を保持するオブジェクト．
Args:
spec (str): KNP出力のうち文節に該当する箇所の文字列
bnst_id (int): 文節ID
newstyle (bool): KNPフォーマットの種類 (公開版KNPの場合はFalse)
Attributes:
bnst_id (int): 文節ID
midasi (str): 見出し
parent (Bunsetsu): 親の文節オブジェクト
parent_id (int): 親の文節ID
children (list): 子の文節オブジェクトのリスト
repname (str): 正規化代表表記 (normalized_repnameに同じ)
normalized_repname (str): 正規化代表表記
head_repname (str): 主辞代表表記
head_prime_repname (str): 主辞’代表表記
fstring (str): feature情報
"""
def __init__(self, spec, bnst_id=0, newstyle=False):
self._mrph_list = MList()
self._tag_list = TList()
self.midasi = ''
self.parent_id = -1
self.parent = None
self.children = []
self.dpndtype = ''
self.fstring = ''
self._pstring = ''
self.bnst_id = bnst_id
spec = spec.strip()
if spec == '*':
pass
elif newstyle:
items = spec.split("\t")
self.parent_id = int(items[2])
self.dpndtype = items[3]
self.fstring = items[17]
self.repname = items[6]
elif re.match(r'\* (-?\d+)([DPIA])(.*)$', spec):
match = re.match(r'\* (-?\d+)([DPIA])(.*)$', spec)
self.parent_id = int(match.group(1))
self.dpndtype = match.group(2)
self.fstring = match.group(3).strip()
else:
raise Exception("Illegal bunsetsu spec: %s" % spec)
self.features = Features(self.fstring)
# Extract 正規化代表表記
if not newstyle:
self.repname = ''
self.normalized_repname = ''
self.head_repname = ''
self.head_prime_repname = ''
normalized_repname = self.features.get("正規化代表表記")
if normalized_repname:
self.repname = normalized_repname
self.normalized_repname = normalized_repname
head_repname = self.features.get("主辞代表表記")
if head_repname:
self.head_repname = head_repname
head_prime_repname = self.features.get("主辞’代表表記")
if head_prime_repname:
self.head_prime_repname = head_prime_repname
def push_mrph(self, mrph):
""" 新しい形態素オブジェクトをセットする """
if len(self._tag_list) > 0:
self._tag_list[-1].push_mrph(mrph)
self._mrph_list.push_mrph(mrph)
def push_tag(self, tag):
""" 新しい基本句オブジェクトをセットする """
if len(self._tag_list) == 0 and len(self._mrph_list) > 0:
raise Exception("Unsafe addition of tags!")
self._tag_list.push_tag(tag)
def set_midasi(self):
""" midasiをセットする """
for i in range(len(self._tag_list)):
self._tag_list[i].set_midasi()
self.midasi = ''.join(mrph.midasi for mrph in self.mrph_list())
def spec(self):
""" 文節に対応するKNP出力 """
return "* %d%s %s\n%s" % (self.parent_id, self.dpndtype, self.fstring,
self._tag_list.spec())
def mrph_list(self):
""" 文節を構成する全形態素オブジェクトを返す
Returns:
list: 形態素オブジェクトMorphemeのリスト
"""
return self._mrph_list
def tag_list(self):
""" 文節を構成する全基本句オブジェクトを返す
Returns:
list: 基本句オブジェクトTagのリスト
"""
return self._tag_list
def pstring(self, string=None):
""" draw_treeしたときに右側に出力する文字列を返す """
if string:
self._pstring = string
else:
return self._pstring
class Bunsetsu(object):
"""
KNP による係り受け解析の単位である文節の各種情報を保持するオブジェクト．
"""
def __init__(self, spec, bnst_id=0, newstyle=False):
self._mrph_list = MList()
self._tag_list = TList()
self.parent_id = -1
self.parent = None
self.children = []
self.dpndtype = ''
self.fstring = ''
self._pstring = ''
self.bnst_id = bnst_id
spec = spec.strip()
if spec == '*':
pass
elif newstyle:
items = spec.split("\t")
self.parent_id = int(items[2])
self.dpndtype = items[3]
self.fstring = items[17]
self.repname = items[6]
elif re.match(r'\* (-?\d+)([DPIA])(.*)$', spec):
match = re.match(r'\* (-?\d+)([DPIA])(.*)$', spec)
self.parent_id = int(match.group(1))
self.dpndtype = match.group(2)
self.fstring = match.group(3).strip()
else:
sys.stderr.write("Illegal bunsetsu spec: %s\n" % spec)
quit(1)
# Extract 正規化代表表記
if not newstyle:
self.repname = ''
match = re.search(r"<正規化代表表記:([^\"\s]+?)>", self.fstring)
if match:
self.repname = match.group(1)
self.hrepname = ''
match = re.search(r"<主辞代表表記:([^\"\s]+?)>", self.fstring)
if match:
self.hrepname = match.group(1)
self.hprepname = ''
match = re.search(r"<主辞’代表表記:([^\"\s]+?)>", self.fstring)
if match:
self.hprepname = match.group(1)
def push_mrph(self, mrph):
if self._tag_list:
self._tag_list[-1].push_mrph(mrph)
self._mrph_list.push_mrph(mrph)
def push_tag(self, tag):
if not self._tag_list and self._mrph_list:
sys.stderr.write("Unsafe addition of tags!\n")
quit(1)
self._tag_list.push_tag(tag)
def spec(self):
return "* %d%s %s\n%s" % (self.parent_id, self.dpndtype, self.fstring,
self._tag_list.spec())
def mrph_list(self):
return self._mrph_list
def tag_list(self):
return self._tag_list
def pstring(self, string=None):
if string:
self._pstring = string
else:
return self._pstring
def bnst_head(self):
if len(self.tag_list()) == 1:
return self.tag_list()[0]
for tag in self.tag_list():
if '文節内' not in tag.features:
return tag
def recursive_children(self):
def __recursive_children(bnst, bs):
assert bnst not in bnst.children
children = bnst.children
if not children:
return []
for c in children:
__recursive_children(c, bs)
bs.append(c)
return bs
return __recursive_children(self, [])
def recursive_adnominals(self):
modifiers = []
for c in self.children:
if '<連体修飾>' in c.fstring:
m_children = c.recursive_children()
m_children.append(c)
modifiers.extend(m_children)
return modifiers
def draw_tag_tree(self, fh=None):
""" タグ列の依存関係を木構造として表現して出力する． """
tlist = TList()
for tag in self.tag_list():
tlist.push_tag(tag)
tlist.draw_tree(fh=fh)
class Bunsetsu(object):
"""
KNP による係り受け解析の単位である文節の各種情報を保持するオブジェクト．
Args:
spec (str): KNP出力のうち文節に該当する箇所の文字列
bnst_id (int): 文節ID
juman_format (JUMAN_FORMAT): Jumanのlattice出力形式
Attributes:
bnst_id (int): 文節ID
midasi (str): 見出し
parent (Bunsetsu): 親の文節オブジェクト
parent_id (int): 親の文節ID
children (list): 子の文節オブジェクトのリスト
repname (str): 正規化代表表記 (normalized_repnameに同じ)
normalized_repname (str): 正規化代表表記
head_repname (str): 主辞代表表記
head_prime_repname (str): 主辞’代表表記
fstring (str): feature情報
"""
def __init__(self, spec, bnst_id=0, juman_format=JUMAN_FORMAT.DEFAULT):
self._mrph_list = MList()
self._tag_list = TList()
self.midasi = ''
self.parent_id = -1
self.parent = None
self.children = []
self.dpndtype = ''
self.fstring = ''
self._pstring = ''
self.bnst_id = bnst_id
spec = spec.strip()
if spec == '*':
pass
elif juman_format != JUMAN_FORMAT.DEFAULT: # TODO
items = spec.split("\t")
self.parent_id = int(items[2])
self.dpndtype = items[3]
self.fstring = items[17]
self.repname = items[6]
elif re.match(r'\* (-?\d+)([DPIA])(.*)$', spec):
match = re.match(r'\* (-?\d+)([DPIA])(.*)$', spec)
self.parent_id = int(match.group(1))
self.dpndtype = match.group(2)
self.fstring = match.group(3).strip()
else:
raise Exception("Illegal bunsetsu spec: %s" % spec)
self.features = Features(self.fstring)
# Extract 正規化代表表記
if juman_format == JUMAN_FORMAT.DEFAULT:
self.repname = ''
self.normalized_repname = ''
self.head_repname = ''
self.head_prime_repname = ''
normalized_repname = self.features.get("正規化代表表記")
if normalized_repname:
self.repname = normalized_repname
self.normalized_repname = normalized_repname
head_repname = self.features.get("主辞代表表記")
if head_repname:
self.head_repname = head_repname
head_prime_repname = self.features.get("主辞’代表表記")
if head_prime_repname:
self.head_prime_repname = head_prime_repname
def push_mrph(self, mrph):
""" 新しい形態素オブジェクトをセットする """
if len(self._tag_list) > 0:
self._tag_list[-1].push_mrph(mrph)
self._mrph_list.push_mrph(mrph)
def push_tag(self, tag):
""" 新しい基本句オブジェクトをセットする """
if len(self._tag_list) == 0 and len(self._mrph_list) > 0:
raise Exception("Unsafe addition of tags!")
self._tag_list.push_tag(tag)
def set_midasi(self):
""" midasiをセットする """
for i in range(len(self._tag_list)):
self._tag_list[i].set_midasi()
self.midasi = ''.join(mrph.midasi for mrph in self.mrph_list())
def spec(self):
""" 文節に対応するKNP出力 """
return "* %d%s %s\n%s" % (self.parent_id, self.dpndtype,
self.fstring, self._tag_list.spec())
def mrph_list(self):
""" 文節を構成する全形態素オブジェクトを返す
Returns:
list: 形態素オブジェクトMorphemeのリスト
"""
return self._mrph_list
def tag_list(self):
""" 文節を構成する全基本句オブジェクトを返す
Returns:
list: 基本句オブジェクトTagのリスト
"""
return self._tag_list
def pstring(self, string=None):
""" draw_treeしたときに右側に出力する文字列を返す """
if string:
self._pstring = string
else:
return self._pstring
class Bunsetsu(object):
"""
KNP による係り受け解析の単位である文節の各種情報を保持するオブジェクト．
"""
def __init__(self, spec, bnst_id=0, newstyle=False):
self._mrph_list = MList()
self._tag_list = TList()
self.parent_id = -1
self.parent = None
self.children = []
self.dpndtype = ''
self.fstring = ''
self._pstring = ''
self.bnst_id = bnst_id
spec = spec.strip()
if spec == '*':
pass
elif newstyle:
items = spec.split(u"\t")
self.parent_id = int(items[2])
self.dpndtype = items[3]
self.fstring = items[17]
self.repname = items[6]
elif re.match(r'\* (-?\d+)([DPIA])(.*)$', spec):
match = re.match(r'\* (-?\d+)([DPIA])(.*)$', spec)
self.parent_id = int(match.group(1))
self.dpndtype = match.group(2)
self.fstring = match.group(3).strip()
else:
sys.stderr.write("Illegal bunsetsu spec: %s\n" % spec)
quit(1)
# Extract 正規化代表表記
if not newstyle:
self.repname = ''
match = re.search(r"<正規化代表表記:([^\"\s]+?)>", self.fstring)
if match:
self.repname = match.group(1)
def push_mrph(self, mrph):
if len(self._tag_list) > 0:
self._tag_list[-1].push_mrph(mrph)
self._mrph_list.push_mrph(mrph)
def push_tag(self, tag):
if len(self._tag_list) == 0 and len(self._mrph_list) > 0:
sys.stderr.write("Unsafe addition of tags!\n")
quit(1)
self._tag_list.push_tag(tag)
def spec(self):
return "* %d%s %s\n%s" % (self.parent_id, self.dpndtype,
self.fstring, self._tag_list.spec())
def mrph_list(self):
return self._mrph_list
def tag_list(self):
return self._tag_list
def pstring(self, string=None):
if string:
self._pstring = string
else:
return self._pstring
def sprint_tag_tree(knp_result):
tlist = TList()
for tag in knp_result.tag_list():
tlist.push_tag(tag)
return tlist.sprint_tree()