def extract_imp_sen(imp_s, pre_s, usedex):
thing = ["こと", "事", "の"]
bp = False
one_sen = ""
for imp_sen in imp_s:
bp = False
cla = seperate_clause(imp_sen)
for context in cla: #[,[]]
for com in context[1]:
word = pas.morph_array(com) #文中の形とそれ以外
if word[0] in thing and word[2] == "非自立":
#print("TRUE：",word[0])
pre_s.append(one_sen)
one_sen = ""
bp = True
break
else:
one_sen = one_sen + word[0]
#print(one_sen)
if bp:
break
else:
# [非自立名詞]が＊＊でないとき
#print("else")
text = ""
num = imp_s.index(imp_sen)
imp = imp_sen.split(usedex[num])
if imp[0] != "":
#前部分にあるとき
cla = seperate_clause(imp[0])
for context in cla:
#print(context)
if context == cla[-1]:
#最後のimporが現れる文節のみ
subject = context[0][3].split("/") #syuji/jutugo
#print(subject)
morph = context[1][0:int(subject[0]) + 1]
for m in morph:
words = pas.morph_array(m)
text += words[0]
else:
for con in context[1]:
words = pas.morph_array(con)
text += words[0]
#print(text)
pre_s.append(text)
text = ""
else:
#後ろ部分にあるとき
print("behind")
cla = seperate_clause(imp[1])
return pre_s
def symbol_check(document):
symbol = ["。", "？", "！", "?"]
mark = ["？", "！", "?"]
sentences = []
string = ""
#[[[info],[morph]],....]
#係り受け解析
clauses = seperate_clause(document)
http = False
for acla in clauses:
for morph in acla[1]:
#1形態素
word = pas.morph_array(morph)
##url処理
if word[0] == "http":
#url処理
http = True
if string != "":
sentences.append(string)
string = ""
break
if http and word[2] == "数":
http = False
break
#ここから普通の処理
#文書中の記号の判
if word[0] in symbol:
if morph != acla[1][-1] and word[0] in mark:
#形態素末でない時
num = acla[1].index(morph)
#次の文節形態素判別
next_w = pas.morph_array(acla[1][num + 1])
if next_w[1] == "助詞":
#文中
string += word[0]
else:
#文区切り
sentences.append(string)
string = ""
else:
sentences.append(string)
string = ""
else:
string += word[0]
#print(string)
#最後が記号で終わる場合
if string != "":
sentences.append(string)
return sentences
def symbol_check(document) : #文内かっこの処理がまだ symbol = ["。","？","！"] mark = ["？","！"] sentences = [] string = "" #[[[info],[morph]],....] #係り受け解析 if isinstance(document,str): clauses = seperate_clause(document) else: clauses = document for acla in clauses : for morph in acla[1] : #1形態素 word = pas.morph_array(morph) #文書中の記号の判別 if word[0] in symbol : if morph != acla[1][-1] and word[0] in mark: #形態素末でない時 num = acla[1].index(morph) #次の文節形態素判別 next_w = pas.morph_array(acla[1][num + 1]) if next_w[1] == "助詞": #文中 string += word[0] else : #文区切り sentences.append(string) string = "" else : sentences.append(string) string = "" else : string += word[0] #print(string) #最後が記号で終わる場合 if string != "": sentences.append(string) return sentences
def search_hint_ex(document, usedex, prep_sen, score):
type1 = ["思う", "感じる", "考える"]
type2 = ["重要", "大切", "大事", "必要", "いいですね"]
type3 = ["ないでしょうか", "でしょうかね", "ではありません", "いただきたい", "ばならない", "どうでしょう"]
#一時保存用
hint_ar = [] #手がかり表現
prep_ar = [] #前の部分
#type1
expression = [["と", "助詞", "格助詞", "引用", "*", "*", "*", "と", "ト", "ト"],
["は", "助詞", "係助詞", "*", "*", "*", "*", "は", "ハ", "ワ"]]
direct = ["を", "助詞", "格助詞", "一般", "*", "*", "*", "を", "ヲ", "ヲ"] #をはまた考える
stopper = ["読点", "句点"]
#type2
#type3
sentences = symbol_check(document)
#print(sentences)
for sen in sentences:
hint_ar = []
prep_ar = []
#type1
stopper = ["読点", "句点"]
clauses = seperate_clause(sen)
end_point = -len(clauses)
clause = clauses #保険
count = -1
exist = False
if len(clauses) == 0:
continue
while True:
#後ろから探す
cla = clause[count] #[文節説明,[その要素,...]]
if exist:
#print("think exist")
#文字列生成処理
out_top = cla[1][-1] #文節末の要素
array = out_top.split(",")
if array[1] in stopper: #文末が、か。
#次の文へ
exist = False
else:
for com in reversed(cla[1]):
word = com.split("\t") #文中の形とそれ以外に分割
#print("add")
prep_ar[-1] = word[0] + prep_ar[-1]
#print(prep_ar)
count += -1
else:
#探す処理
#print("s")
out_end = cla[1][0] #文節頭の要素(思うとかで１文節使用しているから)
#array = out_end.split(",") #要素の属性をリスト化
array = pas.morph_array(out_end) #要素の属性をリスト化
#think要素があるか
if array[-3] in type1:
#表現が現れた時
#print(array)
prep_ar.append("")
exist = True
hint_ar.append(array[-3])
#thinkの直前の要素判定
if count - 1 < end_point:
prep_sen.pop()
break
next_cla = clause[count - 1]
#think表現の前部分を分析
words = pas.morph_array(next_cla[1][-1])
#print("w:",words)
if words in expression:
if words[0] == "は":
wordss = pas.morph_array(next_cla[1][-2])
if wordss in expression:
#とは 削除
clause[count - 1][1].pop()
clause[count - 1][1].pop()
else:
#と を削除
clause[count - 1][1].pop()
#print("rem: ",clause[count - 1][1])
if len(clause[count - 1][1]) == 0:
break
#think表現が動詞として使用されるとき
if words == direct:
#print("direct")
#prep_ar[-1] += array[0]
for c in cla[1]:
cm = pas.morph_array(c)
prep_ar[-1] += cm[0]
#print(prep_ar[-1])
count += -1 #インデントあってる？
if count < end_point:
break
#type2
imp_sentences = []
for impor in type2:
if impor in sen:
#句点について
sep = sen.split("、")
if len(sep) > 1:
#句点区切りがあるとき
for s in reversed(sep):
if impor in s:
#表現が現れている方を代入
sent = s
else:
#ないとき
sent = sen
sep_txt = sent.rsplit(impor, 1)
extr = sep_txt[0]
imp_sentences.append(extr)
hint_ar.append(impor)
if len(imp_sentences) != 0:
#[こと]とかを排除
prep_ar = extract_imp_sen(imp_sentences, prep_ar, hint_ar)
#type3
for esp in type3:
if esp in sen:
#句点について
sep = sen.split("、")
if len(sep) > 1:
#後ろ部分を採用
sent = sep[-1]
else:
sent = sen
sep_txt = sent.rsplit(esp, 1)
#print("sep:",sep_txt)
if esp == "ではありません" and sep_txt[1] == "":
#否定で終わる時
continue
extr = sep_txt[0]
prep_ar.append(extr)
hint_ar.append(esp)
#print("hint:",hint_ar)
#print("prep:",prep_ar)
if len(hint_ar) > 1:
#print("compare")
#文中に複数の手がかり表現
best_h = ""
best_prep = ""
max_h = 0
max_num = 0
for i in range(len(hint_ar)):
#スコア最大
if score[hint_ar[i]] > max_h:
max_h = score[hint_ar[i]]
best_h = hint_ar[i]
max_num = i
'''
if len(prep_ar[i]) < min_p :
min_p = len(prep_ar[i])
shortest_prep = prep_ar[i]
'''
#対応する前部分を選択
best_prep = prep_ar[max_num]
for prep in prep_ar:
if prep in best_prep:
best_prep = prep
usedex.append(best_h)
prep_sen.append(best_prep)
elif len(hint_ar) == 1:
usedex.append(hint_ar[0])
prep_sen.append(prep_ar[0])
#print("ex:",usedex)
#print("pre:",prep_sen)
return usedex, prep_sen
def search_think_expression(document) : #clauses
think = ["思う","感じる","考える"]
expression = [["と","助詞","格助詞","引用","*","*","*","と","ト","ト"],["を","助詞,格助詞,一般,*,*,*,を,ヲ,ヲ"]] #をはまた考える
stopper = ["読点","句点"]
think_sentences = []
count = -1
exist = False
usedex = []
#print(document)
sentences = symbol_check(document)
#print(sentences)
for sentence in sentences :
clauses = seperate_clause(sentence)
end_point = -len(clauses)
clause = clauses #保険
count = -1
exist = False
while True :
#後ろから探す
cla = clause[count] #[文節説明,[その要素,...]]
if exist :
print("exist")
#文字列生成処理
out_top = cla[1][-1] #文節末の要素
array = out_top.split(",")
if array[1] in stopper : #文末が、か。
#次の文へ
exist = False
else :
for com in reversed(cla[1]) :
word = com.split("\t") #文中の形とそれ以外に分割
#print("add")
think_sentences[-1] = word[0] + think_sentences[-1]
#print(think_sentences)
count += -1
else :
#探す処理
out_end = cla[1][0] #文節頭の要素(思うとかで１文節使用しているから)
array = out_end.split(",") #要素の属性をリスト化
#think要素があるか
if array[-3] in think :
#print(array[-3])
think_sentences.append("")
exist = True
usedex.append(array[-3])
#thinkの直前の要素判定
if count - 1 < end_point :
think_sentences.pop()
break
next_cla = clause[count - 1]
words = pas.morph_array(next_cla[1][-1])
if words == expression[0] :
#と を削除
clause[count - 1][1].pop()
#print(clause[count - 1][1])
if len(clause[count - 1][1]) == 0:
break
count += -1
if count < end_point :
break
if len(think_sentences) == 0:
print("nothing thinking")
else :
nothing = 0
#print(think_sentences)
return (usedex,think_sentences)
def search_importance(document) : #clause
importance = ["重要","必要","大事","大切", "必須"]
#sep = ["句点","読点"]
imp_sentences = [] #表現を含む文
imp_flag = False
one_sen = ""
end = False #文末かどうか
usedex = []
sentences = symbol_check(document)
for sen in sentences :
for impor in importance :
if impor in sen :
#句点について
sep = sen.split("、")
if len(sep) > 1:
for s in reversed(sep) :
if impor in s :
#表現が現れている方を代入
sent = s
else :
sent = sen
sep_txt = sent.rsplit(impor,1)
extr = sep_txt[0]
imp_sentences.append(extr)
usedex.append(impor)
#print(imp_sentences)
#-----------------------------------------------
important_thing = [] #その内容
thing = ["こと","事","の"]
bp = False
for imp_sen in imp_sentences :
bp = False
cla = seperate_clause(imp_sen)
for context in cla: #[,[]]
for com in context[1] :
word = pas.morph_array(com) #文中の形とそれ以外
if word[0] in thing and word[2] == "非自立":
print("TRUE：",word[0])
important_thing.append(one_sen)
#print(important_thing)
one_sen = ""
bp = True
break
else :
one_sen = one_sen + word[0]
#print(one_sen)
if bp :
break
else :
#print("else")
#print(important_thing)
text = ""
num = imp_sentences.index(imp_sen)
imp = imp_sen.split(usedex[num])
if imp[0] != "" :
#前部分にあるとき
cla = pas.seperate_clause(imp[0])
for context in cla :
#print(context)
if context == cla[-1] :
#最後のimporが現れる文節のみ
subject = context[0][3].split("/") #syuji/jutugo
#print(subject)
morph = context[1][0:int(subject[0]) + 1]
for m in morph :
words = pas.morph_array(m)
text += words[0]
else :
for con in context[1] :
words = pas.morph_array(con)
text += words[0]
#print(text)
important_thing.append(text)
text = ""
else :
#後ろ部分にあるとき
cla = seperate_clause(imp[1])
#mitei
if len(important_thing) == 0:
print("nothing important")
else :
nothing = 0
#print(important_thing)
return (usedex,important_thing)