Compare commits
No commits in common. "mecab" and "master" have entirely different histories.
3 changed files with 97 additions and 117 deletions
27
README.md
27
README.md
|
|
@ -1,24 +1,32 @@
|
|||
# Pseudo Chinese (with MeCab)
|
||||
# Pseudo Chinese
|
||||
Convert Japanese to pseudo-Chinese.
|
||||
|
||||
## Description
|
||||
This tool will automatically generate fake Chinese from Japanese sentences.
|
||||
|
||||
Using MeCab to parse and word-tag Japanese sentences instead of COTOHA API.
|
||||
|
||||
## Demo
|
||||
私は本日定時退社します -> 我本日定時退社
|
||||
私は本日定時退社します -> 我本日定時退社也
|
||||
|
||||
私はお酒を飲みたい -> 我御酒飲欲
|
||||
私はお酒を飲みたい -> 我飲酒希望
|
||||
|
||||
## Requirement
|
||||
- Python 3.5.1
|
||||
- mecab-python3
|
||||
- unidic-lite
|
||||
- [COTOHA API](https://api.ce-cotoha.com/contents/index.html)
|
||||
|
||||
You need to register for a COTOHA API account before you can run this tool.
|
||||
|
||||
Once you have registered your COTOHA API account, you will set your Client ID and Client Secret to `env.json` .
|
||||
|
||||
```json
|
||||
{
|
||||
"client_id": "yourclinetid",
|
||||
"client_secret": "yourclinetsecret"
|
||||
}
|
||||
```
|
||||
|
||||
## Usage
|
||||
```
|
||||
$ python -u pseudo-chinese.py [sentence]
|
||||
$ python -u pseudo-chinese.py
|
||||
```
|
||||
|
||||
## Contribution
|
||||
|
|
@ -34,5 +42,4 @@ MIT
|
|||
|
||||
## Author
|
||||
|
||||
- [Shoichiro Kono](https://github.com/k2font) (orig. creater)
|
||||
- [Tan Kian-ting](https://github.com/yoxem) (porting to MeCab, and modified it.)
|
||||
[Shoichiro Kono](https://github.com/k2font)
|
||||
|
|
|
|||
4
env.json
Normal file
4
env.json
Normal file
|
|
@ -0,0 +1,4 @@
|
|||
{
|
||||
"client_id": "<Client ID>",
|
||||
"client_secret": "<Client Secret>"
|
||||
}
|
||||
|
|
@ -1,52 +1,48 @@
|
|||
import functools
|
||||
import MeCab
|
||||
import requests
|
||||
import json
|
||||
import sys
|
||||
import re
|
||||
|
||||
BASE_URL = "https://api.ce-cotoha.com/api/dev/nlp/"
|
||||
|
||||
# アクセストークンを取得する関数
|
||||
# Function to get the access token.
|
||||
# 获取访问令牌的函数
|
||||
def auth(client_id, client_secret):
|
||||
token_url = "https://api.ce-cotoha.com/v1/oauth/accesstokens"
|
||||
headers = {
|
||||
"Content-Type": "application/json",
|
||||
"charset": "UTF-8"
|
||||
}
|
||||
|
||||
data = {
|
||||
"grantType": "client_credentials",
|
||||
"clientId": client_id,
|
||||
"clientSecret": client_secret
|
||||
}
|
||||
|
||||
r = requests.post(token_url,headers=headers,data=json.dumps(data))
|
||||
|
||||
return r.json()["access_token"]
|
||||
|
||||
# 形態素解析する関数
|
||||
# Function for morphological analysis.
|
||||
# 形态学分析功能
|
||||
def parse(sentence):
|
||||
mecab_tagger = MeCab.Tagger()
|
||||
raw_result = mecab_tagger.parse(sentence).split('\n')
|
||||
result = []
|
||||
for i in raw_result[:-2]:
|
||||
j = i.split('\t')
|
||||
item = dict()
|
||||
item['form'] = j[0] # 食べ
|
||||
#print(j)
|
||||
if len(j) > 1:
|
||||
item['lemma'] = j[3] # 食べる
|
||||
item['pos'] = j[4] # 動詞-一般
|
||||
item['features'] = j[6] # 連用形-一般
|
||||
else:
|
||||
item['lemma'] = j[0]
|
||||
item["pos"] = ""
|
||||
item["features"] = ""
|
||||
def parse(sentence, access_token):
|
||||
base_url = BASE_URL
|
||||
|
||||
result.append(item)
|
||||
return result
|
||||
headers = {
|
||||
"Content-Type": "application/json",
|
||||
"charset": "UTF-8",
|
||||
"Authorization": "Bearer {}".format(access_token)
|
||||
}
|
||||
|
||||
data = {
|
||||
"sentence": sentence,
|
||||
"type": "default"
|
||||
}
|
||||
|
||||
def is_hira(string):
|
||||
if isinstance(string, str):
|
||||
string = list(string)
|
||||
if len(string) == 0:
|
||||
return False
|
||||
elif len(string) == 1:
|
||||
return (("ぁ" <= string[0]) and (string[0] <= "ん"))
|
||||
if len(string) > 1:
|
||||
return functools.reduce((lambda x, y: (is_hira(x) and is_hira(y))) , string)
|
||||
|
||||
def contain_kanji(str):
|
||||
if len(str) == 0:
|
||||
return False
|
||||
elif len(str) == 1:
|
||||
return re.match(r"[一-龯]", str)
|
||||
if len(str) > 1:
|
||||
return functools.reduce(lambda x, y: contain_kanji(x) or contain_kanji(y) , str)
|
||||
|
||||
r = requests.post(base_url + "v1/parse",headers=headers,data=json.dumps(data))
|
||||
return r.json()
|
||||
|
||||
# ひらがなを削除する関数
|
||||
# Function to delete hiragana.
|
||||
|
|
@ -55,81 +51,54 @@ def hira_to_blank(str):
|
|||
return "".join(["" if ("ぁ" <= ch <= "ん") else ch for ch in str])
|
||||
|
||||
if __name__ == "__main__":
|
||||
|
||||
envjson = open('env.json', 'r')
|
||||
json_load = json.load(envjson)
|
||||
CLIENT_ID = json_load["client_id"]
|
||||
CLIENT_SECRET = json_load["client_secret"]
|
||||
|
||||
document = "私は明日、伊豆大島に行きたい"
|
||||
args = sys.argv
|
||||
if len(args) >= 2:
|
||||
document = str(args[1])
|
||||
|
||||
parse_document = parse(document)
|
||||
#print(parse_document)
|
||||
access_token = auth(CLIENT_ID, CLIENT_SECRET)
|
||||
parse_document = parse(document, access_token)
|
||||
print(parse_document)
|
||||
result_list = list()
|
||||
|
||||
for i, token in enumerate(parse_document):
|
||||
|
||||
for chunks in parse_document['result']:
|
||||
for token in chunks["tokens"]:
|
||||
# 形態素解析結果に置き換えルールを適用する
|
||||
if (token["pos"] != "助詞-格助詞"
|
||||
and token["pos"] != "助詞-接続助詞"
|
||||
and token["pos"] != "助詞-終助詞"
|
||||
and token["pos"] != "助詞-接続助詞" ):
|
||||
if '終止形-一般' in token["features"]:
|
||||
if ("為る" in token["lemma"]) or ("ます" in token["lemma"]):
|
||||
prime = "" # don't translate it.
|
||||
if (token["pos"] != "連用助詞"
|
||||
and token["pos"] != "引用助詞"
|
||||
and token["pos"] != "終助詞"
|
||||
and token["pos"] != "接続接尾辞"
|
||||
and token["pos"] != "動詞活用語尾"):
|
||||
if token["pos"] == "動詞接尾辞" and '終止' in token["features"]:
|
||||
if ("する" in token["lemma"]) or ("ます" in token["lemma"]):
|
||||
prime = "也"
|
||||
elif "たい" in token["lemma"]:
|
||||
prime = "欲"
|
||||
elif token["lemma"] in ["ない", "無い"]:
|
||||
prime = "無"
|
||||
elif token['lemma'] == 'た':
|
||||
prime = "希望"
|
||||
elif token['lemma'] != 'ない':
|
||||
prime = "了"
|
||||
else:
|
||||
print(is_hira(token['lemma']))
|
||||
if is_hira(token['lemma']):
|
||||
|
||||
prime = "実行"
|
||||
else:
|
||||
prime = token["form"]
|
||||
else:
|
||||
prime = token["lemma"]
|
||||
else:
|
||||
|
||||
if is_hira(token["lemma"]) and contain_kanji(token["form"]):
|
||||
prime=token["form"]
|
||||
else:
|
||||
prime = token["lemma"]
|
||||
|
||||
|
||||
if (token['lemma'] == '君' or token['lemma'] == '貴方' or token['lemma'] == 'お前'):
|
||||
prime = '你'
|
||||
|
||||
if token['lemma'] == '為る' and parse_document[i-1]['pos'] == '名詞-普通名詞-サ変可能':
|
||||
prime = ''
|
||||
|
||||
|
||||
compound_matched = re.match("([^-]+)-([^-]+)", token['lemma'])
|
||||
if compound_matched:
|
||||
prime = compound_matched.group(1)
|
||||
|
||||
if token['lemma'] == '私-代名詞':
|
||||
if token['lemma'] == '私':
|
||||
prime = '我'
|
||||
|
||||
if (token['lemma'] == '君' or token['lemma'] == 'あなた' or token['lemma'] == 'お前'):
|
||||
prime = '你'
|
||||
|
||||
if len(token["features"]) != 0:
|
||||
if "連体形-一般" in token['features']:
|
||||
if token['lemma'] == 'ない':
|
||||
prime = "無之"
|
||||
else:
|
||||
prime = prime + "之"
|
||||
|
||||
|
||||
|
||||
if "SURU" in token["features"][0] :
|
||||
prime = "実行"
|
||||
elif "連体" in token['features'][0]:
|
||||
prime = "的"
|
||||
elif "疑問符" in token["features"][0]:
|
||||
prime = "如何?"
|
||||
|
||||
result_list.append(hira_to_blank(prime))
|
||||
|
||||
|
||||
|
||||
if token['lemma'] == 'の' and token['pos'] == "助詞-格助詞":
|
||||
prime = "之"
|
||||
result_list.append(hira_to_blank(prime))
|
||||
if token["form"] == "か" and token['pos'] == '助詞-終助詞':
|
||||
prime = "乎"
|
||||
result_list.append(hira_to_blank(prime))
|
||||
|
||||
print(''.join(result_list))
|
||||
print(''.join(result_list))
|
||||
|
|
|
|||
Loading…
Reference in a new issue