async def translate(self, ctx):
msg = ctx.message
words = msg.content.split(" ")
to_language = words[-1].lower()
text = str()
arr = msg.content.split('"')
if len(arr) == 3:
text = arr[1]
else:
text = " ".join(words[2:-1])
try:
# LANGCODES has keys as languages, values as codes
# LANGUAGES has keys as codes, values as languages
code = None
if to_language in LANGCODES.keys(): # if language name was passed
# print("lang name was passed!!")
code = LANGCODES[to_language]
elif to_language in LANGCODES.values(): # if code was passed
# print("lang code was passed!!")
code = to_language
else:
await ctx.send(
"Please enter a valid language!" +
"\nUsage: ```pepe translate <text> <language-name>```" +
'\nFor a list of supported languages, use "pepe languages"'
)
trans = Translator()
translated_text = trans.translate(text, dest=code)
await ctx.send(translated_text.text) # , tts=True)
except:
traceback.print_exc()
def toTranslateBox(self):
if self.ui.toTranslateBox.currentIndex() == 0:
self.destLanguage = 'en' #default destination language is English
self.ui.toTranslateBox.setCurrentText("English")
return
else:
lang = self.ui.toTranslateBox.currentText().lower()
# print('to', lang)
self.destLanguage = LANGCODES.get(lang, None)
def _changSettings():
lang = event.message.text[4:]
dest = LANGCODES.get(lang)
user_lang_modified = _changeUserTranslationDest(user_profile,
dest=dest)
print(
"Translated destination language is successfully changed to: %s" %
user_lang_modified.get("dest"))
return user_lang_modified
def list_questions(language):
if language in LANGCODES.values():
translator = Translator()
return_data = [
Question(id=int(question[1].get('id')),
created=question[1].get('created'),
topic=int(question[1].get('topic')),
question=translator.translate(question[1].get('question'),
dest=language).text,
answer=translator.translate(question[1].get('answer'),
dest=language).text)
for question in data.items()
]
return JSONResponse(return_data, status_code=200)
return JSONResponse({}, status_code=400)
def __init__(self):
self.translation_client: Translator = Translator()
self.translation_cache: Dict[str, Dict[str, str]] = dict()
self.hyphenated_linebreak_regex = re.compile(r"\s*-\n\s*")
self.linebreak_in_the_midst_of_text_regex = re.compile(r"\n[ ]*(\S+)")
self.valid_languages = list(LANGCODES.keys()) + list(LANGUAGES.keys())
'Floods': 'other',
'Shootings': 'crimeviolence',
'Derailment': 'other',
'Wildfire': 'other',
'Haze': 'other'
}
docroot = '/afs/crc.nd.edu/user/n/nsmith9/NLP_final_project/googleTrain'
trans = Translator()
scorrect = True
with open(sys.argv[1]) as json_data:
d = json.load(json_data)
lkeys = list(d.keys())
tmp = trans.detect(lkeys[0])
print(tmp.lang)
if tmp.lang not in LANGCODES.values():
scorrect = False
def words(text):
return re.findall(r'\w+', text.lower())
WORDS = Counter(words(docroot + '/' + tmp.lang + '.txt'))
def P(word, N=sum(WORDS.values())):
"Probability of `word`."
return WORDS[word] / N
from googletrans import Translator, LANGCODES
import random
lc = list(LANGCODES.values())
translator = Translator()
def add_hooks(client, table):
table["mangle"] = mangle
table["translate"] = translate
async def mangle(args, client, message):
try:
times = int(args[1])
text = " ".join(args[2:-1]) + "\n" + args[-1]
except:
times = 10
text = " ".join(args[1:-1] + "\n" + args[-1])
tmp_msg = "Translating..."
tmp = await client.send_message(message.channel, tmp_msg)
for i in range(times):
text = translator.translate(text, dest=random.choice(lc)).text
tmp_msg += str(i + 1) + "..."
await client.edit_message(tmp, tmp_msg)
text = translator.translate(text, dest="en").text
await client.edit_message(tmp, text)
async def translate(args, client, message):
def get_fdiedb():
translator = Translator()
language = choice(list(LANGCODES.keys()))
print(language)
result = translator.translate("fick die deutsche Bahn", dest=language)
return language + ": " + result.text
def fromTranslateBox(self):
if self.ui.fromTranslateBox.currentIndex() == 0:
self.srcLanguage = 'auto' #let google(translation service) determine the source language
else:
lang = self.ui.fromTranslateBox.currentText().lower()
self.srcLanguage = LANGCODES.get(lang, None)
async def translate(ctx):
message = ctx.message
# do not want the bot to reply to itself so
if message.author == client.user:
return
request = '{}'.format(message.content)
try:
request = request.split("!translate ")[1]
except IndexError:
print(
"E - No message provided to translate, please refer to the help command (-h) for more info: {}"
.format(request))
return await ctx.send(
"Error - No message provided to translate, please refer to the help command (-h) for more info: {}"
.format(request))
print("Request: {}".format(request))
print("=" * 50)
print("S - New request received")
args = request.split(" ")
l_ind = find_in_list(args, "-l")
n_ind = find_in_list(args, "-n")
t_ind = find_in_list(args, "-t")
h_ind = find_in_list(args, "-h")
if h_ind != -1:
print("S - Help information requested")
return await ctx.send(HELP_TEXT)
await ctx.send("Translation request received (⌐■_■)")
if max(t_ind, l_ind, n_ind) == -1:
text = request
n = 10
langs = ['de', 'ko', 'la', 'ja', 'eo'] # default
else:
if t_ind == -1 or len(args[t_ind + 1:]) == 0:
print("E - No text provided (-t): {}".format(request))
return await ctx.send(
"Error - No text provided (-t): {}".format(request))
text = " ".join(args[t_ind + 1:])
if max(l_ind, n_ind, t_ind) != t_ind:
print(
"E - Optional arguments (-l,-n) should be positioned before the text argument (-t), please refer to the help command (-h) for more info: {}"
.format(request))
return await ctx.send(
"Error - Optional arguments (-l,-n) should be positioned before the text argument (-t), please refer to the help command (-h) for more info: {}"
.format(request))
if n_ind == -1:
n = 10
else:
try:
n = int(args[n_ind + 1])
except ValueError:
print(
"E - Invalid no. of translations (-n), please refer to the help command (-h) for more info: {}"
.format(request))
return await ctx.send(
"Error - Invalid no. of translations (-n), please refer to the help command (-h) for more info: {}"
.format(request))
if l_ind == -1:
langs = ['de', 'ko', 'la', 'ja', 'eo'] # default
else:
ind = l_ind
while not ind in (n_ind, t_ind, h_ind):
ind += 1
langs = args[l_ind + 1:ind]
if len(langs) == 0:
print(
"E - No language codes provided (-l), please refer to the help command (-h) for valid codes : {}"
.format(request))
return await ctx.send(
"Error - No language codes provided (-l), please refer to the help command (-h) for valid codes : {}"
.format(request))
invalid = [
lang for lang in langs if not lang in list(LANGCODES.values())
]
if len(invalid) != 0:
print(
"E - Invalid language code(s) provided (-l): {}. Please refer to the help command (-h) for valid codes: {}"
.format(" ".join(invalid), request))
return await ctx.send(
"Error - Invalid language code(s) provided (-l): {}. Please refer to the help command (-h) for valid codes: {}"
.format(" ".join(invalid), request))
print("S - Languages: {}".format(langs))
print("S - No. of iterations: {}".format(n))
print("S - Text to be translated: {}".format(text))
await background_translate(ctx, text, langs, n)
def test_language_translation(language, string):
if language in LANGCODES.values():
translator = Translator()
return translator.translate(string, dest=language).text
return None
def main():
st.sidebar.write('StanzaGraphs ')
mes = '''
a Multilingual STANZA-based Summary and Keyword Extractor and Question-Answering System using TextGraphs and Neural Networks
'''
st.sidebar.write(mes)
langList = ['', 'Auto detect']
langList += list(LANGCODES.keys())
langFull = st.sidebar.selectbox('Translate to?', langList)
if langFull != "Auto detect":
lang = LANGCODES.get(langFull)
else:
lang = langFull
st.write('Selected language: ', langFull)
uploaded_file = st.sidebar.file_uploader('Select a File',
type=['txt', 'pdf'])
title = None
action = None
if uploaded_file is not None:
fpath = save_uploaded_file(uploaded_file)
if fpath[-4:] == '.pdf':
pdf2txt(fpath[:-4])
text = file2text(fpath[:-4] + '.txt')
data_lang = langid.classify(text)[0]
if data_lang == 'zh':
data_lang = 'zh-cn'
elif data_lang == 'jv':
data_lang = 'jw'
data_fulllang = LANGUAGES.get(data_lang)
st.sidebar.write('Language:', data_fulllang)
fname = fpath[:-4]
print("fname: ")
print(fname)
if not title or title != uploaded_file.name:
title = uploaded_file.name
action = st.sidebar.selectbox("Choose an action",
["Summarize", "Ask a question"])
if action == "Summarize":
proceed = st.sidebar.button('Run with selected option!')
if proceed:
if langFull == "Auto detect":
st.write('Translate summary from ', fname, 'to ',
data_fulllang)
else:
st.write('Translate summary from ', fname, 'to ',
langFull)
summary(fname=fname, lang=lang)
pass
elif action == "Ask a question":
st.write('Query Answering')
question = st.text_input('Input your question here:')
if question:
st.write('Answers:')
answer(fname, question, lang)
else:
st.info("Please select a text file to upload")
channel_access_token = conf_json.get("channel_access_token") # 回傳訊息給 Line 時使用
secret_key = conf_json.get("secret_key") # WebhookHandler驗證事件消息是否來自Line使用
linebot_developer_id = conf_json.get("self_user_id")
server_url = "https://" + conf_json.get("server_url")
rich_menu_id = conf_json.get("rich_menu_id")
# load mongodb settings
with open("../mongo_client_settings", "r", encoding="utf-8") as infile:
mongo_client_settings = json.loads(infile.read())
line_bot_api = LineBotApi(channel_access_token) # 回傳訊息給 Line 時使用
handler = WebhookHandler(secret_key) # WebhookHandler驗證事件消息是否來自Line使用
# Make the arranged lang_list
LANGCODES["chinese_tw"] = LANGCODES.pop("chinese (traditional)")
LANGCODES["chinese_cn"] = LANGCODES.pop("chinese (simplified)")
LANGCODES["kurdish"] = LANGCODES.pop("kurdish (kurmanji)")
LANGCODES["myanmar"] = LANGCODES.pop("myanmar (burmese)")
LANGCODES.pop("Filipino")
LANGCODES.pop("Hebrew")
lang_list = sorted([key for key in LANGCODES])
# New a tranlator to deal with translation requests
translator = Translator(service_urls=[
'translate.google.com', 'translate.google.com.tw',
'translate.google.co.kr', 'translate.google.co.jp'
])
app = Flask(__name__, static_url_path="/res", static_folder="../res")
async def languages(self, ctx):
text = [(lang, code) for lang, code in LANGCODES.items()]
await ctx.send("\n".join([f"{code}: {lang}" for code, lang in text]))
def specify(s):
return LANGCODES.get(redirs.get(s, s), redirs.get(s, s))
def languageList():
return LANGCODES.values()
def __init__(self, setting):
langList = list(LANGCODES.keys())
wx.Frame.__init__(self,
None,
-1,
'CopyTranslator ' + version,
size=(465, 345))
self.SetIcon(wx.Icon(logopath, wx.BITMAP_TYPE_ICO))
self.SetWindowStyle(MainFrame.mainStyle)
self.setting = setting
TextPanel = wx.Panel(self, -1)
buttonPanel = wx.Panel(self, -1)
# 始终置顶按钮
self.topCheck = wx.CheckBox(buttonPanel, -1, 'Stay on top')
self.Bind(wx.EVT_CHECKBOX, self.setting.ReverseStayTop, self.topCheck)
# 自动检测语言按钮
self.detectCheck = wx.CheckBox(buttonPanel, -1, 'Auto detect language')
self.Bind(wx.EVT_CHECKBOX, self.setting.ReverseDete, self.detectCheck)
# 监听剪贴板选框
self.listenCheck = wx.CheckBox(buttonPanel, -1, 'Listen on Clipboard')
self.Bind(wx.EVT_CHECKBOX, self.setting.ReverseListen,
self.listenCheck)
# 自动复制选框
self.copyCheck = wx.CheckBox(buttonPanel, -1, 'Auto copy')
self.Bind(wx.EVT_CHECKBOX, self.setting.ReverseCopy, self.copyCheck)
# 切换模式
self.switchBtn = wx.Button(buttonPanel, -1, "Switch Mode")
self.Bind(wx.EVT_BUTTON, self.setting.SwitchMode, self.switchBtn)
self.transBtn = wx.Button(buttonPanel, -1, "Translate")
self.Bind(wx.EVT_BUTTON, self.setting.translateCtrl, self.transBtn)
self.transBtn.SetDefault()
# 原文本
self.srcLabel = wx.StaticText(TextPanel, -1, "Source:")
self.srcText = wx.TextCtrl(TextPanel,
-1,
"",
size=(300, 125),
style=wx.TE_MULTILINE) # 创建一个文本控件
self.copyBtn = wx.Button(buttonPanel, -1, "Copy result")
self.Bind(wx.EVT_BUTTON, self.setting.Copy, self.copyBtn)
# 目标文本
self.dstLabel = wx.StaticText(TextPanel, -1, "Result:")
self.destText = wx.TextCtrl(TextPanel,
-1,
"",
size=(300, 125),
style=wx.TE_MULTILINE) # 创建一个文本控件
self.fromlabel = wx.StaticText(buttonPanel, -1, 'Source language')
self.fromchoice = wx.Choice(buttonPanel, -1, choices=langList)
self.fromchoice.SetSelection(self.fromchoice.FindString('english'))
tolabel = wx.StaticText(buttonPanel, -1, 'Target language :')
self.tochoice = wx.Choice(buttonPanel, -1, choices=langList)
self.tochoice.SetSelection(
self.tochoice.FindString('chinese (simplified)'))
panel1sizer = wx.FlexGridSizer(4, 1, 6, 6)
panel1sizer.AddMany(
[self.srcLabel, self.srcText, self.dstLabel, self.destText])
TextPanel.SetSizer(panel1sizer)
panel2sizer = wx.FlexGridSizer(11, 1, 6, 0)
panel2sizer.AddMany([
self.topCheck, self.listenCheck, self.detectCheck, self.copyCheck,
self.fromlabel, self.fromchoice, tolabel, self.tochoice,
self.switchBtn, self.transBtn, self.copyBtn
])
buttonPanel.SetSizer(panel2sizer)
sizer = wx.FlexGridSizer(1, 2, 0, 0)
sizer.AddMany([TextPanel, buttonPanel])
self.SetSizer(sizer)
self.Fit()
# 创建定时器
self.timer = wx.Timer(self) # 创建定时器
self.Bind(wx.EVT_TIMER, self.setting.OnTimer, self.timer) # 绑定一个定时器事件
# 绑定事件
self.Bind(wx.EVT_CLOSE, self.OnClose)
self.Bind(
wx.EVT_ICONIZE, self.OnIconfiy
) # 窗口最小化时，调用OnIconfiy,注意Wx窗体上的最小化按钮，触发的事件是 wx.EVT_ICONIZE,而根本就没有定义什么wx.EVT_MINIMIZE,但是最大化，有个wx.EVT_MAXIMIZE。
from googletrans import Translator as BaseTranslator
from googletrans.client import Translated
from utils.formatters import BetterEmbed
if TYPE_CHECKING:
from asyncio import AbstractEventLoop
from aiohttp import ClientSession
# outside of the class to use it as a converter
def to_language(arg: str) -> Union[str, None]:
"""Converts a string into a valid google trans language"""
if (low := arg.lower()) in LANGUAGES:
return low
elif res := LANGCODES.get(low):
return res
raise BadArgument(message=f"Couldn't find the language {arg}")
def check_length(arg: str) -> str:
"""Checks the initial text and returns it"""
if len(arg) > 200:
raise BadArgument(message="Cannot translate texts longer than 200 characters")
return arg
class AioTranslator(BaseTranslator):
"""Translates stuff using google translate's api"""
def __init__(self, *args, **kwargs):
