Spaces:
Sleeping
Sleeping
Upload Summary.py
Browse files- pages/Summary.py +131 -0
pages/Summary.py
ADDED
|
@@ -0,0 +1,131 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
import os
|
| 2 |
+
import time
|
| 3 |
+
import re
|
| 4 |
+
|
| 5 |
+
import streamlit as st
|
| 6 |
+
import requests
|
| 7 |
+
from wordcloud import WordCloud
|
| 8 |
+
import matplotlib.pyplot as plt
|
| 9 |
+
|
| 10 |
+
|
| 11 |
+
# Установка API URL и заголовков
|
| 12 |
+
API_URL_TRA = "https://api-inference.huggingface.co" \
|
| 13 |
+
"/models/Helsinki-NLP/opus-mt-en-ru"
|
| 14 |
+
API_URL_KEY = "https://api-inference.huggingface.co" \
|
| 15 |
+
"/models/ml6team/keyphrase-extraction-kbir-inspec"
|
| 16 |
+
API_URL_SUM = "https://api-inference.huggingface.co" \
|
| 17 |
+
"/models/facebook/bart-large-cnn"
|
| 18 |
+
|
| 19 |
+
TOKEN = os.getenv('API_TOKEN')
|
| 20 |
+
HEADERS = {"Authorization": TOKEN}
|
| 21 |
+
|
| 22 |
+
|
| 23 |
+
def hugging_api_request(url, payload):
|
| 24 |
+
response = requests.post(url, headers=HEADERS, json=payload, timeout=120)
|
| 25 |
+
body = response.json()
|
| 26 |
+
if 'error' in body:
|
| 27 |
+
print(response.status_code, body)
|
| 28 |
+
if 'estimated_time' in body:
|
| 29 |
+
time.sleep(body['estimated_time'])
|
| 30 |
+
else:
|
| 31 |
+
return
|
| 32 |
+
hugging_api_request(url, payload)
|
| 33 |
+
return body
|
| 34 |
+
|
| 35 |
+
|
| 36 |
+
# Функция для получения ключевых слов
|
| 37 |
+
def get_key_words(payload):
|
| 38 |
+
return hugging_api_request(API_URL_KEY, payload)
|
| 39 |
+
|
| 40 |
+
|
| 41 |
+
# Функция для перевода слова
|
| 42 |
+
def translate_key_words(payload):
|
| 43 |
+
return hugging_api_request(API_URL_TRA, payload)
|
| 44 |
+
|
| 45 |
+
|
| 46 |
+
# Функция для составления конспекта
|
| 47 |
+
def make_summary(payload):
|
| 48 |
+
return hugging_api_request(API_URL_SUM, payload)
|
| 49 |
+
|
| 50 |
+
|
| 51 |
+
# Очищаем список слов
|
| 52 |
+
def clean_list(words_list):
|
| 53 |
+
cleaned_words_list = []
|
| 54 |
+
for word in words_list:
|
| 55 |
+
word = word.lower()
|
| 56 |
+
word = re.sub(r"[^а-яА-Яa-zA-Z\s]", "", word)
|
| 57 |
+
word = word.lstrip()
|
| 58 |
+
word = word.rstrip()
|
| 59 |
+
cleaned_words_list.append(word)
|
| 60 |
+
return cleaned_words_list
|
| 61 |
+
|
| 62 |
+
|
| 63 |
+
# Настраиваем заголовок и название страницы
|
| 64 |
+
st.set_page_config(layout="wide", page_title="Students' Personal Assistant")
|
| 65 |
+
st.markdown(' # :female-student: Персональный помощник для студентов')
|
| 66 |
+
|
| 67 |
+
st.divider()
|
| 68 |
+
st.markdown('# :blue_book: Конспект на английском языке')
|
| 69 |
+
|
| 70 |
+
col1, col2 = st.columns(2)
|
| 71 |
+
text_from_tarea = col1.text_area('Введите тект статьи на английском языке',
|
| 72 |
+
key='t_area', height=500)
|
| 73 |
+
|
| 74 |
+
button_start = st.button('Обработать текст')
|
| 75 |
+
key_words_list = []
|
| 76 |
+
|
| 77 |
+
if button_start:
|
| 78 |
+
with st.spinner('Составляем конспект...'):
|
| 79 |
+
# Составляем конспект
|
| 80 |
+
summary_text = make_summary({"inputs": text_from_tarea})
|
| 81 |
+
col2.text_area('Конспект статьи', height=500,
|
| 82 |
+
key='sum_area', value=summary_text[0]['summary_text'])
|
| 83 |
+
|
| 84 |
+
with st.spinner('Получаем ключевые слова...'):
|
| 85 |
+
# Извлекаем ключевые слова
|
| 86 |
+
kew_words = get_key_words({"inputs": text_from_tarea})
|
| 87 |
+
for key_word in kew_words:
|
| 88 |
+
key_words_list.append(key_word['word'].lower())
|
| 89 |
+
|
| 90 |
+
sorted_keywords = set(sorted(key_words_list))
|
| 91 |
+
sorted_keywords = clean_list(sorted_keywords)
|
| 92 |
+
|
| 93 |
+
with st.spinner('Переводим ключевые слова...'):
|
| 94 |
+
# Переводим ключевые слова
|
| 95 |
+
translated_words_dict = translate_key_words(
|
| 96 |
+
{"inputs": sorted_keywords})
|
| 97 |
+
translated_words_list = [
|
| 98 |
+
word['translation_text'] for word in translated_words_dict]
|
| 99 |
+
|
| 100 |
+
# Создаем карточки
|
| 101 |
+
cleaned_words_list_ru = clean_list(translated_words_list)
|
| 102 |
+
cards_list = []
|
| 103 |
+
for item1, item2 in zip(sorted_keywords, cleaned_words_list_ru):
|
| 104 |
+
cards_list.append([item1, item2])
|
| 105 |
+
|
| 106 |
+
st.success('Готово')
|
| 107 |
+
|
| 108 |
+
with st.spinner('Создаем WordCloud...'):
|
| 109 |
+
# Выводим Word Cloud
|
| 110 |
+
st.set_option('deprecation.showPyplotGlobalUse', False)
|
| 111 |
+
words_str = ', '.join(sorted_keywords)
|
| 112 |
+
w = WordCloud(background_color="white",
|
| 113 |
+
width=1600, height=800).generate(words_str)
|
| 114 |
+
plt.imshow(w, interpolation='bilinear')
|
| 115 |
+
plt.imshow(w)
|
| 116 |
+
plt.axis("off")
|
| 117 |
+
st.pyplot()
|
| 118 |
+
|
| 119 |
+
# Выводим карточки
|
| 120 |
+
st.markdown('# :bookmark_tabs: Карточки из ключевых слов')
|
| 121 |
+
col1, col2, col3 = st.columns(3)
|
| 122 |
+
columns = [col1, col2, col3]
|
| 123 |
+
for index, el in enumerate(cards_list):
|
| 124 |
+
with columns[(index + 1) % 3]:
|
| 125 |
+
with st.container(border=True):
|
| 126 |
+
col4, col5 = st.columns([0.1, 0.9])
|
| 127 |
+
with col4:
|
| 128 |
+
st.write("# :flower_playing_cards:")
|
| 129 |
+
with col5:
|
| 130 |
+
st.write(f'## :green[{el[0]}]')
|
| 131 |
+
st.write(f'### :blue[{el[1]}]')
|