Spaces:
Runtime error
Runtime error
Commit
·
6f5c495
1
Parent(s):
c3991a5
Create app.py
Browse files
app.py
ADDED
@@ -0,0 +1,132 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
|
2 |
+
|
3 |
+
|
4 |
+
#python.exe -m pip install --upgrade pip
|
5 |
+
#pip install feedparser
|
6 |
+
#pip install newspaper3k
|
7 |
+
#pip install streamlit
|
8 |
+
|
9 |
+
# streamlit run <app.py> for one app
|
10 |
+
# python run_all.py <folder_path> for multiple apps
|
11 |
+
|
12 |
+
|
13 |
+
import streamlit as st
|
14 |
+
import feedparser
|
15 |
+
from urllib.parse import quote
|
16 |
+
import streamlit.state as state
|
17 |
+
|
18 |
+
|
19 |
+
st.set_page_config(layout="wide")
|
20 |
+
#st.markdown("<h2 style='text-align: center; font-family: Verdana; color: red;'>Tin tức tổng hợp 24h qua</h2><br>", unsafe_allow_html=True)
|
21 |
+
#st.markdown("<h4 style='text-align: center;'>Tìm kiếm theo từ khóa hoặc tên báo (ví dụ: Hà Nội, Vaccine, Mạng xã hội, Trí tuệ nhân tạo, cafebiz.vn, Mỹ...)</h4><br>", unsafe_allow_html=True)
|
22 |
+
#-----------------------------------------------------------------------------------
|
23 |
+
# Báo tiếng Việt
|
24 |
+
#-----------------------------------------------------------------------------------
|
25 |
+
st.sidebar.title("Báo tiếng Việt")
|
26 |
+
# Custom keywords
|
27 |
+
keyword = st.sidebar.text_input("Nhập/Xóa từ khóa", "Hà Nội")
|
28 |
+
|
29 |
+
# Select websites
|
30 |
+
list = ['cafebiz.vn', 'cafef.vn', 'thanhnien.vn','vnexpress.net','soha.vn','zingnews.vn','tuoitre.vn','laodong.vn', 'All']
|
31 |
+
domain = st.sidebar.multiselect("Chọn websites:",list,default=["All"])
|
32 |
+
|
33 |
+
sites = ""
|
34 |
+
if "All" in domain:
|
35 |
+
domain = ""
|
36 |
+
if domain:
|
37 |
+
for i in domain[:-1]:
|
38 |
+
sites = sites + "site:{} OR ".format(i)
|
39 |
+
sites = sites + "site:" + domain[-1]
|
40 |
+
|
41 |
+
query = keyword + " " + sites + " "
|
42 |
+
|
43 |
+
if keyword == "" and sites == "":
|
44 |
+
query = "Hà Nội"
|
45 |
+
|
46 |
+
#st.sidebar.write(query)
|
47 |
+
|
48 |
+
|
49 |
+
|
50 |
+
#-----------------------------------------------------------------------------------
|
51 |
+
if st.sidebar.button('Tìm kiếm 🔎'):
|
52 |
+
|
53 |
+
url = 'https://news.google.com/rss/search?q=' + quote(query) + 'when:1d&hl=vi'
|
54 |
+
#st.sidebar.write(url)
|
55 |
+
|
56 |
+
feed = feedparser.parse(url)
|
57 |
+
|
58 |
+
i = 1
|
59 |
+
html = ""
|
60 |
+
for post in feed.entries:
|
61 |
+
|
62 |
+
html = html + f"{post.title[:120]} <a href='{post.link}' target='_blank' > Link </a><br>"
|
63 |
+
|
64 |
+
|
65 |
+
if (i % 5 == 0):
|
66 |
+
html = html + "<br>"
|
67 |
+
i = i+1
|
68 |
+
|
69 |
+
html1 = f"""
|
70 |
+
<p style='font-size: 17px; font-family: Segoe UI;color:#202124'>
|
71 |
+
{html}
|
72 |
+
</p>
|
73 |
+
"""
|
74 |
+
st.markdown("<h2 style='text-align: center; font-family: Verdana; color: red;'>Tin tức 24h</h2><br>", unsafe_allow_html=True)
|
75 |
+
st.markdown(html1, unsafe_allow_html=True)
|
76 |
+
|
77 |
+
|
78 |
+
|
79 |
+
#-----------------------------------------------------------------------------------
|
80 |
+
# Báo tiếng Anh
|
81 |
+
#-----------------------------------------------------------------------------------
|
82 |
+
st.sidebar.write("-------------------------------")
|
83 |
+
st.sidebar.title("English News")
|
84 |
+
# Custom keywords
|
85 |
+
keyword_en = st.sidebar.text_input("Input/Delete keyword", "AI")
|
86 |
+
|
87 |
+
# Select websites
|
88 |
+
list_en = ['bbc.com','martechseries.com','bloomberg.com', 'cnn.com','cnbc.com','reuters.com','foxnews.com','.entrepreneur.com','techradar.com', 'All']
|
89 |
+
domain_en = st.sidebar.multiselect("Select websites:",list_en,default=["All"])
|
90 |
+
|
91 |
+
sites_en = ""
|
92 |
+
if "All" in domain_en:
|
93 |
+
domain_en = ""
|
94 |
+
if domain_en:
|
95 |
+
for i in domain[:-1]:
|
96 |
+
sites_en = sites_en + "site:{} OR ".format(i)
|
97 |
+
sites_en = sites_en + "site:" + domain_en[-1]
|
98 |
+
|
99 |
+
query_en = keyword_en + " " + sites_en + " "
|
100 |
+
|
101 |
+
if keyword_en == "" and sites_en == "":
|
102 |
+
query = "Vaccine"
|
103 |
+
|
104 |
+
#-----------------------------------------------------------------------------------
|
105 |
+
if st.sidebar.button('Search 🔎'):
|
106 |
+
|
107 |
+
url_en = 'https://news.google.com/rss/search?q=' + quote(query_en) + 'when:1d&hl=en&gl=US'
|
108 |
+
#st.sidebar.write(url)
|
109 |
+
|
110 |
+
feed_en = feedparser.parse(url_en)
|
111 |
+
|
112 |
+
i = 1
|
113 |
+
html = ""
|
114 |
+
for post_en in feed_en.entries:
|
115 |
+
|
116 |
+
html = html + f"{post_en.title[:120]} <a href='{post_en.link}' target='_blank' > Link </a><br>"
|
117 |
+
|
118 |
+
|
119 |
+
if (i % 5 == 0):
|
120 |
+
html = html + "<br>"
|
121 |
+
i = i+1
|
122 |
+
|
123 |
+
html1 = f"""
|
124 |
+
<p style='font-size: 17px; font-family: Segoe UI;color:#202124'>
|
125 |
+
{html}
|
126 |
+
</p>
|
127 |
+
"""
|
128 |
+
st.markdown("<h2 style='text-align: center; font-family: Verdana; color: red;'>News 24h</h2><br>", unsafe_allow_html=True)
|
129 |
+
st.markdown(html1, unsafe_allow_html=True)
|
130 |
+
|
131 |
+
|
132 |
+
#st.sidebar.title(url_en)
|