Spaces:
Building
Building
Delete app-backup.py
Browse files- app-backup.py +0 -202
app-backup.py
DELETED
@@ -1,202 +0,0 @@
|
|
1 |
-
import gradio as gr
|
2 |
-
import requests
|
3 |
-
import json
|
4 |
-
import os
|
5 |
-
from datetime import datetime, timedelta
|
6 |
-
from bs4 import BeautifulSoup # ์น ํ์ด์ง์์ ํ
์คํธ๋ฅผ ์ถ์ถํ๊ธฐ ์ํด ์ฌ์ฉ
|
7 |
-
from huggingface_hub import InferenceClient # LLM ์ฌ์ฉ์ ์ํด ํ์
|
8 |
-
|
9 |
-
# ํ์ํ ํจํค์ง ์ค์น (ํ์ํ ๊ฒฝ์ฐ ์ฃผ์์ ์ ๊ฑฐํ๊ณ ์คํ)
|
10 |
-
# !pip install bs4 huggingface_hub
|
11 |
-
|
12 |
-
# ํ๊ฒฝ ๋ณ์์์ API ํค ๊ฐ์ ธ์ค๊ธฐ (API ํค๋ ์์ ํ๊ฒ ๊ด๋ฆฌ๋์ด์ผ ํฉ๋๋ค)
|
13 |
-
API_KEY = os.getenv("SERPHOUSE_API_KEY") # ๋ณธ์ธ์ SerpHouse API ํค๋ฅผ ํ๊ฒฝ ๋ณ์๋ก ์ค์ ํ์ธ์.
|
14 |
-
HF_TOKEN = os.getenv("HF_TOKEN") # Hugging Face API ํ ํฐ์ ํ๊ฒฝ ๋ณ์๋ก ์ค์ ํ์ธ์.
|
15 |
-
|
16 |
-
MAJOR_COUNTRIES = [
|
17 |
-
"United States", "United Kingdom", "Canada", "Australia", "Germany",
|
18 |
-
"France", "Japan", "South Korea", "China", "India",
|
19 |
-
"Brazil", "Mexico", "Russia", "Italy", "Spain",
|
20 |
-
"Netherlands", "Sweden", "Switzerland", "Norway", "Denmark",
|
21 |
-
"Finland", "Belgium", "Austria", "New Zealand", "Ireland",
|
22 |
-
"Singapore", "Hong Kong", "Israel", "United Arab Emirates", "Saudi Arabia",
|
23 |
-
"South Africa", "Turkey", "Egypt", "Poland", "Czech Republic",
|
24 |
-
"Hungary", "Greece", "Portugal", "Argentina", "Chile",
|
25 |
-
"Colombia", "Peru", "Venezuela", "Thailand", "Malaysia",
|
26 |
-
"Indonesia", "Philippines", "Vietnam", "Pakistan", "Bangladesh"
|
27 |
-
]
|
28 |
-
|
29 |
-
def search_serphouse(query, country, page=1, num_result=10):
|
30 |
-
url = "https://api.serphouse.com/serp/live"
|
31 |
-
|
32 |
-
now = datetime.utcnow()
|
33 |
-
yesterday = now - timedelta(days=1)
|
34 |
-
date_range = f"{yesterday.strftime('%Y-%m-%d')},{now.strftime('%Y-%m-%d')}"
|
35 |
-
|
36 |
-
payload = {
|
37 |
-
"data": {
|
38 |
-
"q": query,
|
39 |
-
"domain": "google.com",
|
40 |
-
"loc": country,
|
41 |
-
"lang": "en",
|
42 |
-
"device": "desktop",
|
43 |
-
"serp_type": "news",
|
44 |
-
"page": str(page),
|
45 |
-
"verbatim": "1",
|
46 |
-
"num": str(num_result),
|
47 |
-
"date_range": date_range
|
48 |
-
}
|
49 |
-
}
|
50 |
-
|
51 |
-
headers = {
|
52 |
-
"accept": "application/json",
|
53 |
-
"content-type": "application/json",
|
54 |
-
"authorization": f"Bearer {API_KEY}"
|
55 |
-
}
|
56 |
-
|
57 |
-
try:
|
58 |
-
response = requests.post(url, json=payload, headers=headers)
|
59 |
-
response.raise_for_status()
|
60 |
-
return response.json()
|
61 |
-
except requests.RequestException as e:
|
62 |
-
error_msg = f"Error: {str(e)}"
|
63 |
-
if response.text:
|
64 |
-
error_msg += f"\nResponse content: {response.text}"
|
65 |
-
return {"error": error_msg}
|
66 |
-
|
67 |
-
def format_results_from_raw(results):
|
68 |
-
try:
|
69 |
-
if isinstance(results, dict) and "error" in results:
|
70 |
-
return "Error: " + results["error"], []
|
71 |
-
|
72 |
-
if not isinstance(results, dict):
|
73 |
-
raise ValueError("๊ฒฐ๊ณผ๊ฐ ์ฌ์ ํ์์ด ์๋๋๋ค.")
|
74 |
-
|
75 |
-
# 'results' ํค ๋ด๋ถ์ ๊ตฌ์กฐ ํ์ธ (์ค์ฒฉ๋ 'results' ์ฒ๋ฆฌ)
|
76 |
-
if 'results' in results:
|
77 |
-
results_content = results['results']
|
78 |
-
if 'results' in results_content:
|
79 |
-
results_content = results_content['results']
|
80 |
-
# 'news' ํค ํ์ธ
|
81 |
-
if 'news' in results_content:
|
82 |
-
news_results = results_content['news']
|
83 |
-
else:
|
84 |
-
news_results = []
|
85 |
-
else:
|
86 |
-
news_results = []
|
87 |
-
else:
|
88 |
-
news_results = []
|
89 |
-
|
90 |
-
if not news_results:
|
91 |
-
return "๊ฒ์ ๊ฒฐ๊ณผ๊ฐ ์์ต๋๋ค.", []
|
92 |
-
|
93 |
-
articles = []
|
94 |
-
|
95 |
-
for idx, result in enumerate(news_results, 1):
|
96 |
-
title = result.get("title", "์ ๋ชฉ ์์")
|
97 |
-
link = result.get("url", result.get("link", "#"))
|
98 |
-
snippet = result.get("snippet", "๋ด์ฉ ์์")
|
99 |
-
channel = result.get("channel", result.get("source", "์ ์ ์์"))
|
100 |
-
time = result.get("time", result.get("date", "์ ์ ์๋ ์๊ฐ"))
|
101 |
-
image_url = result.get("img", result.get("thumbnail", ""))
|
102 |
-
|
103 |
-
articles.append({
|
104 |
-
"title": title,
|
105 |
-
"link": link,
|
106 |
-
"snippet": snippet,
|
107 |
-
"channel": channel,
|
108 |
-
"time": time,
|
109 |
-
"image_url": image_url
|
110 |
-
})
|
111 |
-
|
112 |
-
return "", articles
|
113 |
-
|
114 |
-
except Exception as e:
|
115 |
-
error_message = f"๊ฒฐ๊ณผ ์ฒ๋ฆฌ ์ค ์ค๋ฅ ๋ฐ์: {str(e)}"
|
116 |
-
return "Error: " + error_message, []
|
117 |
-
|
118 |
-
def serphouse_search(query, country):
|
119 |
-
# ํ์ด์ง์ ๊ฒฐ๊ณผ ์์ ๊ธฐ๋ณธ๊ฐ์ ์ค์ ํฉ๋๋ค.
|
120 |
-
page = 1
|
121 |
-
num_result = 10
|
122 |
-
results = search_serphouse(query, country, page, num_result)
|
123 |
-
error_message, articles = format_results_from_raw(results)
|
124 |
-
return error_message, articles
|
125 |
-
|
126 |
-
# LLM ์ค์
|
127 |
-
hf_client = InferenceClient("CohereForAI/c4ai-command-r-plus-08-2024", token=HF_TOKEN)
|
128 |
-
|
129 |
-
def summarize_article(url):
|
130 |
-
try:
|
131 |
-
# ์น ํ์ด์ง์์ ํ
์คํธ ์ถ์ถ
|
132 |
-
response = requests.get(url)
|
133 |
-
response.raise_for_status()
|
134 |
-
soup = BeautifulSoup(response.text, 'html.parser')
|
135 |
-
# ๋ชจ๋ ํ
์คํธ๋ฅผ ๏ฟฝ๏ฟฝ๏ฟฝ์ถ (๊ฐ๋จํ ์์)
|
136 |
-
text = ' '.join([p.get_text() for p in soup.find_all('p')])
|
137 |
-
if not text.strip():
|
138 |
-
return "๊ธฐ์ฌ ๋ด์ฉ์ ๊ฐ์ ธ์ฌ ์ ์์ต๋๋ค."
|
139 |
-
|
140 |
-
# ์์ฝ ์์ฑ
|
141 |
-
prompt = f"๋ค์ ์์ด ๊ธฐ์ฌ๋ฅผ ํ๊ตญ์ด๋ก 3๋ฌธ์ฅ์ผ๋ก ์์ฝํ์ธ์:\n{text}"
|
142 |
-
summary = hf_client.text_generation(prompt, max_new_tokens=500)
|
143 |
-
return summary
|
144 |
-
except Exception as e:
|
145 |
-
return f"์์ฝ ์ค ์ค๋ฅ ๋ฐ์: {str(e)}"
|
146 |
-
|
147 |
-
css = """
|
148 |
-
footer {
|
149 |
-
visibility: hidden;
|
150 |
-
}
|
151 |
-
"""
|
152 |
-
|
153 |
-
# Gradio ์ธํฐํ์ด์ค ๊ตฌ์ฑ
|
154 |
-
with gr.Blocks(css=css, title="NewsAI ์๋น์ค") as iface:
|
155 |
-
gr.Markdown("๊ฒ์์ด๋ฅผ ์
๋ ฅํ๊ณ ์ํ๋ ๊ตญ๊ฐ๋ฅผ ์ ํํ๋ฉด, ๊ฒ์์ด์ ์ผ์นํ๋ 24์๊ฐ ์ด๋ด ๋ด์ค๋ฅผ ์ต๋ 10๊ฐ ์ถ๋ ฅํฉ๋๋ค.")
|
156 |
-
|
157 |
-
with gr.Column():
|
158 |
-
with gr.Row():
|
159 |
-
query = gr.Textbox(label="๊ฒ์์ด")
|
160 |
-
country = gr.Dropdown(MAJOR_COUNTRIES, label="๊ตญ๊ฐ", value="South Korea")
|
161 |
-
search_button = gr.Button("๊ฒ์")
|
162 |
-
|
163 |
-
article_outputs = []
|
164 |
-
|
165 |
-
def search_and_display(query, country):
|
166 |
-
error_message, articles = serphouse_search(query, country)
|
167 |
-
if error_message:
|
168 |
-
return gr.update(visible=True, value=error_message)
|
169 |
-
else:
|
170 |
-
# ๊ธฐ์กด ์ถ๋ ฅ๋ฌผ ์ ๊ฑฐ
|
171 |
-
for components in article_outputs:
|
172 |
-
for component in components:
|
173 |
-
component.visible = False
|
174 |
-
article_outputs.clear()
|
175 |
-
|
176 |
-
# ๊ฐ ๊ธฐ์ฌ์ ๋ํด ์ถ๋ ฅ ์์ฑ
|
177 |
-
for article in articles:
|
178 |
-
with gr.Column():
|
179 |
-
title = gr.Markdown(f"### [{article['title']}]({article['link']})")
|
180 |
-
image = gr.Image(value=article['image_url'], visible=bool(article['image_url']), shape=(200, 150))
|
181 |
-
snippet = gr.Markdown(f"**์์ฝ:** {article['snippet']}")
|
182 |
-
info = gr.Markdown(f"**์ถ์ฒ:** {article['channel']} | **์๊ฐ:** {article['time']}")
|
183 |
-
analyze_button = gr.Button("๋ถ์")
|
184 |
-
summary_output = gr.Markdown(visible=False)
|
185 |
-
|
186 |
-
def analyze_article(url):
|
187 |
-
summary = summarize_article(url)
|
188 |
-
summary_output.update(value=summary, visible=True)
|
189 |
-
|
190 |
-
analyze_button.click(analyze_article, inputs=gr.State(article['link']), outputs=summary_output)
|
191 |
-
|
192 |
-
article_outputs.append([title, image, snippet, info, analyze_button, summary_output])
|
193 |
-
|
194 |
-
return gr.update()
|
195 |
-
|
196 |
-
search_button.click(
|
197 |
-
search_and_display,
|
198 |
-
inputs=[query, country],
|
199 |
-
outputs=[]
|
200 |
-
)
|
201 |
-
|
202 |
-
iface.launch(auth=("gini", "pick"))
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|