kijeoung commited on
Commit
1fbeb24
ยท
verified ยท
1 Parent(s): e08bb4a

Create app.py

Browse files
Files changed (1) hide show
  1. app.py +131 -0
app.py ADDED
@@ -0,0 +1,131 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ import gradio as gr
2
+ import openai
3
+ import os
4
+ import requests
5
+ from bs4 import BeautifulSoup
6
+
7
+ # OpenAI API ํด๋ผ์ด์–ธํŠธ ์„ค์ •
8
+ openai.api_key = os.getenv("OPENAI_API_KEY")
9
+ if not openai.api_key:
10
+ raise ValueError("OpenAI API ํ† ํฐ(OPENAI_API_KEY)์ด ์„ค์ •๋˜์ง€ ์•Š์•˜์Šต๋‹ˆ๋‹ค.")
11
+
12
+ #############################
13
+ # OpenAI API ํ˜ธ์ถœ ํ•จ์ˆ˜
14
+ #############################
15
+ def call_openai_api(content: str, system_message: str, max_tokens: int, temperature: float, top_p: float) -> str:
16
+ try:
17
+ response = openai.ChatCompletion.create(
18
+ model="gpt-4o-mini",
19
+ messages=[
20
+ {"role": "system", "content": system_message},
21
+ {"role": "user", "content": content},
22
+ ],
23
+ max_tokens=max_tokens,
24
+ temperature=temperature,
25
+ top_p=top_p,
26
+ )
27
+ return response.choices[0].message['content']
28
+ except Exception as e:
29
+ return f"์˜ค๋ฅ˜๊ฐ€ ๋ฐœ์ƒํ–ˆ์Šต๋‹ˆ๋‹ค: {str(e)}"
30
+
31
+ #############################
32
+ # OpenAI ์„ค์ •
33
+ #############################
34
+ OPENAI_SYSTEM_MESSAGE = """๋ฐ˜๋“œ์‹œ ํ•œ๊ธ€๋กœ ๋‹ต๋ณ€ํ•  ๊ฒƒ.
35
+ ๋„ˆ๋Š” ์ตœ๊ณ ์˜ ๋น„์„œ์ด๋‹ค.
36
+ ๋‚ด๊ฐ€ ์š”๊ตฌํ•˜๋Š” ๊ฒƒ๋“ค์„ ์ตœ๋Œ€ํ•œ ์ž์„ธํ•˜๊ณ  ์ •ํ™•ํ•˜๊ฒŒ ๋‹ต๋ณ€ํ•˜๋ผ.
37
+ ##[๊ธฐ๋ณธ๊ทœ์น™]
38
+ 1. ๋ฐ˜๋“œ์‹œ ํ•œ๊ตญ์–ด(ํ•œ๊ธ€)๋กœ ์ž‘์„ฑํ•˜๋ผ.
39
+ 2. ๋„ˆ๋Š” ๊ฐ€์žฅ ์ฃผ๋ชฉ๋ฐ›๋Š” ๋งˆ์ผ€ํ„ฐ์ด๋ฉฐ ๋ธ”๋กœ๊ทธ ๋งˆ์ผ€ํŒ… ์ „๋ฌธ๊ฐ€์ด๋‹ค.
40
+ 3. ํŠนํžˆ ๋„ˆ๋Š” '์ •๋ณด์„ฑ(Informative)' ์ „๋ฌธ ๋ธ”๋กœ๊ทธ ๋งˆ์ผ€ํŒ… ์ „๋ฌธ๊ฐ€์ด๋‹ค.
41
+ 4. ์ •๋ณด ์ œ๊ณต์— ์ดˆ์ ์„ ๋งž์ถ”์–ด ์ž‘์„ฑํ•œ๋‹ค.
42
+ ##[ํ…์ŠคํŠธ ์ž‘์„ฑ ๊ทœ์น™]
43
+ 1. ์†Œ์ฃผ์ œ๋ฅผ 5๊ฐœ๋กœ ๊ตฌ๋ถ„ํ•˜์—ฌ 2000์ž ์ด์ƒ๋˜๋„๋ก ์ž‘์„ฑํ•˜๋ผ.
44
+ 2. ์ „์ฒด ๋งฅ๋ฝ์„ ์ดํ•ดํ•˜๊ณ  ๋ฌธ์žฅ์˜ ์ผ๊ด€์„ฑ์„ ์œ ์ง€ํ•˜๋ผ.
45
+ 3. ์ ˆ๋Œ€๋กœ ์ฐธ๊ณ ๊ธ€์„ ํ•œ๋ฌธ์žฅ ์ด์ƒ ๊ทธ๋Œ€๋กœ ์ถœ๋ ฅํ•˜์ง€ ๋ง ๊ฒƒ.
46
+ 4. ์ฃผ์ œ์™€ ์ƒํ™ฉ์— ๋งž๋Š” ์ ์ ˆํ•œ ์–ดํœ˜๋ฅผ ์„ ํƒํ•˜๋ผ.
47
+ 5. ํ•œ๊ธ€ ์–ดํœ˜์˜ ๋‚œ์ด๋„๋Š” ์‰ฝ๊ฒŒ ์ž‘์„ฑํ•˜๋ผ.
48
+ 6. ์ ˆ๋Œ€ ๋ฌธ์žฅ์˜ ๋์— '๋‹ต๋‹ˆ๋‹ค'๋ฅผ ์‚ฌ์šฉํ•˜์ง€ ๋ง ๊ฒƒ.
49
+ ###[์ •๋ณด์„ฑ ๋ธ”๋กœ๊ทธ ์ž‘์„ฑ ๊ทœ์น™]
50
+ 1. ๋…์ž๊ฐ€ ์–ป๊ณ ์ž ํ•˜๋Š” ์œ ์šฉํ•œ ์ •๋ณด์™€ ํฅ๋ฏธ๋กœ์šด ์ •๋ณด๋ฅผ ์ œ๊ณตํ•˜๋„๋ก ์ž‘์„ฑํ•˜๋ผ.
51
+ 2. ๋…์ž์˜ ๊ณต๊ฐ์„ ์ด๋Œ์–ด๋‚ด๊ณ  ๊ถ๊ธˆ์ฆ์„ ํ•ด๊ฒฐํ•˜๋„๋ก ์ž‘์„ฑํ•˜๋ผ.
52
+ 3. ๋…์ž์˜ ๊ด€์‹ฌ์‚ฌ๋ฅผ ์ถฉ์กฑ์‹œํ‚ค๋„๋ก ์ž‘์„ฑํ•˜๋ผ.
53
+ 4. ๋…์ž์—๊ฒŒ ์ด๋“์ด ๋˜๋Š” ์ •๋ณด๋ฅผ ์ž‘์„ฑํ•˜๋ผ.
54
+ ##[์ œ์™ธ ๊ทœ์น™]
55
+ 1. ๋ฐ˜๋“œ์‹œ ๋น„์†์–ด ๋ฐ ์š•์„ค(expletive, abusive language, slang)์€ ์ œ์™ธํ•˜๋ผ.
56
+ 2. ๋ฐ˜๋“œ์‹œ ์ฐธ๊ณ ๊ธ€์˜ ๋งํฌ(URL)๋Š” ์ œ์™ธํ•˜๋ผ.
57
+ 3. ์ฐธ๊ณ ๊ธ€์—์„œ '๋งํฌ๋ฅผ ํ™•์ธํ•ด์ฃผ์„ธ์š”'์™€ ๊ฐ™์€ ๋งํฌ ์ด๋™์˜ ๋ฌธ๊ตฌ๋Š” ์ œ์™ธํ•˜๋ผ.
58
+ 4. ์ฐธ๊ณ ๊ธ€์— ์žˆ๋Š” ์ž‘์„ฑ์ž, ํ™”์ž, ์œ ํŠœ๋ฒ„, ๊ธฐ์ž์˜ ์ด๋ฆ„, ์• ์นญ, ๋‹‰๋„ค์ž„์€ ๋ฐ˜๋“œ์‹œ ์ œ์™ธํ•˜๋ผ.
59
+ 5. ๋ฐ˜๋“œ์‹œ ๋ฌธ์žฅ์˜ ๋๋ถ€๋ถ„์ด ์–ด์ƒ‰ํ•œ ํ•œ๊ตญ์–ด ํ‘œํ˜„์€ ์ œ์™ธํ•˜๋ผ('์˜ˆ์š”', '๋‹ต๋‹ˆ๋‹ค', 'ํ•ด์š”', 'ํ•ด์ฃผ์ฃ ', '๋์ฃ ', '๋์–ด์š”', '๊ณ ์š”' ๋“ฑ.)
60
+ """
61
+ OPENAI_MAX_TOKENS = 4000
62
+ OPENAI_TEMPERATURE = 0.7
63
+ OPENAI_TOP_P = 0.95
64
+
65
+ #############################
66
+ # ๋„ค์ด๋ฒ„ ๋ธ”๋กœ๊ทธ ์Šคํฌ๋ž˜ํ•‘
67
+ #############################
68
+ def convert_to_mobile_url(url):
69
+ if "m.blog.naver.com" not in url:
70
+ if "blog.naver.com" in url:
71
+ url_parts = url.split("/")
72
+ if len(url_parts) >= 5:
73
+ user_id = url_parts[3]
74
+ post_id = url_parts[4]
75
+ return f"https://m.blog.naver.com/{user_id}/{post_id}"
76
+ return url
77
+
78
+ def scrape_naver_blog(url):
79
+ try:
80
+ mobile_url = convert_to_mobile_url(url)
81
+ response = requests.get(mobile_url)
82
+ response.raise_for_status()
83
+ soup = BeautifulSoup(response.text, 'html.parser')
84
+ title_element = soup.find("div", class_="se-module se-module-text se-title-text")
85
+ title = title_element.get_text(strip=True) if title_element else "์ œ๋ชฉ์„ ์ฐพ์„ ์ˆ˜ ์—†์Œ"
86
+ content_elements = soup.find_all("div", class_="se-module se-module-text")
87
+ content = "\n".join(
88
+ elem.get_text(strip=True) for elem in content_elements
89
+ ) if content_elements else "๋‚ด์šฉ์„ ์ฐพ์„ ์ˆ˜ ์—†์Œ"
90
+ return f"์ œ๋ชฉ: {title}\n\n๋‚ด์šฉ: {content}"
91
+ except Exception as e:
92
+ return f"Error: {e}"
93
+
94
+ #############################
95
+ # Gradio UI ๊ตฌ์„ฑ (ํƒญ ๊ตฌ์กฐ ์ ์šฉ)
96
+ #############################
97
+ with gr.Blocks() as demo:
98
+ gr.Markdown("# ๋‹ค๊ธฐ๋Šฅ ๋„๊ตฌ")
99
+ with gr.Tabs():
100
+ with gr.Tab("๋ธ”๋กœ๊ทธ ์ƒ์„ฑ๊ธฐ"):
101
+ tone_radio = gr.Radio(
102
+ label="๋งํˆฌ๋ฐ”๊พธ๊ธฐ",
103
+ choices=["์นœ๊ทผํ•˜๊ฒŒ", "์ผ๋ฐ˜์ ์ธ", "์ „๋ฌธ์ ์ธ"],
104
+ value="์ผ๋ฐ˜์ ์ธ"
105
+ )
106
+ ref1 = gr.Textbox(label="์ฐธ์กฐ๊ธ€ 1")
107
+ ref2 = gr.Textbox(label="์ฐธ์กฐ๊ธ€ 2")
108
+ ref3 = gr.Textbox(label="์ฐธ์กฐ๊ธ€ 3")
109
+ output_box = gr.Textbox(label="๊ฒฐ๊ณผ", lines=20, interactive=False)
110
+ generate_button = gr.Button("์ƒ์„ฑํ•˜๊ธฐ")
111
+ generate_button.click(
112
+ fn=lambda t, r1, r2, r3: call_openai_api(
113
+ f"๋งํˆฌ: {t}\n์ฐธ์กฐ๊ธ€1: {r1}\n์ฐธ์กฐ๊ธ€2: {r2}\n์ฐธ์กฐ๊ธ€3: {r3}\n",
114
+ OPENAI_SYSTEM_MESSAGE, OPENAI_MAX_TOKENS, OPENAI_TEMPERATURE, OPENAI_TOP_P
115
+ ),
116
+ inputs=[tone_radio, ref1, ref2, ref3],
117
+ outputs=output_box
118
+ )
119
+
120
+ with gr.Tab("๋„ค์ด๋ฒ„ ๋ธ”๋กœ๊ทธ ์Šคํฌ๋ž˜ํ•‘"):
121
+ blog_url = gr.Textbox(label="๋„ค์ด๋ฒ„ ๋ธ”๋กœ๊ทธ URL")
122
+ scrape_output = gr.Textbox(label="์Šคํฌ๋ž˜ํ•‘ ๊ฒฐ๊ณผ", lines=10, interactive=False)
123
+ scrape_button = gr.Button("์Šคํฌ๋ž˜ํ•‘ํ•˜๊ธฐ")
124
+ scrape_button.click(
125
+ fn=scrape_naver_blog,
126
+ inputs=blog_url,
127
+ outputs=scrape_output
128
+ )
129
+
130
+ if __name__ == "__main__":
131
+ demo.launch()