Spaces:
Sleeping
Sleeping
File size: 13,185 Bytes
a76e765 8c1d98e 1966659 2df08a1 a76e765 22bef0a 12d804b 1966659 c5a9055 12d804b 1966659 ad0ddc3 1966659 22fe9e6 22bef0a 1966659 12d804b 1966659 c5a9055 1966659 c5a9055 1966659 c5a9055 f0a45c3 c5a9055 2f65748 c5a9055 12d804b c5a9055 f0a45c3 a76e765 8c1d98e a76e765 421135d a76e765 421135d a76e765 421135d a76e765 421135d a76e765 ed14ca2 a76e765 12d804b 22bef0a 12d804b a76e765 |
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66 67 68 69 70 71 72 73 74 75 76 77 78 79 80 81 82 83 84 85 86 87 88 89 90 91 92 93 94 95 96 97 98 99 100 101 102 103 104 105 106 107 108 109 110 111 112 113 114 115 116 117 118 119 120 121 122 123 124 125 126 127 128 129 130 131 132 133 134 135 136 137 138 139 140 141 142 143 144 145 146 147 148 149 150 151 152 153 154 155 156 157 158 159 160 161 162 163 164 165 166 167 168 169 170 171 172 173 174 175 176 177 178 179 180 181 182 183 184 185 186 187 188 189 190 191 192 193 194 195 196 197 198 199 200 201 202 203 204 205 206 207 208 209 210 211 212 213 214 215 216 217 218 219 220 221 222 223 224 225 226 227 228 229 230 231 232 233 234 235 236 237 238 239 240 241 242 243 244 245 246 247 248 249 250 251 252 253 254 255 256 257 258 259 260 261 262 263 264 265 266 267 268 269 270 271 272 273 274 275 276 277 278 279 280 281 282 283 284 285 286 287 288 289 290 291 292 293 294 295 296 297 298 299 300 301 302 303 304 305 306 307 308 309 310 311 312 313 314 |
import os
import random
import time
import re
import json
import requests
from bs4 import BeautifulSoup
from requests.adapters import HTTPAdapter
from requests.packages.urllib3.util.retry import Retry
import openai
import gradio as gr
from fpdf import FPDF as FPDF2
from datetime import datetime
from zoneinfo import ZoneInfo
from langchain.chat_models import ChatOpenAI
from langchain.prompts import ChatPromptTemplate
from langchain.chains import LLMChain
from langchain.callbacks import get_openai_callback
import sys
# API ํค ์ค์
OPENAI_API_KEY = os.getenv("OPENAI_API_KEY")
# OpenAI ์ค์
openai.api_key = OPENAI_API_KEY
def setup_session():
try:
session = requests.Session()
retries = Retry(total=5, backoff_factor=1, status_forcelist=[502, 503, 504])
session.mount('https://', HTTPAdapter(max_retries=retries))
return session
except Exception as e:
return None
def generate_naver_search_url(query):
base_url = "https://search.naver.com/search.naver?"
params = {"ssc": "tab.blog.all", "sm": "tab_jum", "query": query}
url = base_url + "&".join(f"{key}={value}" for key, value in params.items())
return url
def crawl_blog_content(url, session):
try:
headers = {
"User-Agent": "Mozilla/5.0 (Windows NT 10.0; Win64; x64) AppleWebKit/537.36 (KHTML, like Gecko) Chrome/58.0.3029.110 Safari/537.3",
"Accept": "text/html,application/xhtml+xml,application/xml;q=0.9,image/webp,image/apng,*/*;q=0.8",
"Accept-Language": "ko-KR,ko;q=0.9,en-US;q=0.8,en;q=0.7",
"Accept-Encoding": "gzip, deflate, br",
"Connection": "keep-alive",
"Referer": "https://search.naver.com/search.naver",
}
delay = random.uniform(1, 2)
time.sleep(delay)
response = session.get(url, headers=headers)
if response.status_code != 200:
return ""
soup = BeautifulSoup(response.content, "html.parser")
content = soup.find("div", attrs={'class': 'se-main-container'})
if content:
return clean_text(content.get_text())
else:
return ""
except Exception as e:
return ""
def crawl_naver_search_results(url, session):
try:
headers = {
"User-Agent": "Mozilla/5.0 (Windows NT 10.0; Win64; x64) AppleWebKit/537.36 (KHTML, like Gecko) Chrome/58.0.3029.110 Safari/537.3",
"Accept": "text/html,application/xhtml+xml,application/xml;q=0.9,image/webp,image/apng,*/*;q=0.8",
"Accept-Language": "ko-KR,ko;q=0.9,en-US;q=0.8,en;q=0.7",
"Accept-Encoding": "gzip, deflate, br",
"Connection": "keep-alive",
"Referer": "https://search.naver.com/search.naver",
}
response = session.get(url, headers=headers)
if response.status_code != 200:
return []
soup = BeautifulSoup(response.content, "html.parser")
results = []
count = 0
for li in soup.find_all("li", class_=re.compile("bx.*")):
if count >= 10:
break
for div in li.find_all("div", class_="detail_box"):
for div2 in div.find_all("div", class_="title_area"):
title = div2.text.strip()
for a in div2.find_all("a", href=True):
link = a["href"]
if "blog.naver" in link:
link = link.replace("https://", "https://m.")
results.append({"์ ๋ชฉ": title, "๋งํฌ": link})
count += 1
if count >= 10:
break
if count >= 10:
break
if count >= 10:
break
return results
except Exception as e:
return []
def clean_text(text):
text = re.sub(r'\s+', ' ', text).strip()
return text
def fetch_references(topic):
search_url = generate_naver_search_url(topic)
session = setup_session()
if session is None:
return ["์ธ์
์ค์ ์คํจ"] * 3
results = crawl_naver_search_results(search_url, session)
if len(results) < 3:
return ["์ถฉ๋ถํ ๊ฒ์ ๊ฒฐ๊ณผ๋ฅผ ์ฐพ์ง ๋ชปํ์ต๋๋ค."] * 3
selected_results = random.sample(results, 3)
references = []
for result in selected_results:
content = crawl_blog_content(result['๋งํฌ'], session)
references.append(f"์ ๋ชฉ: {result['์ ๋ชฉ']}\n๋ด์ฉ: {content}")
return references
def fetch_crawl_results(query):
references = fetch_references(query)
return references[0], references[1], references[2]
def generate_blog_post(query, prompt_template):
try:
target_length = 1500 # ๋ด๋ถ์ ์ผ๋ก ๋ชฉํ ๊ธ์์ ์ค์
max_attempts = 2 # ์ต๋ 2๋ฒ ์คํ (์ด๊ธฐ 1๋ฒ + ์ฌ์๋ 1๋ฒ)
references = fetch_references(query)
ref1, ref2, ref3 = references
chat = ChatOpenAI(
model_name="gpt-4o-mini",
temperature=0.85,
max_tokens=10000,
top_p=0.9,
frequency_penalty=0.5,
presence_penalty=0,
n=1,
request_timeout=60
)
prompt = ChatPromptTemplate.from_template(
prompt_template + """
์ฃผ์ : {query}
์ฐธ๊ณ ๊ธ1: {ref1}
์ฐธ๊ณ ๊ธ2: {ref2}
์ฐธ๊ณ ๊ธ3: {ref3}
๋ค์ ํํ์ ์ฌ์ฉํ์ง ๋ง์ธ์: ์ฌ๋ฌ๋ถ, ๋ง์ง๋ง์ผ๋ก, ๊ฒฐ๋ก ์ ์ผ๋ก, ๊ฒฐ๊ตญ, ์ข
ํฉ์ ์ผ๋ก, ๋ฐ๋ผ์, ๋ง๋ฌด๋ฆฌ, ์์ฝ
์ฝ {target_length}์๋ก ์์ฑํด์ฃผ์ธ์.
"""
)
chain = LLMChain(llm=chat, prompt=prompt)
unwanted_patterns = [
r'\b์ฌ๋ฌ๋ถ[,.]?\s*',
r'\b(๋ง์ง๋ง์ผ๋ก|๊ฒฐ๋ก ์ ์ผ๋ก|๊ฒฐ๊ตญ|์ข
ํฉ์ ์ผ๋ก|๋ฐ๋ผ์|๋ง๋ฌด๋ฆฌ|์์ฝ)[,.]?\s*'
]
for attempt in range(max_attempts):
with get_openai_callback() as cb:
result = chain.run(query=query, ref1=ref1, ref2=ref2, ref3=ref3, target_length=target_length)
generated_post = result.strip()
# ๋ชฉํ ๊ธ์์๋ฅผ ์ถฉ์กฑํ๊ณ ์์น ์๋ ํํ์ด ์์ผ๋ฉด ๋ฃจํ ์ข
๋ฃ
if len(generated_post) >= target_length and not any(re.search(pattern, generated_post, re.IGNORECASE) for pattern in unwanted_patterns):
break
# ์ฒซ ๋ฒ์งธ ์๋ ํ ์ฌ์๋ ์ ํ๋กฌํํธ ์์
if attempt == 0:
if len(generated_post) < target_length:
prompt.template += f"\n\nํ์ฌ ๊ธ์์๋ {len(generated_post)}์์
๋๋ค. ์ฝ {target_length - len(generated_post)}์๋ฅผ ์ถ๊ฐ๋ก ์์ฑํ์ฌ ์ด {target_length}์๊ฐ ๋๋๋ก ํด์ฃผ์ธ์."
if any(re.search(pattern, generated_post, re.IGNORECASE) for pattern in unwanted_patterns):
prompt.template += "\n\n์์น ์๋ ํํ์ด ํฌํจ๋์ด ์์ต๋๋ค. ํด๋น ํํ์ ์ ๊ฑฐํ๊ณ ์์ฐ์ค๋ฝ๊ฒ ๊ธ์ ๋ค์ ์์ฑํด์ฃผ์ธ์."
final_post = f"์ฃผ์ : {query}\n\n{generated_post}"
actual_length = len(generated_post)
return final_post, ref1, ref2, ref3, actual_length
except Exception as e:
return f"๋ธ๋ก๊ทธ ๊ธ ์์ฑ ์ค ์ค๋ฅ ๋ฐ์: {str(e)}", "", "", "", 0
# PDF ํด๋์ค ๋ฐ ๊ด๋ จ ํจ์ ์ ์
class PDF(FPDF2):
def __init__(self):
super().__init__()
current_dir = os.path.dirname(__file__)
self.add_font("NanumGothic", "", os.path.join(current_dir, "NanumGothic.ttf"))
self.add_font("NanumGothic", "B", os.path.join(current_dir, "NanumGothicBold.ttf"))
self.add_font("NanumGothicExtraBold", "", os.path.join(current_dir, "NanumGothicExtraBold.ttf"))
self.add_font("NanumGothicLight", "", os.path.join(current_dir, "NanumGothicLight.ttf"))
def header(self):
self.set_font('NanumGothic', '', 10)
def footer(self):
self.set_y(-15)
self.set_font('NanumGothic', '', 8)
self.cell(0, 10, f'Page {self.page_no()}', 0, 0, 'C')
def save_to_pdf(blog_post, user_topic):
pdf = PDF()
pdf.add_page()
lines = blog_post.split('\n')
title = lines[0].strip()
content = '\n'.join(lines[1:]).strip()
# ํ์ฌ ๋ ์ง์ ์๊ฐ์ ๊ฐ์ ธ์ต๋๋ค (๋ํ๋ฏผ๊ตญ ์๊ฐ ๊ธฐ์ค)
now = datetime.now(ZoneInfo("Asia/Seoul"))
date_str = now.strftime("%y%m%d")
time_str = now.strftime("%H%M")
# ํ์ผ๋ช
์์ฑ
filename = f"{date_str}_{time_str}_{format_filename(user_topic)}.pdf"
pdf.set_font("NanumGothic", 'B', size=14)
pdf.cell(0, 10, title, ln=True, align='C')
pdf.ln(10)
pdf.set_font("NanumGothic", '', size=11)
pdf.multi_cell(0, 5, content)
print(f"Saving PDF as: {filename}")
pdf.output(filename)
return filename
def format_filename(text):
text = re.sub(r'[^\w\s-]', '', text)
return text[:50].strip()
def save_content_to_pdf(blog_post, user_topic):
return save_to_pdf(blog_post, user_topic)
# ๊ธฐ๋ณธ ํ๋กฌํํธ ํ
ํ๋ฆฟ
DEFAULT_PROMPT_TEMPLATE = """
[๋ธ๋ก๊ทธ ๊ธ ์์ฑ ๊ธฐ๋ณธ ๊ท์น]
1. ๋ฐ๋์ ํ๊ธ๋ก ์์ฑํ๋ผ
2. ์ฃผ์ด์ง ์ฐธ๊ณ ๊ธ์ ๋ฐํ์ผ๋ก 1๊ฐ์ ์ํ๋ฆฌ๋ทฐํ(Product Review) ๋ธ๋ก๊ทธ๋ฅผ ์์ฑ
3. ์ฃผ์ ์ ์ ๋ชฉ์ ์ ์ธํ ๊ธ์ด 1500๋จ์ด ์ด์์ด ๋๋๋ก ์์ฑ
4. ๊ธ์ ์ ๋ชฉ์ ์ํ๋ฆฌ๋ทฐํ ๋ธ๋ก๊ทธ ํํ์ ๋ง๋ ์ ์ ํ ์ ๋ชฉ์ผ๋ก ์ถ๋ ฅ
- ์ฐธ๊ณ ๊ธ์ ์ ๋ชฉ๋ ์ฐธ๊ณ ํ๋, ๋์ผํ๊ฒ ์์ฑํ์ง ๋ง ๊ฒ
5. ๋ฐ๋์ ๋งํฌ๋ค์ด ํ์์ด ์๋ ์์ํ ํ
์คํธ๋ก๋ง ์ถ๋ ฅํ๋ผ
6. ๋ค์ํ๋ฒ ์ฐธ๊ณ ๊ธ์ ๊ฒํ ํ์ฌ ๋ด์ฉ์ ์ถฉ๋ถํ ๋ฐ์ํ๋, ์ฐธ๊ณ ๊ธ์ ๊ธ์ ๊ทธ๋๋ก ์ฌ์์ฑํ์ง๋ ๋ง ๊ฒ
[๋ธ๋ก๊ทธ ๊ธ ์์ฑ ์ธ๋ถ ๊ท์น]
1. ์ฌ์ฉ์๊ฐ ์
๋ ฅํ ์ฃผ์ ์ ์ฃผ์ด์ง ์ฐธ๊ณ ๊ธ 3๊ฐ๋ฅผ ๋ฐํ์ผ๋ก ์ํ๋ฆฌ๋ทฐํ ๋ธ๋ก๊ทธ ๊ธ 1๊ฐ๋ฅผ ์์ฑํ๋ผ
2. ์ฃผ์ด์ง ๋ชจ๋ ๊ธ์ ๋ถ์ํ์ฌ ํ๋์ ๋์ฃผ์ ๋ฅผ ์ ์ ํ๋ผ(1๊ฐ์ ์ฐธ๊ณ ๊ธ์ ์น์ฐ์น์ง ๋ง๊ณ ๋ค์ํ ๋ด์ฉ์ ๋ด์๊ฒ)
3. ์ฌ๋ฌ๊ฐ์ง ์ํ์ด๋ผ๋ฉด ์ํ 1๊ฐ์ ์น์ฐ์น ๋ฆฌ๋ทฐ๋ฅผ ์์ฑํ์ง ๋ง ๊ฒ.
4. ๋์ฃผ์ ์ ๋ง๊ฒ ๊ธ์ ๋งฅ๋ฝ์ ์ ์งํ๋ผ
5. ์ฐธ๊ณ ๊ธ์ ์์ฑ๋ ์ํ๊ณผ ๊ธฐ๋ฅ์ ์ง์คํ์ฌ ์์ฑํ๋ผ
6. ์ค์ ๋ด๊ฐ ์ฌ์ฉํด๋ณด๊ณ ๊ฒฝํํ ๋ด์ฉ์ ์์ฑํ ๋ฆฌ๋ทฐ ํํ๋ก ๊ธ์ ์์ฑ
7. ๋ด์ฉ์ ๊ธ์ ์ ์ผ๋ก ์์ฑํ๋, ์ํ์ด ๋๋ณด์ด๋๋ก ์์ฑ(์ ํ์ด ์ฌ๋ฌ๊ฐ์ผ ๊ฒฝ์ฐ, ํ๋์ ์ํ์ ์น์ฐ์น์ง ๋ง ๊ฒ)
8. ์ํ์ ๊ฐ์น๋ฅผ ๊ณ ๊ฐ์๊ฒ ์ดํํ๋ผ.
9. ๊ธ์ ์, ๋ค ๋ฌธ์ฅ์ด ์์ฐ์ค๋ฝ๊ฒ ์ด์ด์ง๋๋ก ์์ฑ
10. ์ดํฌ๋ ์ฃผ์ด์ง ์ฐธ๊ณ ๊ธ 3๊ฐ์ง์ ์ดํฌ๋ฅผ ์ ์ ํ ๋ฐ์ํ๋ผ
- ํนํ ๋ฌธ์ฅ์ ๋ ๋ถ๋ถ์ ์ ์ ํ ๋ฐ์(๊ฐ๊ธ์ '~์'๋ก ๋๋๋๋ก ์์ฑ)
- ๋๋ฌด ๋ฑ๋ฑํ์ง ์๊ฒ ํธ์ํ๊ฒ ์ฝ์ ์ ์๋๋ก ์์ฐ์ค๋ฌ์ด ๋ํ์ฒด๋ฅผ ๋ฐ์
- ๋จ์ด ์ ํ์ ์ฌ์ด ํ๊ตญ์ด ์ดํ๋ฅผ ์ฌ์ฉํ๊ณ ์ฌ์ ์ํํ, ์ค๋๋ ํํ์ ์ ์ธํ๋ผ
[์ ์ธ ๊ท์น]
1. ๋ฐ๋์ ์ฐธ๊ณ ๊ธ์ ํฌํจ๋ ๋งํฌ(URL)๋ ์ ์ธ
2. ์ฐธ๊ณ ๊ธ์์ '๋งํฌ๋ฅผ ํ์ธํด์ฃผ์ธ์'์ ๊ฐ์ ๋งํฌ ์ด๋์ ๋ฌธ๊ตฌ๋ ์ ์ธ
3. ์ฐธ๊ณ ๊ธ์ ์๋ ์์ฑ์, ํ์, ์ ํ๋ฒ, ๊ธฐ์(Writer, speaker, YouTuber, reporter)์ ์ด๋ฆ, ์ ์นญ, ๋๋ค์(Name, Nkickname)์ ๋ฐ๋์ ์ ์ธ
4. '์
์ฒด๋ก ๋ถํฐ ์ ๊ณต ๋ฐ์์ ์์ฑ', '์ฟ ํก ํํธ๋์ค'๋ฑ์ ํํ์ ๋ฐ๋์ ์ ์ธํ๋ผ.
5. ๊ธ์ ๊ตฌ์กฐ๊ฐ ๋๋ฌ๋๊ฒ ์์ฑํ์ง ๋ง ๊ฒ(์์, ๋์ ๋ํ ํํ)
- ์ฌ๋ฌ๋ถ,
- ๋ง์ง๋ง์ผ๋ก, ๊ฒฐ๋ก ์ ์ผ๋ก, ๊ฒฐ๊ตญ, ์ข
ํฉ์ ์ผ๋ก, ๋ฐ๋ผ์, ๋ง๋ฌด๋ฆฌ, ์์ฝ,
"""
# Gradio ์ฑ ์์ฑ
with gr.Blocks() as iface:
gr.Markdown("# ๋ธ๋ก๊ทธ ๊ธ ์์ฑ๊ธฐ_๋ฆฌ๋ทฐ_๊ธฐ๋ฅ์ง์คํ")
gr.Markdown("์ฃผ์ ๋ฅผ ์
๋ ฅํ๊ณ ๋ธ๋ก๊ทธ ๊ธ ์์ฑ ๋ฒํผ์ ๋๋ฅด๋ฉด ์๋์ผ๋ก ๋ธ๋ก๊ทธ ๊ธ์ ์์ฑํฉ๋๋ค.")
query_input = gr.Textbox(lines=1, placeholder="๋ธ๋ก๊ทธ ๊ธ์ ์ฃผ์ ๋ฅผ ์
๋ ฅํด์ฃผ์ธ์...", label="์ฃผ์ ")
prompt_input = gr.Textbox(lines=10, value=DEFAULT_PROMPT_TEMPLATE, label="ํ๋กฌํํธ ํ
ํ๋ฆฟ", visible=True)
generate_button = gr.Button("๋ธ๋ก๊ทธ ๊ธ ์์ฑ")
output_text = gr.Textbox(label="์์ฑ๋ ๋ธ๋ก๊ทธ ๊ธ")
ref1_text = gr.Textbox(label="์ฐธ๊ณ ๊ธ 1", lines=10, visible=True)
ref2_text = gr.Textbox(label="์ฐธ๊ณ ๊ธ 2", lines=10, visible=True)
ref3_text = gr.Textbox(label="์ฐธ๊ณ ๊ธ 3", lines=10, visible=True)
save_pdf_button = gr.Button("PDF๋ก ์ ์ฅ")
pdf_output = gr.File(label="์์ฑ๋ PDF ํ์ผ")
generate_button.click(
generate_blog_post,
inputs=[query_input, prompt_input],
outputs=[output_text, ref1_text, ref2_text, ref3_text],
show_progress=True
)
save_pdf_button.click(
save_content_to_pdf,
inputs=[output_text, query_input],
outputs=[pdf_output],
show_progress=True
)
# Gradio ์ฑ ์คํ
if __name__ == "__main__":
iface.launch() |