Delete app-backup1.py
Browse files- app-backup1.py +0 -415
app-backup1.py
DELETED
@@ -1,415 +0,0 @@
|
|
1 |
-
import gradio as gr
|
2 |
-
import requests
|
3 |
-
import json
|
4 |
-
import os
|
5 |
-
from datetime import datetime, timedelta
|
6 |
-
from huggingface_hub import InferenceClient
|
7 |
-
|
8 |
-
|
9 |
-
|
10 |
-
API_KEY = os.getenv("SERPHOUSE_API_KEY")
|
11 |
-
# hf_client = InferenceClient("CohereForAI/c4ai-command-r-plus-08-2024", token=os.getenv("HF_TOKEN"))
|
12 |
-
|
13 |
-
# 국가별 언어 코드 매핑
|
14 |
-
COUNTRY_LANGUAGES = {
|
15 |
-
"United States": "en",
|
16 |
-
"United Kingdom": "en",
|
17 |
-
"Taiwan": "zh-TW", # 대만어(번체 중국어)
|
18 |
-
"Canada": "en",
|
19 |
-
"Australia": "en",
|
20 |
-
"Germany": "de",
|
21 |
-
"France": "fr",
|
22 |
-
"Japan": "ja",
|
23 |
-
"South Korea": "ko",
|
24 |
-
"China": "zh",
|
25 |
-
"India": "hi",
|
26 |
-
"Brazil": "pt",
|
27 |
-
"Mexico": "es",
|
28 |
-
"Russia": "ru",
|
29 |
-
"Italy": "it",
|
30 |
-
"Spain": "es",
|
31 |
-
"Netherlands": "nl",
|
32 |
-
"Singapore": "en",
|
33 |
-
"Hong Kong": "zh-HK",
|
34 |
-
"Indonesia": "id",
|
35 |
-
"Malaysia": "ms",
|
36 |
-
"Philippines": "tl",
|
37 |
-
"Thailand": "th",
|
38 |
-
"Vietnam": "vi",
|
39 |
-
"Belgium": "nl",
|
40 |
-
"Denmark": "da",
|
41 |
-
"Finland": "fi",
|
42 |
-
"Ireland": "en",
|
43 |
-
"Norway": "no",
|
44 |
-
"Poland": "pl",
|
45 |
-
"Sweden": "sv",
|
46 |
-
"Switzerland": "de",
|
47 |
-
"Austria": "de",
|
48 |
-
"Czech Republic": "cs",
|
49 |
-
"Greece": "el",
|
50 |
-
"Hungary": "hu",
|
51 |
-
"Portugal": "pt",
|
52 |
-
"Romania": "ro",
|
53 |
-
"Turkey": "tr",
|
54 |
-
"Israel": "he",
|
55 |
-
"Saudi Arabia": "ar",
|
56 |
-
"United Arab Emirates": "ar",
|
57 |
-
"South Africa": "en",
|
58 |
-
"Argentina": "es",
|
59 |
-
"Chile": "es",
|
60 |
-
"Colombia": "es",
|
61 |
-
"Peru": "es",
|
62 |
-
"Venezuela": "es",
|
63 |
-
"New Zealand": "en",
|
64 |
-
"Bangladesh": "bn",
|
65 |
-
"Pakistan": "ur",
|
66 |
-
"Egypt": "ar",
|
67 |
-
"Morocco": "ar",
|
68 |
-
"Nigeria": "en",
|
69 |
-
"Kenya": "sw",
|
70 |
-
"Ukraine": "uk",
|
71 |
-
"Croatia": "hr",
|
72 |
-
"Slovakia": "sk",
|
73 |
-
"Bulgaria": "bg",
|
74 |
-
"Serbia": "sr",
|
75 |
-
"Estonia": "et",
|
76 |
-
"Latvia": "lv",
|
77 |
-
"Lithuania": "lt",
|
78 |
-
"Slovenia": "sl",
|
79 |
-
"Luxembourg": "fr",
|
80 |
-
"Malta": "mt",
|
81 |
-
"Cyprus": "el",
|
82 |
-
"Iceland": "is"
|
83 |
-
}
|
84 |
-
|
85 |
-
COUNTRY_LOCATIONS = {
|
86 |
-
"United States": "United States",
|
87 |
-
"United Kingdom": "United Kingdom",
|
88 |
-
"Taiwan": "Taiwan", # 국가명 사용
|
89 |
-
"Canada": "Canada",
|
90 |
-
"Australia": "Australia",
|
91 |
-
"Germany": "Germany",
|
92 |
-
"France": "France",
|
93 |
-
"Japan": "Japan",
|
94 |
-
"South Korea": "South Korea",
|
95 |
-
"China": "China",
|
96 |
-
"India": "India",
|
97 |
-
"Brazil": "Brazil",
|
98 |
-
"Mexico": "Mexico",
|
99 |
-
"Russia": "Russia",
|
100 |
-
"Italy": "Italy",
|
101 |
-
"Spain": "Spain",
|
102 |
-
"Netherlands": "Netherlands",
|
103 |
-
"Singapore": "Singapore",
|
104 |
-
"Hong Kong": "Hong Kong",
|
105 |
-
"Indonesia": "Indonesia",
|
106 |
-
"Malaysia": "Malaysia",
|
107 |
-
"Philippines": "Philippines",
|
108 |
-
"Thailand": "Thailand",
|
109 |
-
"Vietnam": "Vietnam",
|
110 |
-
"Belgium": "Belgium",
|
111 |
-
"Denmark": "Denmark",
|
112 |
-
"Finland": "Finland",
|
113 |
-
"Ireland": "Ireland",
|
114 |
-
"Norway": "Norway",
|
115 |
-
"Poland": "Poland",
|
116 |
-
"Sweden": "Sweden",
|
117 |
-
"Switzerland": "Switzerland",
|
118 |
-
"Austria": "Austria",
|
119 |
-
"Czech Republic": "Czech Republic",
|
120 |
-
"Greece": "Greece",
|
121 |
-
"Hungary": "Hungary",
|
122 |
-
"Portugal": "Portugal",
|
123 |
-
"Romania": "Romania",
|
124 |
-
"Turkey": "Turkey",
|
125 |
-
"Israel": "Israel",
|
126 |
-
"Saudi Arabia": "Saudi Arabia",
|
127 |
-
"United Arab Emirates": "United Arab Emirates",
|
128 |
-
"South Africa": "South Africa",
|
129 |
-
"Argentina": "Argentina",
|
130 |
-
"Chile": "Chile",
|
131 |
-
"Colombia": "Colombia",
|
132 |
-
"Peru": "Peru",
|
133 |
-
"Venezuela": "Venezuela",
|
134 |
-
"New Zealand": "New Zealand",
|
135 |
-
"Bangladesh": "Bangladesh",
|
136 |
-
"Pakistan": "Pakistan",
|
137 |
-
"Egypt": "Egypt",
|
138 |
-
"Morocco": "Morocco",
|
139 |
-
"Nigeria": "Nigeria",
|
140 |
-
"Kenya": "Kenya",
|
141 |
-
"Ukraine": "Ukraine",
|
142 |
-
"Croatia": "Croatia",
|
143 |
-
"Slovakia": "Slovakia",
|
144 |
-
"Bulgaria": "Bulgaria",
|
145 |
-
"Serbia": "Serbia",
|
146 |
-
"Estonia": "Estonia",
|
147 |
-
"Latvia": "Latvia",
|
148 |
-
"Lithuania": "Lithuania",
|
149 |
-
"Slovenia": "Slovenia",
|
150 |
-
"Luxembourg": "Luxembourg",
|
151 |
-
"Malta": "Malta",
|
152 |
-
"Cyprus": "Cyprus",
|
153 |
-
"Iceland": "Iceland"
|
154 |
-
}
|
155 |
-
|
156 |
-
MAJOR_COUNTRIES = list(COUNTRY_LOCATIONS.keys())
|
157 |
-
|
158 |
-
def translate_query(query, country):
|
159 |
-
try:
|
160 |
-
# 영어 입력 확인
|
161 |
-
if is_english(query):
|
162 |
-
print(f"영어 검색어 감지 - 원본 사용: {query}")
|
163 |
-
return query
|
164 |
-
|
165 |
-
# 선택된 국가가 번역 지원 국가인 경우
|
166 |
-
if country in COUNTRY_LANGUAGES:
|
167 |
-
# South Korea 선택시 한글 입력은 그대로 사용
|
168 |
-
if country == "South Korea":
|
169 |
-
print(f"한국 선택 - 원본 사용: {query}")
|
170 |
-
return query
|
171 |
-
|
172 |
-
target_lang = COUNTRY_LANGUAGES[country]
|
173 |
-
print(f"번역 시도: {query} -> {country}({target_lang})")
|
174 |
-
|
175 |
-
url = f"https://translate.googleapis.com/translate_a/single"
|
176 |
-
params = {
|
177 |
-
"client": "gtx",
|
178 |
-
"sl": "auto",
|
179 |
-
"tl": target_lang,
|
180 |
-
"dt": "t",
|
181 |
-
"q": query
|
182 |
-
}
|
183 |
-
|
184 |
-
response = requests.get(url, params=params)
|
185 |
-
translated_text = response.json()[0][0][0]
|
186 |
-
print(f"번역 완료: {query} -> {translated_text} ({country})")
|
187 |
-
return translated_text
|
188 |
-
|
189 |
-
return query
|
190 |
-
|
191 |
-
except Exception as e:
|
192 |
-
print(f"번역 오류: {str(e)}")
|
193 |
-
return query
|
194 |
-
|
195 |
-
def translate_to_korean(text):
|
196 |
-
try:
|
197 |
-
url = "https://translate.googleapis.com/translate_a/single"
|
198 |
-
params = {
|
199 |
-
"client": "gtx",
|
200 |
-
"sl": "auto",
|
201 |
-
"tl": "ko",
|
202 |
-
"dt": "t",
|
203 |
-
"q": text
|
204 |
-
}
|
205 |
-
|
206 |
-
response = requests.get(url, params=params)
|
207 |
-
translated_text = response.json()[0][0][0]
|
208 |
-
return translated_text
|
209 |
-
except Exception as e:
|
210 |
-
print(f"한글 번역 오류: {str(e)}")
|
211 |
-
return text
|
212 |
-
|
213 |
-
def is_english(text):
|
214 |
-
return all(ord(char) < 128 for char in text.replace(' ', '').replace('-', '').replace('_', ''))
|
215 |
-
|
216 |
-
def is_korean(text):
|
217 |
-
return any('\uAC00' <= char <= '\uD7A3' for char in text)
|
218 |
-
|
219 |
-
def search_serphouse(query, country, page=1, num_result=10):
|
220 |
-
url = "https://api.serphouse.com/serp/live"
|
221 |
-
|
222 |
-
now = datetime.utcnow()
|
223 |
-
yesterday = now - timedelta(days=1)
|
224 |
-
date_range = f"{yesterday.strftime('%Y-%m-%d')},{now.strftime('%Y-%m-%d')}"
|
225 |
-
|
226 |
-
translated_query = translate_query(query, country)
|
227 |
-
print(f"Original query: {query}")
|
228 |
-
print(f"Translated query: {translated_query}")
|
229 |
-
|
230 |
-
payload = {
|
231 |
-
"data": {
|
232 |
-
"q": translated_query,
|
233 |
-
"domain": "google.com",
|
234 |
-
"loc": COUNTRY_LOCATIONS.get(country, "United States"),
|
235 |
-
"lang": COUNTRY_LANGUAGES.get(country, "en"),
|
236 |
-
"device": "desktop",
|
237 |
-
"serp_type": "news",
|
238 |
-
"page": "1",
|
239 |
-
"num": "10",
|
240 |
-
"date_range": date_range,
|
241 |
-
"sort_by": "date"
|
242 |
-
}
|
243 |
-
}
|
244 |
-
|
245 |
-
headers = {
|
246 |
-
"accept": "application/json",
|
247 |
-
"content-type": "application/json",
|
248 |
-
"authorization": f"Bearer {API_KEY}"
|
249 |
-
}
|
250 |
-
|
251 |
-
try:
|
252 |
-
response = requests.post(url, json=payload, headers=headers)
|
253 |
-
print("Request payload:", json.dumps(payload, indent=2, ensure_ascii=False))
|
254 |
-
print("Response status:", response.status_code)
|
255 |
-
|
256 |
-
response.raise_for_status()
|
257 |
-
return {"results": response.json(), "translated_query": translated_query}
|
258 |
-
except requests.RequestException as e:
|
259 |
-
return {"error": f"Error: {str(e)}", "translated_query": query}
|
260 |
-
|
261 |
-
def format_results_from_raw(response_data):
|
262 |
-
if "error" in response_data:
|
263 |
-
return "Error: " + response_data["error"], []
|
264 |
-
|
265 |
-
try:
|
266 |
-
results = response_data["results"]
|
267 |
-
translated_query = response_data["translated_query"]
|
268 |
-
|
269 |
-
news_results = results.get('results', {}).get('results', {}).get('news', [])
|
270 |
-
if not news_results:
|
271 |
-
return "검색 결과가 없습니다.", []
|
272 |
-
|
273 |
-
articles = []
|
274 |
-
for idx, result in enumerate(news_results, 1):
|
275 |
-
articles.append({
|
276 |
-
"index": idx,
|
277 |
-
"title": result.get("title", "제목 없음"),
|
278 |
-
"link": result.get("url", result.get("link", "#")),
|
279 |
-
"snippet": result.get("snippet", "내용 없음"),
|
280 |
-
"channel": result.get("channel", result.get("source", "알 수 없음")),
|
281 |
-
"time": result.get("time", result.get("date", "알 수 없는 시간")),
|
282 |
-
"image_url": result.get("img", result.get("thumbnail", "")),
|
283 |
-
"translated_query": translated_query
|
284 |
-
})
|
285 |
-
return "", articles
|
286 |
-
except Exception as e:
|
287 |
-
return f"결과 처리 중 오류 발생: {str(e)}", []
|
288 |
-
|
289 |
-
def serphouse_search(query, country):
|
290 |
-
response_data = search_serphouse(query, country)
|
291 |
-
return format_results_from_raw(response_data)
|
292 |
-
|
293 |
-
css = """
|
294 |
-
footer {visibility: hidden;}
|
295 |
-
"""
|
296 |
-
|
297 |
-
with gr.Blocks(theme="Nymbo/Nymbo_Theme", css=css, title="NewsAI 서비스") as iface:
|
298 |
-
gr.Markdown("검색어를 입력하고 원하는 국가(67개국)를 선택하면, 검색어와 일치하는 24시간 이내 뉴스를 최대 100개 출력합니다.")
|
299 |
-
gr.Markdown("국가 선택후 검색어에 '한글'을 입력하면 현지 언어로 번역되어 검색합니다. 예: 'Taiwan' 국가 선택후 '삼성' 입력시 '三星'으로 자동 검색")
|
300 |
-
|
301 |
-
with gr.Column():
|
302 |
-
with gr.Row():
|
303 |
-
query = gr.Textbox(label="검색어")
|
304 |
-
country = gr.Dropdown(MAJOR_COUNTRIES, label="국가", value="South Korea")
|
305 |
-
|
306 |
-
# 검색 상태 메시지
|
307 |
-
status_message = gr.Markdown("", visible=True)
|
308 |
-
|
309 |
-
# 번역 결과를 보여주는 컴포넌트
|
310 |
-
translated_query_display = gr.Markdown(visible=False)
|
311 |
-
|
312 |
-
search_button = gr.Button("검색", variant="primary")
|
313 |
-
|
314 |
-
progress = gr.Progress()
|
315 |
-
articles_state = gr.State([])
|
316 |
-
|
317 |
-
article_components = []
|
318 |
-
for i in range(100):
|
319 |
-
with gr.Group(visible=False) as article_group:
|
320 |
-
title = gr.Markdown()
|
321 |
-
image = gr.Image(width=200, height=150)
|
322 |
-
snippet = gr.Markdown()
|
323 |
-
info = gr.Markdown()
|
324 |
-
|
325 |
-
article_components.append({
|
326 |
-
'group': article_group,
|
327 |
-
'title': title,
|
328 |
-
'image': image,
|
329 |
-
'snippet': snippet,
|
330 |
-
'info': info,
|
331 |
-
'index': i,
|
332 |
-
})
|
333 |
-
|
334 |
-
def search_and_display(query, country, articles_state, progress=gr.Progress()):
|
335 |
-
# 검색 상태 메시지 업데이트
|
336 |
-
status_msg = "검색을 진행중입니다. 잠시만 기다리세요..."
|
337 |
-
|
338 |
-
progress(0, desc="검색어 번역 중...")
|
339 |
-
|
340 |
-
# 검색어 번역
|
341 |
-
translated_query = translate_query(query, country)
|
342 |
-
translated_display = f"**원본 검색어:** {query}\n**번역된 검색어:** {translated_query}" if translated_query != query else f"**검색어:** {query}"
|
343 |
-
|
344 |
-
progress(0.2, desc="검색 시작...")
|
345 |
-
error_message, articles = serphouse_search(query, country)
|
346 |
-
progress(0.5, desc="결과 처리 중...")
|
347 |
-
|
348 |
-
outputs = []
|
349 |
-
outputs.append(gr.update(value=status_msg, visible=True)) # 상태 메시지
|
350 |
-
outputs.append(gr.update(value=translated_display, visible=True)) # 번역 결과
|
351 |
-
|
352 |
-
if error_message:
|
353 |
-
outputs.append(gr.update(value=error_message, visible=True))
|
354 |
-
for comp in article_components:
|
355 |
-
outputs.extend([
|
356 |
-
gr.update(visible=False), gr.update(), gr.update(),
|
357 |
-
gr.update(), gr.update()
|
358 |
-
])
|
359 |
-
articles_state = []
|
360 |
-
else:
|
361 |
-
outputs.append(gr.update(value="", visible=False))
|
362 |
-
total_articles = len(articles)
|
363 |
-
for idx, comp in enumerate(article_components):
|
364 |
-
progress((idx + 1) / total_articles, desc=f"결과 표시 중... {idx + 1}/{total_articles}")
|
365 |
-
if idx < len(articles):
|
366 |
-
article = articles[idx]
|
367 |
-
image_url = article['image_url']
|
368 |
-
image_update = gr.update(value=image_url, visible=True) if image_url and not image_url.startswith('data:image') else gr.update(value=None, visible=False)
|
369 |
-
|
370 |
-
# 요약 내용 한글 번역
|
371 |
-
korean_summary = translate_to_korean(article['snippet'])
|
372 |
-
|
373 |
-
outputs.extend([
|
374 |
-
gr.update(visible=True),
|
375 |
-
gr.update(value=f"### [{article['title']}]({article['link']})"),
|
376 |
-
image_update,
|
377 |
-
gr.update(value=f"**요약:** {article['snippet']}\n\n**한글 요약:** {korean_summary}"),
|
378 |
-
gr.update(value=f"**출처:** {article['channel']} | **시간:** {article['time']}")
|
379 |
-
])
|
380 |
-
else:
|
381 |
-
outputs.extend([
|
382 |
-
gr.update(visible=False), gr.update(), gr.update(),
|
383 |
-
gr.update(), gr.update()
|
384 |
-
])
|
385 |
-
articles_state = articles
|
386 |
-
|
387 |
-
progress(1.0, desc="완료!")
|
388 |
-
outputs.append(articles_state)
|
389 |
-
|
390 |
-
# 검색 완료 후 상태 메시지 숨김
|
391 |
-
outputs[0] = gr.update(value="", visible=False)
|
392 |
-
|
393 |
-
return outputs
|
394 |
-
|
395 |
-
search_outputs = [
|
396 |
-
status_message,
|
397 |
-
translated_query_display,
|
398 |
-
gr.Markdown(visible=False)
|
399 |
-
]
|
400 |
-
|
401 |
-
for comp in article_components:
|
402 |
-
search_outputs.extend([
|
403 |
-
comp['group'], comp['title'], comp['image'],
|
404 |
-
comp['snippet'], comp['info']
|
405 |
-
])
|
406 |
-
search_outputs.append(articles_state)
|
407 |
-
|
408 |
-
search_button.click(
|
409 |
-
search_and_display,
|
410 |
-
inputs=[query, country, articles_state],
|
411 |
-
outputs=search_outputs,
|
412 |
-
show_progress=True
|
413 |
-
)
|
414 |
-
|
415 |
-
iface.launch(auth=("it1","chosun1"))
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|