Spaces:
Sleeping
Sleeping
File size: 28,536 Bytes
211685a 8ece039 1483b6e f88d400 190609e ef0c882 0203e2d c1fcc1b 5780309 c1fcc1b 5780309 c1fcc1b d15d37b c1fcc1b 211685a aa2ecd6 bd5cb51 d15d37b 5780309 d15d37b 5780309 1483b6e 9c59925 25f393b 74b563d f88d400 9c59925 f74acac f88d400 9c59925 74b563d f88d400 1483b6e 425f676 9c59925 f74acac d15d37b 425f676 0203e2d 425f676 0203e2d 74b563d d15d37b 74b563d 9c59925 f74acac 0203e2d 35c6f88 f74acac d15d37b 0203e2d 425f676 0203e2d f74acac d15d37b 0203e2d 425f676 35c6f88 f12be04 0203e2d c1fcc1b 425f676 5780309 d15d37b 425f676 0203e2d 43b28bf c1fcc1b 0203e2d 5780309 d15d37b 425f676 d15d37b f88d400 9c59925 0203e2d 60ca19e 0203e2d c1fcc1b 5780309 d15d37b 0203e2d 74b563d 0203e2d 74b563d 0203e2d f74acac d15d37b 35c6f88 f74acac 25f393b 74b563d 25f393b aa98710 43b28bf c1fcc1b 43b28bf f74acac d15d37b 43b28bf 25f393b 74b563d 25f393b 43b28bf f12be04 43b28bf aa98710 c1fcc1b 43b28bf f74acac d15d37b 35c6f88 f74acac aa2ecd6 f74acac aa2ecd6 f74acac f12be04 25f393b 74b563d f12be04 25f393b c1fcc1b 74b563d f12be04 25f393b f12be04 f74acac d15d37b f12be04 aa2ecd6 f12be04 aa2ecd6 f74acac f12be04 25f393b 74b563d f12be04 25f393b c1fcc1b 74b563d f12be04 25f393b f12be04 5780309 d15d37b 5780309 9c59925 d15d37b 5780309 9c59925 5780309 d15d37b 5780309 9c59925 5780309 9c59925 5780309 9c59925 5780309 f74acac 5780309 f74acac 5780309 74b563d 9c59925 5780309 9c59925 5780309 9c59925 5780309 9c59925 5780309 5f14908 f74acac 5f14908 f74acac 74b563d 5780309 6d93a47 35c6f88 6d93a47 35c6f88 2b4df81 6d93a47 2b4df81 35c6f88 6d93a47 c1fcc1b 74b563d 6d93a47 35c6f88 6d93a47 35c6f88 2b4df81 6d93a47 2b4df81 35c6f88 6d93a47 c1fcc1b aa2ecd6 5780309 aa2ecd6 5780309 bd5cb51 aa2ecd6 c1fcc1b 5780309 74b563d 5780309 9c59925 5780309 74b563d 13068e8 2b4df81 35c6f88 2b4df81 35c6f88 2b4df81 35c6f88 2b4df81 35c6f88 2b4df81 35c6f88 2b4df81 35c6f88 2b4df81 35c6f88 2b4df81 35c6f88 2b4df81 35c6f88 2b4df81 74b563d 35c6f88 6e53f81 c206255 2b4df81 13068e8 c206255 87563ed 2b4df81 c206255 2b4df81 87563ed 6e53f81 c724ab8 6e53f81 a7a6666 c724ab8 c206255 8b73ee4 a4c87fe 8b73ee4 bd5cb51 8b73ee4 6e53f81 8b73ee4 6e53f81 2340130 6e53f81 8b73ee4 c724ab8 6e53f81 c724ab8 6e53f81 c724ab8 8b73ee4 c724ab8 6e53f81 c724ab8 6e53f81 c724ab8 13068e8 8b73ee4 c724ab8 2340130 8b73ee4 d15d37b 5780309 8b73ee4 5780309 d15d37b 8b73ee4 5780309 8b73ee4 d15d37b 5780309 8b73ee4 5780309 9a199bb 1483b6e aa2ecd6 |
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 99 100 101 102 103 104 105 106 107 108 109 110 111 112 113 114 115 116 117 118 119 120 121 122 123 124 125 126 127 128 129 130 131 132 133 134 135 136 137 138 139 140 141 142 143 144 145 146 147 148 149 150 151 152 153 154 155 156 157 158 159 160 161 162 163 164 165 166 167 168 169 170 171 172 173 174 175 176 177 178 179 180 181 182 183 184 185 186 187 188 189 190 191 192 193 194 195 196 197 198 199 200 201 202 203 204 205 206 207 208 209 210 211 212 213 214 215 216 217 218 219 220 221 222 223 224 225 226 227 228 229 230 231 232 233 234 235 236 237 238 239 240 241 242 243 244 245 246 247 248 249 250 251 252 253 254 255 256 257 258 259 260 261 262 263 264 265 266 267 268 269 270 271 272 273 274 275 276 277 278 279 280 281 282 283 284 285 286 287 288 289 290 291 292 293 294 295 296 297 298 299 300 301 302 303 304 305 306 307 308 309 310 311 312 313 314 315 316 317 318 319 320 321 322 323 324 325 326 327 328 329 330 331 332 333 334 335 336 337 338 339 340 341 342 343 344 345 346 347 348 349 350 351 352 353 354 355 356 357 358 359 360 361 362 363 364 365 366 367 368 369 370 371 372 373 374 375 376 377 378 379 380 381 382 383 384 385 386 387 388 389 390 391 392 393 394 395 396 397 398 399 400 401 402 403 404 405 406 407 408 409 410 411 412 413 414 415 416 417 418 419 420 421 422 423 424 425 426 427 428 429 430 431 432 433 434 435 436 437 438 439 440 441 442 443 444 445 446 447 448 449 450 451 452 453 454 455 456 457 458 459 460 461 462 463 464 465 466 467 468 469 470 471 472 473 474 475 476 477 478 479 480 481 482 483 484 485 486 487 488 489 490 491 492 493 494 495 496 497 498 499 500 501 502 503 504 505 506 507 508 509 510 511 512 513 514 515 516 517 518 519 520 521 522 523 524 525 526 527 528 529 530 531 532 533 534 535 536 537 538 539 540 541 542 543 544 545 546 547 548 549 550 551 552 553 554 555 556 557 558 559 560 561 562 563 564 565 566 567 568 569 570 571 572 573 574 575 576 577 578 579 580 581 582 583 584 585 586 587 588 589 590 591 |
import gradio as gr
import pandas as pd
import openpyxl
import datetime
import tempfile
import os
from openpyxl.utils.dataframe import dataframe_to_rows
import openai
# ===================== OpenAI API ํด๋ผ์ด์ธํธ ์ค์ =====================
openai.api_key = os.getenv("OPENAI_API_KEY")
def call_api(content, system_message, max_tokens, temperature, top_p):
"""
ChatGPT API ํธ์ถ ํจ์.
"""
response = openai.ChatCompletion.create(
model="gpt-4o-mini",
messages=[
{"role": "system", "content": system_message},
{"role": "user", "content": content}
],
max_tokens=max_tokens,
temperature=temperature,
top_p=top_p
)
return response.choices[0].message['content']
def respond_gemini_qna(question, system_message, max_tokens, temperature, top_p, model_id):
"""
Gemini Flash ๋ชจ๋ธ์ ์ด์ฉํด ์ง๋ฌธ(question)์ ๋ํ ๋ต๋ณ์ ๋ฐํํ๋ ํจ์.
"""
try:
import google.generativeai as genai
except ModuleNotFoundError:
return ("์ค๋ฅ๊ฐ ๋ฐ์ํ์ต๋๋ค: 'google-generativeai' ๋ชจ๋์ ์ฐพ์ ์ ์์ต๋๋ค. "
"ํด๊ฒฐ ๋ฐฉ๋ฒ: 'pip install --upgrade google-generativeai' ๋ฅผ ์คํํ์ฌ ์ค์นํด์ฃผ์ธ์.")
gemini_api_key = os.getenv("GEMINI_API_KEY")
if not gemini_api_key:
return "Gemini API ํ ํฐ์ด ํ์ํฉ๋๋ค."
genai.configure(api_key=gemini_api_key)
prompt = f"{system_message}\n\n{question}"
try:
model = genai.GenerativeModel(model_name=model_id)
response = model.generate_content(prompt)
return response.text
except Exception as e:
return f"์ค๋ฅ๊ฐ ๋ฐ์ํ์ต๋๋ค: {str(e)}"
def respond_o1mini_qna(question, system_message, max_tokens, temperature):
"""
o1-mini ๋ชจ๋ธ์ ์ด์ฉํด ํ ๋ฒ์ ์ง๋ฌธ(question)์ ๋ํ ๋ต๋ณ์ ๋ฐํํ๋ ํจ์.
o1-mini์์๋ 'system' ๋ฉ์์ง๋ฅผ ์ง์ํ์ง ์์ผ๋ฏ๋ก system_message์ question์ ํ๋์ 'user' ๋ฉ์์ง๋ก ํฉ์ณ ์ ๋ฌํฉ๋๋ค.
๋ํ, o1-mini์์๋ 'max_tokens' ๋์ 'max_completion_tokens'๋ฅผ ์ฌ์ฉํ๋ฉฐ, temperature๋ ๊ณ ์ ๊ฐ 1๋ง ์ง์ํฉ๋๋ค.
"""
openai_token = os.getenv("OPENAI_API_KEY")
if not openai_token:
return "OpenAI API ํ ํฐ์ด ํ์ํฉ๋๋ค."
openai.api_key = openai_token
combined_message = f"{system_message}\n\n{question}"
messages = [{"role": "user", "content": combined_message}]
try:
response = openai.ChatCompletion.create(
model="o1-mini",
messages=messages,
max_completion_tokens=max_tokens,
temperature=1, # ๊ณ ์ ๋ ๊ฐ 1 ์ฌ์ฉ
)
assistant_message = response.choices[0].message['content']
return assistant_message
except Exception as e:
return f"์ค๋ฅ๊ฐ ๋ฐ์ํ์ต๋๋ค: {str(e)}"
# ===================== (1) ๋ฆฌ๋ทฐ์ต์
๋ถ์: ๋ฏธ์
1~9 ์ฒ๋ฆฌ =====================
def analyze_options(uploaded_file, selected_year, llm_model_choice):
"""
์
๋ก๋๋ ํ์ผ๋ก๋ถํฐ ๋ฏธ์
1~9(์ต์
/ํ์ ์ง๊ณ)๋ฅผ ์ํํ๊ณ ,
์์ ์์
ํ์ผ๊ณผ ์ ํ๋
๋ Top20 ์ต์
๋ฆฌ์คํธ(์ ์ฒด์ต์
๋ถ์ ํฌํจ)๋ฅผ ๋ฐํํ๋ค.
"""
if uploaded_file is None:
return None, ["์ ์ฒด์ต์
๋ถ์"]
try:
df_upload = pd.read_excel(uploaded_file)
except Exception as e:
return None, ["์ ์ฒด์ต์
๋ถ์"]
template_file = "๋ฆฌ๋ทฐ๋ถ์.ver1.0.xlsx"
if not os.path.exists(template_file):
return None, ["์ ์ฒด์ต์
๋ถ์"]
try:
wb = openpyxl.load_workbook(template_file)
except Exception as e:
return None, ["์ ์ฒด์ต์
๋ถ์"]
# (๋ฏธ์
1) ์๋ณธ๋ฐ์ดํฐ ์ํธ ์์ฑ
if "์๋ณธ๋ฐ์ดํฐ" in wb.sheetnames:
ws_source = wb["์๋ณธ๋ฐ์ดํฐ"]
wb.remove(ws_source)
ws_source = wb.create_sheet("์๋ณธ๋ฐ์ดํฐ")
else:
ws_source = wb.create_sheet("์๋ณธ๋ฐ์ดํฐ")
for r_idx, row in enumerate(dataframe_to_rows(df_upload, index=False, header=True), start=1):
for c_idx, value in enumerate(row, start=1):
ws_source.cell(row=r_idx, column=c_idx, value=value)
# ๋ฆฌ๋ทฐ ๋ ์ง ์ฒ๋ฆฌ
try:
df_upload['๋ฆฌ๋ทฐ๋ ์ง'] = pd.to_datetime(df_upload.iloc[:, 1], errors='coerce')
df_upload['๋ฆฌ๋ทฐ๋ ์ง'] = df_upload['๋ฆฌ๋ทฐ๋ ์ง'].apply(
lambda d: d.replace(tzinfo=None) if pd.notnull(d) and d.tzinfo is not None else d
)
df_valid = df_upload.dropna(subset=['๋ฆฌ๋ทฐ๋ ์ง']).copy()
except Exception as e:
return None, ["์ ์ฒด์ต์
๋ถ์"]
now = datetime.datetime.now()
current_year = now.year
start_year_val = current_year - 2
end_year_val = current_year
# ๋์๋ณด๋๋ฐ์ดํฐ ์ํธ
if "๋์๋ณด๋๋ฐ์ดํฐ" in wb.sheetnames:
ws_dashboard = wb["๋์๋ณด๋๋ฐ์ดํฐ"]
else:
ws_dashboard = wb.create_sheet("๋์๋ณด๋๋ฐ์ดํฐ")
# (๋ฏธ์
2) ์ต๊ทผ 3๋
'๋
์' ์ถ์ด
ws_dashboard["C2"] = str(current_year)[-2:]
row_idx = 5
for year in range(start_year_val, end_year_val + 1):
for month in range(1, 13):
count = df_valid[(df_valid['๋ฆฌ๋ทฐ๋ ์ง'].dt.year == year) & (df_valid['๋ฆฌ๋ทฐ๋ ์ง'].dt.month == month)].shape[0]
ws_dashboard.cell(row=row_idx, column=3, value=count)
row_idx += 1
# (๋ฏธ์
3) ์ต๊ทผ 3๋
'๋
๋' ์ถ์ด
ws_dashboard["E2"] = str(current_year)[-2:]
row_year = 5
for y in [current_year, current_year - 1, current_year - 2]:
count = df_valid[df_valid['๋ฆฌ๋ทฐ๋ ์ง'].dt.year == y].shape[0]
ws_dashboard.cell(row=row_year, column=6, value=count)
row_year += 1
# (๋ฏธ์
4) ์ ํ๋ ๋
๋ '์๋ณ' ์ถ์ด
try:
selected_year_int = int("20" + selected_year[:2])
except Exception as e:
return None, ["์ ์ฒด์ต์
๋ถ์"]
ws_dashboard["I2"] = selected_year[:2]
for month in range(1, 13):
count = df_valid[(df_valid['๋ฆฌ๋ทฐ๋ ์ง'].dt.year == selected_year_int) & (df_valid['๋ฆฌ๋ทฐ๋ ์ง'].dt.month == month)].shape[0]
ws_dashboard.cell(row=4 + month, column=9, value=count)
# (๋ฏธ์
5) ์ ํ๋ ๋
๋ '์์ผ' ์ถ์ด
start_date = datetime.date(selected_year_int, 1, 1)
end_date = datetime.date(selected_year_int, 12, 31)
df_selected = df_valid[df_valid['๋ฆฌ๋ทฐ๋ ์ง'].dt.year == selected_year_int].copy()
df_selected['๋ฆฌ๋ทฐ๋ ์ง_์ผ'] = df_selected['๋ฆฌ๋ทฐ๋ ์ง'].dt.date
daily_counts = df_selected.groupby('๋ฆฌ๋ทฐ๋ ์ง_์ผ').size().to_dict()
row_day = 5
current_day = start_date
while current_day <= end_date:
ws_dashboard.cell(row=row_day, column=12, value=daily_counts.get(current_day, 0))
row_day += 1
current_day += datetime.timedelta(days=1)
# (๋ฏธ์
6) ์ต๊ทผ 3๋
์ ์ฒด์ต์
๋ฐ์ดํฐ ์
๋ ฅ
df_recent = df_valid[(df_valid['๋ฆฌ๋ทฐ๋ ์ง'].dt.year >= start_year_val) & (df_valid['๋ฆฌ๋ทฐ๋ ์ง'].dt.year <= end_year_val)].copy()
if df_recent.shape[1] >= 3:
df_recent['์ต์
์๋ณธ'] = df_recent.iloc[:, 2].astype(str).fillna('')
def get_first_option(opt_str):
return opt_str.split(' / ')[0].strip()
df_recent['์ต์
1'] = df_recent['์ต์
์๋ณธ'].apply(get_first_option)
option_counts_3y = df_recent['์ต์
1'].value_counts()
top20_3y = option_counts_3y.head(20)
sum_others_3y = option_counts_3y.iloc[20:].sum() if len(option_counts_3y) > 20 else 0
row_opt = 5
for opt_name, cnt in top20_3y.items():
ws_dashboard.cell(row=row_opt, column=14, value=opt_name)
ws_dashboard.cell(row=row_opt, column=15, value=cnt)
row_opt += 1
ws_dashboard.cell(row=25, column=15, value=sum_others_3y)
else:
ws_dashboard.cell(row=5, column=14, value="๊ตฌ๋งค์ต์
์ด์ด ์์ต๋๋ค.")
ws_dashboard.cell(row=5, column=15, value=0)
# (๋ฏธ์
7) ์ ํ๋
๋ ๊ธฐ์ค ์ต์
๋ฐ์ดํฐ (๋ด๋ฆผ์ฐจ์)
df_selected_year = df_valid[df_valid['๋ฆฌ๋ทฐ๋ ์ง'].dt.year == selected_year_int].copy()
if df_selected_year.shape[1] >= 3:
df_selected_year['์ต์
์๋ณธ'] = df_selected_year.iloc[:, 2].astype(str).fillna('')
def get_first_option(opt_str):
return opt_str.split(' / ')[0].strip()
df_selected_year['์ต์
1'] = df_selected_year['์ต์
์๋ณธ'].apply(get_first_option)
option_counts_selected = df_selected_year['์ต์
1'].value_counts()
top20_selected = option_counts_selected.head(20)
sum_others_selected = option_counts_selected.iloc[20:].sum() if len(option_counts_selected) > 20 else 0
row_opt = 5
for opt_name, cnt in top20_selected.items():
ws_dashboard.cell(row=row_opt, column=18, value=opt_name)
ws_dashboard.cell(row=row_opt, column=19, value=cnt)
row_opt += 1
ws_dashboard.cell(row=25, column=19, value=sum_others_selected)
else:
ws_dashboard.cell(row=5, column=18, value="๊ตฌ๋งค์ต์
์ด์ด ์์ต๋๋ค.")
ws_dashboard.cell(row=5, column=19, value=0)
# (๋ฏธ์
8) ์ต๊ทผ3๋
ํ์ ํํฉ - top 20 ์ต์
df_recent_score = df_valid[(df_valid['๋ฆฌ๋ทฐ๋ ์ง'].dt.year >= start_year_val) & (df_valid['๋ฆฌ๋ทฐ๋ ์ง'].dt.year <= end_year_val)].copy()
if df_recent_score.shape[1] >= 5:
df_recent_score['์ต์
์๋ณธ'] = df_recent_score.iloc[:, 2].astype(str).fillna('')
def get_first_option(opt_str):
return opt_str.split(' / ')[0].strip()
df_recent_score['์ต์
1'] = df_recent_score['์ต์
์๋ณธ'].apply(get_first_option)
df_recent_score['ํ์ '] = pd.to_numeric(df_recent_score.iloc[:, 4], errors='coerce')
df_recent_score = df_recent_score.dropna(subset=['ํ์ '])
group_8 = df_recent_score.groupby(['์ต์
1', 'ํ์ ']).size().unstack(fill_value=0)
group_8 = group_8.reindex(columns=[5,4,3,2,1], fill_value=0)
group_8['total_count'] = group_8.sum(axis=1)
group_8 = group_8.sort_values(by='total_count', ascending=False)
top20_8 = group_8.head(20).copy()
others_8 = group_8.iloc[20:].copy()
sum_others_5 = others_8[5].sum()
sum_others_4 = others_8[4].sum()
sum_others_3 = others_8[3].sum()
sum_others_2 = others_8[2].sum()
sum_others_1 = others_8[1].sum()
row_v = 5
for opt_name, row_data in top20_8.iterrows():
ws_dashboard.cell(row=row_v, column=22, value=opt_name)
ws_dashboard.cell(row=row_v, column=24, value=row_data[5])
ws_dashboard.cell(row=row_v, column=25, value=row_data[4])
ws_dashboard.cell(row=row_v, column=26, value=row_data[3])
ws_dashboard.cell(row=row_v, column=27, value=row_data[2])
ws_dashboard.cell(row=row_v, column=28, value=row_data[1])
row_v += 1
ws_dashboard.cell(row=25, column=24, value=sum_others_5)
ws_dashboard.cell(row=25, column=25, value=sum_others_4)
ws_dashboard.cell(row=25, column=26, value=sum_others_3)
ws_dashboard.cell(row=25, column=27, value=sum_others_2)
ws_dashboard.cell(row=25, column=28, value=sum_others_1)
else:
ws_dashboard.cell(row=5, column=22, value="์ต์
๋๋ ํ์ ์ด์ด ๋ถ์กฑํฉ๋๋ค.")
# (๋ฏธ์
9) ์ ํ๋
๋ ํ์ ํํฉ - top 20 ์ต์
df_selected_score = df_valid[df_valid['๋ฆฌ๋ทฐ๋ ์ง'].dt.year == selected_year_int].copy()
if df_selected_score.shape[1] >= 5:
df_selected_score['์ต์
์๋ณธ'] = df_selected_score.iloc[:, 2].astype(str).fillna('')
def get_first_option(opt_str):
return opt_str.split(' / ')[0].strip()
df_selected_score['์ต์
1'] = df_selected_score['์ต์
์๋ณธ'].apply(get_first_option)
df_selected_score['ํ์ '] = pd.to_numeric(df_selected_score.iloc[:, 4], errors='coerce')
df_selected_score = df_selected_score.dropna(subset=['ํ์ '])
group_9 = df_selected_score.groupby(['์ต์
1', 'ํ์ ']).size().unstack(fill_value=0)
group_9 = group_9.reindex(columns=[5,4,3,2,1], fill_value=0)
group_9['total_count'] = group_9.sum(axis=1)
group_9 = group_9.sort_values(by='total_count', ascending=False)
top20_9 = group_9.head(20).copy()
others_9 = group_9.iloc[20:].copy()
sum_others_5 = others_9[5].sum()
sum_others_4 = others_9[4].sum()
sum_others_3 = others_9[3].sum()
sum_others_2 = others_9[2].sum()
sum_others_1 = others_9[1].sum()
row_ad = 5
for opt_name, row_data in top20_9.iterrows():
ws_dashboard.cell(row=row_ad, column=30, value=opt_name)
ws_dashboard.cell(row=row_ad, column=32, value=row_data[5])
ws_dashboard.cell(row=row_ad, column=33, value=row_data[4])
ws_dashboard.cell(row=row_ad, column=34, value=row_data[3])
ws_dashboard.cell(row=row_ad, column=35, value=row_data[2])
ws_dashboard.cell(row=row_ad, column=36, value=row_data[1])
row_ad += 1
ws_dashboard.cell(row=25, column=32, value=sum_others_5)
ws_dashboard.cell(row=25, column=33, value=sum_others_4)
ws_dashboard.cell(row=25, column=34, value=sum_others_3)
ws_dashboard.cell(row=25, column=35, value=sum_others_2)
ws_dashboard.cell(row=25, column=36, value=sum_others_1)
else:
ws_dashboard.cell(row=5, column=30, value="์ต์
๋๋ ํ์ ์ด์ด ๋ถ์กฑํฉ๋๋ค.")
# ์์ ์์
ํ์ผ ์ ์ฅ
temp_file = tempfile.NamedTemporaryFile(delete=False, suffix=".xlsx")
wb.save(temp_file.name)
temp_file.close()
# ์ ํ๋
๋ Top20 ์ต์
๋ฆฌ์คํธ ์์ฑ (์ ์ฒด์ต์
๋ถ์ ํฌํจ)
top20_option_list = ["์ ์ฒด์ต์
๋ถ์"]
if df_selected_year.shape[1] >= 3:
option_counts_selected = df_selected_year['์ต์
1'].value_counts().head(20)
for opt_name, cnt in option_counts_selected.items():
top20_option_list.append(f"{opt_name}({cnt})")
return temp_file.name, top20_option_list
# ===================== (2) ๋ฆฌ๋ทฐ๋ถ์: ๋ฏธ์
10~12 ์คํ =====================
def analyze_reviews(partial_file, selected_option, llm_model_choice):
"""
'๋ฆฌ๋ทฐ์ต์
๋ถ์' ๊ฒฐ๊ณผ ์์
๊ณผ ์ ํ ์ต์
(๋๋ ์ ์ฒด์ต์
๋ถ์)์ ๋ฐํ์ผ๋ก
๋ฏธ์
10~12(์ฃผ์๊ธ์ /๋ถ์ ๋ฆฌ๋ทฐ ์ถ์ถ + LLM ๋ถ์)๋ฅผ ์ํํ๊ณ ,
์ต์ข
์์
ํ์ผ๊ณผ ํ
์คํธ ๋ถ์ ๊ฒฐ๊ณผ๋ฅผ ๋ฐํํ๋ค.
"""
if partial_file is None:
return None, "", "", "", ""
try:
wb = openpyxl.load_workbook(partial_file)
except Exception as e:
return None, "", "", "", ""
if "์๋ณธ๋ฐ์ดํฐ" not in wb.sheetnames:
return None, "", "", "", ""
ws_source = wb["์๋ณธ๋ฐ์ดํฐ"]
data = ws_source.values
cols = next(data)
df_upload = pd.DataFrame(data, columns=cols)
# (๋ฏธ์
10) ์ฃผ์ ๋ฆฌ๋ทฐ ๋ด์ฉ ์ํธ ์์ฑ
if "์ฃผ์ ๋ฆฌ๋ทฐ ๋ด์ฉ" in wb.sheetnames:
ws_review = wb["์ฃผ์ ๋ฆฌ๋ทฐ ๋ด์ฉ"]
wb.remove(ws_review)
ws_review = wb.create_sheet("์ฃผ์ ๋ฆฌ๋ทฐ ๋ด์ฉ")
else:
ws_review = wb.create_sheet("์ฃผ์ ๋ฆฌ๋ทฐ ๋ด์ฉ")
if df_upload.shape[1] < 5:
ws_review.cell(row=1, column=1, value="๋ฆฌ๋ทฐ๋ด์ฉ ๋ฐ ํ์ ์ด์ด ๋ถ์กฑํฉ๋๋ค.")
return None, "", "", "", ""
df_upload['์ต์
์๋ณธ'] = df_upload.iloc[:, 2].astype(str).fillna('')
if selected_option == "์ ์ฒด์ต์
๋ถ์":
df_filtered = df_upload.copy()
else:
selected_opt_name = selected_option.rsplit("(", 1)[0].strip()
def get_first_option(opt_str):
return opt_str.split(' / ')[0].strip()
df_upload['์ต์
1'] = df_upload['์ต์
์๋ณธ'].apply(get_first_option)
df_filtered = df_upload[df_upload['์ต์
1'] == selected_opt_name].copy()
# (๋ฏธ์
10) ์ฃผ์ ๋ฆฌ๋ทฐ ๋ด์ฉ(ํ์ =3 ์ ์ธ) + ๊ธ์์ ์ด ์์ฑ โ ๋ด๋ฆผ์ฐจ์ ์ ๋ ฌ
df_review = df_filtered.copy()
df_review = df_review[df_review.iloc[:,4] != 3]
df_review['๊ธ์์'] = df_review.iloc[:,3].astype(str).apply(len)
df_review = df_review.sort_values(by='๊ธ์์', ascending=False)
for r_idx, row in enumerate(dataframe_to_rows(df_review, index=False, header=True), start=1):
for c_idx, value in enumerate(row, start=1):
ws_review.cell(row=r_idx, column=c_idx, value=value)
# (๋ฏธ์
11) ๊ธ์ /๋ถ์ ๋ฆฌ๋ทฐ ์ถ์ถ
df_positive = df_review[(df_review.iloc[:,4] == 5) & (df_review['๊ธ์์'] < 500)].copy()
if df_positive.shape[0] < 20:
df_positive_4 = df_review[(df_review.iloc[:,4] == 4) & (df_review['๊ธ์์'] < 500)]
df_positive = pd.concat([df_positive, df_positive_4])
df_positive = df_positive.head(20)
positive_reviews = ""
for idx, row in df_positive.iterrows():
positive_reviews += f"์์ด๋: {row.iloc[0]}, ์ ์: {row.iloc[4]}, ๊ธ์์: {row['๊ธ์์']}\n๋ฆฌ๋ทฐ: {row.iloc[3]}\n\n"
df_negative = df_review[(df_review.iloc[:,4] == 1) & (df_review['๊ธ์์'] < 500)].copy()
if df_negative.shape[0] < 30:
df_negative_2 = df_review[(df_review.iloc[:,4] == 2) & (df_review['๊ธ์์'] < 500)]
df_negative = pd.concat([df_negative, df_negative_2])
df_negative = df_negative.head(30)
negative_reviews = ""
for idx, row in df_negative.iterrows():
negative_reviews += f"์์ด๋: {row.iloc[0]}, ์ ์: {row.iloc[4]}, ๊ธ์์: {row['๊ธ์์']}\n๋ฆฌ๋ทฐ: {row.iloc[3]}\n\n"
# (๋ฏธ์
12) LLM์ ์ด์ฉํ ๋ฆฌ๋ทฐ ๋ถ์
positive_system_msg = (
"๐ ๊ธ์ ๋ฆฌ๋ทฐ ๋ถ์:\n"
"๋๋ ๋ฆฌ๋ทฐ ๋ฐ์ดํฐ๋ฅผ ๋ถ์ํ๋ ๋น
๋ฐ์ดํฐ ๋ถ์๊ฐ์ด๋ค. ๊ณ ๊ฐ์ ๋ฆฌ๋ทฐ ๋ฐ์ดํฐ๋ฅผ ๋ฐํ์ผ๋ก ๊ธ์ ์ ์ธ ์๊ฒฌ๋ง์ ๋ถ์ํด๋ผ. ๋ฐ๋์ ์ ๊ณต๋ ๋ฆฌ๋ทฐ ๋ฐ์ดํฐ์์๋ง ๋ถ์ํ๋ฉฐ, ๋์ ์๊ฐ์ ํฌํจํ์ง ๋ง ๊ฒ.\n"
"[๋ถ์ ์กฐ๊ฑด]\n"
"- ๋ฐ๋์ ๊ธ์ ์ ์ธ ์๊ฒฌ๋ง ๋ถ์ํ๊ณ , ๋ถ์ ์ ์ธ ์๊ฒฌ์ ์ ์ธํ ๊ฒ.\n"
"- ๊ธฐ๋ฅ ๋ฐ ์ฑ๋ฅ, ๊ฐ์ฑ, ์ค์ ์ฌ์ฉ, ๋ฐฐ์ก, ํ๊ฒ๋ณ ๊ด์ ์ผ๋ก ๋ถ์ํ ๊ฒ.\n"
"- ๋ง์ผํ
์ ํ์ฉํ ์ ์๋ ๊ณ ๊ฐ์ ์ค์ ๋ฆฌ๋ทฐ ๋จ์ด๋ฅผ ๋ฐ๋์ ํฌํจํ ๊ฒ.\n"
"[์ถ๋ ฅ ํํ]\n"
"- ๊ฐ๊ฐ์ ์ ๋ชฉ ์์๋ '๐' ์ด๋ชจ์ง๋ฅผ ์ฌ์ฉํ๋ฉฐ, '#'๋ '##'์ ์ฌ์ฉํ์ง ๋ง ๊ฒ.\n"
"- ๊ฐ์ฅ ๋ง์ง๋ง์๋ \"๐์ข
ํฉ์๊ฒฌ\"์ด๋ผ๋ ์ ๋ชฉ์ผ๋ก ์ข
ํฉ ์๊ฒฌ์ ์์ฑํ๋ผ.\n"
" - ์ข
ํฉ์๊ฒฌ์๋ ํญ๋ชฉ๋ณ ์ ๋ชฉ์ ์ ์ธํ๊ณ ์์ ์ ๋ฌธ์ฅ์ผ๋ก ์์ฑํ ๊ฒ.\n"
" - ์ด์ด์ '๐น ๊ฐ์ '๊ณผ '๐น ๊ธฐํ' ์ ๋ชฉ์ผ๋ก SWOT ๋ถ์ ๊ฒฐ๊ณผ๋ฅผ ์ ๊ณตํ ๊ฒ.\n"
"- ์ค์ ๊ณ ๊ฐ์ ๋ฆฌ๋ทฐ์์ ์ฌ์ฉ๋ ๋จ์ด๋ฅผ ๋ฐ๋์ ํฌํจํ ๊ฒ."
)
positive_user_msg = "๋ค์ ๋ฆฌ๋ทฐ๋ฅผ ๋ถ์ํด ์ฃผ์ธ์:\n" + positive_reviews
negative_system_msg = (
"๐ ๋ถ์ ๋ฆฌ๋ทฐ ๋ถ์:\n"
"๋๋ ๋ฆฌ๋ทฐ ๋ฐ์ดํฐ๋ฅผ ๋ถ์ํ๋ ๋น
๋ฐ์ดํฐ ๋ถ์๊ฐ์ด๋ค. ๊ณ ๊ฐ์ ๋ฆฌ๋ทฐ ๋ฐ์ดํฐ๋ฅผ ๋ฐํ์ผ๋ก ๋ถ์ ์ ์ธ ์๊ฒฌ๋ง์ ๋ถ์ํด๋ผ. ๋ฐ๋์ ์ ๊ณต๋ ๋ฆฌ๋ทฐ ๋ฐ์ดํฐ์์๋ง ๋ถ์ํ๋ฉฐ, ๋์ ์๊ฐ์ ํฌํจํ์ง ๋ง ๊ฒ.\n"
"[๋ถ์ ์กฐ๊ฑด]\n"
"- ๋ถ์ ์ ์ธ ์๊ฒฌ๋ง ๋ถ์ํ๊ณ , ๊ธ์ ์ ์ธ ์๊ฒฌ์ ์ ์ธํ ๊ฒ.\n"
"- ๊ธฐ๋ฅ ๋ฐ ์ฑ๋ฅ, ๊ฐ์ฑ, ์ค์ ์ฌ์ฉ, ๋ฐฐ์ก, ๊ณ ๊ฐ์ ๋ถ๋
ธ ๊ด์ ์ผ๋ก ๋ถ์ํ ๊ฒ.\n"
"- ๋ถ์ ์ ์ธ ๋ฆฌ๋ทฐ ๋ถ์ ๊ฒฐ๊ณผ๋ฅผ ๋ฐํ์ผ๋ก '๊ฐ์ ํ ์ '์ ์ถ๋ ฅํ ๊ฒ.\n"
"[์ถ๋ ฅ ํํ]\n"
"- ๊ฐ๊ฐ์ ์ ๋ชฉ ์์๋ '๐' ์ด๋ชจ์ง๋ฅผ ์ฌ์ฉํ๋ฉฐ, '#'๋ '##'์ ์ฌ์ฉํ์ง ๋ง ๊ฒ.\n"
"- ๊ฐ์ฅ ๋ง์ง๋ง์๋ \"๐ข๊ฐ์ ํ ์ \"์ด๋ผ๋ ์ ๋ชฉ์ผ๋ก ๊ฐ์ ํ ์ ์ ์์ฑํ๋ผ.\n"
" - ๊ฐ์ ํ ์ ์๋ ํญ๋ชฉ๋ณ ์ ๋ชฉ์ ์ ์ธํ๊ณ ์์ ์ ๋ฌธ์ฅ์ผ๋ก ์์ฑํ ๊ฒ.\n"
" - ์ด์ด์ '๐ ์ฝ์ '๊ณผ '๐ ์ํ' ์ ๋ชฉ์ผ๋ก SWOT ๋ถ์ ๊ฒฐ๊ณผ๋ฅผ ์ ๊ณตํ ๊ฒ.\n"
"- ์ค์ ๊ณ ๊ฐ์ ๋ฆฌ๋ทฐ์์ ์ฌ์ฉ๋ ๋จ์ด๋ฅผ ๋ฐ๋์ ํฌํจํ ๊ฒ."
)
negative_user_msg = "๋ค์ ๋ฆฌ๋ทฐ๋ฅผ ๋ถ์ํด ์ฃผ์ธ์:\n" + negative_reviews
if llm_model_choice == "ChatGPT (gpt-4o-mini)":
positive_analysis = call_api(
positive_user_msg, positive_system_msg,
max_tokens=15000, temperature=0.3, top_p=0.95
)
negative_analysis = call_api(
negative_user_msg, negative_system_msg,
max_tokens=15000, temperature=0.3, top_p=0.95
)
elif llm_model_choice == "Gemini Flash (gemini-2.0-flash)":
positive_analysis = respond_gemini_qna(
positive_user_msg, positive_system_msg,
max_tokens=15000, temperature=0.3, top_p=0.95,
model_id="gemini-2.0-flash"
)
negative_analysis = respond_gemini_qna(
negative_user_msg, negative_system_msg,
max_tokens=15000, temperature=0.3, top_p=0.95,
model_id="gemini-2.0-flash"
)
elif llm_model_choice == "o1-mini":
positive_analysis = respond_o1mini_qna(
positive_user_msg, positive_system_msg,
max_tokens=15000, temperature=1
)
negative_analysis = respond_o1mini_qna(
negative_user_msg, negative_system_msg,
max_tokens=15000, temperature=1
)
else:
positive_analysis = "LLM ๋ชจ๋ธ ์ ํ ์ค๋ฅ"
negative_analysis = "LLM ๋ชจ๋ธ ์ ํ ์ค๋ฅ"
if "๋์๋ณด๋๋ฐ์ดํฐ" in wb.sheetnames:
ws_dashboard = wb["๋์๋ณด๋๋ฐ์ดํฐ"]
ws_dashboard["AL5"] = positive_analysis
ws_dashboard["AM5"] = negative_analysis
final_file = tempfile.NamedTemporaryFile(delete=False, suffix=".xlsx")
wb.save(final_file.name)
final_file.close()
return final_file.name, positive_reviews, negative_reviews, positive_analysis, negative_analysis
# ===================== Gradio UI ๊ตฌ์ฑ (HTML/CSS ์ปค์คํฐ๋ง์ด์ง ์ ์ฉ) =====================
custom_css = """
body {
background: #f7f7f7;
font-family: 'Arial', sans-serif;
}
.gradio-container {
padding: 20px;
}
.custom-header {
text-align: center;
font-size: 36px;
font-weight: bold;
margin-bottom: 20px;
color: #333;
}
.custom-frame {
border: 2px solid #ccc;
border-radius: 20px;
padding: 20px;
margin: 10px 0;
background-color: #fff;
}
.custom-button {
border-radius: 20px !important;
background: #4caf50 !important;
color: white !important;
font-size: 20px !important;
padding: 10px 20px !important;
}
.custom-title {
font-size: 28px;
font-weight: bold;
margin-bottom: 10px;
color: #555;
}
.custom-subtitle {
font-size: 22px;
font-weight: bold;
margin-bottom: 8px;
color: #666;
}
.instructions {
background-color: #e0f7fa;
border: 2px dashed #00796b;
border-radius: 15px;
padding: 15px;
font-size: 18px;
color: #004d40;
margin-bottom: 20px;
text-align: left;
}
"""
# ๊ฐ๋จํ๊ณ ์ข์ธก ์ ๋ ฌ๋ ๋จ๊ณ๋ณ ์ฌ์ฉ์ค๋ช
์ (๋จผ์ ์ ์)
usage_instructions = """
<div class="instructions">
<p style="font-size:24px; font-weight:bold;">๐ ์ฌ์ฉ์ค๋ช
์</p>
<p style="font-size:16px;">
1๋จ๊ณ: ์ข์ธก "๐ ๋ฐ์ดํฐ ์
๋ ฅ"์์ ํ์ผ ์
๋ก๋์ ๋ถ์๋
๋ ์ ํ ํ, ๋ฆฌ๋ทฐ์ต์
๋ถ์ ๋ฒํผ ํด๋ฆญ<br>
2๋จ๊ณ: ํ๋จ "๐ ๋ฆฌ๋ทฐ๋ถ์ ์ต์
์ ํ"์์ LLM ๋ชจ๋ธ๊ณผ ์์ดํ
์ต์
์ ํ ํ, ๋ฆฌ๋ทฐ๋ถ์ ์ต์
์ ํ ๋ฒํผ ํด๋ฆญ<br>
3๋จ๊ณ: ์ฐ์ธก "๐๋ถ์๋ณด๊ณ ์ ๋ค์ด๋ก๋"์์ ๋ณด๊ณ ์๋ฅผ ๋ค์ด๋ก๋ํ๊ณ ๊ฒฐ๊ณผ ํ์ธ<br>
</p>
</div>
"""
with gr.Blocks(css=custom_css, title="๋ฆฌ๋ทฐ ๋ถ์ ์๋น์ค") as demo:
gr.HTML("<div class='custom-header'>๐ ๊ณ ๊ฐ ๋ฆฌ๋ทฐ ๋ถ์ ์๋น์ค ๐</div>")
gr.HTML(usage_instructions)
# [๋ฐ์ดํฐ ์
๋ ฅ ๋ฐ ๋ถ์๋ณด๊ณ ์ ๋ค์ด๋ก๋ ํ๋ ์] (์ข/์ฐ ๋ฐฐ์น)
with gr.Row():
with gr.Column(elem_classes="custom-frame"):
gr.HTML("<div class='custom-title'>๐ ๋ฐ์ดํฐ ์
๋ ฅ</div>")
file_input = gr.File(label="์๋ณธ ์์
ํ์ผ ์
๋ก๋", file_types=[".xlsx"])
year_radio = gr.Radio(
choices=[f"{str(y)[-2:]}๋
" for y in range(datetime.datetime.now().year, datetime.datetime.now().year-5, -1)],
label="๋ถ์๋
๋ ์ ํ",
value=f"{str(datetime.datetime.now().year)[-2:]}๋
"
)
analyze_button = gr.Button("๋ฆฌ๋ทฐ์ต์
๋ถ์", elem_classes="custom-button")
with gr.Column(elem_classes="custom-frame"):
gr.HTML("<div class='custom-title'>๐๋ถ์๋ณด๊ณ ์ ๋ค์ด๋ก๋</div>")
download_final_output = gr.File(label="๋ณด๊ณ ์ ๋ค์ด๋ก๋")
# [๋ฆฌ๋ทฐ๋ถ์ ํ๋ ์] (๋ฐ์ดํฐ ์
๋ ฅ ํ๋ ์๊ณผ ๋ณ๋; ์ด๊ธฐ์๋ ์จ๊น)
with gr.Column(elem_classes="custom-frame", visible=False) as review_analysis_frame:
gr.HTML("<div class='custom-title'>๋ฆฌ๋ทฐ๋ถ์ ์ต์
์ ํ</div>")
llm_model_radio = gr.Radio(
choices=["ChatGPT (gpt-4o-mini)", "Gemini Flash (gemini-2.0-flash)", "o1-mini"],
label="LLM ๋ชจ๋ธ ์ ํ",
value="ChatGPT (gpt-4o-mini)"
)
top20_dropdown = gr.Dropdown(
label="์์ดํ
์ต์
๋ถ์",
choices=["์ ์ฒด์ต์
๋ถ์"],
value="์ ์ฒด์ต์
๋ถ์"
)
review_button = gr.Button("๐ ๋ฆฌ๋ทฐ๋ถ์ ์ต์
์ ํ", elem_classes="custom-button")
# [๋ถ์ ๊ฒฐ๊ณผ ํ๋ ์] - ์๋จ ์ ๋ชฉ ์ถ๊ฐ: "๐ ์ต์
๋ณ ๋ฆฌ๋ทฐ๋ถ์"
with gr.Column(elem_classes="custom-frame"):
gr.HTML("<div class='custom-title'>๐ ์ต์
๋ณ ๋ฆฌ๋ทฐ๋ถ์</div>")
with gr.Row():
with gr.Column(elem_classes="custom-frame"):
gr.HTML("<div class='custom-subtitle'>โจ ์ฃผ์๊ธ์ ๋ฆฌ๋ทฐ</div>")
positive_output = gr.Textbox(label="๊ธ์ ๋ฆฌ๋ทฐ๋ฆฌ์คํธ(20๊ฐ)", lines=10)
with gr.Column(elem_classes="custom-frame"):
gr.HTML("<div class='custom-subtitle'>โจ ์ฃผ์๋ถ์ ๋ฆฌ๋ทฐ</div>")
negative_output = gr.Textbox(label="๋ถ์ ๋ฆฌ๋ทฐ๋ฆฌ์คํธ(30๊ฐ)", lines=10)
with gr.Row():
with gr.Column(elem_classes="custom-frame"):
gr.HTML("<div class='custom-subtitle'>๐ข ๊ธ์ ๋ฆฌ๋ทฐ ๋ถ์</div>")
positive_analysis_output = gr.Textbox(label="๊ธ์ ๋ฆฌ๋ทฐ ๋ถ์", lines=8)
with gr.Column(elem_classes="custom-frame"):
gr.HTML("<div class='custom-subtitle'>๐ข ๋ถ์ ๋ฆฌ๋ทฐ ๋ถ์</div>")
negative_analysis_output = gr.Textbox(label="๋ถ์ ๋ฆฌ๋ทฐ ๋ถ์", lines=8)
# hidden state: ๋ฆฌ๋ทฐ์ต์
๋ถ์ ๊ฒฐ๊ณผ ์์
ํ์ผ ์ ์ฅ
partial_file_state = gr.State()
# [๋ฐ์ดํฐ ์
๋ ฅ] - ๋ฆฌ๋ทฐ์ต์
๋ถ์ ๋ฒํผ ํด๋ฆญ ์ ์คํ:
# ๋ฆฌ๋ทฐ์ต์
๋ถ์ ๊ฒฐ๊ณผ๋ฅผ ์ ์ฅํ๊ณ , ๋ฆฌ๋ทฐ๋ถ์ ํ๋ ์์ ๋ณด์ด๊ฒ ํ๋ฉฐ ์์ดํ
์ต์
๋ถ์ ๋๋กญ๋ค์ด ์
๋ฐ์ดํธ
def on_click_analyze_options(uploaded_file, selected_year):
partial_file, top20_list = analyze_options(uploaded_file, selected_year, "ChatGPT (gpt-4o-mini)")
return partial_file, gr.update(visible=True), gr.update(choices=top20_list, value="์ ์ฒด์ต์
๋ถ์")
analyze_button.click(
fn=on_click_analyze_options,
inputs=[file_input, year_radio],
outputs=[partial_file_state, review_analysis_frame, top20_dropdown]
)
# [๋ฆฌ๋ทฐ๋ถ์] - ๋ฆฌ๋ทฐ๋ถ์ ๋ฒํผ ํด๋ฆญ ์ ์คํ
def on_click_analyze_reviews(partial_file, selected_option, llm_model):
final_file, pos_reviews, neg_reviews, pos_analysis, neg_analysis = analyze_reviews(
partial_file, selected_option, llm_model
)
return final_file, pos_reviews, neg_reviews, pos_analysis, neg_analysis
review_button.click(
fn=on_click_analyze_reviews,
inputs=[partial_file_state, top20_dropdown, llm_model_radio],
outputs=[download_final_output, positive_output, negative_output, positive_analysis_output, negative_analysis_output]
)
if __name__ == "__main__":
demo.launch()
|