Update app.py
Browse files
app.py
CHANGED
@@ -1,337 +1,28 @@
|
|
1 |
-
from transformers import pipeline
|
2 |
-
from dataclasses import dataclass, field
|
3 |
-
from typing import List, Optional, Dict, Any
|
4 |
-
import re
|
5 |
-
from datetime import datetime
|
6 |
-
import logging
|
7 |
-
import html
|
8 |
-
from uuid import uuid4
|
9 |
-
import torch
|
10 |
import gradio as gr
|
11 |
-
import
|
12 |
-
|
13 |
-
# Настройка логирования
|
14 |
-
logging.basicConfig(
|
15 |
-
level=logging.INFO,
|
16 |
-
format='%(asctime)s - %(name)s - %(levelname)s - %(message)s'
|
17 |
-
)
|
18 |
-
logger = logging.getLogger(__name__)
|
19 |
-
|
20 |
-
@dataclass
|
21 |
-
class Comment:
|
22 |
-
"""Представляет комментарий Instagram со всеми метаданными"""
|
23 |
-
id: str = field(default_factory=lambda: str(uuid4()))
|
24 |
-
username: str = ""
|
25 |
-
time: str = ""
|
26 |
-
content: str = ""
|
27 |
-
likes: int = 0
|
28 |
-
level: int = 0
|
29 |
-
parent_id: Optional[str] = None
|
30 |
-
replies: List['Comment'] = field(default_factory=list)
|
31 |
-
is_verified: bool = False
|
32 |
-
mentions: List[str] = field(default_factory=list)
|
33 |
-
hashtags: List[str] = field(default_factory=list)
|
34 |
-
is_deleted: bool = False
|
35 |
-
sentiment: Optional[str] = None
|
36 |
-
language: Optional[str] = None
|
37 |
-
emojis: List[str] = field(default_factory=list)
|
38 |
-
|
39 |
-
def __post_init__(self):
|
40 |
-
if len(self.content) > 2200:
|
41 |
-
logger.warning(f"Comment content exceeds 2200 characters for user {self.username}")
|
42 |
-
self.content = self.content[:2200] + "..."
|
43 |
-
|
44 |
-
class InstagramCommentAnalyzer:
|
45 |
-
"""Анализатор комментариев Instagram с расширенной функциональностью"""
|
46 |
-
|
47 |
-
COMMENT_PATTERN = r'''
|
48 |
-
(?P<username>[\w\u0400-\u04FF.-]+)\s*
|
49 |
-
(?P<time>(?:\d+\s+(?:нед|мин|ч|д|мес|год|sec|min|h|d|w|mon|y)\.?))\s*
|
50 |
-
(?P<content>.*?)
|
51 |
-
(?:(?:Отметки|Likes)\s*"?Нравится"?:\s*(?P<likes>\d+))?
|
52 |
-
(?:Ответить|Reply)?(?:Показать\sперевод|Show\stranslation)?(?:Нравится|Like)?
|
53 |
-
'''
|
54 |
-
|
55 |
-
TIME_MAPPING = {
|
56 |
-
'нед': 'week', 'мин': 'minute', 'ч': 'hour',
|
57 |
-
'д': 'day', 'мес': 'month', 'год': 'year',
|
58 |
-
'w': 'week', 'h': 'hour', 'd': 'day',
|
59 |
-
'mon': 'month', 'y': 'year'
|
60 |
-
}
|
61 |
-
|
62 |
-
def __init__(self, max_depth: int = 10, max_comment_length: int = 2200):
|
63 |
-
"""Инициализация анализатора"""
|
64 |
-
self.check_dependencies()
|
65 |
-
self.max_depth = max_depth
|
66 |
-
self.max_comment_length = max_comment_length
|
67 |
-
self.pattern = re.compile(self.COMMENT_PATTERN, re.VERBOSE | re.DOTALL)
|
68 |
-
self.comments: List[Comment] = []
|
69 |
-
self.stats = self.initialize_stats()
|
70 |
-
self.sentiment_analyzer = self.load_sentiment_model()
|
71 |
-
|
72 |
-
def initialize_stats(self) -> Dict[str, int]:
|
73 |
-
"""Инициализация статистики"""
|
74 |
-
return {
|
75 |
-
'total_comments': 0,
|
76 |
-
'deleted_comments': 0,
|
77 |
-
'empty_comments': 0,
|
78 |
-
'max_depth_reached': 0,
|
79 |
-
'truncated_comments': 0,
|
80 |
-
'processed_mentions': 0,
|
81 |
-
'processed_hashtags': 0,
|
82 |
-
'processed_emojis': 0,
|
83 |
-
'failed_parses': 0
|
84 |
-
}
|
85 |
-
|
86 |
-
def check_dependencies(self):
|
87 |
-
"""Проверка зависимостей"""
|
88 |
-
required_packages = ['torch', 'transformers', 'emoji']
|
89 |
-
for package in required_packages:
|
90 |
-
try:
|
91 |
-
__import__(package)
|
92 |
-
except ImportError:
|
93 |
-
logger.error(f"Required package {package} is not installed")
|
94 |
-
raise
|
95 |
-
|
96 |
-
def load_sentiment_model(self):
|
97 |
-
"""Загрузка модели анализа тональности"""
|
98 |
-
try:
|
99 |
-
device = "cuda" if torch.cuda.is_available() else "cpu"
|
100 |
-
logger.info(f"Using device: {device}")
|
101 |
-
return pipeline(
|
102 |
-
"sentiment-analysis",
|
103 |
-
model="distilbert-base-uncased-finetuned-sst-2-english",
|
104 |
-
device=device
|
105 |
-
)
|
106 |
-
except Exception as e:
|
107 |
-
logger.error(f"Model loading failed: {str(e)}")
|
108 |
-
raise
|
109 |
-
|
110 |
-
def normalize_text(self, text: str) -> str:
|
111 |
-
"""Улучшенная нормализация текста"""
|
112 |
-
text = html.unescape(text)
|
113 |
-
text = ' '.join(text.split())
|
114 |
-
text = re.sub(r'[\u200b\ufeff\u200c]', '', text)
|
115 |
-
return text
|
116 |
-
|
117 |
-
def extract_emojis(self, text: str) -> List[str]:
|
118 |
-
"""Извлечение эмодзи из текста"""
|
119 |
-
return [c for c in text if c in emoji.EMOJI_DATA]
|
120 |
-
|
121 |
-
def normalize_time(self, time_str: str) -> str:
|
122 |
-
"""Нормализация временных меток"""
|
123 |
-
for rus, eng in self.TIME_MAPPING.items():
|
124 |
-
if rus in time_str:
|
125 |
-
return time_str.replace(rus, eng)
|
126 |
-
return time_str
|
127 |
-
|
128 |
-
def clean_content(self, content: str) -> str:
|
129 |
-
"""Очистка содержимого комментария"""
|
130 |
-
content = content.strip()
|
131 |
-
content = re.sub(r'\s+', ' ', content)
|
132 |
-
if len(content) > self.max_comment_length:
|
133 |
-
self.stats['truncated_comments'] += 1
|
134 |
-
content = content[:self.max_comment_length] + "..."
|
135 |
-
return content
|
136 |
-
|
137 |
-
def extract_metadata(self, comment: Comment) -> None:
|
138 |
-
"""Извлечение метаданных из комментария"""
|
139 |
-
try:
|
140 |
-
# Извлечение упоминаний и хэштегов
|
141 |
-
comment.mentions = re.findall(r'@(\w+)', comment.content)
|
142 |
-
comment.hashtags = re.findall(r'#(\w+)', comment.content)
|
143 |
-
|
144 |
-
# Извлечение эмодзи
|
145 |
-
comment.emojis = self.extract_emojis(comment.content)
|
146 |
-
|
147 |
-
# Обновление статистики
|
148 |
-
self.stats['processed_mentions'] += len(comment.mentions)
|
149 |
-
self.stats['processed_hashtags'] += len(comment.hashtags)
|
150 |
-
self.stats['processed_emojis'] += len(comment.emojis)
|
151 |
-
|
152 |
-
# Проверка верификации
|
153 |
-
comment.is_verified = bool(re.search(r'✓|Подтвержденный', comment.username))
|
154 |
-
except Exception as e:
|
155 |
-
logger.error(f"Metadata extraction failed: {str(e)}")
|
156 |
-
|
157 |
-
def analyze_sentiment(self, text: str) -> str:
|
158 |
-
"""Анализ тональности текста"""
|
159 |
-
try:
|
160 |
-
result = self.sentiment_analyzer(text)
|
161 |
-
return result[0]['label']
|
162 |
-
except Exception as e:
|
163 |
-
logger.error(f"Sentiment analysis failed: {str(e)}")
|
164 |
-
return "UNKNOWN"
|
165 |
-
def process_comment(self, text: str, parent_id: Optional[str] = None, level: int = 0) -> Optional[Comment]:
|
166 |
-
"""Обработка отдельного комментария"""
|
167 |
-
if not self.validate_input(text):
|
168 |
-
return None
|
169 |
-
|
170 |
-
if level > self.max_depth:
|
171 |
-
logger.warning(f"Maximum depth {self.max_depth} exceeded")
|
172 |
-
self.stats['max_depth_reached'] += 1
|
173 |
-
return None
|
174 |
-
|
175 |
-
try:
|
176 |
-
text = self.normalize_text(text)
|
177 |
-
match = self.pattern.match(text)
|
178 |
-
|
179 |
-
if not match:
|
180 |
-
alt_match = self.alternative_parse(text)
|
181 |
-
if not alt_match:
|
182 |
-
raise ValueError(f"Could not parse comment: {text[:100]}...")
|
183 |
-
match = alt_match
|
184 |
-
|
185 |
-
data = match.groupdict()
|
186 |
-
comment = Comment(
|
187 |
-
username=data['username'].strip(),
|
188 |
-
time=self.normalize_time(data['time']),
|
189 |
-
content=self.clean_content(data['content']),
|
190 |
-
likes=self.parse_likes(data.get('likes', '0')),
|
191 |
-
level=level,
|
192 |
-
parent_id=parent_id
|
193 |
-
)
|
194 |
-
|
195 |
-
# Анализ тональности и метаданных
|
196 |
-
comment.sentiment = self.analyze_sentiment(comment.content)
|
197 |
-
self.extract_metadata(comment)
|
198 |
-
|
199 |
-
self.stats['total_comments'] += 1
|
200 |
-
return comment
|
201 |
-
|
202 |
-
except Exception as e:
|
203 |
-
logger.error(f"Error processing comment: {str(e)}", exc_info=True)
|
204 |
-
self.stats['failed_parses'] += 1
|
205 |
-
return self.create_damaged_comment()
|
206 |
-
|
207 |
-
def alternative_parse(self, text: str) -> Optional[re.Match]:
|
208 |
-
"""Альтернативный метод парсинга для сложных случаев"""
|
209 |
-
alternative_patterns = [
|
210 |
-
# Более простой паттерн
|
211 |
-
r'(?P<username>[\w\u0400-\u04FF.-]+)\s*(?P<content>.*?)(?P<time>\d+\s+\w+\.?)(?P<likes>\d+)?',
|
212 |
-
# Паттерн для мобильной версии
|
213 |
-
r'(?P<username>[\w\u0400-\u04FF.-]+)\s*(?P<content>.*?)(?P<time>\d+\s+\w+)(?:Like)?(?P<likes>\d+)?'
|
214 |
-
]
|
215 |
-
|
216 |
-
for pattern in alternative_patterns:
|
217 |
-
try:
|
218 |
-
match = re.compile(pattern, re.VERBOSE | re.DOTALL).match(text)
|
219 |
-
if match:
|
220 |
-
return match
|
221 |
-
except Exception:
|
222 |
-
continue
|
223 |
-
return None
|
224 |
-
|
225 |
-
def parse_likes(self, likes_str: str) -> int:
|
226 |
-
"""Безопасный парсинг количества лайков"""
|
227 |
-
try:
|
228 |
-
return int(re.sub(r'\D', '', likes_str) or 0)
|
229 |
-
except (ValueError, TypeError):
|
230 |
-
return 0
|
231 |
-
|
232 |
-
def create_damaged_comment(self) -> Comment:
|
233 |
-
"""Создание заглушки для поврежденного комментария"""
|
234 |
-
return Comment(
|
235 |
-
username="[damaged]",
|
236 |
-
time="unknown",
|
237 |
-
content="[Поврежденные данные]",
|
238 |
-
is_deleted=True
|
239 |
-
)
|
240 |
-
|
241 |
-
def validate_input(self, text: str) -> bool:
|
242 |
-
"""Валидация входного текста"""
|
243 |
-
if not text or not isinstance(text, str):
|
244 |
-
logger.error("Invalid input: text must be non-empty string")
|
245 |
-
return False
|
246 |
-
if len(text) > 50000:
|
247 |
-
logger.error("Input text too large")
|
248 |
-
return False
|
249 |
-
return True
|
250 |
-
|
251 |
-
def format_comment(self, comment: Comment, index: int) -> str:
|
252 |
-
"""Форматирование комментария для вывода"""
|
253 |
-
try:
|
254 |
-
if comment.is_deleted:
|
255 |
-
return f'{index}. "[УДАЛЕНО]"'
|
256 |
-
|
257 |
-
emoji_str = ' '.join(comment.emojis) if comment.emojis else ''
|
258 |
-
mentions_str = ', '.join(comment.mentions) if comment.mentions else ''
|
259 |
-
hashtags_str = ', '.join(comment.hashtags) if comment.hashtags else ''
|
260 |
-
|
261 |
-
return (
|
262 |
-
f'{index}. "{comment.username}" "{comment.time}" '
|
263 |
-
f'"{comment.content}" "Лайки: {comment.likes}" '
|
264 |
-
f'"Настроение: {comment.sentiment}" '
|
265 |
-
f'"Эмодзи: {emoji_str}" '
|
266 |
-
f'"Упоминания: {mentions_str}" '
|
267 |
-
f'"Хэштеги: {hashtags_str}"'
|
268 |
-
)
|
269 |
-
|
270 |
-
try-except Exception as e:
|
271 |
-
logger.error(f"Error formatting comment: {str(e)}")
|
272 |
-
return f'{index}. "[ОШИБКА ФОРМАТИРОВАНИЯ]"'
|
273 |
-
|
274 |
-
def process_comments(self, text: str) -> List[str]:
|
275 |
-
"""Обработка всех комментариев"""
|
276 |
-
try:
|
277 |
-
self.stats = self.initialize_stats()
|
278 |
-
text = self.normalize_text(text)
|
279 |
-
raw_comments = text.split('ОтветитьНравится')
|
280 |
-
formatted_comments = []
|
281 |
-
|
282 |
-
for i, raw_comment in enumerate(raw_comments, 1):
|
283 |
-
if not raw_comment.strip():
|
284 |
-
continue
|
285 |
-
|
286 |
-
comment = self.process_comment(raw_comment)
|
287 |
-
if comment:
|
288 |
-
formatted_comments.append(self.format_comment(comment, i))
|
289 |
|
290 |
-
|
291 |
-
|
292 |
-
|
293 |
-
|
294 |
|
295 |
-
|
296 |
-
|
297 |
-
analyzer = InstagramCommentAnalyzer()
|
298 |
|
299 |
-
|
300 |
-
|
301 |
-
|
|
|
|
|
302 |
|
303 |
-
|
304 |
-
fn=analyze_text,
|
305 |
-
inputs=gr.Textbox(
|
306 |
-
lines=10,
|
307 |
-
placeholder="Вставьте текст комментариев здесь...",
|
308 |
-
label="Входной текст"
|
309 |
-
),
|
310 |
-
outputs=gr.Textbox(
|
311 |
-
lines=20,
|
312 |
-
placeholder="Результаты анализа будут отображены здесь...",
|
313 |
-
label="Результаты анализа"
|
314 |
-
),
|
315 |
-
title="Instagram Comment Analyzer",
|
316 |
-
description="Анализатор комментариев Instagram с поддержкой эмодзи и мультиязычности",
|
317 |
-
theme="default",
|
318 |
-
analytics_enabled=False,
|
319 |
-
)
|
320 |
-
return iface
|
321 |
|
322 |
-
|
323 |
-
|
324 |
-
|
325 |
-
|
326 |
-
|
327 |
-
|
328 |
-
server_port=7860,
|
329 |
-
share=False,
|
330 |
-
debug=True
|
331 |
-
)
|
332 |
-
except Exception as e:
|
333 |
-
logger.error(f"Application failed to start: {str(e)}")
|
334 |
-
raise
|
335 |
|
336 |
-
|
337 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
import gradio as gr
|
2 |
+
import re
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
3 |
|
4 |
+
# Функция для обработки входного текста
|
5 |
+
def process_text(input_text):
|
6 |
+
# Регулярное выражение для извлечения данных
|
7 |
+
pattern = r"(\S+) (\d+ нед\.)\.([^.]+?)Отметки \"Нравится\": (\d+)"
|
8 |
|
9 |
+
# Поиск всех совпадений в тексте
|
10 |
+
matches = re.findall(pattern, input_text)
|
|
|
11 |
|
12 |
+
# Формируем вывод
|
13 |
+
output = []
|
14 |
+
for i, match in enumerate(matches, 1):
|
15 |
+
username, time, text, likes = match
|
16 |
+
output.append(f'{i}. "{username}" "{time}" "{text.strip()}" "Нравится {likes}"')
|
17 |
|
18 |
+
return "\n".join(output)
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
19 |
|
20 |
+
# Настройка Gradio интерфейса
|
21 |
+
iface = gr.Interface(
|
22 |
+
fn=process_text, # функция для обработки текста
|
23 |
+
inputs=gr.Textbox(lines=10, placeholder="Введите текст сюда..."), # поле для ввода текста
|
24 |
+
outputs=gr.Textbox(lines=10, placeholder="Результат...") # поле для вывода результата
|
25 |
+
)
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
26 |
|
27 |
+
# Запуск интерфейса
|
28 |
+
iface.launch()
|