Spaces:
Paused
Paused
Update sentiment_news.py
Browse files- sentiment_news.py +86 -8
sentiment_news.py
CHANGED
|
@@ -2,14 +2,23 @@ import os, asyncio
|
|
| 2 |
import httpx
|
| 3 |
from gnews import GNews
|
| 4 |
import feedparser
|
| 5 |
-
from datetime import datetime
|
|
|
|
| 6 |
|
|
|
|
|
|
|
|
|
|
| 7 |
CRYPTO_RSS_FEEDS = {
|
| 8 |
"Cointelegraph": "https://cointelegraph.com/rss",
|
| 9 |
"CoinDesk": "https://www.coindesk.com/arc/outboundfeeds/rss/",
|
| 10 |
"CryptoSlate": "https://cryptoslate.com/feed/",
|
| 11 |
"NewsBTC": "https://www.newsbtc.com/feed/",
|
| 12 |
-
"Bitcoin.com": "https://news.bitcoin.com/feed/"
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 13 |
}
|
| 14 |
|
| 15 |
class NewsFetcher:
|
|
@@ -23,14 +32,33 @@ class NewsFetcher:
|
|
| 23 |
'Cache-Control': 'no-cache'
|
| 24 |
}
|
| 25 |
)
|
|
|
|
| 26 |
self.gnews = GNews(language='en', country='US', period='3h', max_results=8)
|
| 27 |
|
| 28 |
async def _fetch_from_gnews(self, symbol: str) -> list:
|
| 29 |
try:
|
| 30 |
base_symbol = symbol.split("/")[0]
|
|
|
|
| 31 |
query = f'"{base_symbol}" cryptocurrency -bitcoin -ethereum -BTC -ETH'
|
|
|
|
|
|
|
| 32 |
news_items = await asyncio.to_thread(self.gnews.get_news, query)
|
| 33 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 34 |
except Exception as e:
|
| 35 |
print(f"Failed to fetch specific news from GNews for {symbol}: {e}")
|
| 36 |
return []
|
|
@@ -40,6 +68,8 @@ class NewsFetcher:
|
|
| 40 |
base_symbol = symbol.split('/')[0]
|
| 41 |
max_redirects = 2
|
| 42 |
current_url = feed_url
|
|
|
|
|
|
|
| 43 |
for attempt in range(max_redirects):
|
| 44 |
try:
|
| 45 |
response = await self.http_client.get(current_url)
|
|
@@ -56,16 +86,41 @@ class NewsFetcher:
|
|
| 56 |
news_items = []
|
| 57 |
search_term = base_symbol.lower()
|
| 58 |
|
| 59 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 60 |
title = entry.title.lower() if hasattr(entry, 'title') else ''
|
| 61 |
summary = entry.summary.lower() if hasattr(entry, 'summary') else entry.description.lower() if hasattr(entry, 'description') else ''
|
| 62 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 63 |
news_items.append({
|
| 64 |
'title': entry.title,
|
| 65 |
'description': summary,
|
| 66 |
'source': source_name,
|
| 67 |
-
'published':
|
| 68 |
})
|
|
|
|
| 69 |
return news_items
|
| 70 |
except Exception as e:
|
| 71 |
print(f"Failed to fetch specific news from {source_name} RSS for {symbol}: {e}")
|
|
@@ -73,6 +128,8 @@ class NewsFetcher:
|
|
| 73 |
|
| 74 |
async def get_news_for_symbol(self, symbol: str) -> str:
|
| 75 |
base_symbol = symbol.split("/")[0]
|
|
|
|
|
|
|
| 76 |
tasks = [self._fetch_from_gnews(symbol)]
|
| 77 |
for name, url in CRYPTO_RSS_FEEDS.items():
|
| 78 |
tasks.append(self._fetch_from_rss_feed(url, name, symbol))
|
|
@@ -83,14 +140,24 @@ class NewsFetcher:
|
|
| 83 |
for result in results:
|
| 84 |
if isinstance(result, Exception):
|
| 85 |
continue
|
|
|
|
| 86 |
for item in result:
|
|
|
|
| 87 |
if self._is_directly_relevant_to_symbol(item, base_symbol):
|
| 88 |
title = item.get('title', 'No Title')
|
| 89 |
description = item.get('description', 'No Description')
|
| 90 |
source = item.get('source', 'Unknown Source')
|
| 91 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
| 92 |
|
| 93 |
news_entry = f"[{source}] {title}. {description}"
|
|
|
|
|
|
|
|
|
|
|
|
|
| 94 |
if published:
|
| 95 |
news_entry += f" (Published: {published})"
|
| 96 |
|
|
@@ -99,25 +166,36 @@ class NewsFetcher:
|
|
| 99 |
if not all_news_text:
|
| 100 |
return f"No specific news found for {base_symbol} in the last 3 hours."
|
| 101 |
|
|
|
|
| 102 |
important_news = all_news_text[:5]
|
| 103 |
return " | ".join(important_news)
|
| 104 |
|
| 105 |
def _is_directly_relevant_to_symbol(self, news_item, base_symbol):
|
|
|
|
|
|
|
|
|
|
|
|
|
| 106 |
title = news_item.get('title', '').lower()
|
| 107 |
description = news_item.get('description', '').lower()
|
| 108 |
symbol_lower = base_symbol.lower()
|
| 109 |
|
|
|
|
| 110 |
if symbol_lower not in title and symbol_lower not in description:
|
| 111 |
return False
|
| 112 |
|
|
|
|
| 113 |
crypto_keywords = [
|
| 114 |
'crypto', 'cryptocurrency', 'token', 'blockchain',
|
| 115 |
'price', 'market', 'trading', 'exchange', 'defi',
|
| 116 |
-
'coin', 'digital currency', 'altcoin'
|
|
|
|
| 117 |
]
|
| 118 |
|
| 119 |
return any(keyword in title or keyword in description for keyword in crypto_keywords)
|
| 120 |
|
|
|
|
|
|
|
|
|
|
| 121 |
class SentimentAnalyzer:
|
| 122 |
def __init__(self, data_manager):
|
| 123 |
self.data_manager = data_manager
|
|
|
|
| 2 |
import httpx
|
| 3 |
from gnews import GNews
|
| 4 |
import feedparser
|
| 5 |
+
from datetime import datetime, timedelta, timezone
|
| 6 |
+
import time
|
| 7 |
|
| 8 |
+
#
|
| 9 |
+
# 🔴 تم التعديل: توسيع قائمة مصادر RSS لتشمل تغطية أوسع للعملات البديلة
|
| 10 |
+
#
|
| 11 |
CRYPTO_RSS_FEEDS = {
|
| 12 |
"Cointelegraph": "https://cointelegraph.com/rss",
|
| 13 |
"CoinDesk": "https://www.coindesk.com/arc/outboundfeeds/rss/",
|
| 14 |
"CryptoSlate": "https://cryptoslate.com/feed/",
|
| 15 |
"NewsBTC": "https://www.newsbtc.com/feed/",
|
| 16 |
+
"Bitcoin.com": "https://news.bitcoin.com/feed/",
|
| 17 |
+
"The Block": "https://www.theblock.co/rss.xml",
|
| 18 |
+
"Decrypt": "https://decrypt.co/feed",
|
| 19 |
+
"AMBCrypto": "https://ambcrypto.com/feed/",
|
| 20 |
+
"CryptoPotato": "https://cryptopotato.com/feed/",
|
| 21 |
+
"U.Today": "https://u.today/rss"
|
| 22 |
}
|
| 23 |
|
| 24 |
class NewsFetcher:
|
|
|
|
| 32 |
'Cache-Control': 'no-cache'
|
| 33 |
}
|
| 34 |
)
|
| 35 |
+
# GNews مفلترة مسبقاً لـ 3 ساعات
|
| 36 |
self.gnews = GNews(language='en', country='US', period='3h', max_results=8)
|
| 37 |
|
| 38 |
async def _fetch_from_gnews(self, symbol: str) -> list:
|
| 39 |
try:
|
| 40 |
base_symbol = symbol.split("/")[0]
|
| 41 |
+
# نستخدم الكلمات المفتاحية السلبية لمحاولة عزل أخبار العملة نفسها
|
| 42 |
query = f'"{base_symbol}" cryptocurrency -bitcoin -ethereum -BTC -ETH'
|
| 43 |
+
|
| 44 |
+
# GNews تعمل بشكل متزامن، لذا نستخدم to_thread
|
| 45 |
news_items = await asyncio.to_thread(self.gnews.get_news, query)
|
| 46 |
+
|
| 47 |
+
#
|
| 48 |
+
# 🔴 تم التعديل: توحيد تنسيق الإخراج ليتضمن مفتاح 'published'
|
| 49 |
+
#
|
| 50 |
+
formatted_items = []
|
| 51 |
+
for item in news_items:
|
| 52 |
+
# GNews توفر تاريخ النشر كنص (وهو مفلتر مسبقاً لـ 3 ساعات)
|
| 53 |
+
published_text = item.get('published date', 'Recent')
|
| 54 |
+
formatted_items.append({
|
| 55 |
+
'title': item.get('title', 'No Title'),
|
| 56 |
+
'description': item.get('description', 'No Description'),
|
| 57 |
+
'source': item.get('source', {}).get('title', 'GNews'),
|
| 58 |
+
'published': published_text # تمرير التاريخ/الوقت
|
| 59 |
+
})
|
| 60 |
+
return formatted_items
|
| 61 |
+
|
| 62 |
except Exception as e:
|
| 63 |
print(f"Failed to fetch specific news from GNews for {symbol}: {e}")
|
| 64 |
return []
|
|
|
|
| 68 |
base_symbol = symbol.split('/')[0]
|
| 69 |
max_redirects = 2
|
| 70 |
current_url = feed_url
|
| 71 |
+
|
| 72 |
+
# منطق معالجة إعادة التوجيه
|
| 73 |
for attempt in range(max_redirects):
|
| 74 |
try:
|
| 75 |
response = await self.http_client.get(current_url)
|
|
|
|
| 86 |
news_items = []
|
| 87 |
search_term = base_symbol.lower()
|
| 88 |
|
| 89 |
+
#
|
| 90 |
+
# 🔴 تم التعديل: حساب الوقت قبل 3 ساعات (باستخدام التوقيت العالمي UTC)
|
| 91 |
+
#
|
| 92 |
+
three_hours_ago = datetime.now(timezone.utc) - timedelta(hours=3)
|
| 93 |
+
|
| 94 |
+
#
|
| 95 |
+
# 🔴 تم التعديل: إزالة قيد '[:15]' والفلترة حسب الوقت
|
| 96 |
+
#
|
| 97 |
+
for entry in feed.entries:
|
| 98 |
title = entry.title.lower() if hasattr(entry, 'title') else ''
|
| 99 |
summary = entry.summary.lower() if hasattr(entry, 'summary') else entry.description.lower() if hasattr(entry, 'description') else ''
|
| 100 |
+
|
| 101 |
+
# التحقق من تاريخ النشر
|
| 102 |
+
published_tuple = entry.get('published_parsed')
|
| 103 |
+
if not published_tuple:
|
| 104 |
+
continue # تخطي الخبر إذا لم يكن له تاريخ نشر
|
| 105 |
+
|
| 106 |
+
# تحويل تاريخ النشر إلى كائن datetime مع دعم التوقيت العالمي (UTC)
|
| 107 |
+
try:
|
| 108 |
+
published_time = datetime.fromtimestamp(time.mktime(published_tuple), timezone.utc)
|
| 109 |
+
except Exception:
|
| 110 |
+
# في حال فشل التحويل، نفترض أنه قديم
|
| 111 |
+
continue
|
| 112 |
+
|
| 113 |
+
#
|
| 114 |
+
# 🔴 تم التعديل: تطبيق الفلتر الزمني (آخر 3 ساعات) وفلتر اسم العملة
|
| 115 |
+
#
|
| 116 |
+
if (search_term in title or search_term in summary) and (published_time >= three_hours_ago):
|
| 117 |
news_items.append({
|
| 118 |
'title': entry.title,
|
| 119 |
'description': summary,
|
| 120 |
'source': source_name,
|
| 121 |
+
'published': published_time.isoformat() # إرسال التاريخ بتنسيق ISO
|
| 122 |
})
|
| 123 |
+
|
| 124 |
return news_items
|
| 125 |
except Exception as e:
|
| 126 |
print(f"Failed to fetch specific news from {source_name} RSS for {symbol}: {e}")
|
|
|
|
| 128 |
|
| 129 |
async def get_news_for_symbol(self, symbol: str) -> str:
|
| 130 |
base_symbol = symbol.split("/")[0]
|
| 131 |
+
|
| 132 |
+
# إنشاء قائمة المهام (GNews + جميع مصادر RSS)
|
| 133 |
tasks = [self._fetch_from_gnews(symbol)]
|
| 134 |
for name, url in CRYPTO_RSS_FEEDS.items():
|
| 135 |
tasks.append(self._fetch_from_rss_feed(url, name, symbol))
|
|
|
|
| 140 |
for result in results:
|
| 141 |
if isinstance(result, Exception):
|
| 142 |
continue
|
| 143 |
+
|
| 144 |
for item in result:
|
| 145 |
+
# نستخدم الفلتر الثانوي للتأكد من أن الخبر يركز فعلاً على العملة
|
| 146 |
if self._is_directly_relevant_to_symbol(item, base_symbol):
|
| 147 |
title = item.get('title', 'No Title')
|
| 148 |
description = item.get('description', 'No Description')
|
| 149 |
source = item.get('source', 'Unknown Source')
|
| 150 |
+
|
| 151 |
+
#
|
| 152 |
+
# 🔴 تم التعديل: هذا المنطق سيعمل الآن بفضل التعديلات السابقة
|
| 153 |
+
#
|
| 154 |
+
published = item.get('published', '') # الحصول على التاريخ/الوقت
|
| 155 |
|
| 156 |
news_entry = f"[{source}] {title}. {description}"
|
| 157 |
+
|
| 158 |
+
#
|
| 159 |
+
# 🔴 تم التعديل: إضافة وقت النشر إلى النص النهائي (تنفيذاً لطلبك)
|
| 160 |
+
#
|
| 161 |
if published:
|
| 162 |
news_entry += f" (Published: {published})"
|
| 163 |
|
|
|
|
| 166 |
if not all_news_text:
|
| 167 |
return f"No specific news found for {base_symbol} in the last 3 hours."
|
| 168 |
|
| 169 |
+
# أخذ أهم 5 أخبار (الأكثر حداثة أو صلة)
|
| 170 |
important_news = all_news_text[:5]
|
| 171 |
return " | ".join(important_news)
|
| 172 |
|
| 173 |
def _is_directly_relevant_to_symbol(self, news_item, base_symbol):
|
| 174 |
+
"""
|
| 175 |
+
فلتر ثانوي للتأكد من أن الخبر ليس مجرد ذكر عابر للعملة،
|
| 176 |
+
بل يتعلق فعلاً بجوانب التداول أو السوق.
|
| 177 |
+
"""
|
| 178 |
title = news_item.get('title', '').lower()
|
| 179 |
description = news_item.get('description', '').lower()
|
| 180 |
symbol_lower = base_symbol.lower()
|
| 181 |
|
| 182 |
+
# يجب أن يكون الرمز موجوداً في العنوان أو الوصف
|
| 183 |
if symbol_lower not in title and symbol_lower not in description:
|
| 184 |
return False
|
| 185 |
|
| 186 |
+
# يجب أن يحتوي الخبر على كلمات مفتاحية تدل على أنه خبر مالي/تداولي
|
| 187 |
crypto_keywords = [
|
| 188 |
'crypto', 'cryptocurrency', 'token', 'blockchain',
|
| 189 |
'price', 'market', 'trading', 'exchange', 'defi',
|
| 190 |
+
'coin', 'digital currency', 'altcoin', 'airdrop', 'listing',
|
| 191 |
+
'partnership', 'update', 'mainnet', 'protocol'
|
| 192 |
]
|
| 193 |
|
| 194 |
return any(keyword in title or keyword in description for keyword in crypto_keywords)
|
| 195 |
|
| 196 |
+
# --- (باقي الملف: SentimentAnalyzer و format_whale_analysis) ---
|
| 197 |
+
# --- (لم يتم المساس بهذه الأجزاء لأنها خارج نطاق الطلب) ---
|
| 198 |
+
|
| 199 |
class SentimentAnalyzer:
|
| 200 |
def __init__(self, data_manager):
|
| 201 |
self.data_manager = data_manager
|