feat: dayli

app.py

@@ -3,6 +3,7 @@ import datetime
 import requests
 import pytz
 import yaml
+from tools.rss_get_papers import HFDaylyPapperTool
 from tools.final_answer import FinalAnswerTool
 from tools.visit_webpage import VisitWebpageTool
 
@@ -37,6 +38,7 @@ def get_current_time_in_timezone(timezone: str) -> str:
 
 final_answer = FinalAnswerTool()
 visit_page = VisitWebpageTool()
+daily_paper = HFDaylyPapperTool()
 
 model = HfApiModel(
 max_tokens=2096,
@@ -54,7 +56,7 @@ with open("prompts.yaml", 'r') as stream:
     
 agent = CodeAgent(
     model=model,
-    tools=[final_answer, image_generation_tool, visit_page], ## add your tools here (don't remove final answer)
+    tools=[final_answer, image_generation_tool, visit_page, daily_paper], ## add your tools here (don't remove final answer)
     max_steps=6,
     verbosity_level=1,
     grammar=None,

requirements.txt

@@ -3,3 +3,4 @@ smolagents
 requests
 duckduckgo_search
 pandas
+feedparser

tools/feed_processor.py

@@ -0,0 +1,117 @@
+import random
+import feedparser
+from dataclasses import dataclass
+from typing import List, Optional, Dict, Set
+from abc import ABC, abstractmethod
+
+@dataclass
+class Article:
+    title: str
+    link: str
+    summary: str
+    published: str
+    authors: str
+    pdf_link: Optional[str] = None
+    source: str = None
+
+class BaseRSSParser(ABC):
+    @abstractmethod
+    def parse_feed(self, feed: feedparser.FeedParserDict) -> List[Article]:
+        """Парсит RSS-ленту и возвращает список объектов Article."""
+        pass
+
+class ArxivRSSParser(BaseRSSParser):
+    def parse_feed(self, feed: feedparser.FeedParserDict) -> List[Article]:
+        articles: List[Article] = []
+        for entry in feed.entries:
+            try:
+                title: str = entry.get('title', 'Без названия')
+                link: str = entry.get('link', '')
+                summary: str = entry.get('summary', '')
+                published: str = entry.get('published', 'Неизвестно')
+                authors_list = entry.get('authors', [])
+                authors: str = ', '.join([author.name for author in authors_list]) if authors_list else 'Неизвестно'
+                pdf_link: Optional[str] = next(
+                    (l.href for l in entry.get('links', []) if l.type == 'application/pdf'), None
+                )
+
+                article = Article(
+                    title=title,
+                    link=link,
+                    summary=summary,
+                    published=published,
+                    authors=authors,
+                    pdf_link=pdf_link
+                )
+                articles.append(article)
+            except Exception as e:
+                print(f"Ошибка при парсинге записи: {e}")
+        return articles
+
+## Пример другого парсера для другого RSS-источника
+class DailyHFRSSParser(BaseRSSParser):
+    def parse_feed(self, feed: feedparser.FeedParserDict) -> List[Article]:
+        # Реализуйте специфическую логику парсинга для другого источника
+        articles: List[Article] = []
+        for entry in feed.entries:
+            # Пример парсинга, замените на актуальные поля
+            title: str = entry.get('title', 'Без названия')
+            link: str = entry.get('link', '')
+            summary: str = entry.get('description', '')
+            published: str = entry.get('pubDate', 'Неизвестно')
+            authors: str = entry.get('author', 'Неизвестно')
+            
+            article = Article(
+                title=title,
+                link=link,
+                summary=summary,
+                published=published,
+                authors=authors,
+                source="Daily papers"
+            )
+            articles.append(article)
+        return articles
+
+
+class RSSFeedFetcher:
+    def __init__(self, feed_url: str):
+        self.feed_url = feed_url
+
+    def fetch_feed(self) -> feedparser.FeedParserDict:
+        """Загружает и парсит RSS-ленту."""
+        try:
+            feed = feedparser.parse(self.feed_url)
+            if feed.bozo:
+                raise ValueError(f"Ошибка при парсинге RSS-ленты: {feed.bozo_exception}")
+            return feed
+        except Exception as e:
+            print(f"Ошибка при загрузке ленты: {e}")
+            return feedparser.FeedParserDict()
+
+class RSSFeedProcessor:
+    def __init__(self):
+        self.feed_parsers: Dict[str, BaseRSSParser] = {}
+        self.feed_urls: Dict[str, str] = {}
+
+    def register_feed(self, source_key: str, feed_url: str, parser: BaseRSSParser):
+        self.feed_parsers[source_key] = parser
+        self.feed_urls[source_key] = feed_url
+
+    def get_latest_articles(self, sources: Set[str], count: int = 1) -> List[Article]:
+        all_articles: List[List[Article]] = []
+        for source_key in sources:
+            parser = self.feed_parsers.get(source_key)
+            feed_url = self.feed_urls.get(source_key)
+            if parser and feed_url:
+                fetcher = RSSFeedFetcher(feed_url)
+                feed = fetcher.fetch_feed()
+                articles = parser.parse_feed(feed)
+                all_articles.append(articles[:count])
+            else:
+                print(f"Источник {source_key} не найден или не имеет парсера")
+
+ 
+        random.shuffle(all_articles)
+        
+  
+        return all_articles[:count*len(sources)]

tools/rss_get_papers.py

@@ -0,0 +1,55 @@
+import re
+from typing import Any, Optional
+from smolagents.tools import Tool
+import requests
+import markdownify
+import smolagents
+
+from feed_processor import  DailyHFRSSParser, RSSFeedProcessor
+
+class HFDaylyPapperTool(Tool):
+    name = "get_actual_ai_news"
+    description = "Return actual news about AI today."
+    output_type = "string"
+    
+    def __init__(self, max_results=10, **kwargs):
+        super().__init__()
+        self.max_results = max_results
+        self.rss_processor = RSSFeedProcessor()
+        self.rss_processor.register_feed(
+            "HuggingFace Daily pappers",
+            "https://jamesg.blog/hf-papers.xml",
+            DailyHFRSSParser(),
+        )
+
+    def forward(self) -> str:
+        try:
+            from markdownify import markdownify
+            from smolagents.utils import truncate_content
+            
+        except ImportError as e:
+            raise ImportError(
+                "You must install packages `markdownify` and `requests` to run this tool: for instance run `pip install markdownify requests`."
+            ) from e
+            
+        try:
+            response = self.rss_processor.get_latest_articles()
+            
+            result = ""
+            for article in response:
+                result+=self.pretty_str_paper(article)
+                
+            markdown_content = markdownify(result).strip()
+            markdown_content = re.sub(r"\n{3,}", "\n\n", markdown_content)
+
+            return truncate_content(markdown_content, 10000)
+
+        except Exception as e:
+            return f"An unexpected error occurred: {str(e)}"
+
+    def pretty_str_paper(article):
+        return     f"*{article.title}*\n" \
+                   f"_Источник_: {article.source}\n" \
+                   f"_Авторы_: {article.authors}\n" \
+                   f"_Опубликовано_: {article.published}\n" \
+                   f"_Описание_: {article.summary}\n\n"