AMdevIA/HFlearningPathAgent

1from smolagents import Tool2from duckduckgo_search import DDGS3import re4from typing import Dict5 6class SearchHfRessourcesTool(Tool):7    def __init__(self):8 9        self.name="search_hf_ressources"10        self.description="Search Hugging Face resources by domain and keywords"11        self.inputs = {12            'domain': {13                'type': 'string', 14                'description': 'Main domain (nlp, computer vision, audio, etc.)'15            },16            'keywords': {17                'type': 'string', 18                'description': 'Additional keywords to refine the search',19                'nullable': True20            },21            'ressource_type': {22                'type': 'string', 23                'description': 'Resource type (all, courses, models, datasets, spaces)',24                'nullable': True25            }26        }27        self.output_type="any"28 29        super().__init__()30        31 32    def forward(self, domain: str, keywords: str = "", ressource_type: str = "all") -> Dict:33        # construction of the research request34        search_query = f"{domain} {keywords} site:huggingface.co"35 36        # filter by ressource type37        if ressource_type != "all":38            search_query += f" {ressource_type}"39 40        # use DuckDuckGo search41        results = []42        try:43            ddgs = DDGS()44            results = list(ddgs.text(search_query, max_results=15))45        except Exception as e:46            print(f"Search error: {e}")47 48        # Analyse and categoryze results49        categorized_results = {50            "courses": [],51            "models": [],52            "spaces": [],53            "datasets": [],54            "tutorials": []55        }56 57        for result in results:58            url = result.get("href", "")59            title = result.get("title", "")60            snippet = result.get("body", "")61 62            # categorization by url63            if "/learn/" in url:64                categorized_results["courses"].append({65                    "title": title,66                    "link": url,67                    "description": snippet,68                    "type": "course"69                })70            elif "/docs/" in url:71                categorized_results["tutorials"].append({72                    "title": title,73                    "link": url,74                    "description": snippet,75                    "type": "tutorial"76                })77            elif "/datasets/" in url:78                categorized_results["datasets"].append({79                    "title": title,80                    "link": url,81                    "description": snippet,82                    "type": "dataset"83                })84            elif "/spaces/" in url:85                categorized_results["spaces"].append({86                    "title": title,87                    "link": url,88                    "description": snippet,89                    "type": "space"90                })91 92        return categorized_results93