feat: add news-aggregator tool

tools/news-aggregator/index.test.ts

@@ -0,0 +1,91 @@
+import { expect } from '@jest/globals';
+import { getToolTestClient } from '../../src/test/utils';
+import * as path from 'path';
+
+interface Article {
+  title: string;
+  url: string;
+  source: string;
+  summary: string;
+  publish_date: string;
+  authors: string[];
+  top_image: string;
+  text: string;
+}
+
+interface NewsAggregatorResult {
+  total_sources_processed: number;
+  total_articles_found: number;
+  failed_sources: string[];
+  articles: Article[];
+  processing_time: number;
+}
+
+describe('News Aggregator Tool', () => {
+  const toolPath = path.join(__dirname, 'tool.py');
+  const client = getToolTestClient();
+
+  it('aggregates news from specific providers', async () => {
+    const result = await client.executeToolFromFile(toolPath, {
+      providers: ['https://reuters.com', 'https://techcrunch.com'],
+      articles_per_source: 3
+    }) as NewsAggregatorResult;
+    console.log(result);
+
+    // Basic checks
+    expect(result).toHaveProperty('total_sources_processed');
+    expect(result).toHaveProperty('total_articles_found');
+    expect(result).toHaveProperty('failed_sources');
+    expect(result).toHaveProperty('articles');
+    expect(result).toHaveProperty('processing_time');
+
+    // Type checks
+    expect(Array.isArray(result.articles)).toBe(true);
+    expect(Array.isArray(result.failed_sources)).toBe(true);
+    expect(typeof result.total_sources_processed).toBe('number');
+    expect(typeof result.total_articles_found).toBe('number');
+    expect(typeof result.processing_time).toBe('number');
+
+    // Content checks
+    if (result.articles.length > 0) {
+      const article = result.articles[0];
+      expect(article).toHaveProperty('title');
+      expect(article).toHaveProperty('url');
+      expect(article).toHaveProperty('source');
+      expect(article).toHaveProperty('summary');
+      expect(article).toHaveProperty('publish_date');
+      expect(article).toHaveProperty('authors');
+      expect(article).toHaveProperty('top_image');
+      expect(article).toHaveProperty('text');
+    }
+  }, 60000);
+
+  it('aggregates news by categories', async () => {
+    const result = await client.executeToolFromFile(toolPath, {
+      providers: [],
+      categories: ['tech'],
+      articles_per_source: 2
+    }) as NewsAggregatorResult;
+
+    expect(result.total_sources_processed).toBeGreaterThan(0);
+    expect(result.articles.length).toBeGreaterThan(0);
+
+    // Verify sources are from tech category
+    const techDomains = ['techcrunch.com', 'theverge.com', 'wired.com'];
+    const hasTechSource = result.articles.some((article: Article) => 
+      techDomains.some(domain => article.source.includes(domain))
+    );
+    expect(hasTechSource).toBe(true);
+  }, 60000);
+
+  it('handles failed sources gracefully', async () => {
+    const result = await client.executeToolFromFile(toolPath, {
+      providers: ['http://invalid-news-source.com'],
+      articles_per_source: 1
+    }) as NewsAggregatorResult;
+
+    expect(result.failed_sources).toContain('http://invalid-news-source.com');
+    expect(result.total_sources_processed).toBe(0);
+    expect(result.articles.length).toBe(0);
+  }, 30000);
+}); 

tools/news-aggregator/metadata.json

@@ -0,0 +1,107 @@
+{
+  "id": "news-aggregator",
+  "name": "News Aggregator",
+  "version": "1.0.0",
+  "description": "Aggregates latest news from multiple sources using newspaper3k, supporting parallel processing and category-based filtering",
+  "author": "Shinkai",
+  "keywords": [
+    "news",
+    "aggregator",
+    "newspaper3k",
+    "scraper",
+    "articles",
+    "multi-source"
+  ],
+  "configurations": {
+    "type": "object",
+    "properties": {
+      "language": {
+        "type": "string",
+        "description": "The language to use for article processing",
+        "default": "en"
+      },
+      "number_threads": {
+        "type": "integer",
+        "description": "Number of threads for article processing",
+        "default": 10
+      },
+      "request_timeout": {
+        "type": "integer",
+        "description": "Timeout in seconds for HTTP requests",
+        "default": 30
+      },
+      "max_concurrent_sources": {
+        "type": "integer",
+        "description": "Maximum number of sources to process concurrently",
+        "default": 5
+      }
+    },
+    "required": []
+  },
+  "parameters": {
+    "type": "object",
+    "properties": {
+      "providers": {
+        "type": "array",
+        "items": {
+          "type": "string"
+        },
+        "description": "List of news provider URLs to aggregate from"
+      },
+      "articles_per_source": {
+        "type": "integer",
+        "description": "Maximum number of articles to fetch per source",
+        "default": 5
+      },
+      "categories": {
+        "type": "array",
+        "items": {
+          "type": "string",
+          "enum": ["general", "tech", "business"]
+        },
+        "description": "Categories to fetch news from (uses default providers if no specific providers given)"
+      }
+    },
+    "required": ["providers"]
+  },
+  "result": {
+    "type": "object",
+    "properties": {
+      "total_sources_processed": {
+        "type": "integer"
+      },
+      "total_articles_found": {
+        "type": "integer"
+      },
+      "failed_sources": {
+        "type": "array",
+        "items": {
+          "type": "string"
+        }
+      },
+      "articles": {
+        "type": "array",
+        "items": {
+          "type": "object",
+          "properties": {
+            "title": {"type": "string"},
+            "url": {"type": "string"},
+            "source": {"type": "string"},
+            "summary": {"type": "string"},
+            "publish_date": {"type": "string"},
+            "authors": {
+              "type": "array",
+              "items": {
+                "type": "string"
+              }
+            },
+            "top_image": {"type": "string"},
+            "text": {"type": "string"}
+          }
+        }
+      },
+      "processing_time": {"type": "number"}
+    },
+    "required": ["total_sources_processed", "total_articles_found", "failed_sources", "articles", "processing_time"]
+  }
+}    

tools/news-aggregator/store.json

@@ -0,0 +1,3 @@
+{
+  "categoryId": "e9e36f6d-a0e3-47e9-b782-acb946b0199e"
+}