Merge pull request #140 from dcSpark/feature/add-wikimedia-page-conte…

…nt-tool

feat: add wikimedia-page-content tool

tools/wikimedia-page-content/index.test.ts

@@ -0,0 +1,49 @@
+import { expect } from '@jest/globals';
+import { getToolTestClient } from '../../src/test/utils';
+import * as path from 'path';
+
+describe('Wikimedia Page Content Tool', () => {
+  const toolPath = path.join(__dirname, 'tool.ts');
+  const client = getToolTestClient();
+
+  it('fetches page content with default parameters', async () => {
+    const response = await client.executeToolFromFile(toolPath, {
+      title: 'Artificial intelligence'
+    });
+    console.log("Response: ", response);
+    expect(response).toHaveProperty('content');
+    expect(response.content).toHaveProperty('title');
+    expect(response.content).toHaveProperty('html');
+    expect(response.content).toHaveProperty('url');
+    expect(response.content).toHaveProperty('lastModified');
+    expect(response.content).toHaveProperty('language');
+
+    expect(response.content.title).toBe('Artificial intelligence');
+    expect(response.content.html).toBeTruthy();
+    expect(response.content.url).toBe('https://en.wikipedia.org/wiki/Artificial_intelligence');
+    expect(response.content.language).toBe('en');
+  }, 30000);
+
+  it('handles custom project and language', async () => {
+    const response = await client.executeToolFromFile(toolPath, {
+      title: 'Intelligence artificielle'
+    }, {
+      project: 'wikipedia',
+      language: 'fr'
+    });
+
+    expect(response.content.url).toBe('https://fr.wikipedia.org/wiki/Intelligence_artificielle');
+    expect(response.content.language).toBe('fr');
+    expect(response.content.html).toBeTruthy();
+  }, 30000);
+
+  it('handles titles with spaces', async () => {
+    const response = await client.executeToolFromFile(toolPath, {
+      title: 'Machine learning'
+    });
+
+    expect(response.content.url).toBe('https://en.wikipedia.org/wiki/Machine_learning');
+    expect(response.content.title).toBe('Machine learning');
+    expect(response.content.html).toBeTruthy();
+  }, 30000);
+});

tools/wikimedia-page-content/metadata.json

@@ -0,0 +1,58 @@
+{
+  "id": "wikimedia-page-content",
+  "version": "1.0.0",
+  "name": "Wikimedia Page Content",
+  "description": "Fetch the full HTML content of a specific Wikimedia page",
+  "author": "Shinkai",
+  "keywords": [
+    "wikimedia",
+    "content",
+    "wikipedia",
+    "page",
+    "html",
+    "article"
+  ],
+  "configurations": {
+    "type": "object",
+    "properties": {
+      "project": {
+        "type": "string",
+        "description": "Wikimedia project (e.g., wikipedia)",
+        "default": "wikipedia"
+      },
+      "language": {
+        "type": "string",
+        "description": "Language code (e.g., en)",
+        "default": "en"
+      }
+    },
+    "required": []
+  },
+  "parameters": {
+    "type": "object",
+    "properties": {
+      "title": {
+        "type": "string",
+        "description": "Title of the page to fetch"
+      }
+    },
+    "required": ["title"]
+  },
+  "result": {
+    "type": "object",
+    "properties": {
+      "content": {
+        "type": "object",
+        "properties": {
+          "title": {"type": "string"},
+          "html": {"type": "string"},
+          "url": {"type": "string"},
+          "lastModified": {"type": "string"},
+          "language": {"type": "string"}
+        },
+        "required": ["title", "html", "url", "lastModified", "language"]
+      }
+    },
+    "required": ["content"]
+  }
+}

tools/wikimedia-page-content/store.json

@@ -0,0 +1,3 @@
+{
+  "categoryId": "cc6ba888-3987-4e2a-af7e-3b137d997262"
+}

tools/wikimedia-page-content/tool.ts

@@ -0,0 +1,68 @@
+import axios from 'npm:axios';
+
+type Configurations = {
+  project?: string;
+  language?: string;
+};
+
+type Parameters = {
+  title: string;
+};
+
+type Result = {
+  content: {
+    title: string;
+    html: string;
+    url: string;
+    lastModified: string;
+    language: string;
+  };
+};
+
+export type Run<C extends Record<string, any>, I extends Record<string, any>, R extends Record<string, any>> = (
+  config: C,
+  inputs: I
+) => Promise<R>;
+
+export const run: Run<Configurations, Parameters, Result> = async (
+  configurations: Configurations,
+  params: Parameters
+): Promise<Result> => {
+  try {
+    const project = configurations?.project || 'wikipedia';
+    const language = configurations?.language || 'en';
+
+    // Using the REST v1 API endpoint for page content
+    const api_url = `https://${language}.${project}.org/api/rest_v1/page/html/${encodeURIComponent(params.title)}`;
+
+    const response = await axios.get(api_url, {
+      headers: {
+        'User-Agent': 'ShinkaiWikimediaPageContent/1.0',
+        'Accept': 'text/html; charset=utf-8',
+        'Api-User-Agent': 'ShinkaiWikimediaPageContent/1.0 (https://github.com/dcSpark/shinkai-tools)'
+      }
+    });
+
+    if (!response.data) {
+      throw new Error('No data received from Wikimedia API');
+    }
+
+    return {
+      content: {
+        title: params.title,
+        html: response.data,
+        url: `https://${language}.${project}.org/wiki/${encodeURIComponent(params.title.replace(/ /g, '_'))}`,
+        lastModified: response.headers['last-modified'] || '',
+        language: language
+      }
+    };
+  } catch (error) {
+    if (axios.isAxiosError(error)) {
+      if (error.response?.status === 404) {
+        throw new Error(`Page '${params.title}' not found`);
+      }
+      throw new Error(`Failed to fetch page content: ${error.response?.data?.detail || error.response?.data?.message || error.message}`);
+    }
+    throw error;
+  }
+};