Update server.js
Browse files
server.js
CHANGED
|
@@ -4,7 +4,7 @@ import { Readability } from '@mozilla/readability';
|
|
| 4 |
import { JSDOM } from 'jsdom';
|
| 5 |
import TurndownService from 'turndown';
|
| 6 |
|
| 7 |
-
|
| 8 |
const turndownService = new TurndownService({
|
| 9 |
headingStyle: 'atx',
|
| 10 |
codeBlockStyle: 'fenced'
|
|
@@ -13,18 +13,15 @@ const turndownService = new TurndownService({
|
|
| 13 |
class WebsiteParser {
|
| 14 |
async fetchAndParse(url) {
|
| 15 |
try {
|
| 16 |
-
// Fetch the webpage
|
| 17 |
const response = await axios.get(url, {
|
| 18 |
headers: {
|
| 19 |
'User-Agent': 'Mozilla/5.0 (compatible; MCPBot/1.0)'
|
| 20 |
}
|
| 21 |
});
|
| 22 |
|
| 23 |
-
// Create a DOM from the HTML
|
| 24 |
const dom = new JSDOM(response.data, { url });
|
| 25 |
const document = dom.window.document;
|
| 26 |
|
| 27 |
-
// Use Readability to extract main content
|
| 28 |
const reader = new Readability(document);
|
| 29 |
const article = reader.parse();
|
| 30 |
|
|
@@ -32,7 +29,6 @@ class WebsiteParser {
|
|
| 32 |
throw new Error('Failed to parse content');
|
| 33 |
}
|
| 34 |
|
| 35 |
-
// Convert HTML to Markdown
|
| 36 |
const markdown = turndownService.turndown(article.content);
|
| 37 |
|
| 38 |
return {
|
|
@@ -43,7 +39,7 @@ class WebsiteParser {
|
|
| 43 |
siteName: article.siteName
|
| 44 |
};
|
| 45 |
} catch (error) {
|
| 46 |
-
throw new Error(
|
| 47 |
}
|
| 48 |
}
|
| 49 |
}
|
|
@@ -52,16 +48,14 @@ const app = express();
|
|
| 52 |
const PORT = process.env.PORT || 7860;
|
| 53 |
const parser = new WebsiteParser();
|
| 54 |
|
| 55 |
-
// Парсинг JSON тела запроса
|
| 56 |
app.use(express.json());
|
| 57 |
|
| 58 |
-
// Эндпоинт для парсинга URL
|
| 59 |
app.post('/parse', async (req, res) => {
|
| 60 |
try {
|
| 61 |
const { url } = req.body;
|
| 62 |
|
| 63 |
if (!url) {
|
| 64 |
-
return res.status(400).json({ error: '
|
| 65 |
}
|
| 66 |
|
| 67 |
const result = await parser.fetchAndParse(url);
|
|
@@ -83,7 +77,7 @@ app.post('/parse', async (req, res) => {
|
|
| 83 |
});
|
| 84 |
|
| 85 |
app.get('/', (req, res) => {
|
| 86 |
-
res.send(
|
| 87 |
});
|
| 88 |
|
| 89 |
app.listen(PORT, () => {
|
|
|
|
| 4 |
import { JSDOM } from 'jsdom';
|
| 5 |
import TurndownService from 'turndown';
|
| 6 |
|
| 7 |
+
|
| 8 |
const turndownService = new TurndownService({
|
| 9 |
headingStyle: 'atx',
|
| 10 |
codeBlockStyle: 'fenced'
|
|
|
|
| 13 |
class WebsiteParser {
|
| 14 |
async fetchAndParse(url) {
|
| 15 |
try {
|
|
|
|
| 16 |
const response = await axios.get(url, {
|
| 17 |
headers: {
|
| 18 |
'User-Agent': 'Mozilla/5.0 (compatible; MCPBot/1.0)'
|
| 19 |
}
|
| 20 |
});
|
| 21 |
|
|
|
|
| 22 |
const dom = new JSDOM(response.data, { url });
|
| 23 |
const document = dom.window.document;
|
| 24 |
|
|
|
|
| 25 |
const reader = new Readability(document);
|
| 26 |
const article = reader.parse();
|
| 27 |
|
|
|
|
| 29 |
throw new Error('Failed to parse content');
|
| 30 |
}
|
| 31 |
|
|
|
|
| 32 |
const markdown = turndownService.turndown(article.content);
|
| 33 |
|
| 34 |
return {
|
|
|
|
| 39 |
siteName: article.siteName
|
| 40 |
};
|
| 41 |
} catch (error) {
|
| 42 |
+
throw new Error(`ошибка парсинга или получения страницы: ${error.message}`);
|
| 43 |
}
|
| 44 |
}
|
| 45 |
}
|
|
|
|
| 48 |
const PORT = process.env.PORT || 7860;
|
| 49 |
const parser = new WebsiteParser();
|
| 50 |
|
|
|
|
| 51 |
app.use(express.json());
|
| 52 |
|
|
|
|
| 53 |
app.post('/parse', async (req, res) => {
|
| 54 |
try {
|
| 55 |
const { url } = req.body;
|
| 56 |
|
| 57 |
if (!url) {
|
| 58 |
+
return res.status(400).json({ error: 'необходимо указать URL!' });
|
| 59 |
}
|
| 60 |
|
| 61 |
const result = await parser.fetchAndParse(url);
|
|
|
|
| 77 |
});
|
| 78 |
|
| 79 |
app.get('/', (req, res) => {
|
| 80 |
+
res.send(`curl -X POST https://prolapse-read.hf.space/parse -H "Content-Type: application/json" -d '{"url": "https://habr.com/ru/companies/serverspace/articles/869252/"}'`);
|
| 81 |
});
|
| 82 |
|
| 83 |
app.listen(PORT, () => {
|