Hugging Face's logo

Spaces:
deepak191z
/
playwright2
Paused

App Files Community
deepak191z commited on
Commit
305000a
·
verified ·
1 Parent(s): b87c704

Update index.mjs

Browse files
Files changed (1) hide show
  1. index.mjs +20 -12
index.mjs CHANGED
@@ -1,5 +1,5 @@
1
  const express = require('express');
2
- const puppeteer = require('puppeteer');
3
  const bodyParser = require('body-parser');
4
  const cors = require('cors');
5
 
@@ -14,7 +14,7 @@ const html = `
14
  <head>
15
  <meta charset="UTF-8">
16
  <meta name="viewport" content="width=device-width, initial-scale=1.0">
17
- <title>YouTube Transcript Generator</title>
18
  <style>
19
  body {
20
  font-family: Arial, sans-serif;
@@ -39,7 +39,7 @@ const html = `
39
  </style>
40
  </head>
41
  <body>
42
- <h1>YouTube Transcript Generator</h1>
43
  <form id="transcriptForm">
44
  <input type="text" id="videoUrl" name="videoUrl" placeholder="YouTube Video URL" required>
45
  <input type="text" id="videoTitle" name="videoTitle" placeholder="Video Title" required>
@@ -90,35 +90,43 @@ app.post('/extract-transcript', async (req, res) => {
90
  if (!videoUrl || !videoTitle) {
91
  return res.status(400).send('videoUrl and videoTitle are required');
92
  }
 
 
 
 
 
93
  try {
94
- const browser = await puppeteer.launch({ headless: true });
95
- const page = await browser.newPage();
96
- await page.goto(videoUrl, { waitUntil: 'networkidle2' });
97
- await page.setViewport({ width: 1920, height: 1080 });
98
 
99
- await page.waitForSelector('tp-yt-paper-button#expand', { timeout: 60000 });
100
  await page.click('tp-yt-paper-button#expand');
101
 
102
- await page.waitForSelector('button[aria-label="Show transcript"]', { timeout: 60000 });
103
  await page.click('button[aria-label="Show transcript"]');
104
 
105
- await page.waitForSelector('ytd-transcript-segment-list-renderer', { timeout: 60000 });
 
106
 
 
107
  const transcript = await page.evaluate(() => {
108
  const elements = Array.from(document.querySelectorAll('ytd-transcript-segment-renderer .segment-text'));
109
  return elements.map(element => element.innerText).join('\n');
110
  });
111
 
112
- await browser.close();
113
  res.json({ transcript });
114
 
115
  } catch (error) {
116
  console.error('Error extracting transcript:', error);
117
  res.status(500).send('Error extracting transcript');
 
 
118
  }
119
  });
120
 
121
- const PORT = process.env.PORT || 7860;
122
  app.listen(PORT, () => {
123
  console.log(`Server is running on port ${PORT}`);
124
  });
 
1
  const express = require('express');
2
+ const { chromium } = require('playwright');
3
  const bodyParser = require('body-parser');
4
  const cors = require('cors');
5
 
 
14
  <head>
15
  <meta charset="UTF-8">
16
  <meta name="viewport" content="width=device-width, initial-scale=1.0">
17
+ <title>YouTube Transcript Generator (Playwright)</title>
18
  <style>
19
  body {
20
  font-family: Arial, sans-serif;
 
39
  </style>
40
  </head>
41
  <body>
42
+ <h1>YouTube Transcript Generator (Playwright)</h1>
43
  <form id="transcriptForm">
44
  <input type="text" id="videoUrl" name="videoUrl" placeholder="YouTube Video URL" required>
45
  <input type="text" id="videoTitle" name="videoTitle" placeholder="Video Title" required>
 
90
  if (!videoUrl || !videoTitle) {
91
  return res.status(400).send('videoUrl and videoTitle are required');
92
  }
93
+
94
+ const browser = await chromium.launch();
95
+ const context = await browser.newContext();
96
+ const page = await context.newPage();
97
+
98
  try {
99
+ await page.goto(videoUrl, { waitUntil: 'networkidle' });
100
+
101
+ // Set viewport size
102
+ await page.setViewportSize({ width: 1920, height: 1080 });
103
 
104
+ // Click the "Expand" button to expand the video description
105
  await page.click('tp-yt-paper-button#expand');
106
 
107
+ // Wait for the "Show transcript" button and click it
108
  await page.click('button[aria-label="Show transcript"]');
109
 
110
+ // Wait for the transcript container to appear
111
+ await page.waitForSelector('ytd-transcript-segment-list-renderer');
112
 
113
+ // Extract the transcript text
114
  const transcript = await page.evaluate(() => {
115
  const elements = Array.from(document.querySelectorAll('ytd-transcript-segment-renderer .segment-text'));
116
  return elements.map(element => element.innerText).join('\n');
117
  });
118
 
 
119
  res.json({ transcript });
120
 
121
  } catch (error) {
122
  console.error('Error extracting transcript:', error);
123
  res.status(500).send('Error extracting transcript');
124
+ } finally {
125
+ await browser.close();
126
  }
127
  });
128
 
129
+ const PORT = 7860;
130
  app.listen(PORT, () => {
131
  console.log(`Server is running on port ${PORT}`);
132
  });