Spaces:
Sleeping
Sleeping
Create app.py
Browse files
app.py
ADDED
|
@@ -0,0 +1,135 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
from flask import Flask, request, jsonify, send_from_directory
|
| 2 |
+
import undetected_chromedriver as uc
|
| 3 |
+
from selenium.webdriver.common.by import By
|
| 4 |
+
from selenium.webdriver.support.ui import WebDriverWait
|
| 5 |
+
from selenium.webdriver.support import expected_conditions as EC
|
| 6 |
+
import os
|
| 7 |
+
import time
|
| 8 |
+
import base64
|
| 9 |
+
from datetime import datetime
|
| 10 |
+
|
| 11 |
+
app = Flask(__name__)
|
| 12 |
+
PUBLIC_DIR = 'public'
|
| 13 |
+
os.makedirs(PUBLIC_DIR, exist_ok=True)
|
| 14 |
+
|
| 15 |
+
def get_driver():
|
| 16 |
+
options = uc.ChromeOptions()
|
| 17 |
+
options.add_argument('--headless')
|
| 18 |
+
options.add_argument('--no-sandbox')
|
| 19 |
+
options.add_argument('--disable-dev-shm-usage')
|
| 20 |
+
options.add_argument('--disable-gpu')
|
| 21 |
+
return uc.Chrome(options=options, use_subprocess=False)
|
| 22 |
+
|
| 23 |
+
@app.route('/')
|
| 24 |
+
def index():
|
| 25 |
+
return jsonify({
|
| 26 |
+
'message': 'Undetected Chrome Scraper API is running',
|
| 27 |
+
'endpoints': {
|
| 28 |
+
'POST /api/scrape': 'Get HTML content from URL',
|
| 29 |
+
'POST /api/execute': 'Execute Python code with Selenium'
|
| 30 |
+
}
|
| 31 |
+
})
|
| 32 |
+
|
| 33 |
+
@app.route('/api/scrape', methods=['POST'])
|
| 34 |
+
def scrape():
|
| 35 |
+
data = request.get_json()
|
| 36 |
+
url = data.get('url')
|
| 37 |
+
wait_time = data.get('wait', 3)
|
| 38 |
+
screenshot = data.get('screenshot', False)
|
| 39 |
+
|
| 40 |
+
if not url:
|
| 41 |
+
return jsonify({'success': False, 'error': 'URL is required'}), 400
|
| 42 |
+
|
| 43 |
+
driver = None
|
| 44 |
+
try:
|
| 45 |
+
driver = get_driver()
|
| 46 |
+
driver.get(url)
|
| 47 |
+
time.sleep(wait_time)
|
| 48 |
+
|
| 49 |
+
html = driver.page_source
|
| 50 |
+
title = driver.title
|
| 51 |
+
current_url = driver.current_url
|
| 52 |
+
|
| 53 |
+
result = {
|
| 54 |
+
'success': True,
|
| 55 |
+
'data': {
|
| 56 |
+
'html': html,
|
| 57 |
+
'title': title,
|
| 58 |
+
'url': current_url,
|
| 59 |
+
'timestamp': datetime.now().isoformat()
|
| 60 |
+
}
|
| 61 |
+
}
|
| 62 |
+
|
| 63 |
+
if screenshot:
|
| 64 |
+
filename = f'screenshot-{int(time.time())}.png'
|
| 65 |
+
filepath = os.path.join(PUBLIC_DIR, filename)
|
| 66 |
+
driver.save_screenshot(filepath)
|
| 67 |
+
result['data']['screenshot'] = f"{request.host_url}files/{filename}"
|
| 68 |
+
|
| 69 |
+
return jsonify(result)
|
| 70 |
+
|
| 71 |
+
except Exception as e:
|
| 72 |
+
return jsonify({'success': False, 'error': str(e)}), 500
|
| 73 |
+
finally:
|
| 74 |
+
if driver:
|
| 75 |
+
driver.quit()
|
| 76 |
+
|
| 77 |
+
@app.route('/api/execute', methods=['POST'])
|
| 78 |
+
def execute():
|
| 79 |
+
data = request.get_json()
|
| 80 |
+
code = data.get('code')
|
| 81 |
+
|
| 82 |
+
if not code:
|
| 83 |
+
return jsonify({'success': False, 'error': 'Code is required'}), 400
|
| 84 |
+
|
| 85 |
+
driver = None
|
| 86 |
+
try:
|
| 87 |
+
driver = get_driver()
|
| 88 |
+
|
| 89 |
+
local_vars = {
|
| 90 |
+
'driver': driver,
|
| 91 |
+
'By': By,
|
| 92 |
+
'WebDriverWait': WebDriverWait,
|
| 93 |
+
'EC': EC,
|
| 94 |
+
'time': time,
|
| 95 |
+
'PUBLIC_DIR': PUBLIC_DIR,
|
| 96 |
+
'os': os
|
| 97 |
+
}
|
| 98 |
+
|
| 99 |
+
exec(code, {}, local_vars)
|
| 100 |
+
|
| 101 |
+
result = local_vars.get('result', None)
|
| 102 |
+
|
| 103 |
+
files = []
|
| 104 |
+
if os.path.exists(PUBLIC_DIR):
|
| 105 |
+
for f in os.listdir(PUBLIC_DIR):
|
| 106 |
+
if f.startswith('screenshot-') and f.endswith('.png'):
|
| 107 |
+
files.append({
|
| 108 |
+
'name': f,
|
| 109 |
+
'publicURL': f"{request.host_url}files/{f}"
|
| 110 |
+
})
|
| 111 |
+
|
| 112 |
+
return jsonify({
|
| 113 |
+
'success': True,
|
| 114 |
+
'data': {
|
| 115 |
+
'result': result,
|
| 116 |
+
'files': files,
|
| 117 |
+
'timestamp': int(time.time())
|
| 118 |
+
}
|
| 119 |
+
})
|
| 120 |
+
|
| 121 |
+
except Exception as e:
|
| 122 |
+
return jsonify({
|
| 123 |
+
'success': False,
|
| 124 |
+
'error': str(e)
|
| 125 |
+
}), 500
|
| 126 |
+
finally:
|
| 127 |
+
if driver:
|
| 128 |
+
driver.quit()
|
| 129 |
+
|
| 130 |
+
@app.route('/files/<path:filename>')
|
| 131 |
+
def serve_file(filename):
|
| 132 |
+
return send_from_directory(PUBLIC_DIR, filename)
|
| 133 |
+
|
| 134 |
+
if __name__ == '__main__':
|
| 135 |
+
app.run(host='0.0.0.0', port=int(os.environ.get('PORT', 7860)))
|