Update app.py
Browse files
app.py
CHANGED
@@ -0,0 +1,45 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
import streamlit as st
|
2 |
+
import requests
|
3 |
+
import html2text
|
4 |
+
import os
|
5 |
+
|
6 |
+
api_key=os.getenv('APIKEY_BRIGHTDATA')
|
7 |
+
st.title("ByPass Capcha & Text Extractor")
|
8 |
+
|
9 |
+
# api_key = st.text_input("Bright Data API Key", type="password")
|
10 |
+
zone = "web_unlocker1"
|
11 |
+
url = st.text_input("Target URL", value="https://2captcha.com/demo/recaptcha-v2")
|
12 |
+
|
13 |
+
# url = url + "?hl=en" # or "?lang=en" or "/en/" depending on the site
|
14 |
+
|
15 |
+
if st.button("Extract Text"):
|
16 |
+
if not url :
|
17 |
+
st.warning("Please enter both API key and URL.")
|
18 |
+
else:
|
19 |
+
headers = {
|
20 |
+
"Authorization": f"Bearer {api_key}",
|
21 |
+
"Content-Type": "application/json",
|
22 |
+
"Accept-Language": "en-US,en;q=0.9"
|
23 |
+
}
|
24 |
+
data = {
|
25 |
+
"zone": zone,
|
26 |
+
"url": url,
|
27 |
+
"format": "raw"
|
28 |
+
}
|
29 |
+
with st.spinner("Fetching page..."):
|
30 |
+
try:
|
31 |
+
response = requests.post(
|
32 |
+
"https://api.brightdata.com/request",
|
33 |
+
json=data,
|
34 |
+
headers=headers,
|
35 |
+
timeout=60
|
36 |
+
)
|
37 |
+
response.raise_for_status()
|
38 |
+
html = response.text
|
39 |
+
# Convert HTML to readable text
|
40 |
+
text = html2text.html2text(html)
|
41 |
+
st.subheader("Extracted Text")
|
42 |
+
st.text_area("Result", text, height=400)
|
43 |
+
st.download_button("Download as .txt", text, file_name="extracted.txt")
|
44 |
+
except Exception as e:
|
45 |
+
st.error(f"Error: {e}")
|