From 7f0a13f44b34d64c461a094e562e3b6f7aa5cfdf Mon Sep 17 00:00:00 2001
From: Nathan Woodburn <github@nathan.woodburn.au>
Date: Thu, 27 Feb 2025 23:35:23 +1100
Subject: [PATCH] feat: Add better cleaning for dprofile
---
server.py | 11 +++++++++-
tools.py | 62 +++++++++++++++++++++++++++++++++++++++++++++----------
2 files changed, 61 insertions(+), 12 deletions(-)
diff --git a/server.py b/server.py
index 535300c..afc9654 100644
--- a/server.py
+++ b/server.py
@@ -94,12 +94,21 @@ def proxy(url: str):
content: requests.Response = tools.proxy(url)
if not content.ok:
+ print(content.text)
return render_template("500.html"), 500
# Get the content type
contentType = content.headers.get("Content-Type")
if "text/html" in contentType:
- return tools.cleanProxyContent(content.text,url,request.host_url)
+ response = make_response(tools.cleanProxyContent(content.text,url,request.host_url))
+ response.headers["Content-Type"] = contentType
+ return response
+
+ # Clean JS
+ if "text/javascript" in contentType or 'application/javascript' in contentType:
+ response = make_response(tools.proxyCleanJS(content.text,url,request.host_url))
+ response.headers["Content-Type"] = contentType
+ return response
response = make_response(content.content)
response.headers["Content-Type"] = contentType
diff --git a/tools.py b/tools.py
index 1e97fe3..1541f98 100644
--- a/tools.py
+++ b/tools.py
@@ -13,7 +13,7 @@ import re
from bs4 import BeautifulSoup
from requests_doh import DNSOverHTTPSSession, add_dns_provider
import urllib3
-
+import socket
resolver = dns.resolver.Resolver()
resolver.nameservers = ["194.50.5.28","194.50.5.27","194.50.5.26"]
@@ -199,21 +199,28 @@ def curl(url: str):
except Exception as e:
return {"success": False, "error": "An error occurred", "message": str(e)}
+
+class ProxyError(Exception):
+ def __init__(self, message):
+ self.message = message
+ self.text = message
+ self.ok = False
+ self.status_code = 500
+ super().__init__(self.message)
+
+
def proxy(url: str) -> requests.Response:
- session = DNSOverHTTPSSession("HNSDoH")
- r = session.get(url,verify=False)
- return r
-
-
-
+ try:
+ session = DNSOverHTTPSSession("HNSDoH")
+ r = session.get(url,verify=False,timeout=30)
+ return r
+ except Exception as e:
+ return ProxyError(str(e))
def cleanProxyContent(htmlContent: str,url:str, proxyHost: str):
# Set proxy host to https if not 127.0.0.1 or localhost
if ":5000" not in proxyHost:
proxyHost = proxyHost.replace("http","https")
-
-
-
# Find all instances of the url in the html
hostUrl = f"{urlparse(url).scheme}://{urlparse(url).netloc}"
@@ -243,9 +250,42 @@ def cleanProxyContent(htmlContent: str,url:str, proxyHost: str):
break
if not ignored:
link.attrs[attrib] = f"{proxyUrl}/{link[attrib]}"
-
+
+ scripts = soup.find_all('script')
+ for script in scripts:
+ if len(script.text) > 0:
+ script.text = proxyCleanJS(script.text,url,proxyHost)
return soup.prettify()
+def proxyCleanJS(jsContent: str, url: str, proxyHost: str):
+ # Set proxy host to https if not 127.0.0.1 or localhost
+ if ":5000" not in proxyHost:
+ proxyHost = proxyHost.replace("http","https")
+
+ hostUrl = f"{urlparse(url).scheme}://{urlparse(url).netloc}"
+ proxyUrl = f"{proxyHost}proxy/{hostUrl}"
+
+ if "dprofile" in url:
+ jsContent = jsContent.replace("window.location.hostname", f"\"{urlparse(url).netloc}\"")
+ jsContent = jsContent.replace("src=\"img", f"src=\"{proxyUrl}/img")
+
+ return jsContent
+
+
+ # Replace all instances of the url with the proxy url
+ hostUrl = f"{urlparse(url).scheme}://{urlparse(url).netloc}"
+ proxyUrl = f"{proxyHost}proxy/{hostUrl}"
+
+ jsContent = jsContent.replace(hostUrl,proxyUrl)
+ # Common ways to get current url
+ for locator in ["window.location.href","window.location","location.href","location"]:
+ jsContent = jsContent.replace(locator,proxyUrl)
+
+
+ return jsContent
+
+
+
# if __name__ == "__main__":
# print(curl("https://dso.dprofile"))
\ No newline at end of file