feat: Add better cleaning for dprofile

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"))