Move function for parsing backend token from url from Protocol to Bac…

…kendToken, and reuse it in Backend code instead of relying on fragile regex. (#154)

carta/backend.py

@@ -58,8 +58,7 @@ class Backend:
         A backend object with a process which has not been started.
 
     """
-    FRONTEND_URL = re.compile(r"CARTA is accessible at (http://.*?:\d+/\?token=(.*))")
-    FRONTEND_URL_NO_AUTH = re.compile(r"CARTA is accessible at (http://(.*?):\d+.*)")
+    FRONTEND_URL = re.compile(r"CARTA is accessible at (http://.*)")
     SESSION_ID = re.compile(r"Session (\d+) \[[\d.]+\] Connected.")
 
     def __init__(self, params, executable_path="carta", remote_host=None, token=None, frontend_url_timeout=10, session_creation_timeout=0):
@@ -96,9 +95,6 @@ def start(self):
         self.proc = subprocess.Popen(self.cmd, stdout=subprocess.PIPE, cwd=pathlib.Path.home(), preexec_fn=os.setpgrp)
         os.set_blocking(self.proc.stdout.fileno(), False)
 
-        frontend_url_re = self.FRONTEND_URL if not self.debug_no_auth else self.FRONTEND_URL_NO_AUTH
-        token_string = None
-
         start = time.time()
 
         while self.frontend_url is None:
@@ -112,15 +108,15 @@ def start(self):
                 return False
 
             for line in self.output:
-                m = frontend_url_re.search(line)
+                m = self.FRONTEND_URL.search(line)
                 if m:
-                    self.frontend_url, token_string = m.groups()
+                    self.frontend_url = m.group(1)
                     break
 
             time.sleep(1)
 
-        if token_string is not None and self.token is None and not self.debug_no_auth:
-            self.token = BackendToken(token_string)
+        if self.frontend_url is not None and self.token is None and not self.debug_no_auth:
+            _, self.token = BackendToken.split_token_from_url(self.frontend_url)
 
         # Only try to parse the session ID if it has been requested
         if self.session_creation_timeout > 0:

carta/protocol.py

@@ -56,7 +56,7 @@ class Protocol:
 
     def __init__(self, frontend_url, token=None, debug_no_auth=False):
         self.frontend_url = frontend_url
-        self.base_url, token_from_url = self.split_token_from_url(frontend_url)
+        self.base_url, token_from_url = BackendToken.split_token_from_url(frontend_url)
 
         if debug_no_auth:
             self.auth = AuthType.NONE
@@ -145,44 +145,6 @@ def request_refresh_token(cls, frontend_url, username, path=None):
 
         return token
 
-    @staticmethod
-    def split_token_from_url(url):
-        """Extract a backend token from a frontend URL.
-
-        Parameters
-        ----------
-        url : string
-            The URL of the frontend.
-
-        Returns
-        -------
-        string
-            The URL with the backend token removed.
-        :obj:`carta.token.BackendToken` object or None
-            The object representing the backend token.
-
-        Raises
-        ------
-        CartaBadUrl
-            If an invalid URL was provided.
-
-        """
-        parsed_url = urllib.parse.urlparse(url)
-
-        if not (parsed_url.scheme and parsed_url.netloc):
-            raise CartaBadUrl(f"Could not parse URL {url}.")
-
-        parsed_query = urllib.parse.parse_qs(parsed_url.query)
-
-        base_url = parsed_url._replace(query='').geturl().rstrip("/")
-
-        if "token" not in parsed_query:
-            token = None
-        else:
-            token = BackendToken(parsed_query["token"][0])
-
-        return base_url, token
-
     @property
     def domain(self):
         """The domain extracted from the frontend URL.

carta/token.py

@@ -4,8 +4,9 @@
 import base64
 import datetime
 import json
+import urllib
 
-from .util import CartaBadToken
+from .util import CartaBadToken, CartaBadUrl
 
 
 class Token:
@@ -38,6 +39,43 @@ class BackendToken(Token):
     def __init__(self, string):
         super().__init__(string)
 
+    @staticmethod
+    def split_token_from_url(url):
+        """Extract a backend token from a frontend URL.
+
+        Parameters
+        ----------
+        url : string
+            The URL of the frontend.
+
+        Returns
+        -------
+        string
+            The URL with the backend token removed.
+        :obj:`carta.token.BackendToken` object or None
+            The object representing the backend token.
+
+        Raises
+        ------
+        CartaBadUrl
+            If an invalid URL was provided.
+        """
+        parsed_url = urllib.parse.urlparse(url)
+
+        if not (parsed_url.scheme and parsed_url.netloc):
+            raise CartaBadUrl(f"Could not parse URL {url}.")
+
+        parsed_query = urllib.parse.parse_qs(parsed_url.query)
+
+        base_url = parsed_url._replace(query='').geturl().rstrip("/")
+
+        if "token" not in parsed_query:
+            token = None
+        else:
+            token = BackendToken(parsed_query["token"][0])
+
+        return base_url, token
+
 
 class ControllerToken(Token):
     """An object representing a security token used by the controller. These tokens are JWTs (JSON Web Tokens) which encode additional information such as an expiration date. The controller uses multiple types of tokens. This interface makes use of long-lived refresh tokens (used to log into the controller and to obtain scripting tokens) and short-lived scripting tokens (used to make scripting requests).