Merge pull request #85 from manics/proxy-base-urls

Configure absolute and server URL prefixes

Author: yuvipanda

.travis.yml

@@ -3,7 +3,11 @@ python:
 - 3.5
 
 script:
-  - true
+  - pip3 install .
+  - pip3 install pytest
+  - JUPYTER_TOKEN=secret jupyter-notebook --config=./tests/resources/jupyter_server_config.py &
+  - sleep 5
+  - pytest
 
 deploy:
   provider: pypi

docs/server-process.rst

@@ -56,6 +56,34 @@ pairs.
    * A callable that takes any :ref:`callable arguments <server-process/callable-argument>`,
      and returns a dictionary of strings that are used & treated same as above.
 
+#. **absolute_url**
+
+   *True* if the URL as seen by the proxied application should be the full URL
+   sent by the user. *False* if the URL as seen by the proxied application should
+   see the URL after the parts specific to jupyter-server-proxy have been stripped.
+
+   For example, with the following config:
+
+   .. code:: python
+
+      c.ServerProxy.servers = {
+        'test-server': {
+          'command': ['python3', '-m', 'http.server', '{port}'],
+          'absolute_url': False
+        }
+      }
+
+   When a user requests ``/test-server/some-url``, the proxied server will see it
+   as a request for ``/some-url`` - the ``/test-server`` part is stripped out.
+
+   If ``absolute_url`` is set to ``True`` instead, the proxied server will see it
+   as a request for ``/test-server/some-url`` instead - without any stripping.
+
+   This is very useful with applications that require a ``base_url`` to be set.
+
+   Defaults to *False*.
+
+
 #. **launcher_entry**
 
    A dictionary with options on if / how an entry in the classic Jupyter Notebook

jupyter_server_proxy/config.py

@@ -9,7 +9,7 @@
 from collections import namedtuple
 from .utils import call_with_asked_args
 
-def _make_serverproxy_handler(name, command, environment, timeout):
+def _make_serverproxy_handler(name, command, environment, timeout, absolute_url):
     """
     Create a SuperviseAndProxyHandler subclass with given parameters
     """
@@ -18,6 +18,8 @@ class _Proxy(SuperviseAndProxyHandler):
         def __init__(self, *args, **kwargs):
             super().__init__(*args, **kwargs)
             self.name = name
+            self.proxy_base = name
+            self.absolute_url = absolute_url
 
         @property
         def process_args(self):
@@ -77,6 +79,7 @@ def make_handlers(base_url, server_processes):
             sp.command,
             sp.environment,
             sp.timeout,
+            sp.absolute_url,
         )
         handlers.append((
             ujoin(base_url, sp.name, r'(.*)'), handler, dict(state={}),
@@ -87,7 +90,8 @@ def make_handlers(base_url, server_processes):
     return handlers
 
 LauncherEntry = namedtuple('LauncherEntry', ['enabled', 'icon_path', 'title'])
-ServerProcess = namedtuple('ServerProcess', ['name', 'command', 'environment', 'timeout', 'launcher_entry'])
+ServerProcess = namedtuple('ServerProcess', [
+    'name', 'command', 'environment', 'timeout', 'absolute_url', 'launcher_entry'])
 
 def make_server_process(name, server_process_config):
     le = server_process_config.get('launcher_entry', {})
@@ -96,6 +100,7 @@ def make_server_process(name, server_process_config):
         command=server_process_config['command'],
         environment=server_process_config.get('environment', {}),
         timeout=server_process_config.get('timeout', 5),
+        absolute_url=server_process_config.get('absolute_url', False),
         launcher_entry=LauncherEntry(
             enabled=le.get('enabled', True),
             icon_path=le.get('icon_path'),
@@ -129,6 +134,10 @@ class ServerProxy(Configurable):
           timeout
             Timeout in seconds for the process to become ready, default 5s.
 
+          absolute_url
+            Proxy requests default to being rewritten to '/'. If this is True,
+            the absolute URL will be sent to the backend instead.
+
           launcher_entry
             A dictionary of various options for entries in classic notebook / jupyterlab launchers.
 
@@ -146,4 +155,4 @@ class ServerProxy(Configurable):
               Title to be used for the launcher entry. Defaults to the name of the server if missing.
         """,
         config=True
-    )
+    )

jupyter_server_proxy/handlers.py

@@ -29,6 +29,12 @@ def get(self, *args):
         self.redirect(urlunparse(dest))
 
 class LocalProxyHandler(WebSocketHandlerMixin, IPythonHandler):
+
+    def __init__(self, *args, **kwargs):
+        self.proxy_base = ''
+        self.absolute_url = kwargs.pop('absolute_url', False)
+        super().__init__(*args, **kwargs)
+
     async def open(self, port, proxied_path=''):
         """
         Called when a client opens a websocket connection.
@@ -39,13 +45,7 @@ async def open(self, port, proxied_path=''):
         if not proxied_path.startswith('/'):
             proxied_path = '/' + proxied_path
 
-        client_uri = '{uri}:{port}{path}'.format(
-            uri='ws://127.0.0.1',
-            port=port,
-            path=proxied_path
-        )
-        if self.request.query:
-            client_uri += '?' + self.request.query
+        client_uri = self.get_client_uri('ws', port, proxied_path)
         headers = self.request.headers
 
         def message_cb(message):
@@ -126,14 +126,63 @@ def _record_activity(self):
         """
         self.settings['api_last_activity'] = utcnow()
 
+    def _get_context_path(self, port):
+        """
+        Some applications need to know where they are being proxied from.
+        This is either:
+        - {base_url}/proxy/{port}
+        - {base_url}/proxy/absolute/{port}
+        - {base_url}/{proxy_base}
+        """
+        if self.proxy_base:
+            return url_path_join(self.base_url, self.proxy_base)
+        if self.absolute_url:
+            return url_path_join(self.base_url, 'proxy', 'absolute', str(port))
+        else:
+            return url_path_join(self.base_url, 'proxy', str(port))
+
+    def get_client_uri(self, protocol, port, proxied_path):
+        context_path = self._get_context_path(port)
+        if self.absolute_url:
+            client_path = url_path_join(context_path, proxied_path)
+        else:
+            client_path = proxied_path
+
+        client_uri = '{protocol}://{host}:{port}{path}'.format(
+            protocol=protocol,
+            host='localhost',
+            port=port,
+            path=client_path
+        )
+        if self.request.query:
+            client_uri += '?' + self.request.query
+
+        return client_uri
+
+    def _build_proxy_request(self, port, proxied_path, body):
+
+        headers = self.proxy_request_headers()
+
+        client_uri = self.get_client_uri('http', port, proxied_path)
+        # Some applications check X-Forwarded-Context and X-ProxyContextPath
+        # headers to see if and where they are being proxied from.
+        if not self.absolute_url:
+            context_path = self._get_context_path(port)
+            headers['X-Forwarded-Context'] = context_path
+            headers['X-ProxyContextPath'] = context_path
+
+        req = httpclient.HTTPRequest(
+            client_uri, method=self.request.method, body=body,
+            headers=headers, **self.proxy_request_options())
+        return req
 
     @web.authenticated
     async def proxy(self, port, proxied_path):
         '''
-        While self.request.uri is
-            (hub)    /user/username/proxy/([0-9]+)/something.
-            (single) /proxy/([0-9]+)/something
-        This serverextension is given {port}/{everything/after}.
+        This serverextension handles:
+            {base_url}/proxy/{port([0-9]+)}/{proxied_path}
+            {base_url}/proxy/absolute/{port([0-9]+)}/{proxied_path}
+            {base_url}/{proxy_base}/{proxied_path}
         '''
 
         if 'Proxy-Connection' in self.request.headers:
@@ -154,29 +203,9 @@ async def proxy(self, port, proxied_path):
             else:
                 body = None
 
-        client_uri = '{uri}:{port}{path}'.format(
-            uri='http://localhost',
-            port=port,
-            path=proxied_path
-        )
-        if self.request.query:
-            client_uri += '?' + self.request.query
-
         client = httpclient.AsyncHTTPClient()
 
-        headers = self.proxy_request_headers()
-
-        # Some applications check X-Forwarded-Context and X-ProxyContextPath
-        # headers to see if and where they are being proxied from. We set
-        # them to be {base_url}/proxy/{port}.
-        headers['X-Forwarded-Context'] = headers['X-ProxyContextPath'] = \
-            url_path_join(self.base_url, 'proxy', str(port))
-
-        req = httpclient.HTTPRequest(
-            client_uri, method=self.request.method, body=body,
-            headers=headers,
-            **self.proxy_request_options())
-
+        req = self._build_proxy_request(port, proxied_path, body)
         response = await client.fetch(req, raise_error=False)
         # record activity at start and end of requests
         self._record_activity()
@@ -378,10 +407,14 @@ def patch(self, path):
     def options(self, path):
         return self.proxy(self.port, path)
 
+
 def setup_handlers(web_app):
     host_pattern = '.*$'
     web_app.add_handlers('.*', [
-        (url_path_join(web_app.settings['base_url'], r'/proxy/(\d+)(.*)'), LocalProxyHandler)
+        (url_path_join(web_app.settings['base_url'], r'/proxy/(\d+)(.*)'),
+         LocalProxyHandler, {'absolute_url': False}),
+        (url_path_join(web_app.settings['base_url'], r'/proxy/absolute/(\d+)(.*)'),
+         LocalProxyHandler, {'absolute_url': True}),
     ])
 
 # vim: set et ts=4 sw=4:

tests/resources/httpinfo.py

@@ -0,0 +1,17 @@
+from http.server import HTTPServer, BaseHTTPRequestHandler
+import sys
+
+class EchoRequestInfo(BaseHTTPRequestHandler):
+    def do_GET(self):
+        self.send_response(200)
+        self.send_header("Content-type", "text/plain")
+        self.end_headers()
+        self.wfile.write('{}\n'.format(self.requestline).encode())
+        self.wfile.write('{}\n'.format(self.headers).encode())
+
+
+if __name__ == '__main__':
+    port = int(sys.argv[1])
+    server_address = ('', port)
+    httpd = HTTPServer(server_address, EchoRequestInfo)
+    httpd.serve_forever()

tests/resources/jupyter_server_config.py

@@ -0,0 +1,10 @@
+c.ServerProxy.servers = {
+    'python-http': {
+        'command': ['python3', './tests/resources/httpinfo.py', '{port}'],
+    },
+    'python-http-abs': {
+        'command': ['python3', './tests/resources/httpinfo.py', '{port}'],
+        'absolute_url': True
+    },
+}
+#c.Application.log_level = 'DEBUG'

tests/test_proxies.py

@@ -0,0 +1,29 @@
+import os
+from http.client import HTTPConnection
+
+PORT = os.getenv('TEST_PORT', 8888)
+TOKEN = os.getenv('JUPYTER_TOKEN', 'secret')
+
+
+def request_get(port, path, token):
+    h = HTTPConnection('localhost', port, 10)
+    h.request('GET', '{}?token={}'.format(path, token))
+    return h.getresponse()
+
+
+def test_server_proxy_non_absolute():
+    r = request_get(PORT, '/python-http/abc', TOKEN)
+    assert r.code == 200
+    s = r.read().decode('ascii')
+    assert s.startswith('GET /abc?token=')
+    assert 'X-Forwarded-Context: /python-http\n' in s
+    assert 'X-Proxycontextpath: /python-http\n' in s
+
+
+def test_server_proxy_absolute():
+    r = request_get(PORT, '/python-http-abs/def', TOKEN)
+    assert r.code == 200
+    s = r.read().decode('ascii')
+    assert s.startswith('GET /python-http-abs/def?token=')
+    assert 'X-Forwarded-Context' not in s
+    assert 'X-Proxycontextpath' not in s