configs: fixed IP extraction in gunicorn

configs/gunicorn_config.py

@@ -11,6 +11,7 @@ import time
 import threading
 import traceback
 import random
+import socket
 from gunicorn.glogging import Logger
 
 
@@ -254,6 +255,74 @@ def post_request(worker, req, environ, resp):
     _check_memory_usage(worker)
 
 
+def _filter_proxy(ip):
+    """
+    Passed in IP addresses in HEADERS can be in a special format of multiple
+    ips. Those comma separated IPs are passed from various proxies in the
+    chain of request processing. The left-most being the original client.
+    We only care about the first IP which came from the org. client.
+
+    :param ip: ip string from headers
+    """
+    if ',' in ip:
+        _ips = ip.split(',')
+        _first_ip = _ips[0].strip()
+        return _first_ip
+    return ip
+
+
+def _filter_port(ip):
+    """
+    Removes a port from ip, there are 4 main cases to handle here.
+    - ipv4 eg. 127.0.0.1
+    - ipv6 eg. ::1
+    - ipv4+port eg. 127.0.0.1:8080
+    - ipv6+port eg. [::1]:8080
+
+    :param ip:
+    """
+    def is_ipv6(ip_addr):
+        if hasattr(socket, 'inet_pton'):
+            try:
+                socket.inet_pton(socket.AF_INET6, ip_addr)
+            except socket.error:
+                return False
+        else:
+            return False
+        return True
+
+    if ':' not in ip:  # must be ipv4 pure ip
+        return ip
+
+    if '[' in ip and ']' in ip:  # ipv6 with port
+        return ip.split(']')[0][1:].lower()
+
+    # must be ipv6 or ipv4 with port
+    if is_ipv6(ip):
+        return ip
+    else:
+        ip, _port = ip.split(':')[:2]  # means ipv4+port
+        return ip
+
+
+def get_ip_addr(environ):
+    proxy_key = 'HTTP_X_REAL_IP'
+    proxy_key2 = 'HTTP_X_FORWARDED_FOR'
+    def_key = 'REMOTE_ADDR'
+    _filters = lambda x: _filter_port(_filter_proxy(x))
+
+    ip = environ.get(proxy_key)
+    if ip:
+        return _filters(ip)
+
+    ip = environ.get(proxy_key2)
+    if ip:
+        return _filters(ip)
+
+    ip = environ.get(def_key, '0.0.0.0')
+    return _filters(ip)
+
+
 class RhodeCodeLogger(Logger):
     """
     Custom Logger that allows some customization that gunicorn doesn't allow
@@ -270,5 +339,58 @@ class RhodeCodeLogger(Logger):
         msecs = int((now - long(now)) * 1000)
         return time.strftime(self.datefmt, time.localtime(now)) + '.{0:03d}'.format(msecs)
 
+    def atoms(self, resp, req, environ, request_time):
+        """ Gets atoms for log formatting.
+        """
+        status = resp.status
+        if isinstance(status, str):
+            status = status.split(None, 1)[0]
+        atoms = {
+            'h': get_ip_addr(environ),
+            'l': '-',
+            'u': self._get_user(environ) or '-',
+            't': self.now(),
+            'r': "%s %s %s" % (environ['REQUEST_METHOD'],
+                               environ['RAW_URI'],
+                               environ["SERVER_PROTOCOL"]),
+            's': status,
+            'm': environ.get('REQUEST_METHOD'),
+            'U': environ.get('PATH_INFO'),
+            'q': environ.get('QUERY_STRING'),
+            'H': environ.get('SERVER_PROTOCOL'),
+            'b': getattr(resp, 'sent', None) is not None and str(resp.sent) or '-',
+            'B': getattr(resp, 'sent', None),
+            'f': environ.get('HTTP_REFERER', '-'),
+            'a': environ.get('HTTP_USER_AGENT', '-'),
+            'T': request_time.seconds,
+            'D': (request_time.seconds * 1000000) + request_time.microseconds,
+            'M': (request_time.seconds * 1000) + int(request_time.microseconds/1000),
+            'L': "%d.%06d" % (request_time.seconds, request_time.microseconds),
+            'p': "<%s>" % os.getpid()
+        }
+
+        # add request headers
+        if hasattr(req, 'headers'):
+            req_headers = req.headers
+        else:
+            req_headers = req
+
+        if hasattr(req_headers, "items"):
+            req_headers = req_headers.items()
+
+        atoms.update({"{%s}i" % k.lower(): v for k, v in req_headers})
+
+        resp_headers = resp.headers
+        if hasattr(resp_headers, "items"):
+            resp_headers = resp_headers.items()
+
+        # add response headers
+        atoms.update({"{%s}o" % k.lower(): v for k, v in resp_headers})
+
+        # add environ variables
+        environ_variables = environ.items()
+        atoms.update({"{%s}e" % k.lower(): v for k, v in environ_variables})
+
+        return atoms
 
 logger_class = RhodeCodeLogger