#2687 - Include mimeo python module

modules/adblockplus/files/mimeo.py

Index: modules/adblockplus/files/mimeo.py
===================================================================
new file mode 100755
--- /dev/null
+++ b/modules/adblockplus/files/mimeo.py
@@ -0,0 +1,89 @@
+#!/usr/bin/env python3
+
+import argparse
+import re
+import sys
+import threading
+import traceback
+
+from http.server import BaseHTTPRequestHandler, HTTPServer
+from string import Template
+
+# The token used for the headers passed by nginx is: http_
+REGEX = r'\$http_\w+\b'
+DEFAULT_LOG = '$remote_addr - - [$time_local] "$request" $status $bytes_sent'
+_lock = threading.Lock()
+
+
+class Handler(BaseHTTPRequestHandler):
+    def get_header_values(self):
+        values = {}
+        headers = re.findall(REGEX, self.format)
+        for name in headers:
+            new_var = name[6:]
+            values[name[1:]] = self.headers.get(new_var, '-')

[mathias, 2017/08/18 08:47:43]

Does this understand translation between i.e. `X-Forwarded-For` and
`x_forwarded_for`?

[f.lopez, 2017/08/19 00:51:00]

No, one has to match them manually. 
proxy_set_header X-Forwarded-For $http_x_forwarded_for;

Other option could be to use a list/tuple/something to match all cases?
('x_forwarded_for', X_Forwarded_for', 'X-Forwarded-For')

Or set a standard to for how the headers should look like.

What do you think?

[mathias, 2017/08/19 15:06:08]

No idea what you mean. I was referring to this code passing on
lowercase_and_underscore_separated_header_names were one would expect
Titlelized-And-Minus-Separated-Header-Names and asking if self.headers.get() can
indeed handle that or if you need to string-transform your input first. I would
guess the latter, but only replacing underscores with minuses (the starting
letters should be examined in case-insensitive fashion anyway, if you're lucky
that is).

+        return values
+
+    def send_simple_response(self, status, response=None):
+        self.send_response(status)
+        self.end_headers()
+        if response is None:
+            response = bytes(self.responses[status][0], 'UTF-8')
+        self.wfile.write(response)
+
+    def write_info(self, args):
+        with _lock:
+            message = Template(self.format)
+            self.output.write(message.safe_substitute(args) + '\n')

[mathias, 2017/08/18 08:47:44]

Shouldn't the template stuff happen before, so that the lock would only be held
for writing a string and flushing?

[f.lopez, 2017/08/19 00:51:00]

Done.

+            self.output.flush()
+
+    def do_POST(self):
+        status = 200
+        content = bytes(self.response, 'UTF-8')
+        values = {
+            'remote_addr': self.address_string(),
+            'time_local': self.log_date_time_string(),
+            'request': self.requestline,
+            'status': status,
+            'bytes_sent': len(content),
+        }
+        values.update(self.get_header_values())
+        try:
+            self.write_info(values)
+            self.send_simple_response(status, content)
+        except:
+            traceback.print_exc(file=sys.stderr)
+            self.send_simple_response(500)
+
+
+if __name__ == '__main__':
+    parser = argparse.ArgumentParser()
+    parser.add_argument('--port', action='store',
+                        default=8000, type=int,
+                        nargs='?',
+                        help='Port to use [default: 8000]')
+    parser.add_argument('--response', action='store',
+                        type=str, nargs='?', default='OK',
+                        help='The response send to the client')
+    parser.add_argument('--format', action='store',
+                        type=str, nargs='?',
+                        default=DEFAULT_LOG,
+                        help='Format of the log ouput')
+    parser.add_argument('output', action='store',
+                        type=str, nargs='?', default='-',
+                        help='The file where the logs will be written')
+    args = parser.parse_args()
+    if args.output and args.output != '-':
+        fh = open(args.output, 'a')
+    else:
+        fh = sys.stdout

[mathias, 2017/08/18 08:47:43]

If you would `os.open(sys.stdout.fileno(), 'w', 0)` you could avoid conditionals
like the one around `fh.close()` later on.

[Vasily Kuznetsov, 2017/08/18 12:49:57]

This doesn't work (at least in Python 3.6), but what you can indeed do is
`open(sys.stdout.fileno(), 'w', closefd=False)` using builtin open, not the one
from os module. This will give you another file-like object that will use the
same file descriptor as `sys.stdout` under the hood but will not close it when
its closed. This way you can indeed eliminate the closing `if` that Matze
doesn't like.

[f.lopez, 2017/08/19 00:51:00]

Done.

+    Handler.output = fh
+    Handler.format = args.format
+    Handler.response = args.response
+    server_address = ('', args.port)
+    httpd = HTTPServer(server_address, Handler)
+    try:

[mathias, 2017/08/18 08:47:44]

This exception handling should enclose everything since the file has been
opened, at least everything that is not an assignment and can possibly fail
(e.g. the HTTPServer instantiation above).

[f.lopez, 2017/08/19 00:51:01]

Done.

+        httpd.serve_forever()
+    except:
+        if args.output != '-':
+            fh.close()