datatracker/ietf/middleware.py
2020-09-21 12:39:00 +00:00

74 lines
2.9 KiB
Python

# Copyright The IETF Trust 2007-2020, All Rights Reserved
# -*- coding: utf-8 -*-
from django.db import connection
from django.db.utils import OperationalError
from django.shortcuts import render
from django.http import HttpResponsePermanentRedirect
from ietf.utils.log import log, exc_parts
from ietf.utils.mail import log_smtp_exception
import re
import smtplib
import unicodedata
def sql_log_middleware(get_response):
def sql_log(request):
response = get_response(request)
for q in connection.queries:
if re.match('(update|insert)', q['sql'], re.IGNORECASE):
log(q['sql'])
return response
return sql_log
class SMTPExceptionMiddleware(object):
def __init__(self, get_response):
self.get_response = get_response
def __call__(self, request):
return self.get_response(request)
def process_exception(self, request, exception):
if isinstance(exception, smtplib.SMTPException):
(extype, value, tb) = log_smtp_exception(exception)
return render(request, 'email_failed.html',
{'exception': extype, 'args': value, 'traceback': "".join(tb)} )
return None
class Utf8ExceptionMiddleware(object):
def __init__(self, get_response):
self.get_response = get_response
def __call__(self, request):
return self.get_response(request)
def process_exception(self, request, exception):
if isinstance(exception, OperationalError):
extype, e, tb = exc_parts()
if e.args[0] == 1366:
log("Database 4-byte utf8 exception: %s: %s" % (extype, e))
return render(request, 'utf8_4byte_failed.html',
{'exception': extype, 'args': e.args, 'traceback': "".join(tb)} )
return None
def redirect_trailing_period_middleware(get_response):
def redirect_trailing_period(request):
response = get_response(request)
if response.status_code == 404 and request.path.endswith("."):
return HttpResponsePermanentRedirect(request.path.rstrip("."))
return response
return redirect_trailing_period
def unicode_nfkc_normalization_middleware(get_response):
def unicode_nfkc_normalization(request):
"""Do Unicode NFKC normalization to turn ligatures into individual characters.
This was prompted by somebody actually requesting an url for /wg/ipfix/charter
where the 'fi' was composed of an \\ufb01 ligature...
There are probably other elements of a request which may need this normalization
too, but let's put that in as it comes up, rather than guess ahead.
"""
request.META["PATH_INFO"] = unicodedata.normalize('NFKC', request.META["PATH_INFO"])
request.path_info = unicodedata.normalize('NFKC', request.path_info)
response = get_response(request)
return response
return unicode_nfkc_normalization