| 1 | """
|
|---|
| 2 | XHTML Degrader Middleware.
|
|---|
| 3 |
|
|---|
| 4 | When sending contents with the XHTML media type, application/xhtml+xml, this
|
|---|
| 5 | module checks to ensure that the user agent (web browser) is capable of
|
|---|
| 6 | rendering it. If not, it changes the media type to text/html and makes the
|
|---|
| 7 | contents more "HTML-friendly" (as per the XHTML 1.0 HTML Compatibility
|
|---|
| 8 | Guidelines).
|
|---|
| 9 | """
|
|---|
| 10 |
|
|---|
| 11 | import re
|
|---|
| 12 |
|
|---|
| 13 | _MEDIA_TYPE_RE = re.compile(r'application\/xhtml\+xml')
|
|---|
| 14 |
|
|---|
| 15 | _EMPTY_TAG_END_RE = re.compile(r'(?<=\S)\/\>')
|
|---|
| 16 |
|
|---|
| 17 | _PROCESSING_INSTRUCTION_RE = re.compile(r'\<\?.*\?\>')
|
|---|
| 18 |
|
|---|
| 19 | def _supports_xhtml(request):
|
|---|
| 20 | """Examines an HTTP request header to determine whether the user agent
|
|---|
| 21 | supports the XHTML media type (application/xhtml+xml). Returns True or
|
|---|
| 22 | False."""
|
|---|
| 23 | if 'HTTP_ACCEPT' in request.META:
|
|---|
| 24 | accept_header = request.META['HTTP_ACCEPT'].strip().lower()
|
|---|
| 25 | else:
|
|---|
| 26 | # No 'Accept' header; assume XHTML media type is preferable.
|
|---|
| 27 | return True
|
|---|
| 28 | if accept_header == '' or '/xhtml+xml' in accept_header \
|
|---|
| 29 | or 'application/*' in accept_header \
|
|---|
| 30 | or (accept_header == '*/*' and not \
|
|---|
| 31 | ('HTTP_USER_AGENT' in request.META and \
|
|---|
| 32 | request.META['HTTP_USER_AGENT'][0:10] == 'Mozilla/4.')):
|
|---|
| 33 | # User agent probably supports XHTML media type.
|
|---|
| 34 | return True
|
|---|
| 35 | else:
|
|---|
| 36 | # No reference to XHTML support.
|
|---|
| 37 | return False
|
|---|
| 38 |
|
|---|
| 39 | class XhtmlDegraderMiddleware(object):
|
|---|
| 40 | """Django middleware that "degrades" any contents sent as XHTML if the
|
|---|
| 41 | requesting browser doesn't support the XHTML media type. Degrading involves
|
|---|
| 42 | switching the content type to text/html, removing XML processing
|
|---|
| 43 | instructions, etc.
|
|---|
| 44 |
|
|---|
| 45 | If the HTTP response is already set to text/html, or set to any media type
|
|---|
| 46 | other than application/xhtml+xml, this middleware will have no effect.
|
|---|
| 47 | """
|
|---|
| 48 |
|
|---|
| 49 | def process_response(self, request, response):
|
|---|
| 50 | # Check if this is XHTML, and check if the user agent supports XHTML.
|
|---|
| 51 | if response['Content-Type'].split(';')[0] != 'application/xhtml+xml' \
|
|---|
| 52 | or _supports_xhtml(request):
|
|---|
| 53 | # The content is fine, simply return it.
|
|---|
| 54 | return response
|
|---|
| 55 | # The content is XHTML, and the user agent doesn't support it.
|
|---|
| 56 | # Fix the media type:
|
|---|
| 57 | response['Content-Type'] = _MEDIA_TYPE_RE.sub('text/html',
|
|---|
| 58 | response['Content-Type'], 1)
|
|---|
| 59 | if 'charset' not in response['Content-Type']:
|
|---|
| 60 | response['Content-Type'] += '; charset=utf-8'
|
|---|
| 61 | # Modify the response contents as required:
|
|---|
| 62 | # Remove any XML processing instructions:
|
|---|
| 63 | response.content = _PROCESSING_INSTRUCTION_RE.sub('',
|
|---|
| 64 | response.content)
|
|---|
| 65 | # Ensure there's a space before the trailing '/>' of empty elements:
|
|---|
| 66 | response.content = _EMPTY_TAG_END_RE.sub(' />',
|
|---|
| 67 | response.content)
|
|---|
| 68 | # Lose any excess whitespace:
|
|---|
| 69 | response.content = response.content.strip()
|
|---|
| 70 | if response.content[0:9] != '<!DOCTYPE':
|
|---|
| 71 | # Add a DOCTYPE, so that the user agent isn't in "quirks" mode.
|
|---|
| 72 | response.content = '<!DOCTYPE html>\n' + response.content
|
|---|
| 73 | return response
|
|---|