def update_headers(self, resp):
headers = resp.headers
if 'expires' in headers:
return {}
if 'cache-control' in headers and headers['cache-control'] != 'public':
return {}
if resp.status not in self.cacheable_by_default_statuses:
return {}
if 'date' not in headers or 'last-modified' not in headers:
return {}
date = calendar.timegm(parsedate_tz(headers['date']))
last_modified = parsedate(headers['last-modified'])
if date is None or last_modified is None:
return {}
now = time.time()
current_age = max(0, now - date)
delta = date - calendar.timegm(last_modified)
freshness_lifetime = max(0, min(delta / 10, 24 * 3600))
if freshness_lifetime <= current_age:
return {}
expires = date + freshness_lifetime
return {'expires': time.strftime(TIME_FMT, time.gmtime(expires))}
def update_headers(self, resp):
headers = resp.headers
if "expires" in headers:
return {}
if "cache-control" in headers and headers["cache-control"] != "public":
return {}
if resp.status not in self.cacheable_by_default_statuses:
return {}
if "date" not in headers or "last-modified" not in headers:
return {}
date = calendar.timegm(parsedate_tz(headers["date"]))
last_modified = parsedate(headers["last-modified"])
if date is None or last_modified is None:
return {}
now = time.time()
current_age = max(0, now - date)
delta = date - calendar.timegm(last_modified)
freshness_lifetime = max(0, min(delta / 10, 24 * 3600))
if freshness_lifetime <= current_age:
return {}
expires = date + freshness_lifetime
return {"expires": time.strftime(TIME_FMT, time.gmtime(expires))}
def parse_date(value):
"""Parse one of the following date formats into a datetime object:
.. sourcecode:: text
Sun, 06 Nov 1994 08:49:37 GMT ; RFC 822, updated by RFC 1123
Sunday, 06-Nov-94 08:49:37 GMT ; RFC 850, obsoleted by RFC 1036
Sun Nov 6 08:49:37 1994 ; ANSI C's asctime() format
If parsing fails the return value is `None`.
:param value: a string with a supported date format.
:return: a :class:`datetime.datetime` object.
"""
if value:
t = parsedate_tz(value.strip())
if t is not None:
try:
year = t[0]
# unfortunately that function does not tell us if two digit
# years were part of the string, or if they were prefixed
# with two zeroes. So what we do is to assume that 69-99
# refer to 1900, and everything below to 2000
if year >= 0 and year <= 68:
year += 2000
elif year >= 69 and year <= 99:
year += 1900
return datetime(*((year,) + t[1:7])) - \
timedelta(seconds=t[-1] or 0)
except (ValueError, OverflowError):
return None
def cached_request(self, request):
"""
Return a cached response if it exists in the cache, otherwise
return False.
"""
cache_url = self.cache_url(request.url)
logger.debug('Looking up "%s" in the cache', cache_url)
cc = self.parse_cache_control(request.headers)
# Bail out if the request insists on fresh data
if "no-cache" in cc:
logger.debug('Request header has "no-cache", cache bypassed')
return False
if "max-age" in cc and cc["max-age"] == 0:
logger.debug('Request header has "max_age" as 0, cache bypassed')
return False
# Request allows serving from the cache, let's see if we find something
cache_data = self.cache.get(cache_url)
if cache_data is None:
logger.debug("No cache entry available")
return False
# Check whether it can be deserialized
resp = self.serializer.loads(request, cache_data)
if not resp:
logger.warning("Cache entry deserialization failed, entry ignored")
return False
# If we have a cached 301, return it immediately. We don't
# need to test our response for other headers b/c it is
# intrinsically "cacheable" as it is Permanent.
# See:
# https://tools.ietf.org/html/rfc7231#section-6.4.2
#
# Client can try to refresh the value by repeating the request
# with cache busting headers as usual (ie no-cache).
if resp.status == 301:
msg = (
'Returning cached "301 Moved Permanently" response '
"(ignoring date and etag information)"
)
logger.debug(msg)
return resp
headers = CaseInsensitiveDict(resp.headers)
if not headers or "date" not in headers:
if "etag" not in headers:
# Without date or etag, the cached response can never be used
# and should be deleted.
logger.debug("Purging cached response: no date or etag")
self.cache.delete(cache_url)
logger.debug("Ignoring cached response: no date")
return False
now = time.time()
date = calendar.timegm(parsedate_tz(headers["date"]))
current_age = max(0, now - date)
logger.debug("Current age based on date: %i", current_age)
# TODO: There is an assumption that the result will be a
# urllib3 response object. This may not be best since we
# could probably avoid instantiating or constructing the
# response until we know we need it.
resp_cc = self.parse_cache_control(headers)
# determine freshness
freshness_lifetime = 0
# Check the max-age pragma in the cache control header
if "max-age" in resp_cc:
freshness_lifetime = resp_cc["max-age"]
logger.debug("Freshness lifetime from max-age: %i", freshness_lifetime)
# If there isn't a max-age, check for an expires header
elif "expires" in headers:
expires = parsedate_tz(headers["expires"])
if expires is not None:
expire_time = calendar.timegm(expires) - date
freshness_lifetime = max(0, expire_time)
logger.debug("Freshness lifetime from expires: %i", freshness_lifetime)
# Determine if we are setting freshness limit in the
# request. Note, this overrides what was in the response.
if "max-age" in cc:
freshness_lifetime = cc["max-age"]
logger.debug(
"Freshness lifetime from request max-age: %i", freshness_lifetime
)
if "min-fresh" in cc:
min_fresh = cc["min-fresh"]
# adjust our current age by our min fresh
current_age += min_fresh
logger.debug("Adjusted current age from min-fresh: %i", current_age)
# Return entry if it is fresh enough
if freshness_lifetime > current_age:
logger.debug('The response is "fresh", returning cached response')
logger.debug("%i > %i", freshness_lifetime, current_age)
return resp
# we're not fresh. If we don't have an Etag, clear it out
if "etag" not in headers:
logger.debug('The cached response is "stale" with no etag, purging')
self.cache.delete(cache_url)
# return the original handler
return False