class SortedDictTests(SimpleTestCase):
def setUp(self):
self.d1 = SortedDict()
self.d1[7] = 'seven'
self.d1[1] = 'one'
self.d1[9] = 'nine'
self.d2 = SortedDict()
self.d2[1] = 'one'
self.d2[9] = 'nine'
self.d2[0] = 'nil'
self.d2[7] = 'seven'
def test_basic_methods(self):
self.assertEqual(list(six.iterkeys(self.d1)), [7, 1, 9])
self.assertEqual(list(six.itervalues(self.d1)), ['seven', 'one', 'nine'])
self.assertEqual(list(six.iteritems(self.d1)), [(7, 'seven'), (1, 'one'), (9, 'nine')])
def test_overwrite_ordering(self):
""" Overwriting an item keeps its place. """
self.d1[1] = 'ONE'
self.assertEqual(list(six.itervalues(self.d1)), ['seven', 'ONE', 'nine'])
def test_append_items(self):
""" New items go to the end. """
self.d1[0] = 'nil'
self.assertEqual(list(six.iterkeys(self.d1)), [7, 1, 9, 0])
def test_delete_and_insert(self):
"""
Deleting an item, then inserting the same key again will place it
at the end.
"""
del self.d2[7]
self.assertEqual(list(six.iterkeys(self.d2)), [1, 9, 0])
self.d2[7] = 'lucky number 7'
self.assertEqual(list(six.iterkeys(self.d2)), [1, 9, 0, 7])
if not six.PY3:
def test_change_keys(self):
"""
Changing the keys won't do anything, it's only a copy of the
keys dict.
This test doesn't make sense under Python 3 because keys is
an iterator.
"""
k = self.d2.keys()
k.remove(9)
self.assertEqual(self.d2.keys(), [1, 9, 0, 7])
def test_init_keys(self):
"""
Initialising a SortedDict with two keys will just take the first one.
A real dict will actually take the second value so we will too, but
we'll keep the ordering from the first key found.
"""
tuples = ((2, 'two'), (1, 'one'), (2, 'second-two'))
d = SortedDict(tuples)
self.assertEqual(list(six.iterkeys(d)), [2, 1])
real_dict = dict(tuples)
self.assertEqual(sorted(six.itervalues(real_dict)), ['one', 'second-two'])
# Here the order of SortedDict values *is* what we are testing
self.assertEqual(list(six.itervalues(d)), ['second-two', 'one'])
def test_overwrite(self):
self.d1[1] = 'not one'
self.assertEqual(self.d1[1], 'not one')
self.assertEqual(list(six.iterkeys(self.d1)), list(six.iterkeys(self.d1.copy())))
def test_append(self):
self.d1[13] = 'thirteen'
self.assertEqual(
repr(self.d1),
"{7: 'seven', 1: 'one', 9: 'nine', 13: 'thirteen'}"
)
def test_pop(self):
self.assertEqual(self.d1.pop(1, 'missing'), 'one')
self.assertEqual(self.d1.pop(1, 'missing'), 'missing')
# We don't know which item will be popped in popitem(), so we'll
# just check that the number of keys has decreased.
l = len(self.d1)
self.d1.popitem()
self.assertEqual(l - len(self.d1), 1)
def test_dict_equality(self):
d = SortedDict((i, i) for i in range(3))
self.assertEqual(d, {0: 0, 1: 1, 2: 2})
def test_tuple_init(self):
d = SortedDict(((1, "one"), (0, "zero"), (2, "two")))
self.assertEqual(repr(d), "{1: 'one', 0: 'zero', 2: 'two'}")
def test_pickle(self):
self.assertEqual(
pickle.loads(pickle.dumps(self.d1, 2)),
{7: 'seven', 1: 'one', 9: 'nine'}
)
def test_copy(self):
orig = SortedDict(((1, "one"), (0, "zero"), (2, "two")))
copied = copy.copy(orig)
self.assertEqual(list(six.iterkeys(orig)), [1, 0, 2])
self.assertEqual(list(six.iterkeys(copied)), [1, 0, 2])
def test_clear(self):
self.d1.clear()
self.assertEqual(self.d1, {})
self.assertEqual(self.d1.keyOrder, [])
def test_insert(self):
d = SortedDict()
with warnings.catch_warnings(record=True) as w:
warnings.simplefilter("always")
d.insert(0, "hello", "world")
assert w[0].category is PendingDeprecationWarning
def test_value_for_index(self):
d = SortedDict({"a": 3})
with warnings.catch_warnings(record=True) as w:
warnings.simplefilter("always")
self.assertEqual(d.value_for_index(0), 3)
assert w[0].category is PendingDeprecationWarning
def handle(self, *app_labels, **options):
from djangocg.db.models import get_app, get_apps, get_model
format = options.get('format')
indent = options.get('indent')
using = options.get('database')
excludes = options.get('exclude')
show_traceback = options.get('traceback')
use_natural_keys = options.get('use_natural_keys')
use_base_manager = options.get('use_base_manager')
excluded_apps = set()
excluded_models = set()
for exclude in excludes:
if '.' in exclude:
app_label, model_name = exclude.split('.', 1)
model_obj = get_model(app_label, model_name)
if not model_obj:
raise CommandError('Unknown model in excludes: %s' % exclude)
excluded_models.add(model_obj)
else:
try:
app_obj = get_app(exclude)
excluded_apps.add(app_obj)
except ImproperlyConfigured:
raise CommandError('Unknown app in excludes: %s' % exclude)
if len(app_labels) == 0:
app_list = SortedDict((app, None) for app in get_apps() if app not in excluded_apps)
else:
app_list = SortedDict()
for label in app_labels:
try:
app_label, model_label = label.split('.')
try:
app = get_app(app_label)
except ImproperlyConfigured:
raise CommandError("Unknown application: %s" % app_label)
if app in excluded_apps:
continue
model = get_model(app_label, model_label)
if model is None:
raise CommandError("Unknown model: %s.%s" % (app_label, model_label))
if app in app_list.keys():
if app_list[app] and model not in app_list[app]:
app_list[app].append(model)
else:
app_list[app] = [model]
except ValueError:
# This is just an app - no model qualifier
app_label = label
try:
app = get_app(app_label)
except ImproperlyConfigured:
raise CommandError("Unknown application: %s" % app_label)
if app in excluded_apps:
continue
app_list[app] = None
# Check that the serialization format exists; this is a shortcut to
# avoid collating all the objects and _then_ failing.
if format not in serializers.get_public_serializer_formats():
raise CommandError("Unknown serialization format: %s" % format)
try:
serializers.get_serializer(format)
except KeyError:
raise CommandError("Unknown serialization format: %s" % format)
def get_objects():
# Collate the objects to be serialized.
for model in sort_dependencies(app_list.items()):
if model in excluded_models:
continue
if not model._meta.proxy and router.allow_syncdb(using, model):
if use_base_manager:
objects = model._base_manager
else:
objects = model._default_manager
for obj in objects.using(using).\
order_by(model._meta.pk.name).iterator():
yield obj
try:
self.stdout.ending = None
serializers.serialize(format, get_objects(), indent=indent,
use_natural_keys=use_natural_keys, stream=self.stdout)
except Exception as e:
if show_traceback:
raise
raise CommandError("Unable to serialize database: %s" % e)
class CachedFilesMixin(object):
default_template = """url("%s")"""
patterns = (
(
"*.css",
(
r"""(url\(['"]{0,1}\s*(.*?)["']{0,1}\))""",
(r"""(@import\s*["']\s*(.*?)["'])""", """@import url("%s")"""),
),
),
)
def __init__(self, *args, **kwargs):
super(CachedFilesMixin, self).__init__(*args, **kwargs)
try:
self.cache = get_cache("staticfiles")
except InvalidCacheBackendError:
# Use the default backend
self.cache = default_cache
self._patterns = SortedDict()
for extension, patterns in self.patterns:
for pattern in patterns:
if isinstance(pattern, (tuple, list)):
pattern, template = pattern
else:
template = self.default_template
compiled = re.compile(pattern)
self._patterns.setdefault(extension, []).append((compiled, template))
def file_hash(self, name, content=None):
"""
Retuns a hash of the file with the given name and optional content.
"""
if content is None:
return None
md5 = hashlib.md5()
for chunk in content.chunks():
md5.update(chunk)
return md5.hexdigest()[:12]
def hashed_name(self, name, content=None):
parsed_name = urlsplit(unquote(name))
clean_name = parsed_name.path.strip()
opened = False
if content is None:
if not self.exists(clean_name):
raise ValueError("The file '%s' could not be found with %r." % (clean_name, self))
try:
content = self.open(clean_name)
except IOError:
# Handle directory paths and fragments
return name
opened = True
try:
file_hash = self.file_hash(clean_name, content)
finally:
if opened:
content.close()
path, filename = os.path.split(clean_name)
root, ext = os.path.splitext(filename)
if file_hash is not None:
file_hash = ".%s" % file_hash
hashed_name = os.path.join(path, "%s%s%s" % (root, file_hash, ext))
unparsed_name = list(parsed_name)
unparsed_name[2] = hashed_name
# Special casing for a @font-face hack, like url(myfont.eot?#iefix")
# http://www.fontspring.com/blog/the-new-bulletproof-font-face-syntax
if "?#" in name and not unparsed_name[3]:
unparsed_name[2] += "?"
return urlunsplit(unparsed_name)
def cache_key(self, name):
return "staticfiles:%s" % hashlib.md5(force_bytes(name)).hexdigest()
def url(self, name, force=False):
"""
Returns the real URL in DEBUG mode.
"""
if settings.DEBUG and not force:
hashed_name, fragment = name, ""
else:
clean_name, fragment = urldefrag(name)
if urlsplit(clean_name).path.endswith("/"): # don't hash paths
hashed_name = name
else:
cache_key = self.cache_key(name)
hashed_name = self.cache.get(cache_key)
if hashed_name is None:
hashed_name = self.hashed_name(clean_name).replace("\\", "/")
# set the cache if there was a miss
# (e.g. if cache server goes down)
self.cache.set(cache_key, hashed_name)
final_url = super(CachedFilesMixin, self).url(hashed_name)
# Special casing for a @font-face hack, like url(myfont.eot?#iefix")
# http://www.fontspring.com/blog/the-new-bulletproof-font-face-syntax
query_fragment = "?#" in name # [sic!]
if fragment or query_fragment:
urlparts = list(urlsplit(final_url))
if fragment and not urlparts[4]:
urlparts[4] = fragment
if query_fragment and not urlparts[3]:
urlparts[2] += "?"
final_url = urlunsplit(urlparts)
return unquote(final_url)
def url_converter(self, name, template=None):
"""
Returns the custom URL converter for the given file name.
"""
if template is None:
template = self.default_template
def converter(matchobj):
"""
Converts the matched URL depending on the parent level (`..`)
and returns the normalized and hashed URL using the url method
of the storage.
"""
matched, url = matchobj.groups()
# Completely ignore http(s) prefixed URLs,
# fragments and data-uri URLs
if url.startswith(("#", "http:", "https:", "data:", "//")):
return matched
name_parts = name.split(os.sep)
# Using posix normpath here to remove duplicates
url = posixpath.normpath(url)
url_parts = url.split("/")
parent_level, sub_level = url.count(".."), url.count("/")
if url.startswith("/"):
sub_level -= 1
url_parts = url_parts[1:]
if parent_level or not url.startswith("/"):
start, end = parent_level + 1, parent_level
else:
if sub_level:
if sub_level == 1:
parent_level -= 1
start, end = parent_level, 1
else:
start, end = 1, sub_level - 1
joined_result = "/".join(name_parts[:-start] + url_parts[end:])
hashed_url = self.url(unquote(joined_result), force=True)
file_name = hashed_url.split("/")[-1:]
relative_url = "/".join(url.split("/")[:-1] + file_name)
# Return the hashed version to the file
return template % unquote(relative_url)
return converter
def post_process(self, paths, dry_run=False, **options):
"""
Post process the given list of files (called from collectstatic).
Processing is actually two separate operations:
1. renaming files to include a hash of their content for cache-busting,
and copying those files to the target storage.
2. adjusting files which contain references to other files so they
refer to the cache-busting filenames.
If either of these are performed on a file, then that file is considered
post-processed.
"""
# don't even dare to process the files if we're in dry run mode
if dry_run:
return
# where to store the new paths
hashed_paths = {}
# build a list of adjustable files
matches = lambda path: matches_patterns(path, self._patterns.keys())
adjustable_paths = [path for path in paths if matches(path)]
# then sort the files by the directory level
path_level = lambda name: len(name.split(os.sep))
for name in sorted(paths.keys(), key=path_level, reverse=True):
# use the original, local file, not the copied-but-unprocessed
# file, which might be somewhere far away, like S3
storage, path = paths[name]
with storage.open(path) as original_file:
# generate the hash with the original content, even for
# adjustable files.
hashed_name = self.hashed_name(name, original_file)
# then get the original's file content..
if hasattr(original_file, "seek"):
original_file.seek(0)
hashed_file_exists = self.exists(hashed_name)
processed = False
# ..to apply each replacement pattern to the content
if name in adjustable_paths:
content = original_file.read().decode(settings.FILE_CHARSET)
for patterns in self._patterns.values():
for pattern, template in patterns:
converter = self.url_converter(name, template)
content = pattern.sub(converter, content)
if hashed_file_exists:
self.delete(hashed_name)
# then save the processed result
content_file = ContentFile(force_bytes(content))
saved_name = self._save(hashed_name, content_file)
hashed_name = force_text(saved_name.replace("\\", "/"))
processed = True
else:
# or handle the case in which neither processing nor
# a change to the original file happened
if not hashed_file_exists:
processed = True
saved_name = self._save(hashed_name, original_file)
hashed_name = force_text(saved_name.replace("\\", "/"))
# and then set the cache accordingly
hashed_paths[self.cache_key(name.replace("\\", "/"))] = hashed_name
yield name, hashed_name, processed
# Finally set the cache
self.cache.set_many(hashed_paths)