Translation of UI (german and english)

Bugfix for feeds
    - removed categories related and up
    - load new books now working
    - category random now working
login page is free of non accessible elements
boolean custom column is vivible in UI
books with only with certain languages can be shown
book shelfs can be deleted from UI
Anonymous user view is more resticted
Added browse of series in sidebar
Dependencys in vendor folder are updated to newer versions (licencs files are now present)
Bugfix editing Authors names
Made upload on windows working
This commit is contained in:
OzzieIsaacs
2016-11-09 19:24:33 +01:00
parent a6b6700a73
commit bbf6d9b026
1690 changed files with 149106 additions and 35697 deletions
+94 -7
View File
@@ -12,11 +12,19 @@ and `.Resolver`.
from __future__ import absolute_import, division, print_function, with_statement
import array
import inspect
import os
import sys
import zlib
try:
xrange # py2
except NameError:
xrange = range # py3
class ObjectDict(dict):
"""Makes a dictionary behave like an object, with attribute-style access.
"""
@@ -33,7 +41,7 @@ class ObjectDict(dict):
class GzipDecompressor(object):
"""Streaming gzip decompressor.
The interface is like that of `zlib.decompressobj` (without the
The interface is like that of `zlib.decompressobj` (without some of the
optional arguments, but it understands gzip headers and checksums.
"""
def __init__(self):
@@ -42,14 +50,24 @@ class GzipDecompressor(object):
# This works on cpython and pypy, but not jython.
self.decompressobj = zlib.decompressobj(16 + zlib.MAX_WBITS)
def decompress(self, value):
def decompress(self, value, max_length=None):
"""Decompress a chunk, returning newly-available data.
Some data may be buffered for later processing; `flush` must
be called when there is no more input data to ensure that
all data was processed.
If ``max_length`` is given, some input data may be left over
in ``unconsumed_tail``; you must retrieve this value and pass
it back to a future call to `decompress` if it is not empty.
"""
return self.decompressobj.decompress(value)
return self.decompressobj.decompress(value, max_length)
@property
def unconsumed_tail(self):
"""Returns the unconsumed portion left over
"""
return self.decompressobj.unconsumed_tail
def flush(self):
"""Return any remaining buffered data not yet returned by decompress.
@@ -97,16 +115,17 @@ def import_object(name):
if type('') is not type(b''):
def u(s):
return s
bytes_type = bytes
unicode_type = str
basestring_type = str
else:
def u(s):
return s.decode('unicode_escape')
bytes_type = str
unicode_type = unicode
basestring_type = basestring
# Deprecated alias that was used before we dropped py25 support.
# Left here in case anyone outside Tornado is using it.
bytes_type = bytes
if sys.version_info > (3,):
exec("""
@@ -132,6 +151,24 @@ def exec_in(code, glob, loc=None):
""")
def errno_from_exception(e):
"""Provides the errno from an Exception object.
There are cases that the errno attribute was not set so we pull
the errno out of the args but if someone instantiates an Exception
without any args you will get a tuple error. So this function
abstracts all that behavior to give you a safe way to get the
errno.
"""
if hasattr(e, 'errno'):
return e.errno
elif e.args:
return e.args[0]
else:
return None
class Configurable(object):
"""Base class for configurable interfaces.
@@ -166,7 +203,7 @@ class Configurable(object):
impl = cls
args.update(kwargs)
instance = super(Configurable, cls).__new__(impl)
# initialize vs __init__ chosen for compatiblity with AsyncHTTPClient
# initialize vs __init__ chosen for compatibility with AsyncHTTPClient
# singleton magic. If we get rid of that we can switch to __init__
# here too.
instance.initialize(**args)
@@ -201,7 +238,7 @@ class Configurable(object):
some parameters.
"""
base = cls.configurable_base()
if isinstance(impl, (unicode_type, bytes_type)):
if isinstance(impl, (unicode_type, bytes)):
impl = import_object(impl)
if impl is not None and not issubclass(impl, cls):
raise ValueError("Invalid subclass of %s" % cls)
@@ -243,6 +280,16 @@ class ArgReplacer(object):
# Not a positional parameter
self.arg_pos = None
def get_old_value(self, args, kwargs, default=None):
"""Returns the old value of the named argument without replacing it.
Returns ``default`` if the argument is not present.
"""
if self.arg_pos is not None and len(args) > self.arg_pos:
return args[self.arg_pos]
else:
return kwargs.get(self.name, default)
def replace(self, new_value, args, kwargs):
"""Replace the named argument in ``args, kwargs`` with ``new_value``.
@@ -265,6 +312,46 @@ class ArgReplacer(object):
return old_value, args, kwargs
def timedelta_to_seconds(td):
"""Equivalent to td.total_seconds() (introduced in python 2.7)."""
return (td.microseconds + (td.seconds + td.days * 24 * 3600) * 10 ** 6) / float(10 ** 6)
def _websocket_mask_python(mask, data):
"""Websocket masking function.
`mask` is a `bytes` object of length 4; `data` is a `bytes` object of any length.
Returns a `bytes` object of the same length as `data` with the mask applied
as specified in section 5.3 of RFC 6455.
This pure-python implementation may be replaced by an optimized version when available.
"""
mask = array.array("B", mask)
unmasked = array.array("B", data)
for i in xrange(len(data)):
unmasked[i] = unmasked[i] ^ mask[i % 4]
if hasattr(unmasked, 'tobytes'):
# tostring was deprecated in py32. It hasn't been removed,
# but since we turn on deprecation warnings in our tests
# we need to use the right one.
return unmasked.tobytes()
else:
return unmasked.tostring()
if (os.environ.get('TORNADO_NO_EXTENSION') or
os.environ.get('TORNADO_EXTENSION') == '0'):
# These environment variables exist to make it easier to do performance
# comparisons; they are not guaranteed to remain supported in the future.
_websocket_mask = _websocket_mask_python
else:
try:
from tornado.speedups import websocket_mask as _websocket_mask
except ImportError:
if os.environ.get('TORNADO_EXTENSION') == '1':
raise
_websocket_mask = _websocket_mask_python
def doctests():
import doctest
return doctest.DocTestSuite()