tahoe-lafs/tahoe-lafs
tahoe-lafs
/
tahoe-lafs
2
0
Fork 0
Code Issues Pull Requests 23 Packages Projects Releases 1 Wiki Activity

Merge remote-tracking branch 'origin/master' into 3581.unicode_to_argv.1

This commit is contained in:
Jean-Paul Calderone 2021-02-12 13:08:51 -05:00
parent 9b2a9e14ae 0b45c2592d
commit 8ffb083d38
29 changed files with 755 additions and 516 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

44
.circleci/config.yml
View File

@ -29,7 +29,7 @@ workflows:
- "debian-9": &DOCKERHUB_CONTEXT
context: "dockerhub-auth"
- "debian-8":
- "debian-10":
<<: *DOCKERHUB_CONTEXT
requires:
- "debian-9"
@ -86,11 +86,6 @@ workflows:
# integration tests.
- "debian-9"
# Generate the underlying data for a visualization to aid with Python 3
# porting.
- "build-porting-depgraph":
<<: *DOCKERHUB_CONTEXT
- "typechecks":
<<: *DOCKERHUB_CONTEXT
@ -107,7 +102,7 @@ workflows:
- "master"
jobs:
- "build-image-debian-8":
- "build-image-debian-10":
<<: *DOCKERHUB_CONTEXT
- "build-image-debian-9":
<<: *DOCKERHUB_CONTEXT
@ -277,11 +272,11 @@ jobs:
fi
debian-8:
debian-10:
<<: *DEBIAN
docker:
- <<: *DOCKERHUB_AUTH
image: "tahoelafsci/debian:8-py2.7"
image: "tahoelafsci/debian:10-py2.7"
user: "nobody"
@ -451,33 +446,6 @@ jobs:
# them in parallel.
nix-build --cores 3 --max-jobs 2 nix/
# Generate up-to-date data for the dependency graph visualizer.
build-porting-depgraph:
# Get a system in which we can easily install Tahoe-LAFS and all its
# dependencies. The dependency graph analyzer works by executing the code.
# It's Python, what do you expect?
<<: *DEBIAN
steps:
- "checkout"
- add_ssh_keys:
fingerprints:
# Jean-Paul Calderone <exarkun@twistedmatrix.com> (CircleCI depgraph key)
# This lets us push to tahoe-lafs/tahoe-depgraph in the next step.
- "86:38:18:a7:c0:97:42:43:18:46:55:d6:21:b0:5f:d4"
- run:
name: "Setup Python Environment"
command: |
/tmp/venv/bin/pip install -e /tmp/project
- run:
name: "Generate dependency graph data"
command: |
. /tmp/venv/bin/activate
./misc/python3/depgraph.sh
typechecks:
docker:
- <<: *DOCKERHUB_AUTH
@ -529,12 +497,12 @@ jobs:
docker push tahoelafsci/${DISTRO}:${TAG}-py${PYTHON_VERSION}
build-image-debian-8:
build-image-debian-10:
<<: *BUILD_IMAGE
environment:
DISTRO: "debian"
TAG: "8"
TAG: "10"
PYTHON_VERSION: "2.7"

2
misc/operations_helpers/provisioning/provisioning.py
View File

@ -46,7 +46,7 @@ class ProvisioningTool(rend.Page):
req = inevow.IRequest(ctx)
def getarg(name, astype=int):
if req.method != "POST":
if req.method != b"POST":
return None
if name in req.fields:
return astype(req.fields[name].value)

1
newsfragments/3326.installation Normal file
View File

@ -0,0 +1 @@
Debian 8 support has been replaced with Debian 10 support.

0
newsfragments/3592.minor Normal file
View File

0
newsfragments/3596.minor Normal file
View File

0
newsfragments/3600.minor Normal file
View File

0
newsfragments/3612.minor Normal file
View File

18
src/allmydata/test/common.py
View File

@ -454,7 +454,7 @@ class FakeCHKFileNode(object): # type: ignore # incomplete implementation
return self.storage_index
def check(self, monitor, verify=False, add_lease=False):
s = StubServer("\x00"*20)
s = StubServer(b"\x00"*20)
r = CheckResults(self.my_uri, self.storage_index,
healthy=True, recoverable=True,
count_happiness=10,
@ -588,12 +588,12 @@ class FakeMutableFileNode(object): # type: ignore # incomplete implementation
self.file_types[self.storage_index] = version
initial_contents = self._get_initial_contents(contents)
data = initial_contents.read(initial_contents.get_size())
data = "".join(data)
data = b"".join(data)
self.all_contents[self.storage_index] = data
return defer.succeed(self)
def _get_initial_contents(self, contents):
if contents is None:
return MutableData("")
return MutableData(b"")
if IMutableUploadable.providedBy(contents):
return contents
@ -647,7 +647,7 @@ class FakeMutableFileNode(object): # type: ignore # incomplete implementation
def raise_error(self):
pass
def get_writekey(self):
return "\x00"*16
return b"\x00"*16
def get_size(self):
return len(self.all_contents[self.storage_index])
def get_current_size(self):
@ -666,7 +666,7 @@ class FakeMutableFileNode(object): # type: ignore # incomplete implementation
return self.file_types[self.storage_index]
def check(self, monitor, verify=False, add_lease=False):
s = StubServer("\x00"*20)
s = StubServer(b"\x00"*20)
r = CheckResults(self.my_uri, self.storage_index,
healthy=True, recoverable=True,
count_happiness=10,
@ -677,7 +677,7 @@ class FakeMutableFileNode(object): # type: ignore # incomplete implementation
count_recoverable_versions=1,
count_unrecoverable_versions=0,
servers_responding=[s],
sharemap={"seq1-abcd-sh0": [s]},
sharemap={b"seq1-abcd-sh0": [s]},
count_wrong_shares=0,
list_corrupt_shares=[],
count_corrupt_shares=0,
@ -731,7 +731,7 @@ class FakeMutableFileNode(object): # type: ignore # incomplete implementation
def overwrite(self, new_contents):
assert not self.is_readonly()
new_data = new_contents.read(new_contents.get_size())
new_data = "".join(new_data)
new_data = b"".join(new_data)
self.all_contents[self.storage_index] = new_data
return defer.succeed(None)
def modify(self, modifier):
@ -762,7 +762,7 @@ class FakeMutableFileNode(object): # type: ignore # incomplete implementation
def update(self, data, offset):
assert not self.is_readonly()
def modifier(old, servermap, first_time):
new = old[:offset] + "".join(data.read(data.get_size()))
new = old[:offset] + b"".join(data.read(data.get_size()))
new += old[len(new):]
return new
return self.modify(modifier)
@ -881,6 +881,8 @@ class WebErrorMixin(object):
body = yield response.content()
self.assertEquals(response.code, code)
if response_substring is not None:
if isinstance(response_substring, unicode):
response_substring = response_substring.encode("utf-8")
self.assertIn(response_substring, body)
returnValue(body)

8
src/allmydata/test/common_util.py
View File

@ -203,6 +203,14 @@ def flip_one_bit(s, offset=0, size=None):
class ReallyEqualMixin(object):
def failUnlessReallyEqual(self, a, b, msg=None):
self.assertEqual(a, b, msg)
# Make sure unicode strings are a consistent type. Specifically there's
# Future newstr (backported Unicode type) vs. Python 2 native unicode
# type. They're equal, and _logically_ the same type, but have
# different types in practice.
if a.__class__ == future_str:
a = unicode(a)
if b.__class__ == future_str:
b = unicode(b)
self.assertEqual(type(a), type(b), "a :: %r (%s), b :: %r (%s), %r" % (a, type(a), b, type(b), msg))

201
src/allmydata/test/eliotutil.py
View File

@ -6,29 +6,43 @@ Tools aimed at the interaction between tests and Eliot.
# Can't use `builtins.str` because it's not JSON encodable:
# `exceptions.TypeError: <class 'future.types.newstr.newstr'> is not JSON-encodeable`
from past.builtins import unicode as str
from future.utils import PY3
from future.utils import PY2
from six import ensure_text
__all__ = [
"RUN_TEST",
"EliotLoggedRunTest",
"eliot_logged_test",
]
try:
from typing import Callable
except ImportError:
pass
from functools import (
wraps,
partial,
wraps,
)
import attr
from zope.interface import (
implementer,
)
from eliot import (
ActionType,
Field,
MemoryLogger,
ILogger,
)
from eliot.testing import (
swap_logger,
check_for_errors,
)
from eliot.testing import capture_logging
from twisted.internet.defer import (
maybeDeferred,
from twisted.python.monkey import (
MonkeyPatcher,
)
from ..util.jsonbytes import BytesJSONEncoder
@ -48,92 +62,12 @@ RUN_TEST = ActionType(
)
def eliot_logged_test(f):
"""
Decorate a test method to run in a dedicated Eliot action context.
The action will finish after the test is done (after the returned Deferred
fires, if a Deferred is returned). It will note the name of the test
being run.
All messages emitted by the test will be validated. They will still be
delivered to the global logger.
"""
# A convenient, mutable container into which nested functions can write
# state to be shared among them.
class storage(object):
pass
# On Python 3, we want to use our custom JSON encoder when validating
# messages can be encoded to JSON:
if PY3:
capture = lambda f : capture_logging(None, encoder_=BytesJSONEncoder)(f)
# On Python 3, we want to use our custom JSON encoder when validating messages
# can be encoded to JSON:
if PY2:
_memory_logger = MemoryLogger
else:
capture = lambda f : capture_logging(None)(f)
@wraps(f)
def run_and_republish(self, *a, **kw):
# Unfortunately the only way to get at the global/default logger...
# This import is delayed here so that we get the *current* default
# logger at the time the decorated function is run.
from eliot._output import _DEFAULT_LOGGER as default_logger
def republish():
# This is called as a cleanup function after capture_logging has
# restored the global/default logger to its original state. We
# can now emit messages that go to whatever global destinations
# are installed.
# storage.logger.serialize() seems like it would make more sense
# than storage.logger.messages here. However, serialize()
# explodes, seemingly as a result of double-serializing the logged
# messages. I don't understand this.
for msg in storage.logger.messages:
default_logger.write(msg)
# And now that we've re-published all of the test's messages, we
# can finish the test's action.
storage.action.finish()
@capture
def run(self, logger):
# Record the MemoryLogger for later message extraction.
storage.logger = logger
# Give the test access to the logger as well. It would be just
# fine to pass this as a keyword argument to `f` but implementing
# that now will give me conflict headaches so I'm not doing it.
self.eliot_logger = logger
return f(self, *a, **kw)
# Arrange for all messages written to the memory logger that
# `capture_logging` installs to be re-written to the global/default
# logger so they might end up in a log file somewhere, if someone
# wants. This has to be done in a cleanup function (or later) because
# capture_logging restores the original logger in a cleanup function.
# We install our cleanup function here, before we call run, so that it
# runs *after* the cleanup function capture_logging installs (cleanup
# functions are a stack).
self.addCleanup(republish)
# Begin an action that should comprise all messages from the decorated
# test method.
with RUN_TEST(name=self.id()).context() as action:
# When the test method Deferred fires, the RUN_TEST action is
# done. However, we won't have re-published the MemoryLogger
# messages into the global/default logger when this Deferred
# fires. So we need to delay finishing the action until that has
# happened. Record the action so we can do that.
storage.action = action
# Support both Deferred-returning and non-Deferred-returning
# tests.
d = maybeDeferred(run, self)
# Let the test runner do its thing.
return d
return run_and_republish
_memory_logger = lambda: MemoryLogger(encoder=BytesJSONEncoder)
@attr.s
@ -174,10 +108,91 @@ class EliotLoggedRunTest(object):
def id(self):
return self.case.id()
@eliot_logged_test
def run(self, result=None):
def run(self, result):
"""
Run the test case in the context of a distinct Eliot action.
The action will finish after the test is done. It will note the name of
the test being run.
All messages emitted by the test will be validated. They will still be
delivered to the global logger.
"""
# The idea here is to decorate the test method itself so that all of
# the extra logic happens at the point where test/application logic is
# expected to be. This `run` method is more like test infrastructure
# and things do not go well when we add too much extra behavior here.
# For example, exceptions raised here often just kill the whole
# runner.
patcher = MonkeyPatcher()
# So, grab the test method.
name = self.case._testMethodName
original = getattr(self.case, name)
decorated = with_logging(ensure_text(self.case.id()), original)
patcher.addPatch(self.case, name, decorated)
try:
# Patch it in
patcher.patch()
# Then use the rest of the machinery to run it.
return self._run_tests_with_factory(
self.case,
self.handlers,
self.last_resort,
).run(result)
finally:
# Clean up the patching for idempotency or something.
patcher.restore()
def with_logging(
test_id, # type: str
test_method, # type: Callable
):
"""
Decorate a test method with additional log-related behaviors.
1. The test method will run in a distinct Eliot action.
2. Typed log messages will be validated.
3. Logged tracebacks will be added as errors.
:param test_id: The full identifier of the test being decorated.
:param test_method: The method itself.
"""
@wraps(test_method)
def run_with_logging(*args, **kwargs):
validating_logger = _memory_logger()
original = swap_logger(None)
try:
swap_logger(_TwoLoggers(original, validating_logger))
with RUN_TEST(name=test_id):
try:
return test_method(*args, **kwargs)
finally:
check_for_errors(validating_logger)
finally:
swap_logger(original)
return run_with_logging
@implementer(ILogger)
class _TwoLoggers(object):
"""
Log to two loggers.
A single logger can have multiple destinations so this isn't typically a
useful thing to do. However, MemoryLogger has inline validation instead
of destinations. That means this *is* useful to simultaneously write to
the normal places and validate all written log messages.
"""
def __init__(self, a, b):
"""
:param ILogger a: One logger
:param ILogger b: Another logger
"""
self._a = a # type: ILogger
self._b = b # type: ILogger
def write(self, dictionary, serializer=None):
self._a.write(dictionary, serializer)
self._b.write(dictionary, serializer)

68
src/allmydata/test/test_eliotutil.py
View File

@ -18,17 +18,25 @@ if PY2:
from sys import stdout
import logging
from unittest import (
skip,
)
from fixtures import (
TempDir,
)
from testtools import (
TestCase,
)
from testtools import (
TestResult,
)
from testtools.matchers import (
Is,
IsInstance,
MatchesStructure,
Equals,
HasLength,
AfterPreprocessing,
)
from testtools.twistedsupport import (
@ -38,12 +46,16 @@ from testtools.twistedsupport import (
from eliot import (
Message,
MessageType,
fields,
FileDestination,
MemoryLogger,
)
from eliot.twisted import DeferredContext
from eliot.testing import (
capture_logging,
assertHasAction,
swap_logger,
)
from twisted.internet.defer import (
@ -173,6 +185,62 @@ class EliotLoggingTests(TestCase):
),
)
def test_validation_failure(self):
"""
If a test emits a log message that fails validation then an error is added
to the result.
"""
# Make sure we preserve the original global Eliot state.
original = swap_logger(MemoryLogger())
self.addCleanup(lambda: swap_logger(original))
class ValidationFailureProbe(SyncTestCase):
def test_bad_message(self):
# This message does not validate because "Hello" is not an
# int.
MSG = MessageType("test:eliotutil", fields(foo=int))
MSG(foo="Hello").write()
result = TestResult()
case = ValidationFailureProbe("test_bad_message")
case.run(result)
self.assertThat(
result.errors,
HasLength(1),
)
def test_skip_cleans_up(self):
"""
After a skipped test the global Eliot logging state is restored.
"""
# Save the logger that's active before we do anything so that we can
# restore it later. Also install another logger so we can compare it
# to the active logger later.
expected = MemoryLogger()
original = swap_logger(expected)
# Restore it, whatever else happens.
self.addCleanup(lambda: swap_logger(original))
class SkipProbe(SyncTestCase):
@skip("It's a skip test.")
def test_skipped(self):
pass
case = SkipProbe("test_skipped")
case.run()
# Retrieve the logger that's active now that the skipped test is done
# so we can check it against the expected value.
actual = swap_logger(MemoryLogger())
self.assertThat(
actual,
Is(expected),
)
class LogCallDeferredTests(TestCase):
"""
Tests for ``log_call_deferred``.

36
src/allmydata/test/test_hashutil.py
View File

@ -126,6 +126,42 @@ class HashUtilTests(unittest.TestCase):
base32.a2b(b"2ckv3dfzh6rgjis6ogfqhyxnzy"),
)
def test_convergence_hasher_tag(self):
"""
``_convergence_hasher_tag`` constructs the convergence hasher tag from a
unique prefix, the required, total, and segment size parameters, and a
convergence secret.
"""
self.assertEqual(
b"allmydata_immutable_content_to_key_with_added_secret_v1+"
b"16:\x42\x42\x42\x42\x42\x42\x42\x42\x42\x42\x42\x42\x42\x42\x42\x42,"
b"9:3,10,1024,",
hashutil._convergence_hasher_tag(
k=3,
n=10,
segsize=1024,
convergence=b"\x42" * 16,
),
)
def test_convergence_hasher_out_of_bounds(self):
"""
``_convergence_hasher_tag`` raises ``ValueError`` if k or n is not between
1 and 256 inclusive or if k is greater than n.
"""
segsize = 1024
secret = b"\x42" * 16
for bad_k in (0, 2, 257):
with self.assertRaises(ValueError):
hashutil._convergence_hasher_tag(
k=bad_k, n=1, segsize=segsize, convergence=secret,
)
for bad_n in (0, 1, 257):
with self.assertRaises(ValueError):
hashutil._convergence_hasher_tag(
k=2, n=bad_n, segsize=segsize, convergence=secret,
)
def test_known_answers(self):
"""
Verify backwards compatibility by comparing hash outputs for some

10
src/allmydata/test/test_util.py
View File

@ -491,12 +491,16 @@ class JSONBytes(unittest.TestCase):
"""Tests for BytesJSONEncoder."""
def test_encode_bytes(self):
"""BytesJSONEncoder can encode bytes."""
"""BytesJSONEncoder can encode bytes.
Bytes are presumed to be UTF-8 encoded.
"""
snowman = u"def\N{SNOWMAN}\uFF00"
data = {
b"hello": [1, b"cd"],
b"hello": [1, b"cd", {b"abc": [123, snowman.encode("utf-8")]}],
}
expected = {
u"hello": [1, u"cd"],
u"hello": [1, u"cd", {u"abc": [123, snowman]}],
}
# Bytes get passed through as if they were UTF-8 Unicode:
encoded = jsonbytes.dumps(data)

590
src/allmydata/test/web/test_web.py
View File

File diff suppressed because it is too large Load Diff

1
src/allmydata/util/_python3.py
View File

@ -196,5 +196,6 @@ PORTED_TEST_MODULES = [
"allmydata.test.web.test_root",
"allmydata.test.web.test_status",
"allmydata.test.web.test_util",
"allmydata.test.web.test_web",
"allmydata.test.web.test_webish",
]

36
src/allmydata/util/hashutil.py
View File

@ -176,10 +176,44 @@ def convergence_hash(k, n, segsize, data, convergence):
return h.digest()
def convergence_hasher(k, n, segsize, convergence):
def _convergence_hasher_tag(k, n, segsize, convergence):
"""
Create the convergence hashing tag.
:param int k: Required shares (in [1..256]).
:param int n: Total shares (in [1..256]).
:param int segsize: Maximum segment size.
:param bytes convergence: The convergence secret.
:return bytes: The bytestring to use as a tag in the convergence hash.
"""
assert isinstance(convergence, bytes)
if k > n:
raise ValueError(
"k > n not allowed; k = {}, n = {}".format(k, n),
)
if k < 1 or n < 1:
# It doesn't make sense to have zero shares. Zero shares carry no
# information, cannot encode any part of the application data.
raise ValueError(
"k, n < 1 not allowed; k = {}, n = {}".format(k, n),
)
if k > 256 or n > 256:
# ZFEC supports encoding application data into a maximum of 256
# shares. If we ignore the limitations of ZFEC, it may be fine to use
# a configuration with more shares than that and it may be fine to
# construct a convergence tag from such a configuration. Since ZFEC
# is the only supported encoder, though, this is moot for now.
raise ValueError(
"k, n > 256 not allowed; k = {}, n = {}".format(k, n),
)
param_tag = netstring(b"%d,%d,%d" % (k, n, segsize))
tag = CONVERGENT_ENCRYPTION_TAG + netstring(convergence) + param_tag
return tag
def convergence_hasher(k, n, segsize, convergence):
tag = _convergence_hasher_tag(k, n, segsize, convergence)
return tagged_hasher(tag, KEYLEN)

31
src/allmydata/util/jsonbytes.py
View File

@ -13,20 +13,34 @@ from future.utils import PY2
if PY2:
from future.builtins import filter, map, zip, ascii, chr, hex, input, next, oct, open, pow, round, super, bytes, dict, list, object, range, str, max, min # noqa: F401
import json
def _bytes_to_unicode(obj):
"""Convert any bytes objects to unicode, recursively."""
if isinstance(obj, bytes):
return obj.decode("utf-8")
if isinstance(obj, dict):
new_obj = {}
for k, v in obj.items():
if isinstance(k, bytes):
k = k.decode("utf-8")
v = _bytes_to_unicode(v)
new_obj[k] = v
return new_obj
if isinstance(obj, (list, set, tuple)):
return [_bytes_to_unicode(i) for i in obj]
return obj
class BytesJSONEncoder(json.JSONEncoder):
"""
A JSON encoder than can also encode bytes.
The bytes are assumed to be UTF-8 encoded Unicode strings.
"""
def default(self, o):
if isinstance(o, bytes):
return o.decode("utf-8")
return json.JSONEncoder.default(self, o)
def iterencode(self, o, **kwargs):
return json.JSONEncoder.iterencode(self, _bytes_to_unicode(o), **kwargs)
def dumps(obj, *args, **kwargs):
@ -34,13 +48,6 @@ def dumps(obj, *args, **kwargs):
The bytes are assumed to be UTF-8 encoded Unicode strings.
"""
if isinstance(obj, dict):
new_obj = {}
for k, v in obj.items():
if isinstance(k, bytes):
k = k.decode("utf-8")
new_obj[k] = v
obj = new_obj
return json.dumps(obj, cls=BytesJSONEncoder, *args, **kwargs)

2
src/allmydata/web/check_results.py
View File

@ -432,7 +432,7 @@ class DeepCheckResultsRenderer(MultiFormatResource):
return CheckResultsRenderer(self._client,
r.get_results_for_storage_index(si))
except KeyError:
raise WebError("No detailed results for SI %s" % html.escape(name),
raise WebError("No detailed results for SI %s" % html.escape(str(name, "utf-8")),
http.NOT_FOUND)
@render_exception

8
src/allmydata/web/common.py
View File

@ -186,7 +186,7 @@ def convert_children_json(nodemaker, children_json):
children = {}
if children_json:
data = json.loads(children_json)
for (namex, (ctype, propdict)) in data.iteritems():
for (namex, (ctype, propdict)) in data.items():
namex = unicode(namex)
writecap = to_bytes(propdict.get("rw_uri"))
readcap = to_bytes(propdict.get("ro_uri"))
@ -283,8 +283,8 @@ def render_time_attr(t):
# actual exception). The latter is growing increasingly annoying.
def should_create_intermediate_directories(req):
t = get_arg(req, "t", "").strip()
return bool(req.method in ("PUT", "POST") and
t = unicode(get_arg(req, "t", "").strip(), "ascii")
return bool(req.method in (b"PUT", b"POST") and
t not in ("delete", "rename", "rename-form", "check"))
def humanize_exception(exc):
@ -674,7 +674,7 @@ def url_for_string(req, url_string):
and the given URL string.
"""
url = DecodedURL.from_text(url_string.decode("utf-8"))
if url.host == b"":
if not url.host:
root = req.URLPath()
netloc = root.netloc.split(b":", 1)
if len(netloc) == 1:

16
src/allmydata/web/common_py3.py
View File

@ -40,8 +40,12 @@ def get_arg(req, argname, default=None, multiple=False):
results = []
if argname in req.args:
results.extend(req.args[argname])
if req.fields and argname in req.fields:
results.append(req.fields[argname].value)
argname_unicode = unicode(argname, "utf-8")
if req.fields and argname_unicode in req.fields:
value = req.fields[argname_unicode].value
if isinstance(value, unicode):
value = value.encode("utf-8")
results.append(value)
if multiple:
return tuple(results)
if results:
@ -79,7 +83,13 @@ class MultiFormatResource(resource.Resource, object):
if isinstance(t, bytes):
t = unicode(t, "ascii")
renderer = self._get_renderer(t)
return renderer(req)
result = renderer(req)
# On Python 3, json.dumps() returns Unicode for example, but
# twisted.web expects bytes. Instead of updating every single render
# method, just handle Unicode one time here.
if isinstance(result, unicode):
result = result.encode("utf-8")
return result
def _get_renderer(self, fmt):
"""

58
src/allmydata/web/directory.py
View File

@ -1,3 +1,11 @@
"""
TODO: When porting to Python 3, the filename handling logic seems wrong. On
Python 3 filename will _already_ be correctly decoded. So only decode if it's
bytes.
Also there's a lot of code duplication I think.
"""
from past.builtins import unicode
from urllib.parse import quote as url_quote
@ -135,7 +143,7 @@ class DirectoryNodeHandler(ReplaceMeMixin, Resource, object):
terminal = (req.prepath + req.postpath)[-1].decode('utf8') == name
nonterminal = not terminal #len(req.postpath) > 0
t = get_arg(req, b"t", b"").strip()
t = unicode(get_arg(req, b"t", b"").strip(), "ascii")
if isinstance(node_or_failure, Failure):
f = node_or_failure
f.trap(NoSuchChildError)
@ -150,10 +158,10 @@ class DirectoryNodeHandler(ReplaceMeMixin, Resource, object):
else:
# terminal node
terminal_requests = (
("POST", "mkdir"),
("PUT", "mkdir"),
("POST", "mkdir-with-children"),
("POST", "mkdir-immutable")
(b"POST", "mkdir"),
(b"PUT", "mkdir"),
(b"POST", "mkdir-with-children"),
(b"POST", "mkdir-immutable")
)
if (req.method, t) in terminal_requests:
# final directory
@ -182,8 +190,8 @@ class DirectoryNodeHandler(ReplaceMeMixin, Resource, object):
)
return d
leaf_requests = (
("PUT",""),
("PUT","uri"),
(b"PUT",""),
(b"PUT","uri"),
)
if (req.method, t) in leaf_requests:
# we were trying to find the leaf filenode (to put a new
@ -224,7 +232,7 @@ class DirectoryNodeHandler(ReplaceMeMixin, Resource, object):
FIXED_OUTPUT_TYPES = ["", "json", "uri", "readonly-uri"]
if not self.node.is_mutable() and t in FIXED_OUTPUT_TYPES:
si = self.node.get_storage_index()
if si and req.setETag('DIR:%s-%s' % (base32.b2a(si), t or "")):
if si and req.setETag(b'DIR:%s-%s' % (base32.b2a(si), t.encode("ascii") or b"")):
return b""
if not t:
@ -255,7 +263,7 @@ class DirectoryNodeHandler(ReplaceMeMixin, Resource, object):
@render_exception
def render_PUT(self, req):
t = get_arg(req, b"t", b"").strip()
t = unicode(get_arg(req, b"t", b"").strip(), "ascii")
replace = parse_replace_arg(get_arg(req, "replace", "true"))
if t == "mkdir":
@ -364,7 +372,7 @@ class DirectoryNodeHandler(ReplaceMeMixin, Resource, object):
return d
def _POST_upload(self, req):
charset = get_arg(req, "_charset", "utf-8")
charset = unicode(get_arg(req, "_charset", b"utf-8"), "utf-8")
contents = req.fields["file"]
assert contents.filename is None or isinstance(contents.filename, str)
name = get_arg(req, "name")
@ -374,7 +382,7 @@ class DirectoryNodeHandler(ReplaceMeMixin, Resource, object):
if not name:
# this prohibts empty, missing, and all-whitespace filenames
raise WebError("upload requires a name")
assert isinstance(name, str)
if isinstance(name, bytes):
name = name.decode(charset)
if "/" in name:
raise WebError("name= may not contain a slash", http.BAD_REQUEST)
@ -413,7 +421,7 @@ class DirectoryNodeHandler(ReplaceMeMixin, Resource, object):
name = get_arg(req, "name")
if not name:
raise WebError("set-uri requires a name")
charset = get_arg(req, "_charset", "utf-8")
charset = unicode(get_arg(req, "_charset", b"utf-8"), "ascii")
name = name.decode(charset)
replace = parse_replace_arg(get_arg(req, "replace", "true"))
@ -436,8 +444,8 @@ class DirectoryNodeHandler(ReplaceMeMixin, Resource, object):
# a slightly confusing error message if someone does a POST
# without a name= field. For our own HTML this isn't a big
# deal, because we create the 'unlink' POST buttons ourselves.
name = ''
charset = get_arg(req, "_charset", "utf-8")
name = b''
charset = unicode(get_arg(req, "_charset", b"utf-8"), "ascii")
name = name.decode(charset)
d = self.node.delete(name)
d.addCallback(lambda res: "thing unlinked")
@ -453,7 +461,7 @@ class DirectoryNodeHandler(ReplaceMeMixin, Resource, object):
return self._POST_relink(req)
def _POST_relink(self, req):
charset = get_arg(req, "_charset", "utf-8")
charset = unicode(get_arg(req, "_charset", b"utf-8"), "ascii")
replace = parse_replace_arg(get_arg(req, "replace", "true"))
from_name = get_arg(req, "from_name")
@ -624,14 +632,14 @@ class DirectoryNodeHandler(ReplaceMeMixin, Resource, object):
# TODO test handling of bad JSON
raise
cs = {}
for name, (file_or_dir, mddict) in children.iteritems():
for name, (file_or_dir, mddict) in children.items():
name = unicode(name) # json returns str *or* unicode
writecap = mddict.get('rw_uri')
if writecap is not None:
writecap = str(writecap)
writecap = writecap.encode("utf-8")
readcap = mddict.get('ro_uri')
if readcap is not None:
readcap = str(readcap)
readcap = readcap.encode("utf-8")
cs[name] = (writecap, readcap, mddict.get('metadata'))
d = self.node.set_children(cs, replace)
d.addCallback(lambda res: "Okay so I did it.")
@ -1144,8 +1152,8 @@ def _slashify_path(path):
in it
"""
if not path:
return ""
return "/".join([p.encode("utf-8") for p in path])
return b""
return b"/".join([p.encode("utf-8") for p in path])
def _cap_to_link(root, path, cap):
@ -1234,10 +1242,10 @@ class ManifestResults(MultiFormatResource, ReloadMixin):
req.setHeader("content-type", "text/plain")
lines = []
is_finished = self.monitor.is_finished()
lines.append("finished: " + {True: "yes", False: "no"}[is_finished])
lines.append(b"finished: " + {True: b"yes", False: b"no"}[is_finished])
for path, cap in self.monitor.get_status()["manifest"]:
lines.append(_slashify_path(path) + " " + cap)
return "\n".join(lines) + "\n"
lines.append(_slashify_path(path) + b" " + cap)
return b"\n".join(lines) + b"\n"
def render_JSON(self, req):
req.setHeader("content-type", "text/plain")
@ -1290,7 +1298,7 @@ class DeepSizeResults(MultiFormatResource):
+ stats.get("size-mutable-files", 0)
+ stats.get("size-directories", 0))
output += "size: %d\n" % total
return output
return output.encode("utf-8")
render_TEXT = render_HTML
def render_JSON(self, req):
@ -1315,7 +1323,7 @@ class DeepStatsResults(Resource, object):
req.setHeader("content-type", "text/plain")
s = self.monitor.get_status().copy()
s["finished"] = self.monitor.is_finished()
return json.dumps(s, indent=1)
return json.dumps(s, indent=1).encode("utf-8")
@implementer(IPushProducer)

16
src/allmydata/web/filenode.py
View File

@ -127,7 +127,7 @@ class PlaceHolderNodeHandler(Resource, ReplaceMeMixin):
http.NOT_IMPLEMENTED)
if not t:
return self.replace_me_with_a_child(req, self.client, replace)
if t == "uri":
if t == b"uri":
return self.replace_me_with_a_childcap(req, self.client, replace)
raise WebError("PUT to a file: bad t=%s" % t)
@ -188,8 +188,8 @@ class FileNodeHandler(Resource, ReplaceMeMixin, object):
# if the client already has the ETag then we can
# short-circuit the whole process.
si = self.node.get_storage_index()
if si and req.setETag('%s-%s' % (base32.b2a(si), t or "")):
return ""
if si and req.setETag(b'%s-%s' % (base32.b2a(si), t.encode("ascii") or b"")):
return b""
if not t:
# just get the contents
@ -281,7 +281,7 @@ class FileNodeHandler(Resource, ReplaceMeMixin, object):
assert self.parentnode and self.name
return self.replace_me_with_a_child(req, self.client, replace)
if t == "uri":
if t == b"uri":
if not replace:
raise ExistingChildError()
assert self.parentnode and self.name
@ -309,7 +309,7 @@ class FileNodeHandler(Resource, ReplaceMeMixin, object):
assert self.parentnode and self.name
d = self.replace_me_with_a_formpost(req, self.client, replace)
else:
raise WebError("POST to file: bad t=%s" % t)
raise WebError("POST to file: bad t=%s" % unicode(t, "ascii"))
return handle_when_done(req, d)
@ -439,7 +439,7 @@ class FileDownloader(Resource, object):
# bytes we were given in the URL. See the comment in
# FileNodeHandler.render_GET for the sad details.
req.setHeader("content-disposition",
'attachment; filename="%s"' % self.filename)
b'attachment; filename="%s"' % self.filename)
filesize = self.filenode.get_size()
assert isinstance(filesize, (int,long)), filesize
@ -475,8 +475,8 @@ class FileDownloader(Resource, object):
size = contentsize
req.setHeader("content-length", b"%d" % contentsize)
if req.method == "HEAD":
return ""
if req.method == b"HEAD":
return b""
d = self.filenode.read(req, first, size)

11
src/allmydata/web/info.py
View File

@ -1,5 +1,6 @@
import os, urllib
import os
from urllib.parse import quote as urlquote
from twisted.python.filepath import FilePath
from twisted.web.template import tags as T, Element, renderElement, XMLFile, renderer
@ -180,7 +181,7 @@ class MoreInfoElement(Element):
else:
return ""
root = self.get_root(req)
quoted_uri = urllib.quote(node.get_uri())
quoted_uri = urlquote(node.get_uri())
text_plain_url = "%s/file/%s/@@named=/raw.txt" % (root, quoted_uri)
return T.li("Raw data as ", T.a("text/plain", href=text_plain_url))
@ -196,7 +197,7 @@ class MoreInfoElement(Element):
@renderer
def check_form(self, req, tag):
node = self.original
quoted_uri = urllib.quote(node.get_uri())
quoted_uri = urlquote(node.get_uri())
target = self.get_root(req) + "/uri/" + quoted_uri
if IDirectoryNode.providedBy(node):
target += "/"
@ -236,8 +237,8 @@ class MoreInfoElement(Element):
def overwrite_form(self, req, tag):
node = self.original
root = self.get_root(req)
action = "%s/uri/%s" % (root, urllib.quote(node.get_uri()))
done_url = "%s/uri/%s?t=info" % (root, urllib.quote(node.get_uri()))
action = "%s/uri/%s" % (root, urlquote(node.get_uri()))
done_url = "%s/uri/%s?t=info" % (root, urlquote(node.get_uri()))
overwrite = T.form(action=action, method="post",
enctype="multipart/form-data")(
T.fieldset(

7
src/allmydata/web/operations.py
View File

@ -1,3 +1,4 @@
from past.builtins import unicode
import time
from hyperlink import (
@ -101,12 +102,12 @@ class OphandleTable(resource.Resource, service.Service):
def getChild(self, name, req):
ophandle = name
if ophandle not in self.handles:
raise WebError("unknown/expired handle '%s'" % escape(ophandle),
raise WebError("unknown/expired handle '%s'" % escape(unicode(ophandle, "utf-8")),
NOT_FOUND)
(monitor, renderer, when_added) = self.handles[ophandle]
t = get_arg(req, "t", "status")
if t == "cancel" and req.method == "POST":
if t == b"cancel" and req.method == b"POST":
monitor.cancel()
# return the status anyways, but release the handle
self._release_ophandle(ophandle)
@ -151,7 +152,7 @@ class ReloadMixin(object):
@renderer
def refresh(self, req, tag):
if self.monitor.is_finished():
return ""
return b""
tag.attributes["http-equiv"] = "refresh"
tag.attributes["content"] = str(self.REFRESH_TIME)
return tag

15
src/allmydata/web/root.py
View File

@ -1,4 +1,5 @@
from future.utils import PY3
from past.builtins import unicode
import os
import time
@ -97,7 +98,7 @@ class URIHandler(resource.Resource, object):
either "PUT /uri" to create an unlinked file, or
"PUT /uri?t=mkdir" to create an unlinked directory
"""
t = get_arg(req, "t", "").strip()
t = unicode(get_arg(req, "t", "").strip(), "utf-8")
if t == "":
file_format = get_format(req, "CHK")
mutable_type = get_mutable_type(file_format)
@ -120,7 +121,7 @@ class URIHandler(resource.Resource, object):
unlinked file or "POST /uri?t=mkdir" to create a
new directory
"""
t = get_arg(req, "t", "").strip()
t = unicode(get_arg(req, "t", "").strip(), "ascii")
if t in ("", "upload"):
file_format = get_format(req)
mutable_type = get_mutable_type(file_format)
@ -177,7 +178,7 @@ class FileHandler(resource.Resource, object):
@exception_to_child
def getChild(self, name, req):
if req.method not in ("GET", "HEAD"):
if req.method not in (b"GET", b"HEAD"):
raise WebError("/file can only be used with GET or HEAD")
# 'name' must be a file URI
try:
@ -200,7 +201,7 @@ class IncidentReporter(MultiFormatResource):
@render_exception
def render(self, req):
if req.method != "POST":
if req.method != b"POST":
raise WebError("/report_incident can only be used with POST")
log.msg(format="User reports incident through web page: %(details)s",
@ -255,11 +256,11 @@ class Root(MultiFormatResource):
if not path:
# Render "/" path.
return self
if path == "helper_status":
if path == b"helper_status":
# the Helper isn't attached until after the Tub starts, so this child
# needs to created on each request
return status.HelperStatus(self._client.helper)
if path == "storage":
if path == b"storage":
# Storage isn't initialized until after the web hierarchy is
# constructed so this child needs to be created later than
# `__init__`.
@ -293,7 +294,7 @@ class Root(MultiFormatResource):
self._describe_server(server)
for server
in broker.get_known_servers()
))
), key=lambda o: sorted(o.items()))
def _describe_server(self, server):

14
src/allmydata/web/status.py
View File

@ -284,7 +284,7 @@ def _find_overlap(events, start_key, end_key):
rows = []
for ev in events:
ev = ev.copy()
if ev.has_key('server'):
if 'server' in ev:
ev["serverid"] = ev["server"].get_longname()
del ev["server"]
# find an empty slot in the rows
@ -362,8 +362,8 @@ def _find_overlap_requests(events):
def _color(server):
h = hashlib.sha256(server.get_serverid()).digest()
def m(c):
return min(ord(c) / 2 + 0x80, 0xff)
return "#%02x%02x%02x" % (m(h[0]), m(h[1]), m(h[2]))
return min(ord(c) // 2 + 0x80, 0xff)
return "#%02x%02x%02x" % (m(h[0:1]), m(h[1:2]), m(h[2:3]))
class _EventJson(Resource, object):
@ -426,7 +426,7 @@ class DownloadStatusPage(Resource, object):
"""
super(DownloadStatusPage, self).__init__()
self._download_status = download_status
self.putChild("event_json", _EventJson(self._download_status))
self.putChild(b"event_json", _EventJson(self._download_status))
@render_exception
def render_GET(self, req):
@ -1288,14 +1288,14 @@ class Status(MultiFormatResource):
# final URL segment will be an empty string. Resources can
# thus know if they were requested with or without a final
# slash."
if not path and request.postpath != ['']:
if not path and request.postpath != [b'']:
return self
h = self.history
try:
stype, count_s = path.split("-")
stype, count_s = path.split(b"-")
except ValueError:
raise WebError("no '-' in '{}'".format(path))
raise WebError("no '-' in '{}'".format(unicode(path, "utf-8")))
count = int(count_s)
stype = unicode(stype, "ascii")
if stype == "up":

17
src/allmydata/web/unlinked.py
View File

@ -1,5 +1,6 @@
from past.builtins import unicode
import urllib
from urllib.parse import quote as urlquote
from twisted.web import http
from twisted.internet import defer
@ -65,8 +66,8 @@ def POSTUnlinkedCHK(req, client):
# if when_done= is provided, return a redirect instead of our
# usual upload-results page
def _done(upload_results, redir_to):
if "%(uri)s" in redir_to:
redir_to = redir_to.replace("%(uri)s", urllib.quote(upload_results.get_uri()))
if b"%(uri)s" in redir_to:
redir_to = redir_to.replace(b"%(uri)s", urlquote(upload_results.get_uri()).encode("utf-8"))
return url_for_string(req, redir_to)
d.addCallback(_done, when_done)
else:
@ -118,8 +119,8 @@ class UploadResultsElement(status.UploadResultsRendererMixin):
def download_link(self, req, tag):
d = self.upload_results()
d.addCallback(lambda res:
tags.a("/uri/" + res.get_uri(),
href="/uri/" + urllib.quote(res.get_uri())))
tags.a("/uri/" + unicode(res.get_uri(), "utf-8"),
href="/uri/" + urlquote(unicode(res.get_uri(), "utf-8"))))
return d
@ -158,7 +159,7 @@ def POSTUnlinkedCreateDirectory(req, client):
redirect = get_arg(req, "redirect_to_result", "false")
if boolean_of_arg(redirect):
def _then_redir(res):
new_url = "uri/" + urllib.quote(res.get_uri())
new_url = "uri/" + urlquote(res.get_uri())
req.setResponseCode(http.SEE_OTHER) # 303
req.setHeader('location', new_url)
return ''
@ -176,7 +177,7 @@ def POSTUnlinkedCreateDirectoryWithChildren(req, client):
redirect = get_arg(req, "redirect_to_result", "false")
if boolean_of_arg(redirect):
def _then_redir(res):
new_url = "uri/" + urllib.quote(res.get_uri())
new_url = "uri/" + urlquote(res.get_uri())
req.setResponseCode(http.SEE_OTHER) # 303
req.setHeader('location', new_url)
return ''
@ -194,7 +195,7 @@ def POSTUnlinkedCreateImmutableDirectory(req, client):
redirect = get_arg(req, "redirect_to_result", "false")
if boolean_of_arg(redirect):
def _then_redir(res):
new_url = "uri/" + urllib.quote(res.get_uri())
new_url = "uri/" + urlquote(res.get_uri())
req.setResponseCode(http.SEE_OTHER) # 303
req.setHeader('location', new_url)
return ''

44
src/allmydata/webish.py
View File

@ -44,6 +44,43 @@ from .web.storage_plugins import (
StoragePlugins,
)
if PY2:
FileUploadFieldStorage = FieldStorage
else:
class FileUploadFieldStorage(FieldStorage):
"""
Do terrible things to ensure files are still bytes.
On Python 2, uploaded files were always bytes. On Python 3, there's a
heuristic: if the filename is set on a field, it's assumed to be a file
upload and therefore bytes. If no filename is set, it's Unicode.
Unfortunately, we always want it to be bytes, and Tahoe-LAFS also
enables setting the filename not via the MIME filename, but via a
separate field called "name".
Thus we need to do this ridiculous workaround. Mypy doesn't like it
either, thus the ``# type: ignore`` below.
Source for idea:
https://mail.python.org/pipermail/python-dev/2017-February/147402.html
"""
@property # type: ignore
def filename(self):
if self.name == "file" and not self._mime_filename:
# We use the file field to upload files, see directory.py's
# _POST_upload. Lack of _mime_filename means we need to trick
# FieldStorage into thinking there is a filename so it'll
# return bytes.
return "unknown-filename"
return self._mime_filename
@filename.setter
def filename(self, value):
self._mime_filename = value
class TahoeLAFSRequest(Request, object):
"""
``TahoeLAFSRequest`` adds several features to a Twisted Web ``Request``
@ -94,7 +131,8 @@ class TahoeLAFSRequest(Request, object):
headers['content-length'] = str(self.content.tell())
self.content.seek(0)
self.fields = FieldStorage(self.content, headers, environ={'REQUEST_METHOD': 'POST'})
self.fields = FileUploadFieldStorage(
self.content, headers, environ={'REQUEST_METHOD': 'POST'})
self.content.seek(0)
self._tahoeLAFSSecurityPolicy()
@ -211,7 +249,7 @@ class WebishServer(service.MultiService):
# use to test ophandle expiration.
self._operations = OphandleTable(clock)
self._operations.setServiceParent(self)
self.root.putChild("operations", self._operations)
self.root.putChild(b"operations", self._operations)
self.root.putChild(b"storage-plugins", StoragePlugins(client))
@ -220,7 +258,7 @@ class WebishServer(service.MultiService):
self.site = TahoeLAFSSite(tempdir, self.root)
self.staticdir = staticdir # so tests can check
if staticdir:
self.root.putChild("static", static.File(staticdir))
self.root.putChild(b"static", static.File(staticdir))
if re.search(r'^\d', webport):
webport = "tcp:"+webport # twisted warns about bare "0" or "3456"
# strports must be native strings.

1
tox.ini
View File

@ -114,6 +114,7 @@ commands =
[testenv:typechecks]
basepython = python3
skip_install = True
deps =
mypy