* feat(ooniprobe): discard lists not in selected categories
One day we may make an integration mistake and for any reason
we may end up with URLs that do not belong to the categories
originally selected by the user. If that happens, it's nice to
have a safety net where we remove URLs that do not belong to
the right category before proceeding with testing.
This diff was conceived while discussing the robustness of
https://github.com/ooni/probe/issues/1299 with @hellais.
* fix behavior and add unit test
* more robust
* fetch RiseupVPN CA cert with MultiGetter. It allows us to write better tests and ensures this test step is added in the logs
* Implement TransportStatus for RiseupVPN tests. It indicates if a whole transport is blocked, which is considered as a test anomaly
* Redesign unit tests for RiseupVPN. Instead of a real backend, mocked server responses are used. Tests for invalid CA certs and for TransportStatus are added.
* Update internal/engine/experiment/riseupvpn/riseupvpn.go
Co-authored-by: Simone Basso <bassosimone@gmail.com>
We cannot control anymore the maximum number of URLs using the API
because now we are using check-in, that has no such limit.
We could theoretically clamp the number of URLs to measure after
the call to check-in, and still honour the setting.
Yet, the right thing to do seems to introduce a max runtime variable
because that is what desktop and mobile do.
Thus, introduce code that warns the user about the change in the
settings, should they have set the URL limit to nonzero.
We are going to do a best effort conversion from the URL limit to
the maximum runtime for the rest of 2021.
Since then, we will silently ignore the URL limit.
This work is part of https://github.com/ooni/probe/issues/1299.
This change allows us to have all fasts tests together. They are
mostly unit tests or integration tests that do not require the
network. The advantage of this strategy is the following. We can
now run all these tests with a single click in VSCode. In turn,
doing that tells us which lines of code we are not covering.
The tests requiring the network are in a separate file, so we can
easily see which lines of code are testing without using the network
and which ones instead depend on that. (Currently, 100% of the
inputloader.go file is tested without using the network.)
While there, rename the other file such that is clear that it
contains tests requiring the network. We now have some tests in
inputloader_test.go that are not strictly unit tests.
This refactoring was identified as useful while working
on https://github.com/ooni/probe/issues/1299.
We used to have an external package called libminiooni so that
third parties could use it. We wrote this such that we could
support github.com/bassosimone/aladdin.
That was actually a not-so-good idea because it added to the APIs
we needed to maintain.
Since the merge of engine into cli, such an API is not public
anymore and aladdin has been deprecated and archived.
Therefore, we can now cleanup the situation and merge libminiooni
into miniooni again, thus making the codebase more local.
This cleanup has been identified while working on
https://github.com/ooni/probe/issues/1299.
* ongoing work
* reduce diff with master
* feat(inputloader): use the check-in API
Part of https://github.com/ooni/probe/issues/1299
* fix: better naming for a variable
* chore: add more tests
* fix: add one more TODO
* feat(session): expose CheckIn method
It seems to me the right thing to do is to query the CheckIn API
from the Session rather than querying it from InputLoader.
Then, InputLoader could just take a reference to a Session-like
interface that allows this functionality.
So, this diff exposes the Session.CheckIn method.
Doing that, in turn, required some refactoring to allow for
more and better unit tests.
While doing that, I also noticed that Session required a mutex
to be a well-behaving type, so I did that.
While doing that, I also tried to cover all the lines in session.go
and, as part of that, I have removed unused code.
Reference issue: https://github.com/ooni/probe/issues/1299.
* fix: reinstate comment I shan't have removed
* fix: repair broken test
* fix: a bit more coverage, annotations, etc.
* Update internal/engine/session.go
* Update internal/engine/session_integration_test.go
* Update internal/engine/session_internal_test.go
There was a face-palming error in the implementation causing the proxy
check to be implemented also without a proxy.
This meant that we were ALWAYS skipping http3 and system resolvers.
The bug has been introduced in 3.8.0. So, the currently released
version of the probe, sadly, has this beheavior :-(.
Reference issue https://github.com/ooni/probe/issues/1426.
We use ExperimentOrchestraClient in several places to help us
calling probe-services APIs. We need to call CheckIn because we
want to use CheckIn in InputLoader.
(We also want to remove the URLs API, but that is not something
doable now, since the mobile app is still using this API via
the wrappers at pkg/oonimkall.)
Work part of https://github.com/ooni/probe/issues/1299.
* fix(webconnectivity): expose network events
By not exposing network events in webconnectivity, we are missing
several interesting, explanatory data points.
This diff fixes the issue by:
1. enriching the definition of network events to include extra
data useful for performing (manual) data analysis;
2. adding a tags field to network events such that we can add
tags to specific events and understand where they come from;
3. exposing all the (tagged) network events that happen when running
a webconnectivity experiment.
See https://github.com/ooni/probe-engine/issues/1157.
* progress
* more work towards landing this diff
* Apply suggestions from code review
* Add signal to the im test group
* fix(ipconfig_test.go): disable when running in CI
Reference issue: https://github.com/ooni/probe/issues/1418
* fix(geolocate): remove unused variable
Came across this while looking into this issue with the CI that
is now failing. Guess fixing it here comes across as leaving the
camp slightly less in a bad shape than how I found it.
Co-authored-by: Simone Basso <bassosimone@gmail.com>
* poc: mobile api for running WebConnectivity
Yesterday we had a team meeting where we discussed the importance
of stopping using MK wrappers for running experiments.
Here's a PoC showing how we can do that for WebConnectivity. It
seems to me we probably want to add more tests and merge this code
such that we can experiment with it quite soon.
There seems to be opportunities for auto-generating code, BTW.
While working on this PoC, I've also took a chance to revamp the
external documentation of pkg/oonimkall.
* feat: start making the code more abstract
* chore: write unit tests for new code
* fix(oonimkall): improve naming
* refactor: cosmetic changes
* fix: explain
* fix(geolocate): no proxy when discovering our IP address
The use case of --proxy is that you cannot contact the OONI
backend otherwise. It is wrong, though, using the proxy when
discovering our IP address. The measurement won't use the
proxy anyway. Therefore, we need to use the IP address that
is performing the measurement. Not the one of the proxy.
What's more, stun is not using a proxy. Therefore, it does
not make much sense that http IP resolvers use a proxy. This
leads to inconsistencies. So, here's anothe reason why this
patch is a good thing (TM).
Finally, because knowing the IP address enables us to sanitize
the data, it's important we discover the correct IP.
Now, up until this point, the `--proxy` option has mostly
been a developers toy. But, users have asked us to have the
possibility of configuring a proxy.
This explains why I have been looking into making `--proxy`
right for a couple of hours now.
See https://github.com/ooni/probe/issues/1382
* fix(session): properly configure the IP lookupper
In reality, we are not going to use the sessionresolver when we're
using a proxy (I just tested). But, it nonetheless feels a lot more
robust to write a correct sessionresolver that handles the proxy
in the most correct way. That is, the sessionresolver will now skip
all the entries that cannot use a socks5 proxy (including among them
also the system resolver). What's more, it will construct a child
resolver that propagates the proxy.
We have confidence that this holds true because we have added a test
ensuring that we are really using the configured proxy.
See https://github.com/ooni/probe/issues/1381
This comes just a few days after 3.6.0. It contains small
improvements required by ooni/probe-desktop.
For this reason, I am going to skeep the normal release
process and I am just bumping the version number.
* refactor: signal et al. are now experimental nettests
We move signal into the experimental nettests group. While there,
also start adding dnscheck and stunreachability as well.
It seems there's more work to be done to correctly represent
the results of dnscheck, but this is fine!
The experimental section is here exactly for this reason!
In terms of UI, the new command is `ooniprobe run experimental`.
We will most likely move signal out of experimental soon, since it's
already working quite well. We need to keep it here for one more
cycle because the desktop app is not ready for it.
See the following issues:
1. https://github.com/ooni/probe/issues/1378
2. https://github.com/ooni/probe/issues/1262
* fix(dnscheck): spell check
* fix: improve documentation
* fix(webconnectivity): allow measuring https://1.1.1.1
There were two issues preventing us from doing so:
1. in netx, the address resolver was too later in the resolver
chain. Therefore, its result wasn't added to the events.
2. when building the DNSCache (in httpget.go), we didn't consider
the case where the input is an address. We need to treat this
case specially to make sure there is no DNSCache.
See https://github.com/ooni/probe/issues/1376.
* fix: add unit tests for code making the dnscache
* fix(netx): make sure all tests pass
* chore: bump webconnectivity version
* internal/engine/ooapi: auto-generated API client
* feat: introduce the callers abstraction
* feat: implement API caching on disk
* feat: implement cloneWithToken when we require login
* feat: implement login
* fix: do not cache all APIs
* feat: start making space for more tests
* feat: implement caching policy
* feat: write tests for caching layer
* feat: add integration tests and fix some minor issues
* feat: write much more unit tests
* feat: add some more easy unit tests
* feat: add tests that use a local server
While there, make sure many fields we care about are OK.
* doc: write basic documentation
* fix: tweak sentence
* doc: improve ooapi documentation
* doc(ooapi): other documentation improvements
* fix(ooapi): remove caching for most APIs
We discussed this topic yesterday with @FedericoCeratto. The only
place where we want LRU caching is MeasurementMeta.
* feat(ooapi): improve handling of errors during login
This was also discussed yesterday with @FedericoCeratto
* fix(swaggerdiff_test.go): temporarily disable
Before I work on this, I need to tend onto other tasks.
* fix(ootest): add one more test case
We're going towards 100% coverage of this package, as it ought to be.
* feat(ooapi): test cases for when the probe clock is off
* fix(ooapi): change test to have 100% unittest coverage
* feat: sync server and client APIs definition
Companion PR: https://github.com/ooni/api/pull/218
* fix(ooapi): start testing again against API
* fix(ooapi): only generate each file once
* chore: set version to 3.7.0-alpha
While there, make sure we don't always skip a currently failing
riseupvpn test, and slightly clarify the readme.
* fix(kvstore): less scoped error message
* chore: update dependencies
* chore: update user agent for measurements
* chore: we're now at v3.6.0
* chore: update assets
* chore: update bundled CA
* fix: address some goreportcard.com warnings
* fix(debian/changelog): zap release that breaks out build scripts
We're forcing the content of changelog with `dch`, so it's fine to
not have any specific new release in there.
* fix: make sure tests are passing locally
Notably, I removed a chunk of code where we were checking for network
activity. Now we don't fetch the databases and it's not important. Before,
it was important because the databases are ~large.
* fix: temporarily comment out riseupvn integration tests
See https://github.com/ooni/probe/issues/1354 for work aimed at
reducing the rate of false positives (thanks @cyBerta!)
* feat(sessionresolver): try many and use what works
* fix(sessionresolver): make sure we can use quic
* fix: the config struct is unnecessary
* fix: make kvstore optional
* feat: write simple integration test
* feat: start adding tests
* feat: continue writing tests
* fix(sessionresolver): add more unit tests
* fix(sessionresolver): finish adding tests
* refactor(sessionresolver): changes after code review
* Create the LaunchAgents directory if it does not exist
Fixes: https://github.com/ooni/probe/issues/1368
* Update cmd/ooniprobe/internal/autorun/autorun_darwin.go
Co-authored-by: Simone Basso <bassosimone@gmail.com>
Co-authored-by: Simone Basso <bassosimone@gmail.com>
* feat: use go1.16 embedding for resources
We want to embed everything that can be easily embedded. We should, at a
minimum, replace the downloading of resources and bindata.
Ref: https://github.com/ooni/probe/issues/1367.
* fix: get rid of bindata and use go embed instead
* fix: start unbreaking some automatic tests
* fix: fetch resources as part of the mobile build
* fix: convert more stuff to go1.16
I still expect many breakages, but we'll fix them.
* fix: make the windows CI green
* fix: get resources before running QA
* fix: go1.16 uses modules by default
* hopefully fix all other outstanding issues
* fix(QA/telegram.py): add another DC IP address
* Apply suggestions from code review
* MVP of a signal messenger test
* Add minimal signal test unit tests
* Add Signal test to the im nettest group
* Add test for https://sfu.voip.signal.org/
* Fix bug in client-side determination of blocking status
* Add uptime.signal.org to the test targets
* Add more tests
* Check for invalid CA being passed
* Check that the update function works as expected
* Update internal/engine/experiment/signal/signal_test.go
Co-authored-by: Simone Basso <bassosimone@gmail.com>
* fix: back out URL we shouldn't have changed
When merging probe-engine into probe-cli, we changed too many URLs
and some of them should not have been changed.
I noticed this during the review of Signal and I choose to add
this commit to revert such changes.
While there, make sure the URL of the experiment is OK.
* fix(signal): reach 100% of coverage
Just so that we can focus on areas of the codebase where we need
more coverage, let us avoid missing an easy line to test.
Co-authored-by: Simone Basso <bassosimone@gmail.com>
