Replace the mocked services in scripts/mocked_services with Rails-level
mocked services.
The scripts in scripts/mocked_services were a set of stub Sinatra
servers used to mock the Reportbooru, Recommender, and IQDBs services
during development. They return fake data so you can test pages that use
these services.
Implementing these services in Rails makes it easier to run them. It
also lets us drop a dependency on Sinatra and drop a use of HTTParty.
To use these services, set the following configuration in danbooru_local_config.rb
or .env.local:
* reportbooru_server: http://localhost:3000/mock/reportbooru
* recommender_server: http://localhost:3000/mock/recommender
* iqdbs_server: http://localhost:3000/mock/iqdb
where `http://localhost:300` is the url for your local Danbooru server
(may need to be changed depending on your configuration).
The Nijie login process works like this:
* First we submit our `email` and `password` to `https://nijie.info/login_int.php`.
* Then we save the NIJIEIEID session cookie from the response.
* We optionally retry if login failed. Nijie returns 429 errors with a
`Retry-After: 5` header if we send too many login requests. This can
happen during parallel testing.
* We cache the login cookies for only 1 hour so we don't have to worry
about them becoming invalid if we cache them too long.
Cookies and retrying errors on failure are handled transparently by Danbooru::Http.
Replace http.rb's builtin redirect following option with our own
redirect follower. This fixes an issue with http.rb losing cookies after
following a redirect.
Allow cookies to be saved and sent back when making several requests in
a row. Usage:
http = Danbooru::Http.use(:session)
# saves the foo=42 cookie sent by the response.
http.get("https://httpbin.org/cookies/set/foo/42")
# sends back the foo=42 cookie from the previous request.
http.get("https://httpbin.org/cookies")
Remove the Downloads::File class. Move download methods to
Danbooru::Http instead. This means that:
* HTTParty has been replaced with http.rb for downloading files.
* Downloading is no longer tightly coupled to source strategies. Before
Downloads::File tried to automatically look up the source and download
the full size image instead if we gave it a sample url. Now we can
do plain downloads without source strategies altering the url.
* The Cloudflare Polish check has been changed from checking for a
Cloudflare IP to checking for the CF-Polished header. Looking up the
list of Cloudflare IPs was slow and flaky during testing.
* The SSRF protection code has been factored out so it can be used for
normal http requests, not just for downloads.
* The Webmock gem can be removed, since it was only used for stubbing
out certain HTTParty requests in the download tests. The Webmock gem
is buggy and caused certain tests to fail during CI.
* The retriable gem can be removed, since we no longer autoretry failed
downloads. We assume that if a download fails once then retrying
probably won't help.
Revert back to previous workaround of fetching previous day if current
day returns no result. A terrible hack, really we should convert dates
to Reportbooru's timezone, but that has other complications.
Fix gem version conflicts described in 20abd8a5f. Nokogiri couldn't be
upgraded past 1.10.9 because 1.11.0 causes a build failure in Nokogumbo
2.0.2, but we couldn't stay on 1.10.9 either because it has a hard
requirement on Ruby <2.7 and we require Ruby >=2.7. This made `bundle
update` fail with a Gemfile conflict.
The fix is to disable libxml2 support when building Nokogumbo. Nokogumbo
wants to use the same version of libxml2 as Nokogiri, but Nokogiri
1.11.0 changed how it reports which version of libxml2 it's using, which
causes Nokogumbo's build to fail. Disabling libxml2 may reduce
performance of Nokogumbo ([1]).
While we're at it, we also make Nokogiri use the system version of
libxml2 instead of its own bundled version. Nokogiri really wants
us to use its own patched version of libxml2 instead of the system
version, but the patches it applies look relatively minor and don't seem
relevant to us ([2]). Using the system version reduces build time during CI.
This adds libxml2 and libxslt as OS-level dependencies of Danbooru. You
may need to do `sudo apt-get install libxml2-dev libxslt-dev` to install
these libraries after this commit.
[1]: https://github.com/rubys/nokogumbo#flavors-of-nokogumbo
[2]: https://github.com/sparklemotion/nokogiri/tree/master/patches/libxml2
Increase timeout to 30 seconds when uploading files to IQDB. Previously
we used the default timeout of 3 seconds, which could cause 599 timeout
errors sometimes if the upload took too long.
Fix "cannot determine size of body" errors on upload page. Caused by
exception during IQDB lookup. We were posting the form data wrong, we
need to wrap the file with HTTP::FormData::File and pass it through the
`form` parameter.
Fix the sidebar on the /posts index page sometimes being blank. This
could happen when either the related tag calculation was too slow and
timed out, or when Reporbooru was unavailable and we couldn't fetch the
list of popular tags.
In the tag list would otherwise be blank, we fall back to frequent tags
(the most common tags on the current page of results).
Also change it so that if Reportbooru is unconfigured, we fail
gracefully by returning blank results instead of failing with an
exception. This is so we can still view the popular searches and missed
searches pages during testing (even though they'll be blank).
3cdf67920 changed it so that Danbooru::Http follows redirects by
default. This broke some things in the Nico Seiga strategy, so disable
following redirects in the Nico Seiga API client for now.
Also change it so that Danbooru::Http follows redirects after a POST
request (by setting `strict: false`). Nico Seiga needs this because it
sends a redirect after we POST the login form.
This passed in development but failed in CI because SavedSearch.redis
used the live Redis server, which worked by accident as long as you had
a Redis server running.
* Combine MissedSearchService, PostViewCountService, and
PopularSearchService into single ReportbooruService class.
* Use Danbooru::Http for these services instead of HTTParty.
Fixes parallel tests failing because uploads for different tests were
stored in the same directory and uploads got clobbered when cleaning up
this directory.
`User.find_or_create_by!` calls `User.name_matches` during username
validation, which triggers a deprecation warning for some reason:
DEPRECATION WARNING: Class level methods will no longer inherit scoping
from `create!` in Rails 6.1. To continue using the scoped relation,
pass it into the block directly. To instead access the full set of
models, as Rails 6.1 will, use `User.default_scoped`.
Add minitest-reporters gem. Replace the default output of `bin/rails
test` with a progress bar. The default output just emits periods for
passed tests, which makes it hard to tell how long tests will take and
causes test runner output on Github to appear to hang. The web console
on Github is line buffered, but the default test runner doesn't normally
emit newlines (unless a test fails), so the output can hang for a long
time.
Store the app in /home/danbooru/app instead of in /app so that we have
permission to write inside the app dir. /app was owned by root, which
prevented the danbooru user from writing to it.
Bug: Replacing posts hosted on cdn.donmai.us didn't work.
Cause: Original files on cdn.donmai.us are hosted under /var/www/danbooru/original/, but replacements
were trying to store them directly under /var/www/danbooru, which failed with a permission error.
We were trying to store them in the wrong directory because we didn't respect the `original_subdir`
option when generating file paths.
Refactor models so that we define attribute API permissions in policy
files instead of directly in models.
This is cleaner because a) permissions are better handled by policies
and b) which attributes are visible to the API is an API-level concern
that models shouldn't have to care about.
This fixes an issue with not being able to precompile CSS/JS assets
unless the database was up and running. This was a problem when building
Docker images because we don't have a database at build time. We needed
the database because `api_attributes` was a class-level macro in some
places, which meant it ran at boot time, but this triggered a database
call because api_attributes used database introspection to get the list
of allowed API attributes.
* Inline Javascript in the <head> is a minor bottleneck because it
blocks page rendering.
* Fix antiproxying protection (proxy sites like to strip <script> tags
from the header).
Don't embed config options directly in Javascript. Including them in the
Javascript forces us to recompile the Javascript whenever these options
change, which requires us to redeploy the entire site instead of just
restarting the app server.
evazion
evazion