Prior to commit a9b3a9c, the server implementation for documented
search operators with dashes, also implicitly supported clients
sending those same operators with underscores. This has been the
case sense the server side support for narrow filtering was
introduced in commit 3af2bf345a.
Updates the stricter version of mapping operator strings to `by*`
functions, to also include the underscore version of any operators
that have dashes. Adds a note that these undocumented versions are
tied to the support for the documented versions.
Use the built-in HTML escaping of Markup("…{var}…").format(), in order
to allow Semgrep to detect mistakes like Markup("…{var}…".format())
and Markup(f"…{var}…").
Signed-off-by: Anders Kaseorg <anders@zulip.com>
Updates the logic for identifying the method to use to extend the
query for the given term from a narrow to use a dictionary that
maps the operator string to the by_* method in the NarrowBuilder
class.
Previously, the by_* method was determined by building a string
based on the operator string and replacing dashes with underscores.
Prior to aa032bf62c, QOS prefetch was set on every `publish` and
before every `start_json_consumer` -- which had a large and
unnecessary effect on publishing rates, which don't care about the
prefetch QOS settings at all, much less re-setting them before every
publish.
Unfortunately, that change had the effect of causing prefetch settings
to almost never be respected -- since the configuration happened in
`ensure_queue`s re-check that the connection was still live. The
initial connection is established in `__init__` via `_connect`, and
the consumer only calls `ensure_queue` once, before setting up the
consumer.
Having no prefetch value set causes an unbounded prefetch; this
manifests itself as the server attempting to shove every event down to
the worker as soon as it starts consuming; if the client cannot keep
up, the server closes the connection. The worker observes the
connection has been shut down, and restarts. While this does make
forward progress, it causes large queues to make progress more slowly,
as they suffer from sporadic restarts.
Shift the QOS configuration to when the connection is set up, which is
a more sensible place for it in general -- and ensures that it is set
on consumers and producers alike, but only once per connection
establishment.
`render_markdown_path` renders Markdown, and also (since baff121115)
runs Jinja2 on the resulting HTML.
The `pure_markdown` flag was added in 0a99fa2fd6, and did two
things: retried the path directly in the filesystem if it wasn't found
by the Jinja2 resolver, and also skipped the subsequent Jinja2
templating step (regardless of where the content was found). In this
context, the name `pure_markdown` made some sense. The only two
callsites were the TOS and privacy policy renders, which might have
had user-supplied arbitrary paths, and we wished to handle absolute
paths in addition to ones inside `templates/`.
Unfortunately, the follow-up of 01bd55bbcb did not refactor the
logic -- it changed it, by making `pure_markdown` only do the former
of the two behaviors. Passing `pure_markdown=True` after that commit
still caused it to always run Jinja2, but allowed it to look elsewhere
in the filesystem.
This set the stage for calls, such as the one introduced in
dedea23745, which passed both a context for Jinja2, as well as
`pure_markdown=True` implying that Jinja2 was not to be used.
Split the two previous behaviors of the `pure_markdown` flag, and use
pre-existing data to control them, rather than an explicit flag. For
handling policy information which is stored at an absolute path
outside of the template root, we switch to using the template search
path if and only if the path is relative. This also closes the
potential inconsistency based on CWD when `pure_markdown=True` was
passed and the path was relative, not absolute.
Decide whether to run Jinja2 based on if a context is passed in at
all. This restores the behavior in the initial 0a99fa2fd6 where a
call to `rendar_markdown_path` could be made to just render markdown,
and not some other unmentioned and unrelated templating language as
well.
Previously, `QuerySet` does not support isinstance check since it is
defined to be generic in django-stubs. In a recent update, such check is
possible by using `QuerySetAny`, a non-generic alias of `QuerySet`.
Signed-off-by: Zixuan James Li <p359101898@gmail.com>
This commit updates 'set_user_topic_visibility_policy_in_database'
to not raise an error when deleting a UserTopic row and the user
doesn't have a visibility_policy for the topic yet, or when setting
the visibility_policy to its current value.
Also, it includes the changes to not send unnecessary events
in such cases.
Currently, there is a checkbox setting for whether to
"Include realm name in subject of message notification emails".
This commit replaces the checkbox setting with a dropdown
having values: Automatic [default], Always, Never.
The Automatic option includes the realm name if, and only if,
there are multiple Zulip realms associated with the user's email.
Tests are added and(or) modified.
Fixes: #19905.
This commit refactors 'do_set_user_topic_visibility_policy'
to remove the if/else block and just have a single call to
'set_user_topic_visibility_policy_in_database'.
The branching out behaviour based on the user_topic
visibility_policy is reduced to one place, i.e.,
'set_user_topic_visibility_policy_in_database'.
Updated the title and description in the 'enable-emoticon-translation'
file and renamed the file accordingly. Added a new bullet point for
'time format' in the 'configure-new-user-settings.md' file and updated
the sidebar index by replacing the title 'Use 24-hour time' with
'Change the time format'.
Zulip already has integrations for server-side Sentry integration;
however, it has historically used the Zulip-specific `blueslip`
library for monitoring browser-side errors. However, the latter sends
errors to email, as well optionally to an internal `#errors` stream.
While this is sufficient for low volumes of users, and useful in that
it does not rely on outside services, at higher volumes it is very
difficult to do any analysis or filtering of the errors. Client-side
errors are exceptionally noisy, with many false positives due to
browser extensions or similar, so determining real real errors from a
stream of un-grouped emails or messages in a stream is quite
difficult.
Add a client-side Javascript sentry integration. To provide useful
backtraces, this requires extending the pre-deploy hooks to upload the
source-maps to Sentry. Additional keys are added to the non-public
API of `page_params` to control the DSN, realm identifier, and sample
rates.
b4dd118aa1 changed how the `user_info_str` parsed information out of
the events it received -- but only changed the server errors, not the
browser errors, though both use the same codepath. As a result, all
browser errors since then have been incorrectly marked as being for
anonymous users.
Build and pass in the expected `user` dict into the event.
This commit adds 'visibility_policy' as a
parameter to user_allows_notifications_in_StreamTopic
function.
This adds logic inside the user_allows_notifications_in_StreamTopic
function, to not return False when a stream is muted
but the topic is UNMUTED.
Adds a method `user_id_to_visibility_policy_dict`
to 'StreamTopicTarget' class to fetch
(user_id => visibility_policy) in single db query.
Co-authored-by: Kartik Srivastava <kaushiksri0908@gmail.com>
Co-authored-by: Prakhar Pratyush <prakhar841301@gmail.com>
This commit replaces 'remove_topic_mute' with
'set_user_topic_visibility_policy_in_database' and
updates it to delete UserTopic row with any configured
visibility_policy and not just muting.
In order to support different types of topic visibility policies,
this renames 'add_topic_mute' to
'set_user_topic_visibility_policy_in_database'
and refactors it to accept a parameter 'visibility_policy'.
Create a corresponding UserTopic row for any visibility policy,
not just muting topics.
When a UserTopic row for (user_profile, stream, topic, recipient_id)
exists already, it updates the row with the new visibility_policy.
In the event of a duplicate request, raises a JsonableError.
i.e., new_visibility_policy == existing_visibility_policy.
There is an increase in the database query count in the message-edit
code path.
Reason:
Earlier, 'add_topic_mute' used 'bulk_create' which either
creates or raises IntegrityError -- 1 query.
Now, 'set_user_topic_visibility_policy' uses get_or_create
-- 2 queries in the case of creating new row.
We can't use the previous approach, because now we have to
handle the case of updating the visibility_policy too.
Also, using bulk_* for a single row is not the correct way.
Co-authored-by: Kartik Srivastava <kaushiksri0908@gmail.com>
Co-authored-by: Prakhar Pratyush <prakhar841301@gmail.com>
Creates `MutableJsonResponse` as a subclass of Django's `HttpResponse`
that we can modify for ignored parameters in the response content.
Updates responses to include `ignored_parameters_unsupported` in
the response data through `has_request_variables`. Creates unit
test for this implementation in `test_decorators.py`.
The `method` parameter processed in `rest_dispatch` is not in the
`REQ` framework, so for any tests that pass that parameter, assert
for the ignored parameter with a comment.
Updates OpenAPI documentation for `ignored_parameters_unsupported`
being returned in the JSON success response for all endpoints.
Adds detailed documentation in the error handling article, and
links to that page in relevant locations throughout the API docs.
For the majority of endpoints, the documentation does not include
the array in any examples of return values, and instead links to
the error handling page. The exceptions are the three endpoints
that had previously supported this return value. The changes note
and example for these endpoints is also used in the error
handling page.
Adds `is_webhook_view` boolean field to the RequestNotes class so
that (when implemented) `ignored_parameters_unsupported` feature
is not something that is applied to webhooks.
Actions like deleting realms may leave unreferenced uploads in the
attachment storage backend.
Fix these by walking the complete contents of the attachment storage
backend, and removing files which are no longer present in the
database. This may take quite some time, as it is necessarily O(n) in
the number of files uploaded to the system.
This commit renames reset_emails_in_zulip_realm function to
reset_email_visibility_to_everyone_in_zulip_realm which makes
it more clear to understand what the function actually does.
This commit also adds a comment explaining what this function
does.
The inital Welcome bot message has an extra section if the user is
joining a demo organization, but the link in that section was not
being formatted correctly. Fixes the formatting so that the link
works.
This is the behaviour inherited from Django[^1]. While setting the
password to empty (`email_password = `) in
`/etc/zulip/zulip-secrets.conf` also would suffice, it's unclear what
the user would have been putting into `EMAIL_HOST_USER` in that
context.
Because we previously did not warn when `email_password` was not
present in `zulip-secrets.conf`, having the error message clarify the
correct configuration for disabling SMTP auth is important.
Fixes: #23938.
[^1]: https://docs.djangoproject.com/en/4.1/ref/settings/#std-setting-EMAIL_HOST_USER
This commit adds backend code to set email_address_visibility when
registering a new user. The realm-level default and the value of
source profile gets overridden by the value user selected during
signup.
This lets us simplify the long-ish ‘../../static/js’ paths, and will
remove the need for the ‘zrequire’ wrapper.
Signed-off-by: Anders Kaseorg <anders@zulip.com>
Ever since we started bundling the app with webpack, there’s been less
and less overlap between our ‘static’ directory (files belonging to
the frontend app) and Django’s interpretation of the ‘static’
directory (files served directly to the web).
Split the app out to its own ‘web’ directory outside of ‘static’, and
remove all the custom collectstatic --ignore rules. This makes it
much clearer what’s actually being served to the web, and what’s being
bundled by webpack. It also shrinks the release tarball by 3%.
Signed-off-by: Anders Kaseorg <anders@zulip.com>
This is quite a bit faster:
```
%timeit calendar.timegm(now.timetuple())
2.91 µs ± 361 ns per loop (mean ± std. dev. of 7 runs, 100,000 loops each)
%timeit int(now.timestamp())
539 ns ± 27 ns per loop (mean ± std. dev. of 7 runs, 1,000,000 loops each)
```
This is particularly important for the presence endpoint, which is a
tight loop of serializing datetimes.
As written, the QOS parameters are (re)set every time ensure_queue is
called, which is every time a message is enqueued. This is wasteful --
particularly QOS parameters only apply for consumers, and setting them
takes a RTT to the server.
Switch to only setting the QOS once, when a connection
is (re)established. In profiling, this reduces the time to call
`queue_json_publish("noop", {})` from 878µs to 150µs.
We add stream_permission_group_settings object which is
similar to property_types framework used for realm settings.
This commit also adds GroupPermissionSetting dataclass for
defining settings inside stream_permission_group_settings.
We add "do_change_stream_group_based_setting" function which
is called in loop to update all the group-based stream settings
and it is now used to update 'can_remove_subscribers_group'
setting instead of "do_change_can_remove_subscribers_group".
We also change the variable name for event_type field of
RealmAuditLog objects to STREAM_GROUP_BASED_SETTING_CHANGED
since this will be used for all group-based stream settings.
'property' field is also added to extra_data field to identify
the setting for which RealmAuditLog object was created.
We will add a migration in further commits which will add the
property field to existing RealmAuditLog objects created for
changing can_remove_subscribers_group setting.
This reverts commit 851d68e0fc.
That commit widened how long the transaction is open, which made it
much more likely that after the user was created in the transaction,
and the memcached caches were flushed, some other request will fill
the `get_realm_user_dicts` cache with data which did not include the
new user (because it had not been committed yet).
If a user creation request lost this race, the user would, upon first
request to `/`, get a blank page and a Javascript error:
Unknown user_id in get_by_user_id: 12345
...where 12345 was their own user-id. This error would persist until
the cache expired (in 7 days) or something else expunged it.
Reverting this does not prevent the race, as the post_save hook's call
to flush_user_profile is still in a transaction (and has been since
168f241ff0), and thus leaves the potential race window open.
However, it much shortens the potential window of opportunity, and is
a reasonable short-term stopgap.
This will allow us to re-use this logic later, when we add support for
re-checking notification settings just before sending email/push
notifications to the user.
Also, since this is essentially part of the notifiability logic,
this better belongs to `notification_data.py` and this change will
hopefully reduce the reading complexity of the message-send codepath.
This commits update the code to use user-level email_address_visibility
setting instead of realm-level to set or update the value of UserProfile.email
field and to send the emails to clients.
Major changes are -
- UserProfile.email field is set while creating the user according to
RealmUserDefault.email_address_visbility.
- UserProfile.email field is updated according to change in the setting.
- 'email_address_visibility' is added to person objects in user add event
and in avatar change event.
- client_gravatar can be different for different users when computing
avatar_url for messages and user objects since email available to clients
is dependent on user-level setting.
- For bots, email_address_visibility is set to EVERYONE while creating
them irrespective of realm-default value.
- Test changes are basically setting user-level setting instead of realm
setting and modifying the checks accordingly.
Previously, user objects contained delivery_email field
only when user had access to real email. Also, delivery_email
was not present if visibility setting is set to "everyone"
as email field was itself set to real email.
This commit changes the code to pass "delivery_email" field
always in the user objects with its value being "None" if
user does not have access to real email and real email otherwise.
The "delivery_email" field value is None for logged-out users.
For bots, the "delivery_email" is always set to real email
irrespective of email_address_visibility setting.
Also, since user has access to real email if visibility is set
to "everyone", "delivery_email" field is passed in that case
too.
There is no change in email field and it is same as before.
This commit also adds code to send event to update delivery_email
field when email_address_visibility setting changes to all the
users whose access to emails changes and also changes the code to
send event on changing delivery_email to users who have access
to email.
This commit renames parse_message_content_edit_or_delete_limit
to parse_message_time_limit_setting and also renames
MESSAGE_CONTENT_EDIT_OR_DELETE_LIMIT_SPECIAL_VALUES_MAP to
MESSAGE_TIME_LIMIT_SETTING_SPECIAL_VALUES_MAP.
We do this change since this function and object will also be
used for message move limit and it makes sense to have a more
generic name.
In Zulip, message topics are case-insensitive but case-preserving.
The `get_context_for_message` function erroneously did a
case-sensitive search, and thus only messages whose topic matched
exactly were pulled in as context.
Make the missed-message pipeline aware that message topics are not
case-sensitive. This means that, when collapsing adjacent messages,
we merge messages with topic headers which are "different"; create a
separate explicit "grouping" to know which to collapse.
The Content-Type of user-provided uploads was provided by the browser
at initial upload time, and stored in S3; however, 04cf68b45e
switched to determining the Content-Disposition merely from the
filename. This makes uploads vulnerable to a stored XSS, wherein a
file uploaded with a content-type of `text/html` and an extension of
`.png` would be served to browsers as `Content-Disposition: inline`,
which is unsafe.
The `Content-Security-Policy` headers in the previous commit mitigate
this, but only for browsers which support them.
Revert parts of 04cf68b45e, specifically by allowing S3 to provide
the Content-Disposition header, and using the
`ResponseContentDisposition` argument when necessary to override it to
`attachment`. Because we expect S3 responses to vary based on this
argument, we include it in the cache key; since the query parameter
has dashes in it, we can't use use the helper `$arg_` variables, and
must parse it from the query parameters manually.
Adding the disposition may decrease the cache hit rate somewhat, but
downloads are infrequent enough that it is unlikely to have a
noticeable effect. We take care to not adjust the cache key for
requests which do not specify the disposition.
Fixes the documentation generated from the Markdown macros
{settings_tab|your-bots} and {settings_tab|bot-list-admin} to
match the text labels in the Zulip UI and improves the text of
relative links to explicitly say if we are referring to the Bots
tab of the Personal or Organization settings menu.
Follow-up to #23256.
This code needs to be more flexible to improve the documentation
of items in the Personal and Organization settings menu when
using the `{settings_tab|[setting-name]}` Markdownm macro that
provides relative links or step-by-step instructions.
This commit moves the Markdown formatting code to a new function that
receives tuples from `link_mapping` as input. This is a preliminary
step to offer more flexibility than the current approach.
- Updates `.prettierignore` for the new directory.
- Updates any reference to the API documentation directory for
markdown files to be `api_docs/` instead of `zerver/api/`.
- Removes a reference link from `docs/documentation/api.md` that
hasn't referenced anything in the text since commit 0542c60.
- Update rendering of API documentation for new directory.
- Clean up the language.
- Add a prominent "Go to organization" button.
- Link to guides for new users and admins.
- Fix duplication bug in text email version.
Co-authored-by: Mateusz Mandera <mateusz.mandera@zulip.com>
Black 23 enforces some slightly more specific rules about empty line
counts and redundant parenthesis removal, but the result is still
compatible with Black 22.
(This does not actually upgrade our Python environment to Black 23
yet.)
Signed-off-by: Anders Kaseorg <anders@zulip.com>
Removes `base_path` argument when making the markdown extension for
parameters in documentation for API endpoints.
This seems to have been originally included for API parameters that
were documented in JSON files, which is no longer in use. Now all
API endpoints in the documentation are documented in
`zerver/openapi/zulip.yaml`.
Removes `base_path` argument when making the markdown extension for
return values in documentation for API endpoints.
This seems to have been a copy and paste error in commit d2ee99a2fd
when `zerver/lib/markdown/api_return_values_generator.py` was created.
Until now, custom emojis with "periods" in their name were allowed, even though
they don't really fit the pattern of how we name them, and in fact the Markdown
processor would not render such custom emoji. Fix this by just disallowing the
character.
Also update the error strings accordingly.
Note that this does not include a migration to eliminate any existing custom emoji with this
character in their name.
Fixes#24066.
We change the do_create_user function to use transaction.atomic
decorator instead of using with block. Due to this change, all
send_event calls are made inside transaction.on_commit.
Some other changes -
- Remove transaction.atomic decorator from send_inital_realm_messages
since it is now called inside a transaction.
- Made changes in tests which tests message events and notifications
to make sure on_commit callbacks are executed.
These files are not Jinja2 templates, so there's no reason that they needed
to be inside `templates/zerver`. Moving them to the top level reflects their
importance and also makes it feel nicer to work on editing the help center content,
without it being unnecessary buried deep in the codebase.
The content of a message is truncated to `MAX_MESSAGE_LENGTH`, which
is 1000 characters. Since the email gateway places attachments at the
very end of the extracted body, that means that they are the first
thing to get truncated off.
That is, if an incoming email message contains 1000 `a`s and an image
attachment, the link that attaches the attachment to the message will
get truncated off, leaving it dangling in the database.
Truncate the message body content separately from the attachment links
which are included at the end of the body.
Since we want to use `accounts/new/send_confirm` to know how many
users actually register after visiting the register page, we
added it to Google Tag Manager, but GTM tracks every user
registration separately due <email> in the URL
making it harder to track.
To solve this, we want to pass <email> as a GET parameter which
can be easily filtered inside GTM using a RegEx and all the
registrations can be tracked as one.
A missed message email notification, where the message is the welcome
message sent by the welcome bot on account creation, get sent when
the user somehow not focuses the browser tab during account creation.
No missed message email or push notifications should be sent for the
messages generated by the welcome bot.
'internal_send_private_message' accepts a parameter
'disable_external_notifications' and is set to 'True' when the sender
is 'welcome bot'.
A check is introduced in `trivially_should_not_notify`, not to notify
if `disable_external_notifications` is true.
TestCases are updated to include the `disable_external_notifications`
check in the early (False) return patterns of `is_push_notifiable` and
`is_email_notifiable`.
One query reduced for both `test_create_user_with_multiple_streams`
and `test_register`.
Reason: When welcome bot sends message after user creation
`do_send_messages` calls `get_active_presence_idle_user_ids`,
`user_ids` in `get_active_presence_idle_user_ids` remains empty if
`disable_external_notifications` is true because `is_notifiable` returns
false.
`get_active_presence_idle_user_ids` calls `filter_presence_idle_user_ids`
and since the `user_ids` is empty, the query inside the function doesn't
get executed.
MissedMessageHookTest updated.
Fixes: #22884
A separate function named `trivially_should_not_notify` is added which
extracts the common checks from `get_push_notification_trigger` and
`get_email_notification_trigger` which are users' notification settings
independent and thus don't depend on what type of notification (email/push)
it is.
Documents link to the bot's user card from the bot's name in
Organization settings > Bots, and information in the bot's user card.
Fixes part of #23970.
When the email mirror gateway is sending messages "as" a user (as
triggered by having access to the missed-message email address),
attachments were still created as the Email Gateway bot. Since the
sender (the end-user) was not the owner of those attachments (the
gateway bot), nor were they referenced yet anywhere, this resulted in
the attachments being "orphaned" and not allowed to be accessed by
anyone -- despite the attachment links being embedded in the message.
This was accompanied by the error:
```
WARN [] User 12345 tried to share upload 123/3LkSA4OcoG6OpAknS2I0SFAQ/example.jpf in message 123456, but lacks permission
INFO [zerver.lib.email_mirror] Successfully processed email from user 12345 to example-stream
```
We solve this by creating attachment objects as the users the message
will be sent from.
The max inline preview limit was previously increased to 10 by #20789.
However, as issue #23624 shows, it's still causing confusion for users
when they include more than 10 links.
Bump this limit up to 24, which is a multiple of the 4 image preview
per line logic.
Accessing .realm will cause a fetch query from the database if the
attribute hasn't been fetched already earlier in the codepath. That's
completely redundant if we're just comparing realms, and we should only
access .realm_id attribute. This seems to eliminate a query in some
codepaths, which is nice in this performance-sensitive function.
When file uploads are stored in S3, this means that Zulip serves as a
302 to S3. Because browsers do not cache redirects, this means that
no image contents can be cached -- and upon every page load or reload,
every recently-posted image must be re-fetched. This incurs extra
load on the Zulip server, as well as potentially excessive bandwidth
usage from S3, and on the client's connection.
Switch to fetching the content from S3 in nginx, and serving the
content from nginx. These have `Cache-control: private, immutable`
headers set on the response, allowing browsers to cache them locally.
Because nginx fetching from S3 can be slow, and requests for uploads
will generally be bunched around when a message containing them are
first posted, we instruct nginx to cache the contents locally. This
is safe because uploaded file contents are immutable; access control
is still mediated by Django. The nginx cache key is the URL without
query parameters, as those parameters include a time-limited signed
authentication parameter which lets nginx fetch the non-public file.
This adds a number of nginx-level configuration parameters to control
the caching which nginx performs, including the amount of in-memory
index for he cache, the maximum storage of the cache on disk, and how
long data is retained in the cache. The currently-chosen figures are
reasonable for small to medium deployments.
The most notable effect of this change is in allowing browsers to
cache uploaded image content; however, while there will be many fewer
requests, it also has an improvement on request latency. The
following tests were done with a non-AWS client in SFO, a server and
S3 storage in us-east-1, and with 100 requests after 10 requests of
warm-up (to fill the nginx cache). The mean and standard deviation
are shown.
| | Redirect to S3 | Caching proxy, hot | Caching proxy, cold |
| ----------------- | ------------------- | ------------------- | ------------------- |
| Time in Django | 263.0 ms ± 28.3 ms | 258.0 ms ± 12.3 ms | 258.0 ms ± 12.3 ms |
| Small file (842b) | 586.1 ms ± 21.1 ms | 266.1 ms ± 67.4 ms | 288.6 ms ± 17.7 ms |
| Large file (660k) | 959.6 ms ± 137.9 ms | 609.5 ms ± 13.0 ms | 648.1 ms ± 43.2 ms |
The hot-cache performance is faster for both large and small files,
since it saves the client the time having to make a second request to
a separate host. This performance improvement remains at least 100ms
even if the client is on the same coast as the server.
Cold nginx caches are only slightly slower than hot caches, because
VPC access to S3 endpoints is extremely fast (assuming it is in the
same region as the host), and nginx can pool connections to S3 and
reuse them.
However, all of the 648ms taken to serve a cold-cache large file is
occupied in nginx, as opposed to the only 263ms which was spent in
nginx when using redirects to S3. This means that to overall spend
less time responding to uploaded-file requests in nginx, clients will
need to find files in their local cache, and skip making an
uploaded-file request, at least 60% of the time. Modeling shows a
reduction in the number of client requests by about 70% - 80%.
The `Content-Disposition` header logic can now also be entirely shared
with the local-file codepath, as can the `url_only` path used by
mobile clients. While we could provide the direct-to-S3 temporary
signed URL to mobile clients, we choose to provide the
served-from-Zulip signed URL, to better control caching headers on it,
and greater consistency. In doing so, we adjust the salt used for the
URL; since these URLs are only valid for 60s, the effect of this salt
change is minimal.
Moving `/user_avatars/` to being served partially through Django
removes the need for the `no_serve_uploads` nginx reconfiguring when
switching between S3 and local backends. This is important because a
subsequent commit will move S3 attachments to being served through
nginx, which would make `no_serve_uploads` entirely nonsensical of a
name.
Serve the files through Django, with an offload for the actual image
response to an internal nginx route. In development, serve the files
directly in Django.
We do _not_ mark the contents as immutable for caching purposes, since
the path for avatar images is hashed only by their user-id and a salt,
and as such are reused when a user's avatar is updated.
Importing `upload_backend` directly means that in testing it must also
be mocked where it is imported, in order to correctly test the right
backend. Since `get_avatar_url` is part of the public
`ZulipUploadBackend` API, add another helper method to call that.
The `django-sendfile2` module unfortunately only supports a single
`SENDFILE` root path -- an invariant which subsequent commits need to
break. Especially as Zulip only runs with a single webserver, and
thus sendfile backend, the functionality is simple to inline.
It is worth noting that the following headers from the initial Django
response are _preserved_, if present, and sent unmodified to the
client; all other headers are overridden by those supplied by the
internal redirect[^1]:
- Content-Type
- Content-Disposition
- Accept-Ranges
- Set-Cookie
- Cache-Control
- Expires
As such, we explicitly unset the Content-type header to allow nginx to
set it from the static file, but set Content-Disposition and
Cache-Control as we want them to be.
[^1]: https://www.nginx.com/resources/wiki/start/topics/examples/xsendfile/
Enforcing a consistent `type` helps us double-check that we're not
playing fast-and-loose with any file paths for local files. As noted
in the comment, this is purely for defense-in-depth.
Passing `write_local_file` a consistent `type` requires removing the
"avatars" out of `realm_avatar_and_logo_path` -- which makes it
consistent across upload backends.
This, in turn, requires a compensatory change to zerver.lib.export, to
be explicit that the realm icons are exported from the avatars
directory. This clarity is likely an improvement.
Updates the help center article to match the style and formatting
of "Import from Slack" and replaces existing content with its
corresponding Markdown macro.
Alex Vandiver
Anders Kaseorg
Lauryn Menard
Zixuan James Li
Prakhar Pratyush
David Rosa
Abhijeet Prasad Bodas
palashb01
Kartik Srivastava
Daniil Fadeev
Sahil Batra
Alya Abbott
Mateusz Mandera
Alya Abbott
Lalit
Tim Abbott
Lauryn Menard
Aman Agrawal
Trident Pancake
Josh Klar