Normally I would let `eslint --fix` do this for me, but there's a bug
that causes:
const x = function ()
{
// ...
};
to become:
const x = ()
=> {
// ...
};
which ESLint thinks is a syntax error. (It probably is; I don't know
enough about the automatic semicolon insertion rules to be confident.)
* caching_middleware: fix gzip compression not triggered
* packages: If a client sets `Accept-Encoding: gzip`, the responseCache will
include `Content-Encoding: gzip` in all future responses, even
if a subsequent request does not set `Accept-Encoding` or another client
requests the file without setting `Accept-Encoding`.
Fix that.
* caching_middleware: use `test` instead of `match`
* add tests
* make code easier to understand
* make the regex more clear
Some authentication plugins use the users defined in the `users`
object but ignore the `password` and `hash` properties.
This change deletes all of the filtering logic, including the logic
that filters out users that have both `password` and `hash` properties
defined. I could have kept that check, but decided to remove it
because:
* There's no harm in defining both `hash` and `password`.
* Allowing both makes it easier to transition from one scheme to
another.
* It's fewer lines of code to maintain.
This will be a breaking change for some people.
We removed all internal password control logic. If this affects you, you have two options:
1. Use a plugin for authentication and use session based pad access (recommended).
1. Use a plugin for password setting.
The reasoning for removing this feature is to reduce the overall security footprint of Etherpad. It is unnecessary and cumbersome to keep this feature and with the thousands of available authentication methods available in the world our focus should be on supporting those and allowing more granual access based on their implementations (instead of half assed baking our own).
Rather than reinvent the wheel, use a well-tested library to parse and
write cookies. This should also help prevent XSS vulnerabilities
because the library handles special characters such as semicolon.
Where feasible I put the await at the end of the function to
minimize the impact on latency.
My motivation for this change: Eliminate a race condition in tests I
am writing.
* `src/node/server.js` can now be run as a script (for normal
operation) or imported as a module (for tests).
* Move shutdown actions to `src/node/server.js` to be close to the
startup actions.
* Put startup and shutdown in functions so that tests can call them.
* Use `await` instead of callbacks.
* Block until the HTTP server is listening to avoid races during
test startup.
* Add a new `shutdown` hook.
* Use the `shutdown` hook to:
* close the HTTP server
* call `end()` on the stats collection to cancel its timers
* call `terminate()` on the Threads.Pool to stop the workers
* Exit with exit code 0 (instead of 1) on SIGTERM.
* Export the HTTP server so that tests can get the HTTP server's
port via `server.address().port` when `settings.port` is 0.
Before this change, `promises.timesLimit()` created `concurrency - 1`
too many promises. The only users of this function use a concurrency
of 500, so this meant that 499 extra promises were created each time
it was used. The bug didn't affect correctness, but it did result in a
large number of unnecessary database operations whenever a pad was
deleted. This change fixes that bug.
Also:
* Convert the function to async and have it resolve after all of the
created promises are resolved.
* Reject concurrency of 0 (unless total is 0).
* Document the function.
* Add tests.
package to reduce http requests: nice-select,
pad_automatic_reconnect, skin_variants, scroll, caretPosition
rename unorm in tar.json so it can be included
* Move session validity check and session author ID fetch to a
separate function. This separate function can be used by hooks,
making it easier for them to properly determine the author ID.
* Rewrite the remainder of checkAccess. Benefits:
- The function is more readable and maintainable now.
- Vulnerability fix: Before, the session IDs in sessionCookie
were not validated when checking settings.requireSession. Now,
sessionCookie must identify a valid session for the
settings.requireSession test to pass.
- Bug fix: Before, checkAccess would sometimes use the author ID
associated with the token even if sessionCookie identified a
valid session. Now it always uses the author ID associated
with the session if available.
Includes settings
Includes i18n
Includes a nice notification
Disconnects on rate limit
Includes feeding into metrics/stats
Include console warn to server console.
1. Introduce contentcollector.js backend tests
1. Fix issue with OL LI items not being properly numbered after import
1. Fix issue with nested OL LI items being improperly numbered on export
1. Fix issue with new lines not being introduced after lists in on import #3961
1. Sanitize HTML on the way in (import)
1. Fix ExportHTML CSS because it needs to support OL > LI > OL not OL > OL [The latter being the correct format]
1. Fix backend tests.
4177b3f943 moved the font-face declarations from src/static/css/pad.css to two
imported files (src/static/css/pad/fonts.css, src/static/css/pad/toolbar.css)
in a different directory.
This results in the font files being invoked from CSSes residing in different
directories in the minified and un-minified case. URLs in the src attribute are
relative to the stylesheet path [0], and so we have to start requiring clean-css
to rebase them.
Before this change, the non minified casse worked by chance, because there were
a lot of "..", which ended up resolving to the root of the site anyways.
Fixes#3956
[0] https://developer.mozilla.org/en-US/docs/Web/CSS/@font-face/src
After each Eterpad restart, the clients will request a new version of the
static assets, even if they are not modified. This is the price we pay for
knowing that no stale files are going to be served ever again. We could also
have used a salted hash of the Etherpad version, but we chose the simpler way.
For the rationale behind using a random string at each restart, see #3958.
ACHTUNG: this may prevent caching HTTP proxies to work.
Closes#3955.
Starting with Etherpad 1.8.3 we decided to use Colibris as default skin for new
installs. Without this change, when starting with no settings.json file,
Etherpad would (wrongly) use "no-skin".
This change should have been part of 70bc71c0c3.
This is a departure from previous versions, which did not limit import/export
requests. Now such requests are ALWAYS rate limited. The default is 10 requests
per IP each 90 seconds, and also applies to old instances upgraded to 1.8.3.
Administrators can tune the parameters via settings.importExportRateLimiting.
From Etherpad 1.8.3 onwards, the maximum allowed size for a single imported
file will always be bounded.
The maximum allowed size can be configured via importMaxFileSize.
The old loadSettings.js was a way of customizing settings upon load, because
the Settings module did not offer this functionality. But it did not work well,
since all the default settings were not loaded.
Let's get rid of loadSettings.js for the bulk of the tests (the "backend"
specs). For the "container" specs, we'll keep it in place until/if we rewrite
Settings.js making it less brittle.
This yields better conversion results, but requires the previous change,
otherwise there would have been difficulties in locating the temporary file
name.
In the next commit, we are going to change the conversion method to
"html:XHTML Writer File:UTF8". Without this change, that conversion method name
would end up in the extension of the temporary file that is created as an
intermediate step. In this way, the file extensione will always stay ".html".
No functional changes, hopefully. Only the extension of the temporary file
should change.
The previous syntax caused a deprecation warning on Node 10.
However, due to the very old version of log4js Etherpad is currently using,
customError objects are going to be displayed as { inspect: [Function: inspect] }.
This needs to be addressed later, updating log4js.
Fixes#3834.
The saved revision "star" button appeared in the timeslider toolbar too.
This change introduces a second flag "page" in toolbar.menu(), which controls
whether the toolbar is being drawn for a pad or in the timeslider page.
Fixes#3767.
Revision b480416375 fixed a bug, but introduced a regression, and the "save
revision" "star" button started to appear both on the left and the right
toolbar.
This change introduces a flag "whichMenu" in toolbar.menu(), that controls
whether the left, the right or the timeslider toolbar is being drawn.
By specification [0], the if-modified-since HTTP header sent by browsers does
not include milliseconds.
Before this patch, let's say a file was generate at time:
t_real-file = 2020-03-22T02:15:53.548Z (note the fractional seconds)
When issuing a conditional request, the browser would truncate the fractional
part, and only request an if-modified-since with this contents:
t_if-modified-since = 2020-03-22T02:15:53.000Z
The minify() function would return HTTP/304 only if
t_if-modified-since >= t_real-file, but this would never be true unless, by
chance, a file was generated at XX.000Z.
This resulted in that file being minified/compressed again and resent to the
client for no reason. After this patch, the server correctly responds with
HTTP/304 without doing any computation, and the browser uses the cached file.
[0] https://developer.mozilla.org/en-US/docs/Web/HTTP/Headers/If-Modified-Since
CleanCSS 3.4.19 had a Regex Denial of Service vulnerability and has to be
updated. The major version bump requires the following changes:
1. Disabling rebase is necessary because otherwise the URLs for the web fonts
become wrong;
EXAMPLE 1:
/static/css/src/static/font/fontawesome-etherpad.woff
instead of
/static/font/fontawesome-etherpad.woff
EXAMPLE 2 (this is more surprising):
/p/src/static/font/opendyslexic.otf
instead of
/static/font/opendyslexic.otf
2. CleanCSS.minify() can either receive a string containing the CSS, or an array
of strings. In that case each array element is interpreted as an absolute
local path from which the CSS file is read.
In version 4.x, CleanCSS API was simplified, eliminating the relativeTo
parameter, and thus we cannot use our already loaded "content" argument, but
we have to wrap the absolute path to the CSS in an array and ask the library
to read it by itself.
Fixes#3616.
Before this patch, visiting the read-only URL for a random pad would remove
the "Save Revision" (the "star" icon) from all the other RW pads. The only way
to make it appear again was to restart the server.
This change does not fix the underlying bug: after visiting a read only link
the "star" button would still disapper, but it is explictly reinserted via an
ad-hoc condition.
Fixes#3702
The change that implemented #3648 (7c099fef5e) was incorrect, and resulted
in disabling every user at startup.
The problem was twofold:
1. _.filter() on an object returns an array of the object's enumerable values
and strips out the keys, see: https://stackoverflow.com/questions/11697702/how-to-use-underscore-js-filter-with-an-object
To filter an object, the function that needs to be used is _.pick();
2. The logic condition on userProperties.password was plain wrong (it should
have been an AND instead of an OR).
This change corrects 1) and 2), and writes more specific logs when something
goes wrong.
Closes#3661.
Currently the version is exposed in a 'Server' http headers.
This commit allows to parameterize it in the settings. By defaults it is
not exposed.
Fixes#3423
In this way the only external call to statFile() provides an explicit value for
"dirStatLimit", and thus the initial check on "undefined" at the start of the
function could be removed (just added a comment for now).
All the configuration values can be read from environment variables using the
syntax "${ENV_VAR_NAME}".
This is useful, for example, when running in a Docker container.
EXAMPLE:
"port": "${PORT}"
"minify": "${MINIFY}"
"skinName": "${SKIN_NAME}"
Would read the configuration values for those items from the environment
variables PORT, MINIFY and SKIN_NAME.
REMARKS:
Please note that a variable substitution always needs to be quoted.
"port": 9001, <-- Literal values. When not using substitution,
"minify": false only strings must be quoted: booleans and
"skin": "colibris" numbers must not.
"port": ${PORT} <-- ERROR: this is not valid json
"minify": ${MINIFY}
"skin": ${SKIN_NAME}
"port": "${PORT}" <-- CORRECT: if you want to use a variable
"minify": "${MINIFY}" substitution, put quotes around its name,
"skin": "${SKIN_NAME}" even if the required value is a number or a
boolean.
Etherpad will take care of rewriting it to
the proper type if necessary.
Resolves#3543
Before this commit, when passed a malformed credentials.json the application
crashed with a stack dump. Now we catch the error and fail in a controlled way
(like already done for settings.json).
Example of exception we no longer throw:
MALFORMEDJSON
^
SyntaxError: Unexpected token M in JSON at position 0
at JSON.parse (<anonymous>)
at Object.reloadSettings (<BASEDIR>/src/node/utils/Settings.js:390:24)
at Object.<anonymous> (<BASEDIR>/src/node/utils/Settings.js:543:9)
at Module._compile (module.js:635:30)
at Object.Module._extensions..js (module.js:646:10)
at Module.load (module.js:554:32)
at tryModuleLoad (module.js:497:12)
at Function.Module._load (module.js:489:3)
at Module.require (module.js:579:17)
at require (internal/module.js:11:18)
NB1: needs additional review and testing - no abiword available on my test bed
NB2: in ImportHandler.js, directly delete the file, and handle the eventual
error later: checking before for existence is prone to race conditions,
and does not handle any errors anyway.
