This patch fixes the currently broken redirect on login when people try
to access a site they have no access to, they are redirected to the main
page to log in. After a successful login they should be redirected to
the original note, but instead are redirect to the index page again.
This aptch fixes the typo that causes the behavior and brings people
back to the note they edited.
Thanks to @clvs7-gh on Github[1], who submitted the patch via email.
On their behalf I hereby submit the change.
[1]: https://github.com/clvs7-gh
Note: I had to ajust this patch to work properly.
Signed-off-by: Sheogorath <sheogorath@shivering-isles.com>
Depending on how the system was setup, this bug lead to keep user's data
around even after a successful deletion of user'S account. This patch
will make sure the missing database constraints are implemented and
missed out deletions are executed.
This bug was introduced to insufficent testing after implementing the
feature initially. It was well tested, using the app process itself, but
the migrations where missed out. I'm currently not sure, if there was
also a change in how sequelize handles cassaded deletion, since I'm
unter the impression that before switching to sequelize 5, this feature
has worked. But I haven't verified this.
No matter what, the cleanup process is rather straight forward and will
be invoked on migration, but can also be done manually using the new
`bin/cleanup` script.
This change will result in a release 1.6.1.
Signed-off-by: Sheogorath <sheogorath@shivering-isles.com>
As we noticed in our poll about CDN usage, that most people
intentionally turn it off, but very little intetionally turn it on or
leave it on. [1]
There is also strong indicators that CDNs don't really provide any
benefits in loading time and due to the small deployments of CodiMD,
there is no big savings due to CDNs either. [2]
Therefore this patch changes the CDN default settings to off in order to
reduce the exposed user data.
[1]: https://community.codimd.org/t/poll-on-cdn-usage/28
[2]: https://csswizardry.com/2019/05/self-host-your-static-assets/
Signed-off-by: Sheogorath <sheogorath@shivering-isles.com>
This update of revealJS helps us to get rid of the headjs depedency
integration using webpack. It updates reveal.js to 3.9.2 and updates the
csp hash accordingly for using the slide mode.
Background for this update is the critical security vulnerability
described by snyk in their disclosure:
https://snyk.io/vuln/SNYK-JS-REVEALJS-543841
Signed-off-by: Sheogorath <sheogorath@shivering-isles.com>
Because of circular import problems, this commit also moves the error messages from response.js to errors.js
Signed-off-by: David Mehren <dmehren1@gmail.com>
This makes the references consistent/compatible with GitHub,
GitLab, Pandoc and many other tools.
This behavior can be enabled in config.json with:
```
"linkifyHeaderStyle": "gfm"
```
Signed-off-by: hoijui <hoijui.quaero@gmail.com>
This regression bug was caused by the error page using the `codimd/head`
template. This resulted in error messages like this:
```
ReferenceError: /codimd/public/views/error.ejs:5
3|
4| <head>
>> 5| <%- include codimd/head %>
6| <link rel="stylesheet" href="<%- serverURL %>/css/center.css">
7| </head>
8|
/codimd/public/views/codimd/head.ejs:7
5| <meta name="apple-mobile-web-app-status-bar-style" content="black">
6| <meta name="mobile-web-app-capable" content="yes">
>> 7| <% for (var og in opengraph) { %>
8| <% if (opengraph.hasOwnProperty(og) && opengraph[og].trim() !== '') { %>
9| <meta property="og:<%- og %>" content="<%- opengraph[og] %>">
10| <% }} if (!opengraph.hasOwnProperty('image')) { %>
opengraph is not defined
at eval (eval at compile (/codimd/node_modules/ejs/lib/ejs.js:618:12), <anonymous>:18:23)
at eval (eval at compile (/codimd/node_modules/ejs/lib/ejs.js:618:12), <anonymous>:99:10)
at returnedFn (/codimd/node_modules/ejs/lib/ejs.js:653:17)
at tryHandleCache (/codimd/node_modules/ejs/lib/ejs.js:251:36)
at View.exports.renderFile [as engine] (/codimd/node_modules/ejs/lib/ejs.js:482:10)
at View.render (/codimd/node_modules/express/lib/view.js:135:8)
at tryRender (/codimd/node_modules/express/lib/application.js:640:10)
at Function.render (/codimd/node_modules/express/lib/application.js:592:3)
at ServerResponse.render (/codimd/node_modules/express/lib/response.js:1012:7)
at responseError (/codimd/lib/response.js:57:20)
at Object.errorNotFound (/codimd/lib/response.js:30:5)
at newNote (/codimd/lib/response.js:134:76)
at /codimd/lib/response.js:172:16
at tryCatcher (/codimd/node_modules/bluebird/js/release/util.js:16:23)
at Promise._settlePromiseFromHandler (/codimd/node_modules/bluebird/js/release/promise.js:517:31)
at Promise._settlePromise (/codimd/node_modules/bluebird/js/release/promise.js:574:18)
at Promise._settlePromise0 (/codimd/node_modules/bluebird/js/release/promise.js:619:10)
at Promise._settlePromises (/codimd/node_modules/bluebird/js/release/promise.js:699:18)
at _drainQueueStep (/codimd/node_modules/bluebird/js/release/async.js:138:12)
at _drainQueue (/codimd/node_modules/bluebird/js/release/async.js:131:9)
at Async._drainQueues (/codimd/node_modules/bluebird/js/release/async.js:147:5)
at Immediate.Async.drainQueues (/codimd/node_modules/bluebird/js/release/async.js:17:14)
```
The fix for that is rather trivial. We simply provide an empty array of
metadata when generating the error template.
Signed-off-by: Sheogorath <sheogorath@shivering-isles.com>
As a temporary fix, to keep you and your users save, this patch disables
the PDF export feature. Details of the attack along with a fix for
future versions of CodiMD will be released in future.
I hope you can live with this solution for this release because I'm
super short on time and the alternative would be to ship no fix at all.
This appears to be the better solution for this release.
Signed-off-by: Sheogorath <sheogorath@shivering-isles.com>
It seems like since we switched to camelcase we missed to update some
variable names in the config section. This patch fixes those.
Signed-off-by: Sheogorath <sheogorath@shivering-isles.com>
Add simple SVG image detecetion base on the file extension .svg.
This fixes the SVG being delivered as binary/octet-stream and makes it possible to embedd the SVG.
Signed-off-by: Lennart Weller <lennart.weller@hansemerkur.de>
Do Not Track (DNT) is an old web standard in order to notify pages that
the user doesn't want to be tracked. Even while a lot of pages either
ignore this header or even worse, use it for tracking purposes, the
orignal intention of this header is good and should be adopted.
This patch implements a respect of the DNT header by no longer including
the optional Google Analytics and disqus integrations when sending a DNT
header. This should reduce outside resource usage and help to stay more
private.
This should later-on extended towards other document content (i.e.
iframe based content).
The reason to not change the CDN handling is that CDNs will be
deprecated with next release and removed in long term.
Signed-off-by: Sheogorath <sheogorath@shivering-isles.com>
We have various places with overly simple if statements that could be
handled by our logging library. Also a lot of those logs are not marked
as debug logs but as info logs, which can cause confusion during
debugging.
This patch removed unneeded if clauses around debug logging statements,
reworks debug log messages towards ECMA templates and add some new
logging statements which might be helpful in order to debug things like
image uploads.
Signed-off-by: Sheogorath <sheogorath@shivering-isles.com>
Since we are about to release it's time to finally fix our linting. This
patch basically runs eslint --fix and does some further manual fixes.
Also it sets up eslint to fail on every warning on order to make
warnings visable in the CI process.
There should no functional change be introduced.
Signed-off-by: Sheogorath <sheogorath@shivering-isles.com>
Currently a problem appears when using OpenID for authentication as
there is no method to add a profile picture right now.
This patch makes sure that all undefined login methods get a profile
picture.
Signed-off-by: Sheogorath <sheogorath@shivering-isles.com>
With very low CPU frequency or bad IO situation, as well as not-loaded
JS CodiMD happens to present unneeded "I'm busy"-messages to users.
This patch allows to configure the lag. The default is taken from the
libray but set in our own default configs.
Signed-off-by: Sheogorath <sheogorath@shivering-isles.com>
disableRequestedAuthnContext: true|false
By default only Password authmethod is accepted, this option allows any other method.
Issue and option described here:
https://github.com/bergie/passport-saml/issues/226
Signed-off-by: Emmanuel Ormancey <emmanuel.ormancey@cern.ch>
Add "both" mode to URLs because I assume most people want to straight away see the code when they click the "edit" button in a published note.
Fixes https://github.com/codimd/server/issues/27
Not tested, followed instructions from @ccoenen , please do review! :)
Signed-off-by: Stéphane Guillou <stephane.guillou@member.fsf.org>
After a long discussion, it turned out that CodiMD as community project
and HackMD as a company, have fundamental different views on the project
governance.
Due to this, it came to point where the decision for a fork was made.
After the fork and move towards an own organisation, this patch updates
all links inside the project to the new repositories.
Signed-off-by: Sheogorath <sheogorath@shivering-isles.com>
Since libravatar got a default fallback to Gravatar and in generell
allows federated image hosting for avatars this shouldn't break any
existing implementations.
The federation functionality is not added yet. This would require to use
the libravatar library.
Details:
https://wiki.libravatar.org/api/
Signed-off-by: Sheogorath <sheogorath@shivering-isles.com>
Since Google+ is shutting down soon, we need to get the profile data
from another URL. Since the library already supports it, all we need to
do is adding a single line of code.
Details:
https://github.com/hackmdio/codimd/issues/1160
Signed-off-by: Sheogorath <sheogorath@shivering-isles.com>
To provide a GitLab integration we need the GitLab integration to be
configured. Otherwise we shouldn't show the Snippet button.
This patch adds the requirement to the variable that decides if the
import from snippets button shows up or not.
Signed-off-by: Sheogorath <sheogorath@shivering-isles.com>
Seems like there is a possible problem when a name containing a space is
passed to this function. using urlencode on the name should fix possible
problems here.
Signed-off-by: Sheogorath <sheogorath@shivering-isles.com>
We talked about that during a community call. It turned out that not
everyone likes to have OpenID on their instance.
This patch disables OpenID by default.
Signed-off-by: Sheogorath <sheogorath@shivering-isles.com>
We used `fs.unlink()` to remove the pdf file after we send it out to the
client. This breaks in Node 10, when no function as second parameter is
supplied.
This patches changes it to the `fs.unlinkSync` function that doesn't
have this requirement and this way doesn't crash.
Signed-off-by: Sheogorath <sheogorath@shivering-isles.com>
Looks like GitHub changed their asset system and our CSP prevented them
from getting loaded.
This patch should fix the Gist embedding with enabled CSP by replacing
the old URL `https://assets-cdn.github.com` with the new
`https://github.githubassets.com`.
Signed-off-by: Sheogorath <sheogorath@shivering-isles.com>
Disqus loads it's embed config.js from its root domain
(https://disqus.com). Our CSPs only allow subdomains (e.g.:
https://codimd.disqus.com). This causes the disqus embedding to fail.
This patch should fix this problem by adding https://disqus.com to the
CSP setting. From a security perspective there is no real change. Since
still the same parties are involved.
Signed-off-by: Sheogorath <sheogorath@shivering-isles.com>
Looks like I was wrong in my previous commit to update revealjs.[1]
The speaker notes broke again with the CSPs. So this patch updates the
hash and this way the speaker notes.
[1]: bcebf1e8d2
Signed-off-by: Sheogorath <sheogorath@shivering-isles.com>
We see some issues that are based on not properly configured
`config.serverURL`.
This patch adds a warning when `config.serverURL` is an empty value.
This should provide users direct feedback about how to improve their
configs.
Signed-off-by: Sheogorath <sheogorath@shivering-isles.com>
Fix wrong config options
In `./lib/web/auth/` some config includes still used `config.serverurl` instead of the correct `config.serverURL`. This causes wrong URL in worst case.
This patch should fix those problems and migrate the wrong statements to camelcase.
This commit also refactors the code a bit, and adds a '-' separator
between a filename and its duplicate index.
This commit fixes#1079.
Signed-off-by: Daan Sprenkels <hello@dsprenkels.com>
Since our previous scrypt library is unmaintained since 3 years, it's
time to look for an alternative.
A refactoring towards another password algorithm was worked on and this
is probably still the way to go. But for now the successor of our
previous library should already be enough.
https://www.npmjs.com/package/scrypt (old library)
https://github.com/ml1nk/node-scrypt (new library)
Signed-off-by: Sheogorath <sheogorath@shivering-isles.com>
It seems like the inital work on the hsts module expected milliseconds.
This has either changed or was never true. Either way, it caused that
the current defaults resulted in theory in a 1000 year HSTS policy.
Luckily helmet was smart enough to not go higher than 1 year.
Anyway, this patch fixes the multiplication of the configured size with
1000 by removing this multiplication.
Also to simplify the reading of the defaults, we split them into their
components, 60 times 60 seconds so we get one hour. 24 of those hours so
we get a day and finally 365 days to get our original wanted default of
one year.
Reference:
d69d65ea74
Signed-off-by: Sheogorath <sheogorath@shivering-isles.com>
Add a configuration setting to "hard"-disable creation of notes as
set by the configuration value. This defaults to `['robots.txt',
'favicon.ico']`, because these files are often accidentally created
by bots and browsers.
This commit fixes#1052.
Signed-off-by: Daan Sprenkels <hello@dsprenkels.com>
During the upgrade of winston in
c3584770f2 a the class extension for
streaming was removed.
This caused silent crashes. Somehow winston simply called
`process.exit(1)` whenever `logger.write()` was called. This is really
bad and only easy to debug because of the testing right after upgrading.
However, reimplementing the stream interface as it was, didn't work, due
to the fact that `logger.write()` is already implemented and causes the
mentioned problem. So we extent the object with an `stream` object that
implements `write()` for streams and pass that to morgan.
So this patch fixes unexpected exiting for streaming towards our logging
module.
References:
https://www.digitalocean.com/community/tutorials/how-to-use-winston-to-log-node-js-applicationsc3584770f2https://stackoverflow.com/a/28824464
Signed-off-by: Sheogorath <sheogorath@shivering-isles.com>
most rules degraded to WARN, so we don't go insane. This will
change over time. The aim is to conform to a common style
Signed-off-by: Claudius Coenen <opensource@amenthes.de>
This fixes part of #1056: an error while obtaining the profile
would have `502`-crashed the server.
Signed-off-by: Claudius Coenen <opensource@amenthes.de>
Our log library got a new major version which should be implemented.
That's exactly what this patch does. Implementing the new version of the
logging library.
Signed-off-by: Sheogorath <sheogorath@shivering-isles.com>
Seems like there was some debugging going on some day, this patch should
make sure the right logging is used.
Signed-off-by: Sheogorath <sheogorath@shivering-isles.com>
Right now we use a substr after reading the commit. That's definitely
wrong and leads to wrong commit hashes since the first 5 chars are
missing.
This patch removes the substr usage here and this way fixes the
generated links.
Signed-off-by: Sheogorath <sheogorath@shivering-isles.com>
Currently we only provide the version from `package.json`. This means
that during updates of instances, e.g. the demo instance, which runs
latest master instead of a stable release, changes are not reflected to
the webclient.
This patch adds a fullversion string that contains the current commit
and this way makes that clients are notified about changes.
Signed-off-by: Sheogorath <sheogorath@shivering-isles.com>
We should use the official OS temp directory instead of an own one, to
not run into conflicts. Also various dependencies already use the OS
temp directory, which makes it pointless to use a different for our
internal purposes then. This commit provides the changes needed to use
the OS tmp directory by default.
Signed-off-by: Sheogorath <sheogorath@shivering-isles.com>
With OpenID every OpenID capable provider can provide authentication for
users of a CodiMD instance. This means we have federated
authentication.
Signed-off-by: Sheogorath <sheogorath@shivering-isles.com>
Right now the feature exists but is almost not usable since the only way
to configure it is to know that it exists from reading the source code
and add it to config.json. This patch provides all needed changes so it
can be used by everyone including documentation.
Signed-off-by: Sheogorath <sheogorath@shivering-isles.com>
Seems like the old version of helmet had a problem with `data:`. This
patch upgrades to the latest version and adds the CSP rule to allow
Google Fonts and the offline version of it, to properly include the
fonts and no longer throw ugly error messages at us.
Signed-off-by: Sheogorath <sheogorath@shivering-isles.com>
While paths like `tmpPath` could previously be configured,
they were all interpreted relative to `appRootPath` because
of `path.join`.
Now the configurable paths can be canonical and therefore
independent of the `appRootPath`.
Signed-off-by: WilliButz <wbutz@cyberfnord.de>
Previously it was assumed that `config.json` would be placed in
the same directory as the rest of CodiMD without any optional override.
This allows to override the path to the `config.json` by setting
`CMD_CONFIG_FILE` to the canonical path of the desired config file.
Signed-off-by: WilliButz <wbutz@cyberfnord.de>
Previously calling `app.js` from another directory than
the base directory of CodiMD would result in an error being
thrown because `lib/workers/dmpWorker.js` could not be found.
This change makes the function call independent of the path CodiMD
is started from.
Signed-off-by: WilliButz <wbutz@cyberfnord.de>
We recently introduced a new way to create notes using a post requeest
to the `/new` endpoint. This is not limited in size, other than pasting
a note in the editor. This patch should enforce this limit also on this
way.
Signed-off-by: Sheogorath <sheogorath@shivering-isles.com>
This patch should fix the unneeded warning of the wrong API version,
when gitlab isn't configured at all.
Signed-off-by: Sheogorath <sheogorath@shivering-isles.com>
`markdown-pdf` seems to fail to provide the PDFs on tmpfs. This leads
crashing codimd which expects the file to be there. This patch should
add some proper error handling when expectation and reality don't fit
together.
Signed-off-by: Sheogorath <sheogorath@shivering-isles.com>
The current error handling seems to conflict with some sequelize
versions. So we add a second version of it in our excemptions.
I'm not happy about it, but when it helps to prevent further migration
breaking, it's worth it.
Signed-off-by: Sheogorath <sheogorath@shivering-isles.com>
Apart from the uri versioning, one big change is the snippet visibility post data (visibility_level -> visibility)
Default gitlab api version to v4
Signed-off-by: Cédric Couralet <cedric.couralet@gmail.com>
This does some more in depth check on the error message and minimizes
the log noise that is caused by LZString.
Signed-off-by: Sheogorath <sheogorath@shivering-isles.com>
Right now we still see a lot of LZString parsing errors in the logs.
They probably come from the user history. We should minimize the number
by add the basic length check there as well.
Signed-off-by: Sheogorath <sheogorath@shivering-isles.com>
It seems like some providers return strange types for emails which cause
problems. We default to something that is definitely a string.
Signed-off-by: Sheogorath <sheogorath@shivering-isles.com>
As it turns out, expressjs doesn't detect the right mimetype and it
seems like I didn't bother to test this enough. So lets fix it for the
next release.
Signed-off-by: Sheogorath <sheogorath@shivering-isles.com>
