Conversation
Codecov Report
@@ Coverage Diff @@
## master #1307 +/- ##
============================================
+ Coverage 69.44% 69.45% +0.01%
Complexity 1657 1657
============================================
Files 32 32
Lines 4068 4070 +2
============================================
+ Hits 2825 2827 +2
Misses 1243 1243
Continue to review full report at Codecov.
|
|
Note that this may restrict the range of available characters in vocabulary IDs. Currently only |
|
Kudos, SonarCloud Quality Gate passed!
|
|
The last commit changed the sanitizing. Now it's more minimal and only strips colons from the URL, preventing the smuggling of absolute URLs. |
joelit
left a comment
There was a problem hiding this comment.
LGTM - better to just break absolute URLs this way (instead of limiting the possibility of using non-alphanumeric characters in vocabulary IDs).








Reasons for creating this PR
See issue #1289 - it was possible to trick Skosmos to generate links to an arbitrary URL. This could be a potential security issue together with some social engineering (instruct users to click on the link).
Link to relevant issue(s), if any
Description of the changes in this PR
The sanitizing will cause the "smuggled" URLs to stop working, for example
http://example.combecomeshttp//examplecomand thus the links won't work anymore.Known problems or uncertainties in this PR
The language switching links on the 404 page generated for a page such as
/Skosmos/http://example.comwill be broken. It's no worse than before though, since they were already broken, now they are just less harmful.Checklist