feature request #9114
Handle malformed URIs in user type
40%
Description
The fix for #9111 should be moved to a new URI user type wrapper that is used instead of the current URI class. Otherwise the workaround needs to be applied whereever URIs are used.
Related issues
Associated revisions
ref #9114 adapt URI to wrapper in cdmlib-app
ref #9114 add URI Wrapper and replace all java.net.URI occurrences by the wrapper in cdmlib
ref #9114 preliminary ignore RdfViewTest
ref #9114 adapt URIs to URI wrapper in vaadin
ref #9114 adapt URI to wrapper in cdmlib-app (cont.)
ref #9114 adapt some classes with URI to wrapper in taxeditor
ref #9114 adapt some classes with URI to wrapper in taxeditor (cont.)
ref #9114 add compareTo to URI wrapper
ref #9114 adapt some classes with URI to wrapper in taxeditor (cont.)
ref #9114 adapt some classes with URI to wrapper in taxeditor (cont.)
ref #9114 adapt some classes with URI to wrapper in taxeditor (cont.)
ref #9114 adapt some classes with URI to wrapper in taxeditor (cont.)
ref #9114 adapt some classes with URI to wrapper in taxeditor (cont.)
ref #9114 add toFile() to URI to wrapper
ref #9114 adapt javadoc for new URI wrapper user type
ref #9114 adapt json mapping to new URI wrapper
ref #9114 further adapt some classes to URI wrapper in taxeditor
ref #9114 further adapt jsonConfigurations and add new constructor to URI wrapper
ref #9114 fix RdfViewTest
ref #9114 fix dozer mapping by reverting URI mappings to java.net.URI
History
#1 Updated by Andreas Müller about 1 year ago
- Related to feature request #9111: Handle Malformed URIs added
#2 Updated by Andreas Müller 7 months ago
- Description updated (diff)
- % Done changed from 0 to 40
#3 Updated by Andreas Müller 7 months ago
- Status changed from New to In Progress
I implemented a new class eu.etaxonomy.cdm.common.URI and use this in model but also at other places like import sources.
Unfortunately, simply subclassing java.net.URI was not possible as it is final.
A method getJavaUri() exists to transform the above class to standard java.net.URI.
Currently there are still 2 open issues:
- httpinvoker: within httpinvoker there is still a ClassCastException and therefore TaxEditor fails to build. First I thought this is because CdmModelCache needs to be build anew, but running the script does not change the model cache file so this does not seem to be the reason
RdfViewTest.testMarshalRdf does not run successfully and therefore I set it to ignore. The children type dc:identifier/ do not seem to be build correctly.
#4 Updated by Andreas Müller 7 months ago
- Assignee changed from Andreas Müller to Katja Luther
- Target version changed from Release 5.18 to Release 5.19
Katja, can you please check if you can find out what the httpinvoker problem is. I currently can't start the TaxEditor from the IDE for debugging.
#5 Updated by Andreas Müller 7 months ago
In #9111#note-1 AK suggests to also encode fragment and query parts. This has not been implemented in the original code in #9111 which has been copied to cdmlib.
I think we should also implement it for fragment and query. Probably not for authority and scheme?
Also the handling of backslash needs to be discussed. In the original version backslash was encoded, too. This is probably not wanted, at least not for the path part. Currently a URISyntaxException is thrown if path contains a backslash.
#6 Updated by Andreas Müller 6 months ago
- Target version changed from Release 5.19 to Release 5.21
#7 Updated by Katja Luther 4 months ago
- Target version changed from Release 5.21 to Release 5.22
#8 Updated by Andreas Müller 4 months ago
- Tags set to fast
This ticket seems to be done or needs to be splitted into an open issues ticket. It can be moved back to 5.19(?) afterwards.
#9 Updated by Katja Luther 3 months ago
- Status changed from In Progress to Feedback
- Assignee changed from Katja Luther to Andreas Müller
Andreas Müller wrote:
In #9111#note-1 AK suggests to also encode fragment and query parts. This has not been implemented in the original code in #9111 which has been copied to cdmlib.
I think we should also implement it for fragment and query. Probably not for authority and scheme?
Also the handling of backslash needs to be discussed. In the original version backslash was encoded, too. This is probably not wanted, at least not for the path part. Currently a URISyntaxException is thrown if path contains a backslash.
Did you implement this already?
#10 Updated by Andreas Müller 3 months ago
- Assignee changed from Andreas Müller to Katja Luther
Katja Luther wrote:
Andreas Müller wrote:
In #9111#note-1 AK suggests to also encode fragment and query parts. This has not been implemented in the original code in #9111 which has been copied to cdmlib.
I think we should also implement it for fragment and query. Probably not for authority and scheme?
Also the handling of backslash needs to be discussed. In the original version backslash was encoded, too. This is probably not wanted, at least not for the path part. Currently a URISyntaxException is thrown if path contains a backslash.
Did you implement this already?
No, nees to be finished
#11 Updated by Katja Luther 3 months ago
- Assignee changed from Katja Luther to Andreas Müller
Andreas Müller wrote:
Katja Luther wrote:
Andreas Müller wrote:
In #9111#note-1 AK suggests to also encode fragment and query parts. This has not been implemented in the original code in #9111 which has been copied to cdmlib.
I think we should also implement it for fragment and query. Probably not for authority and scheme?
Also the handling of backslash needs to be discussed. In the original version backslash was encoded, too. This is probably not wanted, at least not for the path part. Currently a URISyntaxException is thrown if path contains a backslash.
Did you implement this already?
No, nees to be finished
Should I do this? Because you worked on this ticket.
#12 Updated by Andreas Müller 3 months ago
- Assignee changed from Andreas Müller to Katja Luther
I only moved the original algorithm to the user type. The remaining issues are about the algorithm. As you implemented the algorithm I thought you will do this.
Or at least give hints why e.g. fragments and query parts were not yet encoded and why backslash was encoded. I didn't want to change the algorithm without feedback on these issues.
Maybe also Andreas K. can give feedback on this.
#13 Updated by Andreas Müller 15 days ago
- Target version changed from Release 5.22 to Release 5.25
This still needs feedback so I move to current milestone. Once it is fixed we can move to 5.19 where most of the work was done or keep it in current milestone.
#14 Updated by Andreas Müller 8 days ago
- Target version changed from Release 5.25 to Release 5.26
