- add new servlet to get meta data from PlantUML diagram
- meta data get not only be requested as text but also as json if you set the `Accept`-header to json
- add `metadata` servlet tests
- GET: like the Proxy where you can pass a URL which the servlet will use to fetch the diagram image
- POST: file upload
- typing `<:` will start the emoji auto complition inside the plantuml editor
- for the sake of simplicity the emoji preview of the completion documentation will fetch the image from the original github repository (not plantuml). The reason is that the images (SVGs) inside plantuml have sometimes removed their svg tag, hence it's difficult to set the correct rendering size.
- expand auto completion (suggestion) documentation by default
- add emoji example GIF and documentation
- set charset to utf-8 for each website
- refactor JSON creation inside UI Helper
Since the update of `bastik` the PDF servlet didn't work anymore.
Problems:
- No PDF UnitTest existed
- Pom dependency `org.apache.xmlgraphics.batik-all` change nearly every dependency to `optional` starting with version `1.15`, hence some important dependencies like the SVG converter were missing
- `DiagramResponse.CONTENT_TYPE` had no mime type value for PDF
Changes:
- add PDF UnitTest
- remove `batik-all` dependency and only include the dependencies PlantUML requires for the PDF generation:
* batik-dom
* batik-svgrasterizer (includes batik-dom)
* batik-svggen
* fop
- remove own `DiagramResponse.CONTENT_TYPE` mapping and use `FileFormat.getMimeType()`
This will change the Graphviz behaviour inside the docker container.
Until now it was always a fixed Graphviz version inside the Dockerfile.
With these changes the docker container will always use the latest Graphviz release if the version was not manually set via the docker build argument `GRAPHVIZ_VERSION`.
If not set as build arg fetch latest version of Graphviz based on their release name by default.
An alternative would be to use `tag_name`.
Also, to avoid the overhead of getting all kinds of relase data, GitLabs GraphQL interface could be used.
```bash
curl -s \
-H "Content-Type:application/json" \
-d '{"query": "{project(fullPath:\"graphviz/graphviz\"){releases(first:5,sort:RELEASED_AT_DESC){nodes{name}}}}"}' \
"https://gitlab.com/api/graphql" \
| jq -r '.data.project.releases.nodes[].name' | sort -V -r | head -n 1
```
In GraphQL the alternative to `name` would be `tagName`.
- add viewport to header (otherwise smartphones will render the desktop version)
- set focus to the code editor after loading the page
- fix some android dark mode theme issues by settings clear default for the light theme as well
There were same base image and naming changes for the jetty and tomcat images plantuml-server uses => plantuml-server uses a pretty old and vulnerable images.
- update base images for jetty and tomcat
- add a alpine version for jetty (not tomcat has no official alpine version)
- add autoremove and clean commands after graphviz installation. This reduces the images by about ~280 MB
- PDF dependency was missing in the pom file the JDK8
* We should think about creating a parent pom - in that case all plantuml dependencies could be in the parent pom and we would only the mantain one pom file.
(It is also possible to drop the Java 8 support.)
* Why do we not have any PDF tests?
- add rule to ignore version update hint with `-dev` followed by a dot and date (e.g. `0.37.0-dev.20230308`)
- migration from JUnit4 to JUnit5
- revert the 4 missing tests, e.g. proxy test from commit 20468f5
- add virtual host name `test.localhost` to embedded jetty server
(JUnit Tests) since localhost and IP-Addresses are no longer
supported by the proxy and use this address inside proxy `src`
- add `test-localhost` support for the docker tests. To support
this the docker hostname need to be set to test.localhost by:
`--hostname=test.localhost` (only for the docker tests)
- proxy: add file format support for PDF
- proxy: add error messages on "bad request" response
- proxy: remove dead code
- old proxy: add error messages on "bad request" response
- fix incorrect README link to docs
- add `HTTP_PROXY_READ_TIMEOUT` option -- close #240
- auto refresh function
- light and dark theme
- monaco editor (vscode) with "apex" as syntax highlighting language
* apex seems to work quite fine (better than no highlighting)
* future possibility: own plantuml language syntax support
* future possibility: autocomplete (to much work but maybe partial)
- implemented example for `!theme ...`
- implemented example for `<&icon>`
* future possibility: code validation
- implemented example for `@start...` and `@end...`:
* should be the first or last command
* should be of the some type (e.g. `@startyaml` and @endyaml)
* should be used exactly once per document/diagram
- editor and preview is splitable into two windows like the
"Extract window" functionality on
(plantuml.com)[https://www.plantuml.com/plantuml]
- multi index / multi paging diagram support
- diagram can be displayed/rended as PNG, SVG, ASCII Art or PDF
- Ctrl+s download the PlantUML Code as code file (diagram.puml)
- Ctrl+, opens the settings and Esc closes the settings
- use html `base` tag containing the context path once instead inside
every single URL/link.
- update and enhance `nginx-contextpath` example
- export javascript code into separated file
- Add TODO note to javascript clipboard check (from PR#250) since
Firefox and Safari do not support the current implementation
Switch from absolute paths `hostpath` to relative paths `contextpath`.
Unfortunately, for the url input javascript is necessary to resolve the
relative url.
Also see Issue #205.
