#1392: smart completions

[Caylipp]

This PR fixes #1392

Implemented changes:

• Add AutoCompletionRegistry for tool-specific completion candidates
• Add ToolArgumentsProperty for completing tool arguments
• Delegate tool argument completion from ToolArgumentsProperty to ToolCommandlet
• Add default empty completion registry in ToolCommandlet
• Add initial Maven completion candidates in Mvn

---

Testing instructions

Please add conscise, understandable instructions on how a reviewer can test/verify the functionality of your contribution here:

1. Create an run configuration for IDEasy.
2. Set the program arguments to: -q complete mvn cl
3. Verify that Maven goals and options are listed, e.g.: clean

Additionally, you can use the end-to-end testing script to test the completion natively in your terminal (#2011).

---

Checklist for this PR

Make sure everything is checked before merging this PR. For further info please also see
our DoD.

• When running mvn clean test locally all tests pass and build is successful
• PR title is of the form #«issue-id»: «brief summary» (e.g. #921: fixed setup.bat). If no issue ID exists, title only.
• PR top-level comment summarizes what has been done and contains link to addressed issue(s)
• PR and issue(s) have suitable labels
• Issue is set to In Progress and assigned to you or there is no issue (might happen for very small PRs)
• You followed all coding conventions
• You have added the issue implemented by your PR in CHANGELOG.adoc unless issue is labeled
  with internal
• You have formulated clear instructions on how to test your contribution under "Testing instructions"

[coveralls]

Coverage Report for CI Build 27742396915

Coverage increased (+0.02%) to 71.299%

Details

• Coverage increased (+0.02%) from the base build.
• Patch coverage: No coverable lines changed in this PR.
• 108 coverage regressions across 6 files.

Uncovered Changes

No uncovered changes found.

Coverage Regressions

108 previously-covered lines in 6 files lost coverage.

File	Lines Losing Coverage	Coverage
com/devonfw/tools/ide/tool/ToolCommandlet.java	57	73.67%
com/devonfw/tools/ide/context/AbstractIdeContext.java	28	66.97%
com/devonfw/tools/ide/tool/mvn/Mvn.java	18	84.69%
com/devonfw/tools/ide/cli/CliArguments.java	2	64.15%
com/devonfw/tools/ide/version/VersionSegment.java	2	89.24%
com/devonfw/tools/ide/tool/ide/IdeToolCommandlet.java	1	78.33%

---

Coverage Stats

Relevant Lines:	16264
Covered Lines:	12097
Line Coverage:	74.38%
Relevant Branches:	7268
Covered Branches:	4681
Branch Coverage:	64.41%
Branches in Coverage %:	Yes
Coverage Strength:	3.15 hits per line

---

💛 - Coveralls

[AdemZarrouki]

Nice Job. You extended the auto complition for mvn.

[AdemZarrouki]

When testing with GraalVM using ide shell, i typed ide mvn -Dexec.m and then pressed tab but the shell was not responding anymore. @Caylipp can you check the cause of this. Aslo tried to run it in a junit test

  @Test
  void testCompleteMavenToolArguments() {

    // arrange
    AbstractIdeContext context = newContext(PROJECT_BASIC, null, false);
    CliArguments args = CliArguments.ofCompletion("mvn", "-Dexec.main");

    // act
    List<CompletionCandidate> candidates = context.complete(args, true);

    // assert
    assertThat(candidates.stream().map(CompletionCandidate::text))
        .contains("-Dexec.mainClass=");
  }

but the test was running for ever

[Caylipp]

When testing with GraalVM using ide shell, i typed ide mvn -Dexec.m and then pressed tab but the shell was not responding anymore. @Caylipp can you check the cause of this. Aslo tried to run it in a junit test

  @Test
  void testCompleteMavenToolArguments() {

    // arrange
    AbstractIdeContext context = newContext(PROJECT_BASIC, null, false);
    CliArguments args = CliArguments.ofCompletion("mvn", "-Dexec.main");

    // act
    List<CompletionCandidate> candidates = context.complete(args, true);

    // assert
    assertThat(candidates.stream().map(CompletionCandidate::text))
        .contains("-Dexec.mainClass=");
  }

but the test was running for ever

Thanks for pointing this out.
I found and fixed the issue. The problem was that -Dexec.m was treated as an option because it starts with -. However, in this case it is not an IDEasy option, but a Maven argument.
If no matching option was found, the completion logic did not advance the current argument and ended up in an endless loop. I added a fallback to value completion for this case.
I tested complete mvn -Dexec.m locally in the shell.

[AdemZarrouki]

I tested the new fix and the endless loop wasn't present anymore. The only problem is that in the issue its stated that
"Also be aware that those "keywords" above that end with = are not complete so they should not add a whitespace after its completion" but when testing the auto-completion with -Dexec.mainClass= and -Dexec.args= i get a whitespace after =. This need to be fixed.

[Caylipp]

I tested the new fix and the endless loop wasn't present anymore. The only problem is that in the issue its stated that "Also be aware that those "keywords" above that end with = are not complete so they should not add a whitespace after its completion" but when testing the auto-completion with -Dexec.mainClass= and -Dexec.args= i get a whitespace after =. This need to be fixed.

You’re right, thanks for pointing this out.
The issue was caused by Bash automatically appending a whitespace after completion. I’ve fixed it in the _ide_completion function by applying compopt -o nospace for candidates ending with =.

[Caylipp]

Also fixed completion for : (e.g. dependency:list) by removing it from COMP_WORDBREAKS to prevent Bash from splitting the argument.

[AdemZarrouki]

Good Job @Caylipp. You fixed the issue. I left a small comment regarding line break in javadoc that need to be removed

[hohwille]

@Caylipp thanks for your PR. Looks really nice now. Great work 👍
I have to admit that I do not fully understand the workaround for the = handling.
Is that maybe because = is by default in COMP_WORDBREAKS?
I guess this is for situations where we have --option=val[TAB] and want to get --option=value. So what is by default braking here? Is a space added before or directly after the equals sign?
Then I get at least the compopt -o nospace workaround.

[Caylipp]

@Caylipp thanks for your PR. Looks really nice now. Great work 👍 I have to admit that I do not fully understand the workaround for the = handling. Is that maybe because = is by default in COMP_WORDBREAKS? I guess this is for situations where we have --option=val[TAB] and want to get --option=value. So what is by default braking here? Is a space added before or directly after the equals sign? Then I get at least the compopt -o nospace workaround.

The whitespace issue is not caused by = being part of COMP_WORDBREAKS. Instead, this is Bash's default behavior: after inserting a completion candidate, Bash appends a space by default because it assumes the token is complete. Since in the issue (#1392) is mentioned "those "keywords" above that end with = are not complete so they should not add a whitespace after its completion", this assumption would be incorrect. Using compopt -o nospace prevents Bash from appending this whitespace.

In case of : (e.g. dependency:list), the problem is different. Here : is part of COMP_WORDBREAKS, so Bash stops completion at dependency:. Therefore, adjusting COMP_WORDBREAKS is required to allow completing the full token dependency:list.

[Caylipp]

I also noticed one additional edge case while retesting: ide mvn --<TAB> currently does not show the expected --... Maven suggestions, while ide mvn -<TAB> works.
This seems to be a separate parsing edge case around --, since -- can also be interpreted as the end-of-options marker.

[Caylipp]

I did some debugging on this. For ideasy -q complete mvn --, the argument -- is already consumed during the initial parsing (before CompleteCommandlet.doRun() is called) so it never reaches the completion logic in CliArguments.ofCompletion(...).
That is why mvn --<TAB> currently does not work, while mvn -<TAB> and mvn --a<TAB> do.
So this seems to be a parsing limitation where -- is treated as end-of-options too early for the completion use case.
Since the release is close and the main Maven smart completion cases are working now, I would suggest tracking this -- edge case separately unless you think it should still be included in this PR.

I also noticed one additional edge case while retesting: ide mvn --<TAB> currently does not show the expected --... Maven suggestions, while ide mvn -<TAB> works. This seems to be a separate parsing edge case around --, since -- can also be interpreted as the end-of-options marker.