RemoveDependency: how to delete multiple things with same groupId?

[svaningelgem]

I have this recipe:

recipeList:
  - org.openrewrite.maven.RemoveDependency:
      groupId: com.example.ws4
      artifactId: security
  - org.openrewrite.maven.RemoveDependency:
      groupId: com.example.ws4
      artifactId: functions

However, when ran I only see one is processed (the first):

[WARNING] Changes have been made to api-impl\pom.xml by:
[WARNING]     org.openrewrite.recommendations.CodeStyle
[WARNING]         org.openrewrite.xml.format.AutoFormat
[WARNING]         org.openrewrite.maven.RemoveDependency: {groupId=com.example.ws4, artifactId=security}
[WARNING]             org.openrewrite.maven.ChangeDependencyGroupIdAndArtifactId: {oldGroupId=com.example.ws4, oldArtifactId=*, newGroupId=com.example.ws5}

Why are the others not processed? (FYI: the other recipe also matches in that same pom file)
Or should I understand it as that the recipe is a set, with the key the class name?
=> that would make sense to remove duplicates from recipes -- but in this case it's not taking into account the parameters - which are different...
=> And if so: that would be a bug imho?

Oh, I also tried to extract them from the recipeList in their own recipe, and include that in the main list like this:

recipeList:
- com.example.removeSecurity
- com.example.removeFunctions

But that didn't work out either (also only security is removed)

[timtebeek]

Hi @svaningelgem ; I've quickly tried to replicate this issue with a unit test, but was not able to using this example:

@Issue("https://github.com/openrewrite/rewrite/discussions/4792")
@Test
void removeTwoDependencies() {
    rewriteRun(
      spec -> spec.recipes(
        new RemoveDependency("junit", "junit", null),
        new RemoveDependency("com.google.guava", "guava", null)
      ),
      pomXml(
        """
          <project>
            <modelVersion>4.0.0</modelVersion>
            <groupId>com.mycompany.app</groupId>
            <artifactId>my-app</artifactId>
            <version>1</version>
            <dependencies>
              <dependency>
                <groupId>com.google.guava</groupId>
                <artifactId>guava</artifactId>
                <version>29.0-jre</version>
              </dependency>
              <dependency>
                <groupId>junit</groupId>
                <artifactId>junit</artifactId>
                <version>4.13.1</version>
                <scope>test</scope>
              </dependency>
            </dependencies>
          </project>
          """,
        """
          <project>
            <modelVersion>4.0.0</modelVersion>
            <groupId>com.mycompany.app</groupId>
            <artifactId>my-app</artifactId>
            <version>1</version>
          </project>
          """
      )
    );
}

Are you able to replicate this with a smaller sample perhaps?

As an implementation detail: the dependency is removed by calling out to a visitor to remove that specific tag:

rewrite/rewrite-maven/src/main/java/org/openrewrite/maven/RemoveDependency.java

Line 83 in 094e714

doAfterVisit(new RemoveContentVisitor<>(tag, true, true));

If that tag can then not be found, perhaps because another recipe modified it in some way, then it will not be removed as it fails to match (different identity).

rewrite/rewrite-xml/src/main/java/org/openrewrite/xml/RemoveContentVisitor.java

Lines 37 to 40 in 094e714

	public Xml visitTag(Xml.Tag tag, P p) {
	Xml.Tag t = (Xml.Tag) super.visitTag(tag, p);
	if (t.getContent() != null) {

[svaningelgem]

That might be possible because I'm changing the groupId from ws4 to ws5. I'll have a look tomorrow if removing that rule fixes it.
But that would mean that I need to duplicate the removal rules. Once for ws4 & once for ws5. Because the order is not guaranteed.

[shanman190]

Also to note, the groupId and artifactId I believe support glob patterns, so you could just throw a ? (match single character only) or * (match all remaining characters) on the end of either value to get back down to a single recipe statement.

[svaningelgem]

Thanks, I'll try that, but for now I have another issue with this removal. So I don't know if the ?/* is actually working.