Spotless confuses real code with simulated code in unit tests
- Dominant language
- Java
- Stars
- 5.6k
- Forks
- 559
- Avg merge
- 1d 13h
- Merged PRs (30d)
- 34
Description
We have a project focused on the application of best practices and code correction using OpenRewrite. We have created a class to implement a custom Recipe that controls and corrects the use of the wildcard "*" in import groupings according to our internally defined standards.
We have a series of unit tests that use the testing libraries provided by OpenRewrite, in which we create code scenarios using string blocks:
```
String before =
"""
import java.util.List;
import java.util.ArrayList;
import java.util.regex.Pattern;
import java.util.Objects;
import java.util.Set;
import java.util.Map;
""";
String after =
"""
import java.util.*;
import java.util.regex.Pattern;
""";
rewriteRun(java(before, after));
```
The problem is that when we run the check command for the Maven plugin com.diffplug.spotless:spotless-maven-plugin:3.8.0, Spotless detects these strings as if they were actual code, despite the use of triple quotes -> """ string code """.
`OrderImportsCustomTest.java:L83 toggle(import java.util.*;) Do not use wildcard imports (e.g. java.util.*) - replace with specific class imports (e.g. java.util.List) as 'spotlessApply' cannot auto-fix this`
We have verified that this behavior only occurs in versions 3.7.0 and above. Version 3.6.0 correctly ignores this simulated code.
We have had to apply a workaround using // spotless:off, but we would like to know if there is a permanent solution to this behavior, which we believe is a false positive.
Poject Stack:
- Java 25
- Openrewrite 8.72.0
- Spottles Maven Plugin 3.8.0
Contributor guide
Research direction
Start with OrderImportsCustomTest.java at line 83 and reproduce the report using the Spotless Maven plugin versions 3.6.0, 3.7.0, and 3.8.0. Trace why the Java text block is treated as source code during the check command, then verify that the simulated import is ignored while real wildcard imports are still reported.
Written by the indexing model from the issue text.
Assessment
- Tech stack
- java
- Domain
- build-system
- Issue type
- Bug
- Difficulty
- 3/5
- Estimated time
- 1-2 days
- Activity status
- Quiet
- Clarity
- Mostly clear
- Newbie friendliness
- 52/100