Skip to content

KOTLIN.md

Latest commit

 

History

History
104 lines (80 loc) · 3.91 KB

KOTLIN.md

File metadata and controls

104 lines (80 loc) · 3.91 KB
Raw

Analyzing Kotlin with MFCQI

MFCQI analyzes Java, Kotlin, and mixed Java/Kotlin repositories with the same 15-metric output contract and the same [0.0, 1.0] geometric-mean score.

Quick start

Install the CLI as described in the README. Kotlin analysis needs no JVM or Kotlin compiler when using a native binary.

# Auto-detect Java, Kotlin, or a mixed repository.
mfcqi analyze .

# Force a mode when analyzing a selected source root.
mfcqi analyze src/main/kotlin --language kotlin
mfcqi analyze . --language mixed

--language (alias --lang) accepts:

Value Behavior
auto Select Java, Kotlin, or mixed analysis from discovered source files
kotlin Analyze .kt files with the Kotlin metric implementations
java Analyze .java files with the Java metric implementations
mixed Combine corresponding Java and Kotlin source metrics

Badge generation and the standalone quality-gate command use the same automatic language selection.

Metric contract

Kotlin reports the same metric keys as Java:

  • Cyclomatic Complexity
  • Cognitive Complexity
  • Halstead Volume
  • Maintainability Index
  • Code Duplication
  • Documentation Coverage
  • security
  • Code Smell Density
  • rfc
  • dit
  • mhf
  • Coupling Between Objects
  • Lack of Cohesion of Methods
  • Dependency Security
  • Secrets Exposure

The source metrics use the same normalization curves and weights as their Java counterparts. Dependency Security and Secrets Exposure are shared language-neutral metrics. In mixed mode, MFCQI averages each corresponding Java and Kotlin normalized source metric, then evaluates the two shared metrics once.

Implementation

Kotlin parsing and duplication detection use PMD 7's Kotlin support, which is based on the official Kotlin grammar. MFCQI calculates its metric formulas over PMD's typed AST and uses PMD CPD for duplication detection. A fingerprinted immutable analysis cache is shared by concurrently executing metrics, so --parallelism does not parse the repository once per metric.

PMD does not embed the Kotlin compiler, and the implementation is verified as part of the GraalVM native CLI build. Kotlin compiler/PSI and Detekt are not runtime dependencies.

Scope and limitations

  • Source discovery includes .kt files. .kts scripts are excluded because PMD's Kotlin source frontend does not parse Gradle or general Kotlin script grammar.
  • Metrics that require semantic type resolution use syntactic Kotlin equivalents. For example, CBO uses declared type references and DIT follows declared inheritance.
  • The security metric is a syntactic heuristic in the source-only CLI, but the Gradle and Maven plugins run real SpotBugs + FindSecBugs bytecode SAST (Kotlin and Java) on the compiled classes — see the plugins in the README.
  • Kotlin Multiplatform source sets are analyzed as ordinary Kotlin source. MFCQI does not yet de-duplicate expect/actual declarations.
  • The LLM tool-output collector is JavaParser-based. Pure Kotlin analysis sends metric scores but does not yet provide Kotlin-specific per-finding file and line context to the recommendation model.

Library use

mfcqi-kotlin is published to Maven Central and depends on pmd-kotlin; no additional repository is required.

repositories { mavenCentral() }
dependencies { implementation("com.integrallis:mfcqi-kotlin:<version>") }
double average = new KotlinCyclomaticComplexity().extract(path);
double normalized = new KotlinCyclomaticComplexity().normalize(average);

List<Metric<?>> fullSuite = KotlinMetrics.all();

Remaining work

  • Add a Kotlin-specific tool-output collector for grounded LLM recommendations.
  • Add KMP source-set awareness and expect/actual de-duplication.
  • Expand real-repository calibration fixtures as Kotlin syntax and PMD evolve.