关于sbt-assembly合并策略去重机制及路径映射的技术咨询

Hey there! Let's break this down step by step since you're new to sbt and sbt-assembly—this stuff can feel overwhelming at first, but once you get the hang of path mapping and merge strategies, it'll click.

Understanding Path Mapping in sbt-assembly #

First, let's clarify how sbt-assembly maps your project's files to the paths it uses for merging:

• When you compile your Scala Play project, source files in src/main/scala/com/example/mypackage get compiled to class files in target/scala-<version>/classes/com/example/mypackage/.
• sbt-assembly treats these compiled class paths (not your source code paths) as the key for merging. So a class like com.example.mypackage.Utils becomes the path com/example/mypackage/Utils.class in the assembly process.
• The PathList helper in merge strategies lets you match these paths by splitting them into a list of segments. For example, PathList("com", "example", "mypackage", _*) matches every file under the com.example.mypackage package and its subpackages.

How Merge Strategies Handle Duplication Errors #

By default, sbt-assembly uses MergeStrategy.singleOrError: if it finds two files with the exact same path (from your project's code, dependencies, or other modules), it throws a duplication error. This is to prevent accidental overwrites of important files.

To fix these errors, you need to tell sbt-assembly exactly what to do when it encounters duplicates for specific paths. Common strategies include:

• MergeStrategy.first: Keep the first version of the file it finds, discard all others.
• MergeStrategy.last: Keep the last version found.
• MergeStrategy.concat: Append the contents of all duplicate files (great for config files like application.conf).
• MergeStrategy.discard: Drop the duplicate file entirely (use this only if you're sure you don't need it).

Configuring Two JARs for Different Packages in a Play Project #

To build two separate JARs targeting different packages, you'll need to define custom assembly tasks in your build.sbt. Here's a concrete example:

import sbtassembly.AssemblyPlugin.autoImport._

// Define assembly task for the first package (e.g., com.example.service)
lazy val assemblyServiceJar = taskKey[File]("Build JAR for service package")
assemblyServiceJar := {
  val jarPath = target.value / "service-assembly.jar"
  assembly.value.copyTo(jarPath)
}
// Configure settings for the service JAR
assemblyServiceJar := assemblyServiceJar.dependsOn(compile in Compile).value
assemblyMergeStrategy in assemblyServiceJar := {
  // Match all files under com.example.service and its subpackages
  case PathList("com", "example", "service", _*) => MergeStrategy.first
  // Handle Play-specific config files
  case "application.conf" => MergeStrategy.concat
  case "reference.conf" => MergeStrategy.concat
  // Fall back to the default strategy for all other paths
  case path =>
    val defaultStrategy = (assemblyMergeStrategy in assembly).value
    defaultStrategy(path)
}

// Define assembly task for the second package (e.g., com.example.client)
lazy val assemblyClientJar = taskKey[File]("Build JAR for client package")
assemblyClientJar := {
  val jarPath = target.value / "client-assembly.jar"
  assembly.value.copyTo(jarPath)
}
// Configure settings for the client JAR
assemblyClientJar := assemblyClientJar.dependsOn(compile in Compile).value
assemblyMergeStrategy in assemblyClientJar := {
  case PathList("com", "example", "client", _*) => MergeStrategy.first
  case "application.conf" => MergeStrategy.concat
  case "reference.conf" => MergeStrategy.concat
  case path =>
    val defaultStrategy = (assemblyMergeStrategy in assembly).value
    defaultStrategy(path)
}

Troubleshooting Duplication Errors #

When you hit a duplication error, start by reading the error message carefully—it will list the conflicting file paths and where they're coming from (e.g., a dependency JAR vs. your project code). Here's how to handle common cases:

1. Duplicate config files: Use MergeStrategy.concat for files like application.conf or reference.conf to combine settings from different sources.
2. Duplicate third-party classes: If two dependencies include the same class, either exclude one of the dependencies (using excludeDependencies) or use MergeStrategy.first to keep the version you need.
3. Duplicate project classes: Double-check your package structure to make sure you don't have identical class names in different modules, or adjust the merge strategy to prioritize the correct version.

内容的提问来源于stack exchange，提问作者David