/* sbt -- Simple Build Tool
 * Copyright 2008, 2009, 2010 Mark Harrah
package sbt

import Resolver.PluginPattern

import java.io.File
import java.net.URI
import java.text.ParseException
import java.util.concurrent.Callable
import java.util.{Collection, Collections => CS}
import CS.singleton

import org.apache.ivy.{core, plugins, util, Ivy}
import core.{IvyPatternHelper, LogOptions}
import core.cache.{CacheMetadataOptions, DefaultRepositoryCacheManager, ModuleDescriptorWriter}
import core.module.descriptor.{Artifact => IArtifact, DefaultArtifact, DefaultDependencyArtifactDescriptor, MDArtifact}
import core.module.descriptor.{DefaultDependencyDescriptor, DefaultModuleDescriptor, DependencyDescriptor, ModuleDescriptor, License}
import core.module.descriptor.{OverrideDependencyDescriptorMediator}
import core.module.id.{ArtifactId,ModuleId, ModuleRevisionId}
import core.resolve.{IvyNode, ResolveData, ResolvedModuleRevision}
import core.settings.IvySettings
import plugins.latest.LatestRevisionStrategy
import plugins.matcher.PatternMatcher
import plugins.parser.m2.PomModuleDescriptorParser
import plugins.repository.ResourceDownloader
import plugins.resolver.{ChainResolver, DependencyResolver}
import plugins.resolver.util.ResolvedResource
import util.{Message, MessageLogger}
import util.extendable.ExtendableItem

import scala.xml.{NodeSeq, Text}

final class IvySbt(val configuration: IvyConfiguration)
		import configuration.baseDirectory

	/** ========== Configuration/Setup ============
	* This part configures the Ivy instance by first creating the logger interface to ivy, then IvySettings, and then the Ivy instance.
	* These are lazy so that they are loaded within the right context.  This is important so that no Ivy XML configuration needs to be loaded,
	* saving some time.  This is necessary because Ivy has global state (IvyContext, Message, DocumentBuilder, ...).
	private def withDefaultLogger[T](logger: MessageLogger)(f: => T): T =
		def action() =
				val originalLogger = Message.getDefaultLogger
				try { f }
				finally { Message.setDefaultLogger(originalLogger) }
		// Ivy is not thread-safe nor can the cache be used concurrently.
		// If provided a GlobalLock, we can use that to ensure safe access to the cache.
		// Otherwise, we can at least synchronize within the JVM.
		//   For thread-safety in particular, Ivy uses a static DocumentBuilder, which is not thread-safe.
		configuration.lock match
			case Some(lock) => lock(ivyLockFile, new Callable[T] { def call = action() })
			case None => action()
	private lazy val settings: IvySettings =
		val is = new IvySettings
		configuration match
			case e: ExternalIvyConfiguration =>
				IvySbt.addResolvers(e.extraResolvers, is, configuration.log)
				IvySbt.loadURI(is, e.uri)
			case i: InlineIvyConfiguration =>
				is.setVariable("ivy.checksums", i.checksums mkString ",")
				i.paths.ivyHome foreach is.setDefaultIvyUserDir
				IvySbt.configureCache(is, i.localOnly, i.resolutionCacheDir)
				IvySbt.setResolvers(is, i.resolvers, i.otherResolvers, i.localOnly, configuration.log)
				IvySbt.setModuleConfigurations(is, i.moduleConfigurations, configuration.log)
	private lazy val ivy: Ivy =
		val i = new Ivy() { private val loggerEngine = new SbtMessageLoggerEngine; override def getLoggerEngine = loggerEngine }
		i.getLoggerEngine.pushLogger(new IvyLoggerInterface(configuration.log))
	// Must be the same file as is used in Update in the launcher
	private lazy val ivyLockFile = new File(settings.getDefaultIvyUserDir, ".sbt.ivy.lock")
	/** ========== End Configuration/Setup ============*/

	/** Uses the configured Ivy instance within a safe context.*/
	def withIvy[T](log: Logger)(f: Ivy => T): T =
		withIvy(new IvyLoggerInterface(log))(f)

	def withIvy[T](log: MessageLogger)(f: Ivy => T): T =
			try { f(ivy) }
			finally {

	final class Module(rawModuleSettings: ModuleSettings)
		val moduleSettings: ModuleSettings = IvySbt.substituteCross(rawModuleSettings)
		def owner = IvySbt.this
		def withModule[T](log: Logger)(f: (Ivy,DefaultModuleDescriptor,String) => T): T =
			withIvy[T](log) { ivy => f(ivy, moduleDescriptor0, defaultConfig0) }

		def moduleDescriptor(log: Logger): DefaultModuleDescriptor = withModule(log)((_,md,_) => md)
		def dependencyMapping(log: Logger): (ModuleRevisionId, ModuleDescriptor) =
			val md = moduleDescriptor(log)
			(md.getModuleRevisionId, md)
		def defaultConfig(log: Logger): String = withModule(log)( (_,_,dc) => dc)
		// these should only be referenced by withModule because lazy vals synchronize on this object
		// withIvy explicitly locks the IvySbt object, so they have to be done in the right order to avoid deadlock
		private[this] lazy val (moduleDescriptor0: DefaultModuleDescriptor, defaultConfig0: String) =
			val (baseModule, baseConfiguration) =
				moduleSettings match
					case ic: InlineConfiguration => configureInline(ic, configuration.log)
					case ec: EmptyConfiguration => configureEmpty(ec)
					case pc: PomConfiguration => configurePom(pc)
					case ifc: IvyFileConfiguration => configureIvyFile(ifc)
			moduleSettings.ivyScala.foreach(IvyScala.checkModule(baseModule, baseConfiguration, configuration.log))
			(baseModule, baseConfiguration)
		private def configureInline(ic: InlineConfiguration, log: Logger) =
			import ic._
			val moduleID = newConfiguredModuleID(module, moduleInfo, configurations)
			IvySbt.setConflictManager(moduleID, conflictManager, ivy.getSettings)
			val defaultConf = defaultConfiguration getOrElse Configurations.config(ModuleDescriptor.DEFAULT_CONFIGURATION)
			log.debug("Using inline dependencies specified in Scala" + (if(ivyXML.isEmpty) "." else " and XML."))

			val parser = IvySbt.parseIvyXML(ivy.getSettings, IvySbt.wrapped(module, ivyXML), moduleID, defaultConf.name, validate)
			IvySbt.addOverrides(moduleID, overrides, ivy.getSettings.getMatcher(PatternMatcher.EXACT))
			val transformedDeps = IvySbt.overrideDirect(dependencies, overrides)
			IvySbt.addDependencies(moduleID, transformedDeps, parser)
			(moduleID, parser.getDefaultConf)
		private def newConfiguredModuleID(module: ModuleID, moduleInfo: ModuleInfo, configurations: Iterable[Configuration]) =
			val mod = new DefaultModuleDescriptor(IvySbt.toID(module), "release", null, false)
			moduleInfo.homepage foreach { h => mod.setHomePage(h.toString) }
			moduleInfo.licenses foreach { l => mod.addLicense(new License(l._1, l._2.toString)) }
			IvySbt.addConfigurations(mod, configurations)
			IvySbt.addArtifacts(mod, module.explicitArtifacts)

		/** Parses the Maven pom 'pomFile' from the given `PomConfiguration`.*/
		private def configurePom(pc: PomConfiguration) =
			val md = CustomPomParser.default.parseDescriptor(settings, toURL(pc.file), pc.validate)
			val dmd = IvySbt.toDefaultModuleDescriptor(md)
			IvySbt.addConfigurations(dmd, Configurations.defaultInternal)
			val defaultConf = Configurations.DefaultMavenConfiguration.name
			for( is <- pc.ivyScala) if(pc.autoScalaTools) {
				val confParser = new CustomXmlParser.CustomParser(settings, Some(defaultConf))
				addScalaToolDependencies(dmd, confParser, is)
			(dmd, defaultConf)
		/** Parses the Ivy file 'ivyFile' from the given `IvyFileConfiguration`.*/
		private def configureIvyFile(ifc: IvyFileConfiguration) =
			val parser = new CustomXmlParser.CustomParser(settings, None)
			val dmd = IvySbt.toDefaultModuleDescriptor(parser.getModuleDescriptor())
			for( is <- ifc.ivyScala ) if(ifc.autoScalaTools)
				addScalaToolDependencies(dmd, parser, is)
			(dmd, parser.getDefaultConf)
		private def addScalaToolDependencies(dmd: DefaultModuleDescriptor, parser: CustomXmlParser.CustomParser, is: IvyScala) {
			IvySbt.addConfigurations(dmd, Configurations.ScalaTool :: Nil)
			IvySbt.addDependencies(dmd, ScalaArtifacts.toolDependencies(is.scalaOrganization, is.scalaFullVersion), parser)
		private def toURL(file: File) = file.toURI.toURL
		private def configureEmpty(ec: EmptyConfiguration) =
			val defaultConf = ModuleDescriptor.DEFAULT_CONFIGURATION
			val mod = newConfiguredModuleID(ec.module, ec.moduleInfo, Seq(Configurations.Default))
			(mod, defaultConf)

private object IvySbt
	val DefaultIvyConfigFilename = "ivysettings.xml"
	val DefaultIvyFilename = "ivy.xml"
	val DefaultMavenFilename = "pom.xml"
	val DefaultChecksums = Seq("sha1", "md5")

	def defaultIvyFile(project: File) = new File(project, DefaultIvyFilename)
	def defaultIvyConfiguration(project: File) = new File(project, DefaultIvyConfigFilename)
	def defaultPOM(project: File) = new File(project, DefaultMavenFilename)

	def loadURI(is: IvySettings, uri: URI)
		if(uri.getScheme == "file")
			is.load(new File(uri)) // IVY-1114

	/** Sets the resolvers for 'settings' to 'resolvers'.  This is done by creating a new chain and making it the default.
	* 'other' is for resolvers that should be in a different chain.  These are typically used for publishing or other actions. */
	private def setResolvers(settings: IvySettings, resolvers: Seq[Resolver], other: Seq[Resolver], localOnly: Boolean, log: Logger)
		def makeChain(label: String, name: String, rs: Seq[Resolver]) = {
			log.debug(label + " repositories:")
			val chain = resolverChain(name, rs, localOnly, settings, log)
		val otherChain = makeChain("Other", "sbt-other", other)
		val mainChain = makeChain("Default", "sbt-chain", resolvers)
	def resolverChain(name: String, resolvers: Seq[Resolver], localOnly: Boolean, settings: IvySettings, log: Logger): DependencyResolver =
		val newDefault = new ChainResolver {
			// Technically, this should be applied to module configurations.
			// That would require custom subclasses of all resolver types in ConvertResolver (a delegation approach does not work).
			// It would be better to get proper support into Ivy.
			// A workaround is to configure the ModuleConfiguration resolver to be a ChainResolver.
			// This method is only used by the pom parsing code in Ivy to find artifacts it doesn't know about.
			// In particular, a) it looks up source and javadoc classifiers b) it looks up a main artifact for packaging="pom"
			// sbt now provides the update-classifiers or requires explicitly specifying classifiers explicitly
			// Providing a main artifact for packaging="pom" does not seem to be correct and the lookup can be expensive.
			// Ideally this could just skip the lookup, but unfortunately several artifacts in practice do not follow the
			// correct behavior for packaging="pom" and so it is only skipped for source/javadoc classifiers.
			override def locate(artifact: IArtifact) = if(hasImplicitClassifier(artifact)) null else super.locate(artifact)

			override def getDependency(dd: DependencyDescriptor, data: ResolveData) =
				if(data.getOptions.getLog != LogOptions.LOG_QUIET)
					Message.info("Resolving " + dd.getDependencyRevisionId + " ...")
				val gd = super.getDependency(dd, data)
		for(sbtResolver <- resolvers) {
			log.debug("\t" + sbtResolver)
			newDefault.add(ConvertResolver(sbtResolver, settings, log))
	def addResolvers(resolvers: Seq[Resolver], settings: IvySettings, log: Logger)
		for(r <- resolvers) {
			log.debug("\t" + r)
			settings.addResolver(ConvertResolver(r, settings, log))
	/** A hack to detect if the given artifact is an automatically generated request for a classifier,
	* as opposed to a user-initiated declaration.  It relies on Ivy prefixing classifier with m:, while sbt uses e:.
	* Clearly, it would be better to have an explicit option in Ivy to control this.*/
	def hasImplicitClassifier(artifact: IArtifact): Boolean =
		import collection.JavaConversions._
		artifact.getQualifiedExtraAttributes.keys.exists(_.asInstanceOf[String] startsWith "m:")
	private def setModuleConfigurations(settings: IvySettings, moduleConfigurations: Seq[ModuleConfiguration], log: Logger)
		val existing = settings.getResolverNames
		for(moduleConf <- moduleConfigurations)
			import moduleConf._
			import IvyPatternHelper._
			import PatternMatcher._
				settings.addResolver(ConvertResolver(resolver, settings, log))
			val attributes = javaMap(Map(MODULE_KEY -> name, ORGANISATION_KEY -> organization, REVISION_KEY -> revision))
			settings.addModuleConfiguration(attributes, settings.getMatcher(EXACT_OR_REGEXP), resolver.name, null, null, null)
	private def configureCache(settings: IvySettings, localOnly: Boolean, resCacheDir: Option[File])
		configureResolutionCache(settings, localOnly, resCacheDir)
		configureRepositoryCache(settings, localOnly)
	private[this] def configureResolutionCache(settings: IvySettings, localOnly: Boolean, resCacheDir: Option[File])
		val base = resCacheDir getOrElse settings.getDefaultResolutionCacheBasedir
		settings.setResolutionCacheManager(new ResolutionCache(base))
	// set the artifact resolver to be the main resolver.
	// this is because sometimes the artifact resolver saved in the cache is not correct
	// the common case is for resolved.getArtifactResolver to be inter-project from a different project's publish-local
	// if there are problems with this, a less aggressive fix might be to only reset the artifact resolver when it is a ProjectResolver
	// a possible problem is that fetching artifacts is slower, due to the full chain being the artifact resolver instead of the specific resolver
	// This also fixes #760, which occurs when metadata exists in a repository, but the artifact doesn't.
	private[this] def resetArtifactResolver(resolved: ResolvedModuleRevision): ResolvedModuleRevision =
		if(resolved eq null)
		else {
			val desc = resolved.getDescriptor
			val updatedDescriptor = CustomPomParser.defaultTransform(desc.getParser, desc)
			new ResolvedModuleRevision(resolved.getResolver, resolved.getResolver, updatedDescriptor, resolved.getReport, resolved.isForce)

	private[this] def configureRepositoryCache(settings: IvySettings, localOnly: Boolean) //, artifactResolver: DependencyResolver)
		val cacheDir = settings.getDefaultRepositoryCacheBasedir()
		val manager = new DefaultRepositoryCacheManager("default-cache", settings, cacheDir) {
			override def findModuleInCache(dd: DependencyDescriptor, revId: ModuleRevisionId, options: CacheMetadataOptions, r: String) = {
				// ignore and reset the resolver- not ideal, but avoids thrashing.
				val resolved = resetArtifactResolver(super.findModuleInCache(dd,revId,options,null))
				// invalidate the cache if the artifact was removed from the local repository
				if(resolved == null) null
				else if(isProjectResolver(resolved.getResolver)) {
				} else {
					val origin = resolved.getReport.getArtifactOrigin
					if(!origin.isLocal) resolved
					else {
						val file = new File(origin.getLocation)
						if(file == null || file.exists) resolved
						else {
			private[this] def isProjectResolver(r: DependencyResolver): Boolean = r match {
				case pr: ProjectResolver => true
				case _ => false
			/** This is overridden to delete outofdate artifacts of changing modules that are not listed in the metadata.
			* This occurs for artifacts with classifiers, for example. */
			override def cacheModuleDescriptor(resolver: DependencyResolver, mdRef: ResolvedResource, dd: DependencyDescriptor, moduleArtifact: IArtifact, downloader: ResourceDownloader, options: CacheMetadataOptions): ResolvedModuleRevision =
				val rmrRaw = super.cacheModuleDescriptor(null, mdRef, dd, moduleArtifact, downloader, options)
				val rmr = resetArtifactResolver(rmrRaw)
				val mrid = moduleArtifact.getModuleRevisionId
				def shouldClear(): Boolean = rmr != null &&
					( (rmr.getReport != null && rmr.getReport.isSearched && isChanging(dd, mrid)) ||
						isProjectResolver(rmr.getResolver) )
				// only handle changing modules whose metadata actually changed.
				// Typically, the publication date in the metadata has to change to get here.
				if(shouldClear()) {
					// this is the locally cached metadata as originally retrieved (e.g. the pom)
					val original = rmr.getReport.getOriginalLocalFile
					if(original != null) {
						// delete all files in subdirectories that are older than the original metadata file's publication date
						//  The publication date is used because the metadata will be redownloaded for changing files,
						//  so the last modified time changes, but the publication date doesn't
						val pubDate = rmrRaw.getPublicationDate
						val lm = if(pubDate eq null) original.lastModified else pubDate.getTime
						val indirectFiles = PathFinder(original.getParentFile).*(DirectoryFilter).**(-DirectoryFilter).get.toList
						val older = indirectFiles.filter(f => f.lastModified < lm).toList
						Message.verbose("Deleting additional old artifacts from cache for changed module " + mrid + older.mkString(":\n\t", "\n\t", ""))
			// ignore the original resolver wherever possible to avoid issues like #704
			override def saveResolvers(descriptor: ModuleDescriptor, metadataResolverName: String, artifactResolverName: String) {}

			def isChanging(dd: DependencyDescriptor, requestedRevisionId: ModuleRevisionId): Boolean =
				!localOnly && (dd.isChanging || requestedRevisionId.getRevision.contains("-SNAPSHOT"))
		manager.setArtifactPattern(PluginPattern + manager.getArtifactPattern)
		manager.setDataFilePattern(PluginPattern + manager.getDataFilePattern)
		manager.setIvyPattern(PluginPattern + manager.getIvyPattern)
	def toIvyConfiguration(configuration: Configuration) =
		import org.apache.ivy.core.module.descriptor.{Configuration => IvyConfig}
		import IvyConfig.Visibility._
		import configuration._
		new IvyConfig(name, if(isPublic) PUBLIC else PRIVATE, description, extendsConfigs.map(_.name).toArray, transitive, null)
	def addExtraNamespace(dmd: DefaultModuleDescriptor): Unit =
		dmd.addExtraAttributeNamespace("e", "http://ant.apache.org/ivy/extra")

	/** Adds the ivy.xml main artifact. */
	private def addMainArtifact(moduleID: DefaultModuleDescriptor)
		val artifact = DefaultArtifact.newIvyArtifact(moduleID.getResolvedModuleRevisionId, moduleID.getPublicationDate)
	private def setConflictManager(moduleID: DefaultModuleDescriptor, conflict: ConflictManager, is: IvySettings)
		val mid = ModuleId.newInstance(conflict.organization, conflict.module)
		val matcher = is.getMatcher(PatternMatcher.EXACT_OR_REGEXP)
		val manager = is.getConflictManager(conflict.name)
		moduleID.addConflictManager(mid, matcher, manager)

	/** Converts the given sbt module id into an Ivy ModuleRevisionId.*/
	def toID(m: ModuleID) =
		import m._
		ModuleRevisionId.newInstance(organization, name, revision, javaMap(extraAttributes))

	private def substituteCross(m: ModuleSettings): ModuleSettings =
		m.ivyScala match {
			case None => m
			case Some(is) => substituteCross(m, is.scalaFullVersion, is.scalaBinaryVersion)
	private def substituteCross(m: ModuleSettings, scalaFullVersion: String, scalaBinaryVersion: String): ModuleSettings =
		val sub = CrossVersion(scalaFullVersion, scalaBinaryVersion)
		m match {
			case ec: EmptyConfiguration => ec.copy(module = sub(ec.module))
			case ic: InlineConfiguration => ic.copy(module = sub(ic.module), dependencies = ic.dependencies map sub, overrides = ic.overrides map sub)
			case _ => m

	private def toIvyArtifact(moduleID: ModuleDescriptor, a: Artifact, allConfigurations: Iterable[String]): MDArtifact =
		val artifact = new MDArtifact(moduleID, a.name, a.`type`, a.extension, null, extra(a, false))
		copyConfigurations(a, artifact.addConfiguration, allConfigurations)
	def getExtraAttributes(revID: ExtendableItem): Map[String,String] =
			import collection.JavaConverters._
	private[sbt] def extra(artifact: Artifact, unqualify: Boolean = false): java.util.Map[String, String] =
		val ea = artifact.classifier match { case Some(c) => artifact.extra("e:classifier" -> c); case None => artifact }
		javaMap(ea.extraAttributes, unqualify)
	private[sbt] def javaMap(m: Map[String,String], unqualify: Boolean = false) =
		val map = if(unqualify) m map { case (k, v) => (k.stripPrefix("e:"), v) } else m
		if(map.isEmpty) null else scala.collection.JavaConversions.mapAsJavaMap(map)

	private object javaMap
		import java.util.{HashMap, Map}
		def apply[K,V](pairs: (K,V)*): Map[K,V] =
			val map = new HashMap[K,V]
			pairs.foreach { case (key, value) => map.put(key, value) }
	/** Creates a full ivy file for 'module' using the 'dependencies' XML as the part after the &lt;info&gt;...&lt;/info&gt; section. */
	private def wrapped(module: ModuleID, dependencies: NodeSeq) =
		<ivy-module version="2.0" xmlns:e="http://ant.apache.org/ivy/extra">
			{ if(hasInfo(module, dependencies))
				addExtraAttributes(defaultInfo(module), module.extraAttributes)
				// this is because Ivy adds a default artifact if none are specified.
				if(dependencies \\ "publications" isEmpty) <publications/> else NodeSeq.Empty
	private[this] def defaultInfo(module: ModuleID): scala.xml.Elem = {
		import module._
		<info organisation={organization} module={name} revision={revision}/>
	private[this] def addExtraAttributes(elem: scala.xml.Elem, extra: Map[String, String]): scala.xml.Elem =
		(elem /: extra) { case (e, (key,value) ) => e % new scala.xml.UnprefixedAttribute(key, value, scala.xml.Null) }
	private def hasInfo(module: ModuleID, x: scala.xml.NodeSeq) =
		val info = <g>{x}</g> \ "info"
			def check(found: NodeSeq, expected: String, label: String) =
					sys.error("Missing " + label + " in inline Ivy XML.")
				else {
					val str = found.text
					if(str != expected) sys.error("Inconsistent " + label + " in inline Ivy XML.  Expected '" + expected + "', got '" + str + "'")
			check(info \ "@organisation", module.organization, "organisation")
			check(info \ "@module", module.name, "name")
			check(info \ "@revision", module.revision, "version")
	/** Parses the given in-memory Ivy file 'xml', using the existing 'moduleID' and specifying the given 'defaultConfiguration'. */
	private def parseIvyXML(settings: IvySettings, xml: scala.xml.NodeSeq, moduleID: DefaultModuleDescriptor, defaultConfiguration: String, validate: Boolean): CustomXmlParser.CustomParser =
		parseIvyXML(settings,  xml.toString, moduleID, defaultConfiguration, validate)
	/** Parses the given in-memory Ivy file 'xml', using the existing 'moduleID' and specifying the given 'defaultConfiguration'. */
	private def parseIvyXML(settings: IvySettings, xml: String, moduleID: DefaultModuleDescriptor, defaultConfiguration: String, validate: Boolean): CustomXmlParser.CustomParser =
		val parser = new CustomXmlParser.CustomParser(settings, Some(defaultConfiguration))

	/** This method is used to add inline dependencies to the provided module. */
	def addDependencies(moduleID: DefaultModuleDescriptor, dependencies: Seq[ModuleID], parser: CustomXmlParser.CustomParser)
		val converted = dependencies map { dependency => convertDependency(moduleID, dependency, parser) }
		val unique = if(hasDuplicateDependencies(converted)) mergeDuplicateDefinitions(converted) else converted
		unique foreach moduleID.addDependency
	/** Determines if there are multiple dependency definitions for the same dependency ID. */
	def hasDuplicateDependencies(dependencies: Seq[DependencyDescriptor]): Boolean =
		val ids = dependencies.map(_.getDependencyRevisionId)
		ids.toSet.size != ids.size

	/** Combines the artifacts, includes, and excludes of duplicate dependency definitions.
	* This is somewhat fragile and is only intended to workaround Ivy (or sbt's use of Ivy) not handling this case properly.
	* In particular, Ivy will create multiple dependency entries when converting a pom with a dependency on a classified artifact and a non-classified artifact:
	*   https://github.com/sbt/sbt/issues/468
	* It will also allow users to declare dependencies on classified modules in different configurations:
	*   https://groups.google.com/d/topic/simple-build-tool/H2MdAARz6e0/discussion
	* as well as basic multi-classifier handling: #285, #419, #480.
	* Multiple dependency definitions should otherwise be avoided as much as possible.
	def mergeDuplicateDefinitions(dependencies: Seq[DependencyDescriptor]): Seq[DependencyDescriptor] =
		// need to preserve basic order of dependencies: can't use dependencies.groupBy 
		val deps = new java.util.LinkedHashMap[ModuleRevisionId, List[DependencyDescriptor]]
		for( dd <- dependencies )
			val id = dd.getDependencyRevisionId
			val updated = deps get id match {
				case null => dd :: Nil
				case v => dd :: v
			deps.put(id, updated)

			import collection.JavaConverters._
		deps.values.asScala.toSeq.flatMap { dds =>
			val mergeable = (dds, dds.tail).zipped.forall( ivyint.MergeDescriptors.mergeable _)
			if(mergeable) dds.reverse.reduceLeft(ivyint.MergeDescriptors.apply _) :: Nil else dds

	/** Transforms an sbt ModuleID into an Ivy DefaultDependencyDescriptor.*/
	def convertDependency(moduleID: DefaultModuleDescriptor, dependency: ModuleID, parser: CustomXmlParser.CustomParser): DefaultDependencyDescriptor =
		val dependencyDescriptor = new DefaultDependencyDescriptor(moduleID, toID(dependency), dependency.isForce, dependency.isChanging, dependency.isTransitive)
		dependency.configurations match
			case None => // The configuration for this dependency was not explicitly specified, so use the default
				parser.parseDepsConfs(parser.getDefaultConf, dependencyDescriptor)
			case Some(confs) => // The configuration mapping (looks like: test->default) was specified for this dependency
				parser.parseDepsConfs(confs, dependencyDescriptor)
		for(artifact <- dependency.explicitArtifacts)
			import artifact.{name, classifier, `type`, extension, url}
			val extraMap = extra(artifact)
			val ivyArtifact = new DefaultDependencyArtifactDescriptor(dependencyDescriptor, name, `type`, extension, url.getOrElse(null), extraMap)
			copyConfigurations(artifact, ivyArtifact.addConfiguration)
			for(conf <- dependencyDescriptor.getModuleConfigurations)
				dependencyDescriptor.addDependencyArtifact(conf, ivyArtifact)
		for(excls <- dependency.exclusions)
			for(conf <- dependencyDescriptor.getModuleConfigurations)
				dependencyDescriptor.addExcludeRule(conf, IvyScala.excludeRule(excls.organization, excls.name, excls.configurations, excls.artifact))
	def copyConfigurations(artifact: Artifact, addConfiguration: String => Unit): Unit =
		copyConfigurations(artifact, addConfiguration, "*" :: Nil)
	private[this] def copyConfigurations(artifact: Artifact, addConfiguration: String => Unit, allConfigurations: Iterable[String]): Unit =
		val confs = if(artifact.configurations.isEmpty) allConfigurations else artifact.configurations.map(_.name)
		confs foreach addConfiguration

	def addOverrides(moduleID: DefaultModuleDescriptor, overrides: Set[ModuleID], matcher: PatternMatcher): Unit =
		overrides foreach addOverride(moduleID, matcher)
	def addOverride(moduleID: DefaultModuleDescriptor, matcher: PatternMatcher)(overrideDef: ModuleID): Unit =
		val overrideID = new ModuleId(overrideDef.organization, overrideDef.name)
		val overrideWith = new OverrideDependencyDescriptorMediator(null, overrideDef.revision)
		moduleID.addDependencyDescriptorMediator(overrideID, matcher, overrideWith)
	/** It is necessary to explicitly modify direct dependencies because Ivy gives
	* "IllegalStateException: impossible to get artifacts when data has not been loaded."
	* when a direct dependency is overridden with a newer version."*/
	def overrideDirect(dependencies: Seq[ModuleID], overrides: Set[ModuleID]): Seq[ModuleID] =
		def key(id: ModuleID) = (id.organization, id.name)
		val overridden = overrides.map(id => (key(id), id.revision)).toMap
		dependencies map { dep =>
			overridden get key(dep) match {
				case Some(rev) => dep.copy(revision = rev)
				case None => dep

	/** This method is used to add inline artifacts to the provided module. */
	def addArtifacts(moduleID: DefaultModuleDescriptor, artifacts: Iterable[Artifact]): Unit =
		for(art <- mapArtifacts(moduleID, artifacts.toSeq); c <- art.getConfigurations)
			moduleID.addArtifact(c, art)

	def addConfigurations(mod: DefaultModuleDescriptor, configurations: Iterable[Configuration]): Unit =
			configurations.foreach(config => mod.addConfiguration(toIvyConfiguration(config)))

	def mapArtifacts(moduleID: ModuleDescriptor, artifacts: Seq[Artifact]): Seq[IArtifact] =
		lazy val allConfigurations = moduleID.getPublicConfigurationsNames
		for(artifact <- artifacts) yield
			toIvyArtifact(moduleID, artifact, allConfigurations)

	/** This code converts the given ModuleDescriptor to a DefaultModuleDescriptor by casting or generating an error.
	* Ivy 2.0.0 always produces a DefaultModuleDescriptor. */
	private def toDefaultModuleDescriptor(md: ModuleDescriptor) =
		md match
			case dmd: DefaultModuleDescriptor => dmd
			case _ => sys.error("Unknown ModuleDescriptor type.")
	def getConfigurations(module: ModuleDescriptor, configurations: Option[Iterable[Configuration]]) =
		configurations match
			case Some(confs) => confs.map(_.name).toList.toArray
			case None => module.getPublicConfigurationsNames