feat: initial attempt at renamer parallelisation

This commit is contained in:
2025-03-13 20:45:57 +00:00
parent 8d8df3357d
commit c73b073f23
3 changed files with 318 additions and 160 deletions

View File

@@ -32,7 +32,10 @@ object ast {
object StrLiter extends ParserBridgePos1Atom[String, StrLiter]
case class PairLiter()(val pos: Position) extends Expr6
object PairLiter extends ParserBridgePos0[PairLiter]
case class Ident(v: String, var uid: Int = -1)(val pos: Position) extends Expr6 with LValue
case class Ident(v: String, var guid: Int = -1, var ty: types.RenamerType = types.?)(
val pos: Position
) extends Expr6
with LValue
object Ident extends ParserBridgePos1Atom[String, Ident] {
def apply(v: String)(pos: Position): Ident = new Ident(v)(pos)
}
@@ -186,7 +189,9 @@ object ast {
/* ============================ STATEMENTS ============================ */
sealed trait Stmt
sealed trait Stmt {
val pos: Position
}
case class Skip()(val pos: Position) extends Stmt
object Skip extends ParserBridgePos0[Skip]
case class VarDecl(varType: Type, name: Ident, value: RValue)(val pos: Position) extends Stmt

View File

@@ -1,6 +1,19 @@
package wacc
import java.io.File
import scala.collection.mutable
import cats.effect.IO
import cats.syntax.all._
import cats.implicits._
import cats.data.Chain
import cats.Foldable
import cats.Functor
import cats.data.NonEmptyList
import parsley.{Failure, Success}
import cats.data.NonEmptyChain
import cats.NonEmptyParallel
private val MAIN = "$main"
object renamer {
import ast._
@@ -11,209 +24,347 @@ object renamer {
case Var
}
private case class ScopeKey(path: String, name: String, identType: IdentType)
private case class ScopeValue(id: Ident, public: Boolean)
private class Scope(
val current: mutable.Map[(String, IdentType), Ident],
val parent: Map[(String, IdentType), Ident]
private val current: mutable.Map[ScopeKey, ScopeValue],
private val parent: Map[ScopeKey, ScopeValue],
guidStart: Int = 0,
val guidInc: Int = 1
) {
private var guid = guidStart
private var immutable = false
private def nextGuid(): Int = {
val id = guid
guid += guidInc
id
}
private def verifyMutable(): Unit = {
if (immutable) throw new IllegalStateException("Cannot modify an immutable scope")
}
/** Create a new scope with the current scope as its parent.
*
* To be used for single-threaded applications.
*
* @return
* A new scope with an empty current scope, and this scope flattened into the parent scope.
*/
def subscope: Scope =
Scope(mutable.Map.empty, Map.empty.withDefault(current.withDefault(parent)))
def withSubscope[T](f: Scope => T): T = {
val subscope = Scope(mutable.Map.empty, Map.empty.withDefault(current.withDefault(parent)), guid, guidInc)
immutable = true
val result = f(subscope)
guid = subscope.guid // Sync GUID
immutable = false
result
}
/** Create new scopes with the current scope as its parent and GUID numbering adjusted
* correctly.
*
* This will permanently mark the current scope as immutable, for thread safety.
*
* To be used for multi-threaded applications.
*
* @return
* New scopes with an empty current scope, and this scope flattened into the parent scope.
*/
def subscopes(n: Int): Seq[Scope] = {
verifyMutable()
immutable = true
(0 until n).map { i =>
Scope(
mutable.Map.empty,
Map.empty.withDefault(current.withDefault(parent)),
guid + i * guidInc,
guidInc * n
)
}
}
/** Attempt to add a new identifier to the current scope. If the identifier already exists in
* the current scope, add an error to the error list.
*
* @param ty
* The semantic type of the variable identifier, or function identifier type.
* @param name
* The name of the identifier.
* @param globalNames
* The global map of identifiers to semantic types - the identifier will be added to this
* map.
* @param globalNumbering
* The global map of identifier names to the number of times they have been declared - will
* used to rename this identifier, and will be incremented.
* @param errors
* The list of errors to append to.
* @return
* An error, if one occurred.
*/
def add(ty: SemType | FuncType, name: Ident)(using
globalNames: mutable.Map[Ident, SemType],
globalFuncs: mutable.Map[Ident, FuncType],
globalNumbering: mutable.Map[String, Int],
errors: mutable.Builder[Error, List[Error]]
) = {
val identType = ty match {
def add(name: Ident, public: Boolean = false): Chain[Error] = {
verifyMutable()
val path = name.pos.file.getCanonicalPath
val identType = name.ty match {
case _: SemType => IdentType.Var
case _: FuncType => IdentType.Func
}
current.get((name.v, identType)) match {
case Some(Ident(_, uid)) =>
errors += Error.DuplicateDeclaration(name)
name.uid = uid
val key = ScopeKey(path, name.v, identType)
current.get(key) match {
case Some(ScopeValue(Ident(_, id, _), _)) =>
name.guid = id
Chain.one(Error.DuplicateDeclaration(name))
case None =>
val uid = globalNumbering.getOrElse(name.v, 0)
name.uid = uid
current((name.v, identType)) = name
ty match {
case semType: SemType =>
globalNames(name) = semType
case funcType: FuncType =>
globalFuncs(name) = funcType
}
globalNumbering(name.v) = uid + 1
name.guid = nextGuid()
current(key) = ScopeValue(name, public)
Chain.empty
}
}
private def get(name: String, identType: IdentType): Option[Ident] =
/** Attempt to add a new identifier as an alias to another to the existing scope.
*
* @param alias
* The (new) alias identifier.
* @param orig
* The (existing) original identifier.
*
* @return
* An error, if one occurred.
*/
def addAlias(alias: Ident, orig: ScopeValue, public: Boolean = false): Chain[Error] = {
verifyMutable()
val path = alias.pos.file.getCanonicalPath
val identType = alias.ty match {
case _: SemType => IdentType.Var
case _: FuncType => IdentType.Func
}
val key = ScopeKey(path, alias.v, identType)
current.get(key) match {
case Some(ScopeValue(Ident(_, id, _), _)) =>
alias.guid = id
Chain.one(Error.DuplicateDeclaration(alias))
case None =>
alias.guid = nextGuid()
current(key) = ScopeValue(orig.id, public)
Chain.empty
}
}
def get(path: String, name: String, identType: IdentType): Option[ScopeValue] =
// Unfortunately map defaults only work with `.apply()`, which throws an error when the key is not found.
// Neither is there a way to check whether a default exists, so we have to use a try-catch.
try {
Some(current.withDefault(parent)((name, identType)))
Some(current.withDefault(parent)(ScopeKey(path, name, identType)))
} catch {
case _: NoSuchElementException => None
}
def getVar(name: String): Option[Ident] = get(name, IdentType.Var)
def getFunc(name: String): Option[Ident] = get(name, IdentType.Func)
def getVar(name: Ident): Option[Ident] = get(name.pos.file.getCanonicalPath, name.v, IdentType.Var).map(_.id)
def getFunc(name: Ident): Option[Ident] = get(name.pos.file.getCanonicalPath, name.v, IdentType.Func).map(_.id)
}
/** Check scoping of all variables and functions in the program. Also generate semantic types for
* all identifiers.
*
* @param prog
* AST of the program
* @param errors
* List of errors to append to
* @return
* Map of all (renamed) identifies to their semantic types
*/
def rename(prog: Program)(using
errors: mutable.Builder[Error, List[Error]]
): (Map[Ident, SemType], Map[Ident, FuncType]) = {
given globalNames: mutable.Map[Ident, SemType] = mutable.Map.empty
given globalFuncs: mutable.Map[Ident, FuncType] = mutable.Map.empty
given globalNumbering: mutable.Map[String, Int] = mutable.Map.empty
val scope = Scope(mutable.Map.empty, Map.empty)
private def prepareGlobalScope(partialProg: PartialProgram)(using scope: Scope): IO[(FuncDecl, Chain[FuncDecl], Chain[Error])] = {
def readImportFile(file: File): IO[String] =
IO.blocking(os.read(os.Path(file.getCanonicalPath)))
def prepareImport(contents: String, file: File)(using scope: Scope): IO[(Chain[FuncDecl], Chain[Error])] = {
parser.parse(contents) match {
case Failure(msg) =>
IO.pure(Chain.empty, Chain.one(Error.SyntaxError(file, msg)))
case Success(fn) =>
val partialProg = fn(file)
for {
(main, chunks, errors) <- prepareGlobalScope(partialProg)
} yield (main +: chunks, errors)
}
}
val PartialProgram(imports, prog) = partialProg
// First prepare this file's functions...
val Program(funcs, main) = prog
funcs
// First add all function declarations to the scope
.map { case FuncDecl(retType, name, params, body) =>
val (funcChunks, funcErrors) = funcs.foldLeft((Chain.empty[FuncDecl], Chain.empty[Error])) {
case ((chunks, errors), func@FuncDecl(retType, name, params, body)) =>
val paramTypes = params.map { param =>
val paramType = SemType(param.paramType)
param.name.ty = paramType
paramType
}
scope.add(FuncType(SemType(retType), paramTypes), name)
(params zip paramTypes, body)
}
// Only then rename the function bodies
// (functions can call one-another regardless of order of declaration)
.foreach { case (params, body) =>
val functionScope = scope.subscope
params.foreach { case (param, paramType) =>
functionScope.add(paramType, param.name)
}
body.toList.foreach(rename(functionScope.subscope)) // body can shadow function params
}
main.toList.foreach(rename(scope))
(globalNames.toMap, globalFuncs.toMap)
name.ty = FuncType(SemType(retType), paramTypes)
(chunks :+ func, errors ++ scope.add(name, public = true))
}
// ...and main body.
val mainBodyIdent = Ident(MAIN, ty = FuncType(?, Nil))(prog.pos)
val mainBodyErrors = scope.add(mainBodyIdent, public = false)
val mainBodyChunk = FuncDecl(IntType()(prog.pos), mainBodyIdent, Nil, main)(prog.pos)
// Now handle imports
val file = prog.pos.file
val preparedImports = imports.foldLeftM[IO, (Chain[FuncDecl], Chain[Error])]((Chain.empty[FuncDecl], Chain.empty[Error])) {
case ((chunks, errors), Import(name, funcs)) =>
val importFile = File(file.getParent, name.v)
for {
contents <- readImportFile(importFile)
(importChunks, importErrors) <- prepareImport(contents, importFile)
importAliasErrors = funcs.foldMap { case ImportedFunc(srcName, aliasName) =>
scope.get(importFile.getCanonicalPath, srcName.v, IdentType.Func) match {
case Some(src) if src.public => scope.addAlias(aliasName, src)
case _ => Chain.one(Error.UndefinedFunction(srcName))
}
}
} yield (chunks ++ importChunks, errors ++ importErrors)
}
for {
(importChunks, importErrors) <- preparedImports
allChunks = importChunks ++ funcChunks
allErrors = importErrors ++ funcErrors ++ mainBodyErrors
} yield (mainBodyChunk, allChunks, allErrors)
}
/** Check scoping of all variables and flatten a program. Also generates semantic types and parses
* any imported files.
*
* @param partialProg
* AST of the program
* @return
* (flattenedProg, errors)
*/
def rename(partialProg: PartialProgram): IO[(Program, Chain[Error])] = {
given scope: Scope = Scope(mutable.Map.empty, Map.empty)
for {
(main, chunks, errors) <- prepareGlobalScope(partialProg)
toRename = (main +: chunks).toList
res = (toRename zip scope.subscopes(toRename.size)).parTraverse { case (func@FuncDecl(retType, name, params, body), subscope) =>
val paramErrors = params.foldMap { param => subscope.add(param.name) }
val bodyErrors = subscope.withSubscope { s => body.foldMap(rename(s)) }
paramErrors ++ bodyErrors
}
} yield (partialProg.self, errors)
}
// /** Check scoping of all variables and functions in the program. Also generate semantic types for
// * all identifiers.
// *
// * @param prog
// * AST of the program
// * @param errors
// * List of errors to append to
// * @return
// * Map of all (renamed) identifies to their semantic types
// */
// def rename(prog: Program)(using
// errors: mutable.Builder[Error, List[Error]]
// ): (Map[Ident, SemType], Map[Ident, FuncType]) = {
// given globalNames: mutable.Map[Ident, SemType] = mutable.Map.empty
// given globalFuncs: mutable.Map[Ident, FuncType] = mutable.Map.empty
// given globalNumbering: mutable.Map[String, Int] = mutable.Map.empty
// val scope = Scope(mutable.Map.empty, Map.empty)
// val Program(funcs, main) = prog
// funcs
// // First add all function declarations to the scope
// .map { case FuncDecl(retType, name, params, body) =>
// val paramTypes = params.map { param =>
// val paramType = SemType(param.paramType)
// paramType
// }
// scope.add(FuncType(SemType(retType), paramTypes), name)
// (params zip paramTypes, body)
// }
// // Only then rename the function bodies
// // (functions can call one-another regardless of order of declaration)
// .foreach { case (params, body) =>
// val functionScope = scope.subscope
// params.foreach { case (param, paramType) =>
// functionScope.add(paramType, param.name)
// }
// body.toList.foreach(rename(functionScope.subscope)) // body can shadow function params
// }
// main.toList.foreach(rename(scope))
// (globalNames.toMap, globalFuncs.toMap)
// }
/** Check scoping of all identifies in a given AST node.
*
* @param scope
* The current scope and flattened parent scope.
* @param node
* The AST node.
* @param globalNames
* The global map of identifiers to semantic types - renamed identifiers will be added to this
* map.
* @param globalNumbering
* The global map of identifier names to the number of times they have been declared - used and
* updated during identifier renaming.
* @param errors
*/
private def rename(scope: Scope)(
node: Ident | Stmt | LValue | RValue | Expr
)(using
globalNames: mutable.Map[Ident, SemType],
globalFuncs: mutable.Map[Ident, FuncType],
globalNumbering: mutable.Map[String, Int],
errors: mutable.Builder[Error, List[Error]]
): Unit = node match {
// These cases are more interesting because the involve making subscopes
// or modifying the current scope.
case VarDecl(synType, name, value) => {
// Order matters here. Variable isn't declared until after the value is evaluated.
rename(scope)(value)
// Attempt to add the new variable to the current scope.
scope.add(SemType(synType), name)
}
case If(cond, thenStmt, elseStmt) => {
rename(scope)(cond)
// then and else both have their own scopes
thenStmt.toList.foreach(rename(scope.subscope))
elseStmt.toList.foreach(rename(scope.subscope))
}
case While(cond, body) => {
rename(scope)(cond)
// while bodies have their own scopes
body.toList.foreach(rename(scope.subscope))
}
// begin-end blocks have their own scopes
case Block(body) => body.toList.foreach(rename(scope.subscope))
private def rename(scope: Scope)(node: Ident | Stmt | LValue | RValue | Expr): Chain[Error] =
node match {
// These cases are more interes/globting because the involve making subscopes
// or modifying the current scope.
case VarDecl(synType, name, value) => {
// Order matters here. Variable isn't declared until after the value is evaluated.
val errors = rename(scope)(value)
// Attempt to add the new variable to the current scope.
name.ty = SemType(synType)
errors ++ scope.add(name)
}
case If(cond, thenStmt, elseStmt) => {
val condErrors = rename(scope)(cond)
// then and else both have their own scopes
val thenErrors = scope.withSubscope(s => thenStmt.foldMap(rename(s)))
val elseErrors = scope.withSubscope(s => elseStmt.foldMap(rename(s)))
condErrors ++ thenErrors ++ elseErrors
}
case While(cond, body) => {
val condErrors = rename(scope)(cond)
// while bodies have their own scopes
val bodyErrors = scope.withSubscope(s => body.foldMap(rename(s)))
condErrors ++ bodyErrors
}
// begin-end blocks have their own scopes
case Block(body) => scope.withSubscope(s => body.foldMap(rename(s)))
// These cases are simpler, mostly just recursive calls to rename()
case Assign(lhs, value) => {
// Variables may be reassigned with their value in the rhs, so order doesn't matter here.
rename(scope)(lhs)
rename(scope)(value)
}
case Read(lhs) => rename(scope)(lhs)
case Free(expr) => rename(scope)(expr)
case Return(expr) => rename(scope)(expr)
case Exit(expr) => rename(scope)(expr)
case Print(expr, _) => rename(scope)(expr)
case NewPair(fst, snd) => {
rename(scope)(fst)
rename(scope)(snd)
}
case Call(name, args) => {
scope.getFunc(name.v) match {
case Some(Ident(_, uid)) => name.uid = uid
case None =>
errors += Error.UndefinedFunction(name)
scope.add(FuncType(?, args.map(_ => ?)), name)
// These cases are simpler, mostly just recursive calls to rename()
case Assign(lhs, value) => {
// Variables may be reassigned with their value in the rhs, so order doesn't matter here.
rename(scope)(lhs) ++ rename(scope)(value)
}
args.foreach(rename(scope))
}
case Fst(elem) => rename(scope)(elem)
case Snd(elem) => rename(scope)(elem)
case ArrayLiter(elems) => elems.foreach(rename(scope))
case ArrayElem(name, indices) => {
rename(scope)(name)
indices.toList.foreach(rename(scope))
}
case Parens(expr) => rename(scope)(expr)
case op: UnaryOp => rename(scope)(op.x)
case op: BinaryOp => {
rename(scope)(op.x)
rename(scope)(op.y)
}
// Default to variables. Only `call` uses IdentType.Func.
case id: Ident => {
scope.getVar(id.v) match {
case Some(Ident(_, uid)) => id.uid = uid
case None =>
errors += Error.UndeclaredVariable(id)
scope.add(?, id)
case Read(lhs) => rename(scope)(lhs)
case Free(expr) => rename(scope)(expr)
case Return(expr) => rename(scope)(expr)
case Exit(expr) => rename(scope)(expr)
case Print(expr, _) => rename(scope)(expr)
case NewPair(fst, snd) => {
rename(scope)(fst) ++ rename(scope)(snd)
}
case Call(name, args) => {
val nameErrors = scope.getFunc(name) match {
case Some(Ident(_, guid, ty)) =>
name.ty = ty
name.guid = guid
Chain.empty
case None =>
name.ty = FuncType(?, args.map(_ => ?))
scope.add(name)
Chain.one(Error.UndefinedFunction(name))
}
val argsErrors = args.foldMap(rename(scope))
nameErrors ++ argsErrors
}
case Fst(elem) => rename(scope)(elem)
case Snd(elem) => rename(scope)(elem)
case ArrayLiter(elems) => elems.foldMap(rename(scope))
case ArrayElem(name, indices) => {
val nameErrors = rename(scope)(name)
val indicesErrors = indices.foldMap(rename(scope))
nameErrors ++ indicesErrors
}
case Parens(expr) => rename(scope)(expr)
case op: UnaryOp => rename(scope)(op.x)
case op: BinaryOp => {
rename(scope)(op.x) ++ rename(scope)(op.y)
}
// Default to variables. Only `call` uses IdentType.Func.
case id: Ident => {
scope.getVar(id) match {
case Some(Ident(_, guid, ty)) =>
id.ty = ty
id.guid = guid
Chain.empty
case None =>
id.ty = ?
scope.add(id)
Chain.one(Error.UndeclaredVariable(id))
}
}
// These literals cannot contain identifies, exit immediately.
case IntLiter(_) | BoolLiter(_) | CharLiter(_) | StrLiter(_) | PairLiter() | Skip() => Chain.empty
}
// These literals cannot contain identifies, exit immediately.
case IntLiter(_) | BoolLiter(_) | CharLiter(_) | StrLiter(_) | PairLiter() | Skip() => ()
}
}

View File

@@ -3,7 +3,9 @@ package wacc
object types {
import ast._
sealed trait SemType {
sealed trait RenamerType
sealed trait SemType extends RenamerType {
override def toString(): String = this match {
case KnownType.Int => "int"
case KnownType.Bool => "bool"
@@ -41,5 +43,5 @@ object types {
}
}
case class FuncType(returnType: SemType, params: List[SemType])
case class FuncType(returnType: SemType, params: List[SemType]) extends RenamerType
}