Generics (#7)

Co-authored-by: venko <alexmflores@gmail.com>
Reviewed-on: cosmonaut/wraith-lang#7
Co-authored-by: cosmonaut <evan@moonside.games>
Co-committed-by: cosmonaut <evan@moonside.games>
main
cosmonaut 2021-06-01 19:58:46 +00:00
parent 473b706ad9
commit d9b01515eb
15 changed files with 1793 additions and 741 deletions

View File

@ -41,12 +41,14 @@ add_executable(
# Source # Source
src/ast.h src/ast.h
src/codegen.h src/codegen.h
src/identcheck.h
src/parser.h src/parser.h
src/validation.h
src/util.h
src/ast.c src/ast.c
src/codegen.c src/codegen.c
src/identcheck.c
src/parser.c src/parser.c
src/validation.c
src/util.c
src/main.c src/main.c
# Generated code # Generated code
${BISON_Parser_OUTPUTS} ${BISON_Parser_OUTPUTS}

13
access.w Normal file
View File

@ -0,0 +1,13 @@
struct G {
Foo(t: bool): bool {
return t;
}
}
struct Program {
static main(): int {
g: G = alloc G;
g.Foo(true);
return 0;
}
}

View File

@ -307,14 +307,38 @@ Body : LEFT_BRACE Statements RIGHT_BRACE
$$ = $2; $$ = $2;
} }
FunctionSignature : Identifier LEFT_PAREN SignatureArguments RIGHT_PAREN COLON Type GenericArgument : Identifier
{ {
$$ = MakeFunctionSignatureNode($1, $6, $3, MakeFunctionModifiersNode(NULL, 0)); $$ = MakeGenericArgumentNode($1, NULL);
} }
| STATIC Identifier LEFT_PAREN SignatureArguments RIGHT_PAREN COLON Type
GenericArguments : GenericArgument
{
$$ = StartGenericArgumentsNode($1);
}
| GenericArguments COMMA GenericArgument
{
$$ = AddGenericArgument($1, $3);
}
GenericArgumentsClause : LESS_THAN GenericArguments GREATER_THAN
{
$$ = $2;
}
|
{
$$ = MakeEmptyGenericArgumentsNode();
}
FunctionSignature : Identifier GenericArgumentsClause LEFT_PAREN SignatureArguments RIGHT_PAREN COLON Type
{
$$ = MakeFunctionSignatureNode($1, $7, $4, MakeFunctionModifiersNode(NULL, 0), $2);
}
| STATIC Identifier GenericArgumentsClause LEFT_PAREN SignatureArguments RIGHT_PAREN COLON Type
{ {
Node *modifier = MakeStaticNode(); Node *modifier = MakeStaticNode();
$$ = MakeFunctionSignatureNode($2, $7, $4, MakeFunctionModifiersNode(&modifier, 1)); $$ = MakeFunctionSignatureNode($2, $8, $5, MakeFunctionModifiersNode(&modifier, 1), $3);
} }
FunctionDeclaration : FunctionSignature Body FunctionDeclaration : FunctionSignature Body

18
generic.w Normal file
View File

@ -0,0 +1,18 @@
struct Foo {
static Func2<U>(u: U) : U {
return u;
}
static Func<T>(t: T): T {
foo: T = t;
return Foo.Func2(foo);
}
}
struct Program {
static Main(): int {
x: int = 4;
y: int = Foo.Func(x);
return x;
}
}

15
ordering.w Normal file
View File

@ -0,0 +1,15 @@
struct Foo {
static Func(): void {
Func2();
}
static Func2(): void {
Func();
}
}
struct Program {
static main(): int {
return 0;
}
}

455
src/ast.c
View File

@ -39,6 +39,12 @@ const char *SyntaxKindString(SyntaxKind syntaxKind)
return "FunctionSignature"; return "FunctionSignature";
case FunctionSignatureArguments: case FunctionSignatureArguments:
return "FunctionSignatureArguments"; return "FunctionSignatureArguments";
case GenericArgument:
return "GenericArgument";
case GenericArguments:
return "GenericArguments";
case GenericTypeNode:
return "GenericTypeNode";
case Identifier: case Identifier:
return "Identifier"; return "Identifier";
case IfStatement: case IfStatement:
@ -271,7 +277,8 @@ Node *MakeFunctionSignatureNode(
Node *identifierNode, Node *identifierNode,
Node *typeNode, Node *typeNode,
Node *arguments, Node *arguments,
Node *modifiersNode) Node *modifiersNode,
Node *genericArgumentsNode)
{ {
Node *node = (Node *)malloc(sizeof(Node)); Node *node = (Node *)malloc(sizeof(Node));
node->syntaxKind = FunctionSignature; node->syntaxKind = FunctionSignature;
@ -279,6 +286,7 @@ Node *MakeFunctionSignatureNode(
node->functionSignature.type = typeNode; node->functionSignature.type = typeNode;
node->functionSignature.arguments = arguments; node->functionSignature.arguments = arguments;
node->functionSignature.modifiers = modifiersNode; node->functionSignature.modifiers = modifiersNode;
node->functionSignature.genericArguments = genericArgumentsNode;
return node; return node;
} }
@ -359,6 +367,54 @@ Node *MakeEmptyFunctionArgumentSequenceNode()
return node; return node;
} }
Node *MakeGenericArgumentNode(Node *identifierNode, Node *constraintNode)
{
Node *node = (Node *)malloc(sizeof(Node));
node->syntaxKind = GenericArgument;
node->genericArgument.identifier = identifierNode;
node->genericArgument.constraint = constraintNode;
return node;
}
Node *StartGenericArgumentsNode(Node *genericArgumentNode)
{
Node *node = (Node *)malloc(sizeof(Node));
node->syntaxKind = GenericArguments;
node->genericArguments.arguments = (Node **)malloc(sizeof(Node *));
node->genericArguments.arguments[0] = genericArgumentNode;
node->genericArguments.count = 1;
return node;
}
Node *AddGenericArgument(Node *genericArgumentsNode, Node *genericArgumentNode)
{
genericArgumentsNode->genericArguments.arguments = (Node **)realloc(
genericArgumentsNode->genericArguments.arguments,
sizeof(Node *) * (genericArgumentsNode->genericArguments.count + 1));
genericArgumentsNode->genericArguments
.arguments[genericArgumentsNode->genericArguments.count] =
genericArgumentNode;
genericArgumentsNode->genericArguments.count += 1;
return genericArgumentsNode;
}
Node *MakeEmptyGenericArgumentsNode()
{
Node *node = (Node *)malloc(sizeof(Node));
node->syntaxKind = GenericArguments;
node->genericArguments.arguments = NULL;
node->genericArguments.count = 0;
return node;
}
Node *MakeGenericTypeNode(char *name)
{
Node *node = (Node *)malloc(sizeof(Node));
node->syntaxKind = GenericTypeNode;
node->genericType.name = strdup(name);
return node;
}
Node *MakeFunctionCallExpressionNode( Node *MakeFunctionCallExpressionNode(
Node *identifierNode, Node *identifierNode,
Node *argumentSequenceNode) Node *argumentSequenceNode)
@ -557,6 +613,7 @@ void PrintNode(Node *node, uint32_t tabCount)
case FunctionSignature: case FunctionSignature:
printf("\n"); printf("\n");
PrintNode(node->functionSignature.identifier, tabCount + 1); PrintNode(node->functionSignature.identifier, tabCount + 1);
PrintNode(node->functionSignature.genericArguments, tabCount + 1);
PrintNode(node->functionSignature.arguments, tabCount + 1); PrintNode(node->functionSignature.arguments, tabCount + 1);
PrintNode(node->functionSignature.type, tabCount + 1); PrintNode(node->functionSignature.type, tabCount + 1);
PrintNode(node->functionSignature.modifiers, tabCount + 1); PrintNode(node->functionSignature.modifiers, tabCount + 1);
@ -572,6 +629,25 @@ void PrintNode(Node *node, uint32_t tabCount)
} }
return; return;
case GenericArgument:
printf("\n");
PrintNode(node->genericArgument.identifier, tabCount + 1);
/* Constraint nodes are not implemented. */
/* PrintNode(node->genericArgument.constraint, tabCount + 1); */
return;
case GenericArguments:
printf("\n");
for (i = 0; i < node->genericArguments.count; i += 1)
{
PrintNode(node->genericArguments.arguments[i], tabCount + 1);
}
return;
case GenericTypeNode:
printf("%s\n", node->genericType.name);
return;
case Identifier: case Identifier:
if (node->typeTag == NULL) if (node->typeTag == NULL)
{ {
@ -630,7 +706,7 @@ void PrintNode(Node *node, uint32_t tabCount)
return; return;
case StringLiteral: case StringLiteral:
printf("%s", node->stringLiteral.string); printf("%s\n", node->stringLiteral.string);
return; return;
case StructDeclaration: case StructDeclaration:
@ -651,6 +727,173 @@ void PrintNode(Node *node, uint32_t tabCount)
} }
} }
void Recurse(Node *node, void (*func)(Node *))
{
uint32_t i;
switch (node->syntaxKind)
{
case AccessExpression:
func(node->accessExpression.accessee);
func(node->accessExpression.accessor);
return;
case AllocExpression:
func(node->allocExpression.type);
return;
case Assignment:
func(node->assignmentStatement.left);
func(node->assignmentStatement.right);
return;
case BinaryExpression:
func(node->binaryExpression.left);
func(node->binaryExpression.right);
return;
case Comment:
return;
case CustomTypeNode:
return;
case Declaration:
func(node->declaration.type);
func(node->declaration.identifier);
return;
case DeclarationSequence:
for (i = 0; i < node->declarationSequence.count; i += 1)
{
func(node->declarationSequence.sequence[i]);
}
return;
case ForLoop:
func(node->forLoop.declaration);
func(node->forLoop.startNumber);
func(node->forLoop.endNumber);
func(node->forLoop.statementSequence);
return;
case FunctionArgumentSequence:
for (i = 0; i < node->functionArgumentSequence.count; i += 1)
{
func(node->functionArgumentSequence.sequence[i]);
}
return;
case FunctionCallExpression:
func(node->functionCallExpression.identifier);
func(node->functionCallExpression.argumentSequence);
return;
case FunctionDeclaration:
func(node->functionDeclaration.functionSignature);
func(node->functionDeclaration.functionBody);
return;
case FunctionModifiers:
for (i = 0; i < node->functionModifiers.count; i += 1)
{
func(node->functionModifiers.sequence[i]);
}
return;
case FunctionSignature:
func(node->functionSignature.identifier);
func(node->functionSignature.type);
func(node->functionSignature.arguments);
func(node->functionSignature.modifiers);
func(node->functionSignature.genericArguments);
return;
case FunctionSignatureArguments:
for (i = 0; i < node->functionSignatureArguments.count; i += 1)
{
func(node->functionSignatureArguments.sequence[i]);
}
return;
case GenericArgument:
func(node->genericArgument.identifier);
func(node->genericArgument.constraint);
return;
case GenericArguments:
for (i = 0; i < node->genericArguments.count; i += 1)
{
func(node->genericArguments.arguments[i]);
}
return;
case GenericTypeNode:
return;
case Identifier:
return;
case IfStatement:
func(node->ifStatement.expression);
func(node->ifStatement.statementSequence);
return;
case IfElseStatement:
func(node->ifElseStatement.ifStatement);
func(node->ifElseStatement.elseStatement);
return;
case Number:
return;
case PrimitiveTypeNode:
return;
case ReferenceTypeNode:
func(node->referenceType.type);
return;
case Return:
func(node->returnStatement.expression);
return;
case ReturnVoid:
return;
case StatementSequence:
for (i = 0; i < node->statementSequence.count; i += 1)
{
func(node->statementSequence.sequence[i]);
}
return;
case StaticModifier:
return;
case StringLiteral:
return;
case StructDeclaration:
func(node->structDeclaration.identifier);
func(node->structDeclaration.declarationSequence);
return;
case Type:
return;
case UnaryExpression:
func(node->unaryExpression.child);
return;
default:
fprintf(
stderr,
"wraith: Unhandled SyntaxKind %s in recurse function.\n",
SyntaxKindString(node->syntaxKind));
return;
}
}
TypeTag *MakeTypeTag(Node *node) TypeTag *MakeTypeTag(Node *node)
{ {
if (node == NULL) if (node == NULL)
@ -698,6 +941,21 @@ TypeTag *MakeTypeTag(Node *node)
->functionSignature.type); ->functionSignature.type);
break; break;
case AllocExpression:
tag = MakeTypeTag(node->allocExpression.type);
break;
case GenericArgument:
tag->type = Generic;
tag->value.genericType =
strdup(node->genericArgument.identifier->identifier.name);
break;
case GenericTypeNode:
tag->type = Generic;
tag->value.genericType = strdup(node->genericType.name);
break;
default: default:
fprintf( fprintf(
stderr, stderr,
@ -706,6 +964,7 @@ TypeTag *MakeTypeTag(Node *node)
SyntaxKindString(node->syntaxKind)); SyntaxKindString(node->syntaxKind));
return NULL; return NULL;
} }
return tag; return tag;
} }
@ -729,11 +988,199 @@ char *TypeTagToString(TypeTag *tag)
{ {
char *inner = TypeTagToString(tag->value.referenceType); char *inner = TypeTagToString(tag->value.referenceType);
size_t innerStrLen = strlen(inner); size_t innerStrLen = strlen(inner);
char *result = malloc(sizeof(char) * (innerStrLen + 5)); char *result = malloc(sizeof(char) * (innerStrLen + 6));
sprintf(result, "Ref<%s>", inner); sprintf(result, "Ref<%s>", inner);
return result; return result;
} }
case Custom: case Custom:
return tag->value.customType; {
char *result =
malloc(sizeof(char) * (strlen(tag->value.customType) + 9));
sprintf(result, "Custom<%s>", tag->value.customType);
return result;
}
case Generic:
{
char *result =
malloc(sizeof(char) * (strlen(tag->value.genericType) + 10));
sprintf(result, "Generic<%s>", tag->value.genericType);
return result;
}
}
}
void LinkParentPointers(Node *node, Node *prev)
{
if (node == NULL)
return;
node->parent = prev;
uint32_t i;
switch (node->syntaxKind)
{
case AccessExpression:
LinkParentPointers(node->accessExpression.accessee, node);
LinkParentPointers(node->accessExpression.accessor, node);
return;
case AllocExpression:
LinkParentPointers(node->allocExpression.type, node);
return;
case Assignment:
LinkParentPointers(node->assignmentStatement.left, node);
LinkParentPointers(node->assignmentStatement.right, node);
return;
case BinaryExpression:
LinkParentPointers(node->binaryExpression.left, node);
LinkParentPointers(node->binaryExpression.right, node);
return;
case Comment:
return;
case CustomTypeNode:
return;
case Declaration:
LinkParentPointers(node->declaration.type, node);
LinkParentPointers(node->declaration.identifier, node);
return;
case DeclarationSequence:
for (i = 0; i < node->declarationSequence.count; i += 1)
{
LinkParentPointers(node->declarationSequence.sequence[i], node);
}
return;
case ForLoop:
LinkParentPointers(node->forLoop.declaration, node);
LinkParentPointers(node->forLoop.startNumber, node);
LinkParentPointers(node->forLoop.endNumber, node);
LinkParentPointers(node->forLoop.statementSequence, node);
return;
case FunctionArgumentSequence:
for (i = 0; i < node->functionArgumentSequence.count; i += 1)
{
LinkParentPointers(
node->functionArgumentSequence.sequence[i],
node);
}
return;
case FunctionCallExpression:
LinkParentPointers(node->functionCallExpression.identifier, node);
LinkParentPointers(node->functionCallExpression.argumentSequence, node);
return;
case FunctionDeclaration:
LinkParentPointers(node->functionDeclaration.functionSignature, node);
LinkParentPointers(node->functionDeclaration.functionBody, node);
return;
case FunctionModifiers:
for (i = 0; i < node->functionModifiers.count; i += 1)
{
LinkParentPointers(node->functionModifiers.sequence[i], node);
}
return;
case FunctionSignature:
LinkParentPointers(node->functionSignature.identifier, node);
LinkParentPointers(node->functionSignature.type, node);
LinkParentPointers(node->functionSignature.arguments, node);
LinkParentPointers(node->functionSignature.modifiers, node);
LinkParentPointers(node->functionSignature.genericArguments, node);
return;
case FunctionSignatureArguments:
for (i = 0; i < node->functionSignatureArguments.count; i += 1)
{
LinkParentPointers(
node->functionSignatureArguments.sequence[i],
node);
}
return;
case GenericArgument:
LinkParentPointers(node->genericArgument.identifier, node);
LinkParentPointers(node->genericArgument.constraint, node);
return;
case GenericArguments:
for (i = 0; i < node->genericArguments.count; i += 1)
{
LinkParentPointers(node->genericArguments.arguments[i], node);
}
return;
case GenericTypeNode:
return;
case Identifier:
return;
case IfStatement:
LinkParentPointers(node->ifStatement.expression, node);
LinkParentPointers(node->ifStatement.statementSequence, node);
return;
case IfElseStatement:
LinkParentPointers(node->ifElseStatement.ifStatement, node);
LinkParentPointers(node->ifElseStatement.elseStatement, node);
return;
case Number:
return;
case PrimitiveTypeNode:
return;
case ReferenceTypeNode:
LinkParentPointers(node->referenceType.type, node);
return;
case Return:
LinkParentPointers(node->returnStatement.expression, node);
return;
case ReturnVoid:
return;
case StatementSequence:
for (i = 0; i < node->statementSequence.count; i += 1)
{
LinkParentPointers(node->statementSequence.sequence[i], node);
}
return;
case StaticModifier:
return;
case StringLiteral:
return;
case StructDeclaration:
LinkParentPointers(node->structDeclaration.identifier, node);
LinkParentPointers(node->structDeclaration.declarationSequence, node);
return;
case Type:
return;
case UnaryExpression:
LinkParentPointers(node->unaryExpression.child, node);
return;
default:
fprintf(
stderr,
"wraith: Unhandled SyntaxKind %s in recurse function.\n",
SyntaxKindString(node->syntaxKind));
return;
} }
} }

View File

@ -1,7 +1,6 @@
#ifndef WRAITH_AST_H #ifndef WRAITH_AST_H
#define WRAITH_AST_H #define WRAITH_AST_H
#include "identcheck.h"
#include <stdint.h> #include <stdint.h>
/* -Wpedantic nameless union/struct silencing */ /* -Wpedantic nameless union/struct silencing */
@ -30,6 +29,9 @@ typedef enum
FunctionModifiers, FunctionModifiers,
FunctionSignature, FunctionSignature,
FunctionSignatureArguments, FunctionSignatureArguments,
GenericArgument,
GenericArguments,
GenericTypeNode,
Identifier, Identifier,
IfStatement, IfStatement,
IfElseStatement, IfElseStatement,
@ -87,7 +89,8 @@ typedef struct TypeTag
Unknown, Unknown,
Primitive, Primitive,
Reference, Reference,
Custom Custom,
Generic
} type; } type;
union union
{ {
@ -97,6 +100,8 @@ typedef struct TypeTag
struct TypeTag *referenceType; struct TypeTag *referenceType;
/* Valid when type = Custom. */ /* Valid when type = Custom. */
char *customType; char *customType;
/* Valid when type = Generic. */
char *genericType;
} value; } value;
} TypeTag; } TypeTag;
@ -192,6 +197,7 @@ struct Node
Node *type; Node *type;
Node *arguments; Node *arguments;
Node *modifiers; Node *modifiers;
Node *genericArguments;
} functionSignature; } functionSignature;
struct struct
@ -200,6 +206,23 @@ struct Node
uint32_t count; uint32_t count;
} functionSignatureArguments; } functionSignatureArguments;
struct
{
Node *identifier;
Node *constraint;
} genericArgument;
struct
{
Node **arguments;
uint32_t count;
} genericArguments;
struct
{
char *name;
} genericType;
struct struct
{ {
char *name; char *name;
@ -276,7 +299,6 @@ struct Node
} unaryExpression; } unaryExpression;
}; };
TypeTag *typeTag; TypeTag *typeTag;
IdNode *idLink;
}; };
const char *SyntaxKindString(SyntaxKind syntaxKind); const char *SyntaxKindString(SyntaxKind syntaxKind);
@ -306,10 +328,16 @@ Node *MakeFunctionSignatureNode(
Node *identifierNode, Node *identifierNode,
Node *typeNode, Node *typeNode,
Node *argumentsNode, Node *argumentsNode,
Node *modifiersNode); Node *modifiersNode,
Node *genericArgumentsNode);
Node *MakeFunctionDeclarationNode( Node *MakeFunctionDeclarationNode(
Node *functionSignatureNode, Node *functionSignatureNode,
Node *functionBodyNode); Node *functionBodyNode);
Node *MakeGenericArgumentNode(Node *identifierNode, Node *constraintNode);
Node *MakeEmptyGenericArgumentsNode();
Node *StartGenericArgumentsNode(Node *genericArgumentNode);
Node *AddGenericArgument(Node *genericArgumentsNode, Node *genericArgumentNode);
Node *MakeGenericTypeNode(char *name);
Node *MakeStructDeclarationNode( Node *MakeStructDeclarationNode(
Node *identifierNode, Node *identifierNode,
Node *declarationSequenceNode); Node *declarationSequenceNode);
@ -337,7 +365,18 @@ Node *MakeForLoopNode(
void PrintNode(Node *node, uint32_t tabCount); void PrintNode(Node *node, uint32_t tabCount);
const char *SyntaxKindString(SyntaxKind syntaxKind); const char *SyntaxKindString(SyntaxKind syntaxKind);
/* Helper function for applying a void function generically over the children of
* an AST node. Used for functions that need to traverse the entire tree but
* only perform operations on a subset of node types. Such functions can match
* the syntaxKinds relevant to their purpose and invoke this function in all
* other cases. */
void Recurse(Node *node, void (*func)(Node *));
void LinkParentPointers(Node *node, Node *prev);
TypeTag *MakeTypeTag(Node *node); TypeTag *MakeTypeTag(Node *node);
char *TypeTagToString(TypeTag *tag); char *TypeTagToString(TypeTag *tag);
Node *LookupIdNode(Node *current, Node *prev, char *target);
#endif /* WRAITH_AST_H */ #endif /* WRAITH_AST_H */

File diff suppressed because it is too large Load Diff

View File

@ -1,485 +0,0 @@
#include <stdbool.h>
#include <stdint.h>
#include <stdio.h>
#include <stdlib.h>
#include <string.h>
#include "ast.h"
#include "identcheck.h"
IdNode *MakeIdNode(NodeType type, char *name, IdNode *parent)
{
IdNode *node = (IdNode *)malloc(sizeof(IdNode));
node->type = type;
node->name = strdup(name);
node->parent = parent;
node->childCount = 0;
node->childCapacity = 0;
node->children = NULL;
node->typeTag = NULL;
return node;
}
void AddChildToNode(IdNode *node, IdNode *child)
{
if (child == NULL)
return;
if (node->children == NULL)
{
node->childCapacity = 2;
node->children =
(IdNode **)malloc(sizeof(IdNode *) * node->childCapacity);
}
else if (node->childCount == node->childCapacity)
{
node->childCapacity *= 2;
node->children = (IdNode **)realloc(
node->children,
sizeof(IdNode *) * node->childCapacity);
}
node->children[node->childCount] = child;
node->childCount += 1;
}
IdNode *MakeIdTree(Node *astNode, IdNode *parent)
{
uint32_t i;
IdNode *mainNode;
switch (astNode->syntaxKind)
{
case AccessExpression:
AddChildToNode(
parent,
MakeIdTree(astNode->accessExpression.accessee, parent));
AddChildToNode(
parent,
MakeIdTree(astNode->accessExpression.accessor, parent));
return NULL;
case AllocExpression:
AddChildToNode(
parent,
MakeIdTree(astNode->allocExpression.type, parent));
return NULL;
case Assignment:
{
if (astNode->assignmentStatement.left->syntaxKind == Declaration)
{
return MakeIdTree(astNode->assignmentStatement.left, parent);
}
else
{
AddChildToNode(
parent,
MakeIdTree(astNode->assignmentStatement.left, parent));
AddChildToNode(
parent,
MakeIdTree(astNode->assignmentStatement.right, parent));
return NULL;
}
}
case BinaryExpression:
AddChildToNode(
parent,
MakeIdTree(astNode->binaryExpression.left, parent));
AddChildToNode(
parent,
MakeIdTree(astNode->binaryExpression.right, parent));
return NULL;
case Declaration:
{
Node *idNode = astNode->declaration.identifier;
mainNode = MakeIdNode(Variable, idNode->identifier.name, parent);
mainNode->typeTag = MakeTypeTag(astNode);
idNode->typeTag = mainNode->typeTag;
break;
}
case DeclarationSequence:
{
mainNode = MakeIdNode(UnorderedScope, "", parent);
for (i = 0; i < astNode->declarationSequence.count; i++)
{
AddChildToNode(
mainNode,
MakeIdTree(astNode->declarationSequence.sequence[i], mainNode));
}
break;
}
case ForLoop:
{
Node *loopDecl = astNode->forLoop.declaration;
Node *loopBody = astNode->forLoop.statementSequence;
mainNode = MakeIdNode(OrderedScope, "for-loop", parent);
AddChildToNode(mainNode, MakeIdTree(loopDecl, mainNode));
AddChildToNode(mainNode, MakeIdTree(loopBody, mainNode));
break;
}
case FunctionArgumentSequence:
for (i = 0; i < astNode->functionArgumentSequence.count; i++)
{
AddChildToNode(
parent,
MakeIdTree(
astNode->functionArgumentSequence.sequence[i],
parent));
}
return NULL;
case FunctionCallExpression:
AddChildToNode(
parent,
MakeIdTree(astNode->functionCallExpression.identifier, parent));
AddChildToNode(
parent,
MakeIdTree(
astNode->functionCallExpression.argumentSequence,
parent));
return NULL;
case FunctionDeclaration:
{
Node *sigNode = astNode->functionDeclaration.functionSignature;
Node *idNode = sigNode->functionSignature.identifier;
char *funcName = idNode->identifier.name;
mainNode = MakeIdNode(Function, funcName, parent);
mainNode->typeTag = MakeTypeTag(astNode);
idNode->typeTag = mainNode->typeTag;
MakeIdTree(sigNode->functionSignature.arguments, mainNode);
MakeIdTree(astNode->functionDeclaration.functionBody, mainNode);
break;
}
case FunctionSignatureArguments:
{
for (i = 0; i < astNode->functionSignatureArguments.count; i++)
{
Node *argNode = astNode->functionSignatureArguments.sequence[i];
AddChildToNode(parent, MakeIdTree(argNode, parent));
}
return NULL;
}
case Identifier:
{
char *name = astNode->identifier.name;
mainNode = MakeIdNode(Placeholder, name, parent);
IdNode *lookupNode = LookupId(mainNode, NULL, name);
if (lookupNode == NULL)
{
fprintf(stderr, "wraith: Could not find IdNode for id %s\n", name);
TypeTag *tag = (TypeTag *)malloc(sizeof(TypeTag));
tag->type = Unknown;
astNode->typeTag = tag;
}
else
{
astNode->typeTag = lookupNode->typeTag;
}
break;
}
case IfStatement:
{
Node *clause = astNode->ifStatement.expression;
Node *stmtSeq = astNode->ifStatement.statementSequence;
mainNode = MakeIdNode(OrderedScope, "if", parent);
MakeIdTree(clause, mainNode);
MakeIdTree(stmtSeq, mainNode);
break;
}
case IfElseStatement:
{
Node *ifNode = astNode->ifElseStatement.ifStatement;
Node *elseStmts = astNode->ifElseStatement.elseStatement;
mainNode = MakeIdNode(OrderedScope, "if-else", parent);
IdNode *ifBranch = MakeIdTree(ifNode, mainNode);
AddChildToNode(mainNode, ifBranch);
IdNode *elseScope = MakeIdNode(OrderedScope, "else", mainNode);
MakeIdTree(elseStmts, elseScope);
AddChildToNode(mainNode, elseScope);
break;
}
case ReferenceTypeNode:
AddChildToNode(parent, MakeIdTree(astNode->referenceType.type, parent));
return NULL;
case Return:
AddChildToNode(
parent,
MakeIdTree(astNode->returnStatement.expression, parent));
return NULL;
case StatementSequence:
{
for (i = 0; i < astNode->statementSequence.count; i++)
{
Node *argNode = astNode->statementSequence.sequence[i];
AddChildToNode(parent, MakeIdTree(argNode, parent));
}
return NULL;
}
case StructDeclaration:
{
Node *idNode = astNode->structDeclaration.identifier;
Node *declsNode = astNode->structDeclaration.declarationSequence;
mainNode = MakeIdNode(Struct, idNode->identifier.name, parent);
mainNode->typeTag = MakeTypeTag(astNode);
for (i = 0; i < declsNode->declarationSequence.count; i++)
{
Node *decl = declsNode->declarationSequence.sequence[i];
AddChildToNode(mainNode, MakeIdTree(decl, mainNode));
}
break;
}
case Type:
AddChildToNode(parent, MakeIdTree(astNode->type.typeNode, parent));
return NULL;
case UnaryExpression:
AddChildToNode(
parent,
MakeIdTree(astNode->unaryExpression.child, parent));
return NULL;
case Comment:
case CustomTypeNode:
case FunctionModifiers:
case FunctionSignature:
case Number:
case PrimitiveTypeNode:
case ReturnVoid:
case StaticModifier:
case StringLiteral:
return NULL;
}
astNode->idLink = mainNode;
return mainNode;
}
void PrintIdNode(IdNode *node)
{
if (node == NULL)
{
fprintf(
stderr,
"wraith: Attempted to call PrintIdNode with null value.\n");
return;
}
switch (node->type)
{
case Placeholder:
printf("Placeholder (%s)\n", node->name);
break;
case OrderedScope:
printf("OrderedScope (%s)\n", node->name);
break;
case UnorderedScope:
printf("UnorderedScope (%s)\n", node->name);
break;
case Struct:
printf("%s : %s\n", node->name, TypeTagToString(node->typeTag));
break;
case Function:
printf(
"%s : Function<%s>\n",
node->name,
TypeTagToString(node->typeTag));
break;
case Variable:
printf("%s : %s\n", node->name, TypeTagToString(node->typeTag));
break;
}
}
void PrintIdTree(IdNode *tree, uint32_t tabCount)
{
if (tree == NULL)
{
fprintf(
stderr,
"wraith: Attempted to call PrintIdTree on a null value.\n");
return;
}
uint32_t i;
for (i = 0; i < tabCount; i++)
{
printf("| ");
}
PrintIdNode(tree);
for (i = 0; i < tree->childCount; i++)
{
PrintIdTree(tree->children[i], tabCount + 1);
}
}
int PrintAncestors(IdNode *node)
{
if (node == NULL)
return -1;
int i;
int indent = 1;
indent += PrintAncestors(node->parent);
for (i = 0; i < indent; i++)
{
printf(" ");
}
PrintIdNode(node);
return indent;
}
IdNode *LookdownId(IdNode *root, NodeType targetType, char *targetName)
{
if (root == NULL)
{
fprintf(
stderr,
"wraith: Attempted to call LookdownId on a null value.\n");
return NULL;
}
IdNode *result = NULL;
IdNode **frontier = (IdNode **)malloc(sizeof(IdNode *));
frontier[0] = root;
uint32_t frontierCount = 1;
while (frontierCount > 0)
{
IdNode *current = frontier[0];
if (current->type == targetType &&
strcmp(current->name, targetName) == 0)
{
result = current;
break;
}
uint32_t i;
for (i = 1; i < frontierCount; i++)
{
frontier[i - 1] = frontier[i];
}
size_t newSize = frontierCount + current->childCount - 1;
if (frontierCount != newSize)
{
frontier = (IdNode **)realloc(frontier, sizeof(IdNode *) * newSize);
}
for (i = 0; i < current->childCount; i++)
{
frontier[frontierCount + i - 1] = current->children[i];
}
frontierCount = newSize;
}
free(frontier);
return result;
}
bool ScopeHasOrdering(IdNode *node)
{
switch (node->type)
{
case OrderedScope:
case Function:
case Variable: /* this is only technically true */
return true;
default:
return false;
}
}
IdNode *LookupId(IdNode *node, IdNode *prev, char *target)
{
if (node == NULL)
{
return NULL;
}
if (strcmp(node->name, target) == 0 && node->type != Placeholder)
{
return node;
}
/* If this is the start of our search, we should not attempt to look at
* child nodes. Only looking up the scope tree is valid at this point.
*
* This has the notable side-effect that this function will return NULL if
* you attempt to look up a struct's internals starting from the node
* representing the struct itself. This is because an IdNode corresponds to
* the location *where an identifier is first declared.* Thus, an identifier
* has no knowledge of identifiers declared "inside" of it.
*/
if (prev == NULL)
{
return LookupId(node->parent, node, target);
}
/* If the current node forms an ordered scope then we want to prevent
* ourselves from looking up identifiers declared after the scope we have
* just come from.
*/
uint32_t idxLimit;
if (ScopeHasOrdering(node))
{
uint32_t i;
for (i = 0, idxLimit = 0; i < node->childCount; i++, idxLimit++)
{
if (node->children[i] == prev)
{
break;
}
}
}
else
{
idxLimit = node->childCount;
}
uint32_t i;
for (i = 0; i < idxLimit; i++)
{
IdNode *child = node->children[i];
if (child == prev || child->type == Placeholder)
{
/* Do not inspect the node we just came from or placeholders. */
continue;
}
if (strcmp(child->name, target) == 0)
{
return child;
}
if (child->type == Struct)
{
uint32_t j;
for (j = 0; j < child->childCount; j++)
{
IdNode *grandchild = child->children[j];
if (strcmp(grandchild->name, target) == 0)
{
return grandchild;
}
}
}
}
return LookupId(node->parent, node, target);
}

View File

@ -1,49 +0,0 @@
/* Validates identifier usage in an AST. */
#ifndef WRAITH_IDENTCHECK_H
#define WRAITH_IDENTCHECK_H
#include <stdint.h>
#include "ast.h"
struct TypeTag;
struct Node;
typedef enum NodeType
{
Placeholder,
UnorderedScope,
OrderedScope,
Struct,
Function,
Variable
} NodeType;
typedef struct IdNode
{
NodeType type;
char *name;
struct TypeTag *typeTag;
struct IdNode *parent;
struct IdNode **children;
uint32_t childCount;
uint32_t childCapacity;
} IdNode;
typedef struct IdStatus
{
enum StatusCode
{
Valid,
} StatusCode;
} IdStatus;
IdNode *MakeIdTree(struct Node *astNode, IdNode *parent);
void PrintIdNode(IdNode *node);
void PrintIdTree(IdNode *tree, uint32_t tabCount);
int PrintAncestors(IdNode *node);
IdNode *LookdownId(IdNode *root, NodeType targetType, char *targetName);
IdNode *LookupId(IdNode *node, IdNode *prev, char *target);
#endif /* WRAITH_IDENTCHECK_H */

View File

@ -2,8 +2,8 @@
#include <stdlib.h> #include <stdlib.h>
#include "codegen.h" #include "codegen.h"
#include "identcheck.h"
#include "parser.h" #include "parser.h"
#include "validation.h"
int main(int argc, char *argv[]) int main(int argc, char *argv[])
{ {
@ -85,12 +85,15 @@ int main(int argc, char *argv[])
} }
else else
{ {
{ LinkParentPointers(rootNode, NULL);
IdNode *idTree = MakeIdTree(rootNode, NULL); /* FIXME: ValidateIdentifiers should return some sort of
PrintIdTree(idTree, /*tabCount=*/0); error status object. */
printf("\n"); ValidateIdentifiers(rootNode);
PrintNode(rootNode, /*tabCount=*/0); TagIdentifierTypes(rootNode);
} ConvertCustomsToGenerics(rootNode);
PrintNode(rootNode, 0);
printf("Beginning codegen.\n");
exitCode = Codegen(rootNode, optimizationLevel); exitCode = Codegen(rootNode, optimizationLevel);
} }
} }

View File

@ -1,7 +1,6 @@
#include "util.h" #include "util.h"
#include <stdlib.h> #include <stdlib.h>
#include <string.h>
char *strdup(const char *s) char *strdup(const char *s)
{ {
@ -15,3 +14,16 @@ char *strdup(const char *s)
memcpy(result, s, slen + 1); memcpy(result, s, slen + 1);
return result; return result;
} }
uint64_t str_hash(char *str)
{
uint64_t hash = 5381;
size_t c;
while ((c = *str++))
{
hash = ((hash << 5) + hash) + c; /* hash * 33 + c */
}
return hash;
}

View File

@ -1,8 +1,10 @@
#ifndef WRAITH_UTIL_H #ifndef WRAITH_UTIL_H
#define WRAITH_UTIL_H #define WRAITH_UTIL_H
#include <stdint.h>
#include <string.h> #include <string.h>
char *strdup(const char *s); char *strdup(const char *s);
uint64_t str_hash(char *str);
#endif /* WRAITH_UTIL_H */ #endif /* WRAITH_UTIL_H */

452
src/validation.c Normal file
View File

@ -0,0 +1,452 @@
#include "validation.h"
#include <stdbool.h>
#include <stdio.h>
#include <stdlib.h>
#include <string.h>
Node *GetIdFromStruct(Node *structDecl)
{
if (structDecl->syntaxKind != StructDeclaration)
{
fprintf(
stderr,
"wraith: Attempted to call GetIdFromStruct on node with kind: "
"%s.\n",
SyntaxKindString(structDecl->syntaxKind));
return NULL;
}
return structDecl->structDeclaration.identifier;
}
Node *GetIdFromFunction(Node *funcDecl)
{
if (funcDecl->syntaxKind != FunctionDeclaration)
{
fprintf(
stderr,
"wraith: Attempted to call GetIdFromFunction on node with kind: "
"%s.\n",
SyntaxKindString(funcDecl->syntaxKind));
return NULL;
}
Node *sig = funcDecl->functionDeclaration.functionSignature;
return sig->functionSignature.identifier;
}
Node *GetIdFromDeclaration(Node *decl)
{
if (decl->syntaxKind != Declaration)
{
fprintf(
stderr,
"wraith: Attempted to call GetIdFromDeclaration on node with kind: "
"%s.\n",
SyntaxKindString(decl->syntaxKind));
}
return decl->declaration.identifier;
}
bool AssignmentHasDeclaration(Node *assign)
{
return (
assign->syntaxKind == Assignment &&
assign->assignmentStatement.left->syntaxKind == Declaration);
}
Node *GetIdFromAssignment(Node *assign)
{
if (assign->syntaxKind != Assignment)
{
fprintf(
stderr,
"wraith: Attempted to call GetIdFromAssignment on node with kind: "
"%s.\n",
SyntaxKindString(assign->syntaxKind));
}
if (AssignmentHasDeclaration(assign))
{
return GetIdFromDeclaration(assign->assignmentStatement.left);
}
return NULL;
}
bool NodeMayHaveId(Node *node)
{
switch (node->syntaxKind)
{
case StructDeclaration:
case FunctionDeclaration:
case Declaration:
case Assignment:
return true;
default:
return false;
}
}
Node *TryGetId(Node *node)
{
switch (node->syntaxKind)
{
case Assignment:
return GetIdFromAssignment(node);
case Declaration:
return GetIdFromDeclaration(node);
case FunctionDeclaration:
return GetIdFromFunction(node);
case StructDeclaration:
return GetIdFromStruct(node);
default:
return NULL;
}
}
Node *LookupFunctionArgId(Node *funcDecl, char *target)
{
Node *args = funcDecl->functionDeclaration.functionSignature
->functionSignature.arguments;
uint32_t i;
for (i = 0; i < args->functionArgumentSequence.count; i += 1)
{
Node *arg = args->functionArgumentSequence.sequence[i];
if (arg->syntaxKind != Declaration)
{
fprintf(
stderr,
"wraith: Encountered %s node in function signature args "
"list.\n",
SyntaxKindString(arg->syntaxKind));
continue;
}
Node *argId = GetIdFromDeclaration(arg);
if (argId != NULL && strcmp(target, argId->identifier.name) == 0)
return argId;
}
return NULL;
}
Node *LookupStructInternalId(Node *structDecl, char *target)
{
Node *decls = structDecl->structDeclaration.declarationSequence;
uint32_t i;
for (i = 0; i < decls->declarationSequence.count; i += 1)
{
Node *match = TryGetId(decls->declarationSequence.sequence[i]);
if (match != NULL && strcmp(target, match->identifier.name) == 0)
return match;
}
return NULL;
}
Node *InspectNode(Node *node, char *target)
{
/* If this node may have an identifier declaration inside it, attempt to
* look up the identifier
* node itself, returning it if it matches the given target name. */
if (NodeMayHaveId(node))
{
Node *candidateId = TryGetId(node);
if (candidateId != NULL &&
strcmp(target, candidateId->identifier.name) == 0)
return candidateId;
}
/* If the candidate node was not the one we wanted, but the node node is a
* function declaration, it's possible that the identifier we want is one of
* the function's parameters rather than the function's name itself. */
if (node->syntaxKind == FunctionDeclaration)
{
Node *match = LookupFunctionArgId(node, target);
if (match != NULL)
return match;
}
/* Likewise if the node node is a struct declaration, inspect the struct's
* internals
* to see if a top-level definition is the one we're looking for. */
if (node->syntaxKind == StructDeclaration)
{
Node *match = LookupStructInternalId(node, target);
if (match != NULL)
return match;
}
return NULL;
}
/* FIXME: Handle staged lookups for AccessExpressions. */
/* FIXME: Similar to above, disallow inspection of struct internals outside of
* AccessExpressions. */
Node *LookupId(Node *current, Node *prev, char *target)
{
if (current == NULL)
return NULL;
Node *match;
/* First inspect the current node to see if it contains the target
* identifier. */
match = InspectNode(current, target);
if (match != NULL)
return match;
/* If this is the start of our search, we should not attempt to look at
* child nodes. Only looking up the AST is valid at this point.
*
* This has the notable side-effect that this function will return NULL if
* you attempt to look up a struct's internals starting from the node
* representing the struct itself. The same is true for functions. */
if (prev == NULL)
return LookupId(current->parent, current, target);
uint32_t i;
uint32_t idxLimit;
switch (current->syntaxKind)
{
case DeclarationSequence:
for (i = 0; i < current->declarationSequence.count; i += 1)
{
Node *decl = current->declarationSequence.sequence[i];
match = InspectNode(decl, target);
if (match != NULL)
return match;
}
break;
case StatementSequence:
idxLimit = current->statementSequence.count;
for (i = 0; i < current->statementSequence.count; i += 1)
{
if (current->statementSequence.sequence[i] == prev)
{
idxLimit = i;
break;
}
}
for (i = 0; i < idxLimit; i += 1)
{
Node *stmt = current->statementSequence.sequence[i];
if (stmt == prev)
break;
match = InspectNode(stmt, target);
if (match != NULL)
return match;
}
break;
}
return LookupId(current->parent, current, target);
}
/* FIXME: This function should be extended to handle multi-stage ID lookups for
* AccessExpression nodes. */
/* FIXME: Make this function return an error status object of some kind.
* A non-OK status should halt compilation. */
void ValidateIdentifiers(Node *node)
{
if (node == NULL)
return;
/* Skip over generic arguments. They contain Identifiers but are not
* actually identifiers, they declare types. */
if (node->syntaxKind == GenericArguments)
return;
if (node->syntaxKind != Identifier)
{
Recurse(node, *ValidateIdentifiers);
return;
}
char *name = node->identifier.name;
Node *decl = LookupId(node, NULL, name);
if (decl == NULL)
{
/* FIXME: Express this case as an error with AST information, see the
* FIXME comment above. */
fprintf(
stderr,
"wraith: Could not find definition of identifier %s.\n",
name);
}
}
/* FIXME: This function should be extended to handle multi-stage ID lookups for
* AccessExpression nodes. */
void TagIdentifierTypes(Node *node)
{
if (node == NULL)
return;
switch (node->syntaxKind)
{
case AllocExpression:
node->typeTag = MakeTypeTag(node);
break;
case Declaration:
node->declaration.identifier->typeTag = MakeTypeTag(node);
break;
case FunctionDeclaration:
node->functionDeclaration.functionSignature->functionSignature
.identifier->typeTag = MakeTypeTag(node);
break;
case StructDeclaration:
node->structDeclaration.identifier->typeTag = MakeTypeTag(node);
break;
case GenericArgument:
node->genericArgument.identifier->typeTag = MakeTypeTag(node);
break;
case Identifier:
{
if (node->typeTag != NULL)
return;
char *name = node->identifier.name;
Node *declaration = LookupId(node, NULL, name);
/* FIXME: Remove this case once ValidateIdentifiers returns error status
* info and halts compilation. See ValidateIdentifiers FIXME. */
if (declaration == NULL)
{
TypeTag *tag = (TypeTag *)malloc(sizeof(TypeTag));
tag->type = Unknown;
node->typeTag = tag;
}
else
{
node->typeTag = declaration->typeTag;
}
break;
}
}
Recurse(node, *TagIdentifierTypes);
}
Node *LookupType(Node *current, char *target)
{
if (current == NULL)
return NULL;
switch (current->syntaxKind)
{
/* If we've encountered a function declaration, check to see if it's generic
* and, if so, if one of its type parameters is the target. */
case FunctionDeclaration:
{
Node *typeArgs = current->functionDeclaration.functionSignature
->functionSignature.genericArguments;
uint32_t i;
for (i = 0; i < typeArgs->genericArguments.count; i += 1)
{
Node *arg = typeArgs->genericArguments.arguments[i];
Node *argId = arg->genericArgument.identifier;
char *argName = argId->identifier.name;
/* note: return the GenericArgument, not the Identifier, so that
* the caller can differentiate between generics and customs. */
if (strcmp(target, argName) == 0)
return arg;
}
return LookupType(current->parent, target);
}
case StructDeclaration:
{
Node *structId = GetIdFromStruct(current);
if (strcmp(target, structId->identifier.name) == 0)
return structId;
return LookupType(current->parent, target);
}
/* If we encounter a declaration sequence, search each of its children for
* struct definitions in case one of them is the target. */
case DeclarationSequence:
{
uint32_t i;
for (i = 0; i < current->declarationSequence.count; i += 1)
{
Node *decl = current->declarationSequence.sequence[i];
if (decl->syntaxKind == StructDeclaration)
{
Node *structId = GetIdFromStruct(decl);
if (strcmp(target, structId->identifier.name) == 0)
return structId;
}
}
return LookupType(current->parent, target);
}
default:
return LookupType(current->parent, target);
}
}
/* FIXME: This function should be modified to handle type parameters over
* structs. */
void ConvertCustomsToGenerics(Node *node)
{
if (node == NULL)
return;
switch (node->syntaxKind)
{
case Declaration:
{
Node *id = node->declaration.identifier;
Node *type = node->declaration.type->type.typeNode;
if (type->syntaxKind == CustomTypeNode)
{
char *target = id->typeTag->value.customType;
Node *typeLookup = LookupType(node, target);
if (typeLookup != NULL && typeLookup->syntaxKind == GenericArgument)
{
id->typeTag->type = Generic;
free(node->declaration.type);
node->declaration.type =
MakeGenericTypeNode(id->typeTag->value.genericType);
}
}
break;
}
case FunctionSignature:
{
Node *id = node->functionSignature.identifier;
Node *type = node->functionSignature.type->type.typeNode;
if (type->syntaxKind == CustomTypeNode)
{
char *target = id->typeTag->value.customType;
Node *typeLookup = LookupType(node, target);
if (typeLookup != NULL && typeLookup->syntaxKind == GenericArgument)
{
id->typeTag->type = Generic;
free(node->functionSignature.type);
node->functionSignature.type =
MakeGenericTypeNode(id->typeTag->value.genericType);
}
}
break;
}
}
Recurse(node, *ConvertCustomsToGenerics);
}

10
src/validation.h Normal file
View File

@ -0,0 +1,10 @@
#ifndef WRAITH_VALIDATION_H
#define WRAITH_VALIDATION_H
#include "ast.h"
void ValidateIdentifiers(Node *node);
void TagIdentifierTypes(Node *node);
void ConvertCustomsToGenerics(Node *node);
#endif /* WRAITH_VALIDATION_H */