mirror of
https://github.com/krgamestudios/Toy.git
synced 2026-04-15 14:54:07 +10:00
525 lines
17 KiB
C
525 lines
17 KiB
C
#include "compiler.h"
|
|
|
|
#include "memory.h"
|
|
|
|
#include "literal.h"
|
|
#include "literal_array.h"
|
|
#include "literal_dictionary.h"
|
|
|
|
#include "console_colors.h"
|
|
|
|
#include <stdio.h>
|
|
|
|
void initCompiler(Compiler* compiler) {
|
|
initLiteralArray(&compiler->literalCache);
|
|
compiler->bytecode = NULL;
|
|
compiler->capacity = 0;
|
|
compiler->count = 0;
|
|
}
|
|
|
|
//separated out, so it can be recursive
|
|
static int writeNodeCompoundToCache(Compiler* compiler, Node* node) {
|
|
int index = -1;
|
|
|
|
//for both, stored as an array
|
|
LiteralArray* store = ALLOCATE(LiteralArray, 1);
|
|
initLiteralArray(store);
|
|
|
|
//emit an array or a dictionary definition
|
|
if (node->compound.literalType == LITERAL_DICTIONARY) {
|
|
//ensure each literal key and value are in the cache, individually
|
|
for (int i = 0; i < node->compound.count; i++) {
|
|
//keys
|
|
switch(node->compound.nodes[i].pair.left->type) {
|
|
case NODE_LITERAL: {
|
|
//keys are literals
|
|
int key = findLiteralIndex(&compiler->literalCache, node->compound.nodes[i].pair.left->atomic.literal);
|
|
if (key < 0) {
|
|
key = pushLiteralArray(&compiler->literalCache, node->compound.nodes[i].pair.left->atomic.literal);
|
|
}
|
|
|
|
pushLiteralArray(store, TO_INTEGER_LITERAL(key));
|
|
}
|
|
break;
|
|
|
|
case NODE_COMPOUND: {
|
|
int key = writeNodeCompoundToCache(compiler, node->compound.nodes[i].pair.left);
|
|
|
|
pushLiteralArray(store, TO_INTEGER_LITERAL(key));
|
|
}
|
|
break;
|
|
|
|
default:
|
|
fprintf(stderr, "[Internal] Unrecognized key node type in writeNodeCompoundToCache()");
|
|
return -1;
|
|
}
|
|
|
|
//values
|
|
switch(node->compound.nodes[i].pair.right->type) {
|
|
case NODE_LITERAL: {
|
|
//values are literals
|
|
int val = findLiteralIndex(&compiler->literalCache, node->compound.nodes[i].pair.right->atomic.literal);
|
|
if (val < 0) {
|
|
val = pushLiteralArray(&compiler->literalCache, node->compound.nodes[i].pair.right->atomic.literal);
|
|
}
|
|
|
|
pushLiteralArray(store, TO_INTEGER_LITERAL(val));
|
|
}
|
|
break;
|
|
|
|
case NODE_COMPOUND: {
|
|
int val = writeNodeCompoundToCache(compiler, node->compound.nodes[i].pair.right);
|
|
|
|
pushLiteralArray(store, TO_INTEGER_LITERAL(val));
|
|
}
|
|
break;
|
|
|
|
default:
|
|
fprintf(stderr, "[Internal] Unrecognized value node type in writeNodeCompoundToCache()");
|
|
return -1;
|
|
}
|
|
}
|
|
|
|
//push the store to the cache, with instructions about how pack it
|
|
index = pushLiteralArray(&compiler->literalCache, TO_DICTIONARY_LITERAL(store)); //WARNING: pushed as a dictionary, so below can recognize it
|
|
}
|
|
else if (node->compound.literalType == LITERAL_ARRAY) {
|
|
//ensure each literal value is in the cache, individually
|
|
for (int i = 0; i < node->compound.count; i++) {
|
|
switch(node->compound.nodes[i].type) {
|
|
case NODE_LITERAL: {
|
|
//values
|
|
int val = findLiteralIndex(&compiler->literalCache, node->compound.nodes[i].atomic.literal);
|
|
if (val < 0) {
|
|
val = pushLiteralArray(&compiler->literalCache, node->compound.nodes[i].atomic.literal);
|
|
}
|
|
|
|
pushLiteralArray(store, TO_INTEGER_LITERAL(val));
|
|
}
|
|
break;
|
|
|
|
case NODE_COMPOUND: {
|
|
int val = writeNodeCompoundToCache(compiler, &node->compound.nodes[i]);
|
|
|
|
index = pushLiteralArray(store, TO_INTEGER_LITERAL(val));
|
|
}
|
|
break;
|
|
|
|
default:
|
|
fprintf(stderr, "[Internal] Unrecognized node type in writeNodeCompoundToCache()");
|
|
return -1;
|
|
}
|
|
}
|
|
|
|
//push the store to the cache, with instructions about how pack it
|
|
index = pushLiteralArray(&compiler->literalCache, TO_ARRAY_LITERAL(store));
|
|
}
|
|
else {
|
|
fprintf(stderr, "[Internal] Unrecognized compound type in writeNodeCompoundToCache()");
|
|
}
|
|
|
|
return index;
|
|
}
|
|
|
|
static int writeLiteralTypeToCache(LiteralArray* literalCache, Literal literal) {
|
|
//I don't like storing types in an array, but it's the easiest and most straight forward method
|
|
LiteralArray* store = ALLOCATE(LiteralArray, 1);
|
|
initLiteralArray(store);
|
|
|
|
//store the base literal in the store
|
|
pushLiteralArray(store, literal);
|
|
|
|
//if it's a compound type, recurse and store the results
|
|
if (AS_TYPE(literal).typeOf == LITERAL_ARRAY || AS_TYPE(literal).typeOf == LITERAL_DICTIONARY) {
|
|
for (int i = 0; i < AS_TYPE(literal).count; i++) {
|
|
//write the values to the cache, and the indexes to the store
|
|
int subIndex = writeLiteralTypeToCache(literalCache, ((Literal*)(AS_TYPE(literal).subtypes))[i]);
|
|
pushLiteralArray(store, TO_INTEGER_LITERAL(subIndex));
|
|
}
|
|
}
|
|
|
|
//push the store to the cache, tweaking the type
|
|
Literal lit = TO_ARRAY_LITERAL(store);
|
|
lit.type = LITERAL_TYPE_INTERMEDIATE; //NOTE: tweaking the type usually isn't a good idea
|
|
return pushLiteralArray(literalCache, lit);
|
|
}
|
|
|
|
void writeCompiler(Compiler* compiler, Node* node) {
|
|
//grow if the bytecode space is too small
|
|
if (compiler->capacity < compiler->count + 1) {
|
|
int oldCapacity = compiler->capacity;
|
|
|
|
compiler->capacity = GROW_CAPACITY(oldCapacity);
|
|
compiler->bytecode = GROW_ARRAY(unsigned char, compiler->bytecode, oldCapacity, compiler->capacity);
|
|
}
|
|
|
|
//determine node type
|
|
switch(node->type) {
|
|
//TODO: more types, like variables, etc.
|
|
case NODE_ERROR: {
|
|
fprintf(stderr, ERROR "[Internal] NODE_ERROR encountered in writeCompiler()\n" RESET);
|
|
compiler->bytecode[compiler->count++] = OP_EOF; //1 byte
|
|
}
|
|
break;
|
|
|
|
case NODE_LITERAL: {
|
|
//ensure the literal is in the cache
|
|
int index = findLiteralIndex(&compiler->literalCache, node->atomic.literal);
|
|
if (index < 0) {
|
|
index = pushLiteralArray(&compiler->literalCache, node->atomic.literal);
|
|
}
|
|
|
|
//push the node opcode to the bytecode
|
|
if (index >= 256) {
|
|
//push a "long" index
|
|
compiler->bytecode[compiler->count++] = OP_LITERAL_LONG; //1 byte
|
|
*((unsigned short*)(compiler->bytecode + compiler->count)) = (unsigned short)index; //2 bytes
|
|
|
|
compiler->count += sizeof(unsigned short);
|
|
}
|
|
else {
|
|
//push the index
|
|
compiler->bytecode[compiler->count++] = OP_LITERAL; //1 byte
|
|
compiler->bytecode[compiler->count++] = (unsigned char)index; //1 byte
|
|
}
|
|
}
|
|
break;
|
|
|
|
case NODE_UNARY:
|
|
//pass to the child node, then embed the unary command (print, negate, etc.)
|
|
writeCompiler(compiler, node->unary.child);
|
|
compiler->bytecode[compiler->count++] = (unsigned char)node->unary.opcode; //1 byte
|
|
break;
|
|
|
|
case NODE_BINARY:
|
|
//pass to the child nodes, then embed the binary command (math, etc.)
|
|
writeCompiler(compiler, node->binary.left);
|
|
writeCompiler(compiler, node->binary.right);
|
|
compiler->bytecode[compiler->count++] = (unsigned char)node->binary.opcode; //1 byte
|
|
break;
|
|
|
|
case NODE_GROUPING:
|
|
compiler->bytecode[compiler->count++] = (unsigned char)OP_GROUPING_BEGIN; //1 byte
|
|
writeCompiler(compiler, node->grouping.child);
|
|
compiler->bytecode[compiler->count++] = (unsigned char)OP_GROUPING_END; //1 byte
|
|
break;
|
|
|
|
case NODE_BLOCK:
|
|
compiler->bytecode[compiler->count++] = (unsigned char)OP_SCOPE_BEGIN; //1 byte
|
|
|
|
for (int i = 0; i < node->block.count; i++) {
|
|
writeCompiler(compiler, &(node->block.nodes[i]));
|
|
}
|
|
|
|
compiler->bytecode[compiler->count++] = (unsigned char)OP_SCOPE_END; //1 byte
|
|
break;
|
|
|
|
case NODE_COMPOUND: {
|
|
int index = writeNodeCompoundToCache(compiler, node);
|
|
|
|
//push the node opcode to the bytecode
|
|
if (index >= 256) {
|
|
//push a "long" index
|
|
compiler->bytecode[compiler->count++] = OP_LITERAL_LONG; //1 byte
|
|
*((unsigned short*)(compiler->bytecode + compiler->count)) = (unsigned short)index; //2 bytes
|
|
|
|
compiler->count += sizeof(unsigned short);
|
|
}
|
|
else {
|
|
//push the index
|
|
compiler->bytecode[compiler->count++] = OP_LITERAL; //1 byte
|
|
compiler->bytecode[compiler->count++] = (unsigned char)index; //1 byte
|
|
}
|
|
}
|
|
break;
|
|
|
|
case NODE_PAIR:
|
|
fprintf(stderr, ERROR "[Internal] NODE_PAIR encountered in writeCompiler()\n" RESET);
|
|
break;
|
|
|
|
case NODE_VAR_TYPES: { //TODO: the "type" keyword
|
|
int index = writeLiteralTypeToCache(&compiler->literalCache, node->varTypes.typeLiteral);
|
|
|
|
//embed the info into the bytecode
|
|
if (index >= 256) {
|
|
//push a "long" index
|
|
compiler->bytecode[compiler->count++] = OP_TYPE_DECL_LONG; //1 byte
|
|
*((unsigned short*)(compiler->bytecode + compiler->count)) = (unsigned short)index; //2 bytes
|
|
|
|
compiler->count += sizeof(unsigned short);
|
|
}
|
|
else {
|
|
//push the index
|
|
compiler->bytecode[compiler->count++] = OP_TYPE_DECL; //1 byte
|
|
compiler->bytecode[compiler->count++] = (unsigned char)index; //1 byte
|
|
}
|
|
}
|
|
break;
|
|
|
|
case NODE_VAR_DECL: {
|
|
//first, embed the expression (leaves it on the stack)
|
|
writeCompiler(compiler, node->varDecl.expression);
|
|
|
|
//write each piece of the declaration to the bytecode
|
|
int identifierIndex = findLiteralIndex(&compiler->literalCache, node->varDecl.identifier);
|
|
if (identifierIndex < 0) {
|
|
identifierIndex = pushLiteralArray(&compiler->literalCache, node->varDecl.identifier);
|
|
}
|
|
|
|
int typeIndex = writeLiteralTypeToCache(&compiler->literalCache, node->varDecl.typeLiteral);
|
|
|
|
//embed the info into the bytecode
|
|
if (identifierIndex >= 256 || typeIndex >= 256) {
|
|
//push a "long" declaration
|
|
compiler->bytecode[compiler->count++] = OP_VAR_DECL_LONG; //1 byte
|
|
|
|
*((unsigned short*)(compiler->bytecode + compiler->count)) = (unsigned short)identifierIndex; //2 bytes
|
|
compiler->count += sizeof(unsigned short);
|
|
|
|
*((unsigned short*)(compiler->bytecode + compiler->count)) = (unsigned short)typeIndex; //2 bytes
|
|
compiler->count += sizeof(unsigned short);
|
|
}
|
|
else {
|
|
//push a declaration
|
|
compiler->bytecode[compiler->count++] = OP_VAR_DECL; //1 byte
|
|
compiler->bytecode[compiler->count++] = (unsigned char)identifierIndex; //1 byte
|
|
compiler->bytecode[compiler->count++] = (unsigned char)typeIndex; //1 byte
|
|
}
|
|
}
|
|
break;
|
|
|
|
//TODO: OP_VAR_ASSIGN
|
|
}
|
|
}
|
|
|
|
void freeCompiler(Compiler* compiler) {
|
|
freeLiteralArray(&compiler->literalCache);
|
|
FREE(unsigned char, compiler->bytecode);
|
|
compiler->bytecode = NULL;
|
|
compiler->capacity = 0;
|
|
compiler->count = 0;
|
|
}
|
|
|
|
static void emitByte(unsigned char** collationPtr, int* capacityPtr, int* countPtr, unsigned char byte) {
|
|
//grow the array
|
|
if (*countPtr + 1 > *capacityPtr) {
|
|
int oldCapacity = *capacityPtr;
|
|
*capacityPtr = GROW_CAPACITY(*capacityPtr);
|
|
*collationPtr = GROW_ARRAY(unsigned char, *collationPtr, oldCapacity, *capacityPtr);
|
|
}
|
|
|
|
//append to the collation
|
|
(*collationPtr)[(*countPtr)++] = byte;
|
|
}
|
|
|
|
static void emitShort(unsigned char** collationPtr, int* capacityPtr, int* countPtr, unsigned short bytes) {
|
|
char* ptr = (char*)&bytes;
|
|
|
|
emitByte(collationPtr, capacityPtr, countPtr, *ptr);
|
|
ptr++;
|
|
emitByte(collationPtr, capacityPtr, countPtr, *ptr);
|
|
}
|
|
|
|
static void emitInt(unsigned char** collationPtr, int* capacityPtr, int* countPtr, int bytes) {
|
|
char* ptr = (char*)&bytes;
|
|
|
|
emitByte(collationPtr, capacityPtr, countPtr, *ptr);
|
|
ptr++;
|
|
emitByte(collationPtr, capacityPtr, countPtr, *ptr);
|
|
ptr++;
|
|
emitByte(collationPtr, capacityPtr, countPtr, *ptr);
|
|
ptr++;
|
|
emitByte(collationPtr, capacityPtr, countPtr, *ptr);
|
|
}
|
|
|
|
static void emitFloat(unsigned char** collationPtr, int* capacityPtr, int* countPtr, float bytes) {
|
|
char* ptr = (char*)&bytes;
|
|
|
|
emitByte(collationPtr, capacityPtr, countPtr, *ptr);
|
|
ptr++;
|
|
emitByte(collationPtr, capacityPtr, countPtr, *ptr);
|
|
ptr++;
|
|
emitByte(collationPtr, capacityPtr, countPtr, *ptr);
|
|
ptr++;
|
|
emitByte(collationPtr, capacityPtr, countPtr, *ptr);
|
|
}
|
|
|
|
//return the result
|
|
unsigned char* collateCompiler(Compiler* compiler, int* size) {
|
|
int capacity = GROW_CAPACITY(0);
|
|
int count = 0;
|
|
unsigned char* collation = ALLOCATE(unsigned char, capacity);
|
|
|
|
//embed the header with version information
|
|
emitByte(&collation, &capacity, &count, TOY_VERSION_MAJOR);
|
|
emitByte(&collation, &capacity, &count, TOY_VERSION_MINOR);
|
|
emitByte(&collation, &capacity, &count, TOY_VERSION_PATCH);
|
|
|
|
//embed the build info
|
|
if (strlen(TOY_VERSION_BUILD) + count + 1 > capacity) {
|
|
int oldCapacity = capacity;
|
|
capacity = strlen(TOY_VERSION_BUILD) + count + 1; //full header size
|
|
collation = GROW_ARRAY(unsigned char, collation, oldCapacity, capacity);
|
|
}
|
|
|
|
memcpy(&collation[count], TOY_VERSION_BUILD, strlen(TOY_VERSION_BUILD));
|
|
count += strlen(TOY_VERSION_BUILD);
|
|
collation[count++] = '\0'; //terminate the build string
|
|
|
|
emitByte(&collation, &capacity, &count, OP_SECTION_END); //terminate header
|
|
|
|
//embed the data section (first short is the number of literals)
|
|
emitShort(&collation, &capacity, &count, compiler->literalCache.count);
|
|
|
|
//emit each literal by type
|
|
for (int i = 0; i < compiler->literalCache.count; i++) {
|
|
//literal Opcode
|
|
// emitShort(&collation, &capacity, &count, OP_LITERAL); //This isn't needed
|
|
|
|
//literal type, followed by literal value
|
|
switch(compiler->literalCache.literals[i].type) {
|
|
case LITERAL_NULL:
|
|
emitByte(&collation, &capacity, &count, LITERAL_NULL);
|
|
//null has no following value
|
|
break;
|
|
|
|
case LITERAL_BOOLEAN:
|
|
emitByte(&collation, &capacity, &count, LITERAL_BOOLEAN);
|
|
emitByte(&collation, &capacity, &count, AS_BOOLEAN(compiler->literalCache.literals[i]));
|
|
break;
|
|
|
|
case LITERAL_INTEGER:
|
|
emitByte(&collation, &capacity, &count, LITERAL_INTEGER);
|
|
emitInt(&collation, &capacity, &count, AS_INTEGER(compiler->literalCache.literals[i]));
|
|
break;
|
|
|
|
case LITERAL_FLOAT:
|
|
emitByte(&collation, &capacity, &count, LITERAL_FLOAT);
|
|
emitFloat(&collation, &capacity, &count, AS_FLOAT(compiler->literalCache.literals[i]));
|
|
break;
|
|
|
|
case LITERAL_STRING: {
|
|
emitByte(&collation, &capacity, &count, LITERAL_STRING);
|
|
|
|
Literal str = compiler->literalCache.literals[i];
|
|
|
|
for (int c = 0; c < STRLEN(str); c++) {
|
|
emitByte(&collation, &capacity, &count, AS_STRING(str)[c]);
|
|
}
|
|
|
|
emitByte(&collation, &capacity, &count, '\0'); //terminate the string
|
|
}
|
|
break;
|
|
|
|
case LITERAL_ARRAY: {
|
|
emitByte(&collation, &capacity, &count, LITERAL_ARRAY);
|
|
|
|
LiteralArray* ptr = AS_ARRAY(compiler->literalCache.literals[i]);
|
|
|
|
//length of the array, as a short
|
|
emitShort(&collation, &capacity, &count, ptr->count);
|
|
|
|
//each element of the array
|
|
for (int i = 0; i < ptr->count; i++) {
|
|
emitShort(&collation, &capacity, &count, (unsigned short)AS_INTEGER(ptr->literals[i])); //shorts representing the indexes of the values
|
|
}
|
|
|
|
freeLiteralArray(ptr);
|
|
FREE(LiteralArray, ptr);
|
|
}
|
|
break;
|
|
|
|
case LITERAL_DICTIONARY: {
|
|
emitByte(&collation, &capacity, &count, LITERAL_DICTIONARY);
|
|
|
|
LiteralArray* ptr = AS_ARRAY(compiler->literalCache.literals[i]); //used an array for storage above
|
|
|
|
//length of the array, as a short
|
|
emitShort(&collation, &capacity, &count, ptr->count); //count is the array size, NOT the dictionary size
|
|
|
|
//each element of the array
|
|
for (int i = 0; i < ptr->count; i++) {
|
|
emitShort(&collation, &capacity, &count, (unsigned short)AS_INTEGER(ptr->literals[i])); //shorts representing the indexes of the values
|
|
}
|
|
|
|
freeLiteralArray(ptr);
|
|
FREE(LiteralArray, ptr);
|
|
}
|
|
break;
|
|
|
|
//TODO: function
|
|
|
|
case LITERAL_IDENTIFIER: {
|
|
emitByte(&collation, &capacity, &count, LITERAL_IDENTIFIER);
|
|
|
|
Literal identifier = compiler->literalCache.literals[i];
|
|
|
|
for (int c = 0; c < STRLEN_I(identifier); c++) {
|
|
emitByte(&collation, &capacity, &count, AS_IDENTIFIER(identifier)[c]);
|
|
}
|
|
|
|
emitByte(&collation, &capacity, &count, '\0'); //terminate the string
|
|
}
|
|
break;
|
|
|
|
case LITERAL_TYPE: {
|
|
//push a raw type
|
|
emitByte(&collation, &capacity, &count, LITERAL_TYPE);
|
|
|
|
Literal typeLiteral = compiler->literalCache.literals[i];
|
|
|
|
//what type this literal represents
|
|
emitByte(&collation, &capacity, &count, AS_TYPE(typeLiteral).typeOf);
|
|
emitByte(&collation, &capacity, &count, AS_TYPE(typeLiteral).constant); //if it's constant
|
|
}
|
|
break;
|
|
|
|
case LITERAL_TYPE_INTERMEDIATE: {
|
|
emitByte(&collation, &capacity, &count, LITERAL_TYPE_INTERMEDIATE);
|
|
|
|
LiteralArray* ptr = AS_ARRAY(compiler->literalCache.literals[i]); //used an array for storage above
|
|
|
|
//the base literal
|
|
Literal typeLiteral = ptr->literals[0];
|
|
|
|
//what type this literal represents
|
|
emitByte(&collation, &capacity, &count, AS_TYPE(typeLiteral).typeOf);
|
|
emitByte(&collation, &capacity, &count, AS_TYPE(typeLiteral).constant); //if it's constant
|
|
|
|
//each element of the array, If they exist, representing sub-types already in the cache
|
|
if (AS_TYPE(typeLiteral).typeOf == LITERAL_ARRAY || AS_TYPE(typeLiteral).typeOf == LITERAL_DICTIONARY) {
|
|
//the type will represent how many to expect in the array
|
|
for (int i = 1; i < ptr->count; i++) {
|
|
emitShort(&collation, &capacity, &count, (unsigned short)AS_INTEGER(ptr->literals[i])); //shorts representing the indexes of the types
|
|
}
|
|
}
|
|
|
|
freeLiteralArray(ptr);
|
|
FREE(LiteralArray, ptr);
|
|
}
|
|
break;
|
|
|
|
default:
|
|
fprintf(stderr, "[Internal] Unknown literal type encountered within literal cache: %d\n", compiler->literalCache.literals[i].type);
|
|
return NULL;
|
|
}
|
|
}
|
|
|
|
emitByte(&collation, &capacity, &count, OP_SECTION_END); //terminate data
|
|
|
|
//code section
|
|
for (int i = 0; i < compiler->count; i++) {
|
|
emitByte(&collation, &capacity, &count, compiler->bytecode[i]);
|
|
}
|
|
|
|
emitByte(&collation, &capacity, &count, OP_SECTION_END); //terminate code
|
|
|
|
emitByte(&collation, &capacity, &count, OP_EOF); //terminate bytecode
|
|
|
|
//finalize
|
|
SHRINK_ARRAY(unsigned char, collation, capacity, count);
|
|
|
|
*size = count;
|
|
|
|
return collation;
|
|
} |