Started working on a decompiler, called 'bytecode inspector'

It only has a few instructions for now, but I can flesh it out over
time.

repl/bytecode_inspector.c

@@ -0,0 +1,178 @@
+#include "bytecode_inspector.h"
+#include "toy_console_colors.h"
+#include "toy_opcodes.h"
+#include "toy_value.h"
+#include "toy_string.h"
+
+#include <stdio.h>
+#include <stdlib.h>
+
+int inspect_instruction(unsigned char* bytecode, unsigned int pc, unsigned int jumps_addr, unsigned int data_addr);
+int inspect_read(unsigned char* bytecode, unsigned int pc, unsigned int jumps_addr, unsigned int data_addr);
+
+// void inspect_jumps(unsigned char* bytecode, unsigned int pc, unsigned int size);
+// void inspect_param(unsigned char* bytecode, unsigned int pc, unsigned int size);
+// void inspect_data(unsigned char* bytecode, unsigned int pc, unsigned int size);
+// void inspect_subs(unsigned char* bytecode, unsigned int pc, unsigned int size);
+
+#define MARKER_VALUE(pc, type) \
+	(pc * sizeof(type))
+
+#define MARKER "\033[" TOY_CC_FONT_BLACK "m" " %lu\t" TOY_CC_RESET
+
+//exposed functions
+void inspect_bytecode(unsigned char* bytecode) {
+	//TODO: handle version info
+
+	unsigned int const header_size = 0;
+	unsigned int const header_jumps = 1;
+	unsigned int const header_param = 2;
+	unsigned int const header_data = 3;
+	unsigned int const header_subs = 4;
+
+	//header size
+	printf(MARKER TOY_CC_NOTICE "Bytecode Size: \t\t%u" TOY_CC_RESET "\n", MARKER_VALUE(header_size, unsigned int), ((unsigned int*)(bytecode))[header_size]);
+
+	//header counts
+	printf(MARKER TOY_CC_NOTICE "Jumps Size:\t\t%u" TOY_CC_RESET "\n", MARKER_VALUE(header_jumps, unsigned int), ((unsigned int*)(bytecode))[header_jumps]);
+	printf(MARKER TOY_CC_NOTICE "Param Size:\t\t%u" TOY_CC_RESET "\n", MARKER_VALUE(header_param, unsigned int), ((unsigned int*)(bytecode))[header_param]);
+	printf(MARKER TOY_CC_NOTICE "Data Size:\t\t%u" TOY_CC_RESET "\n", MARKER_VALUE(header_data, unsigned int), ((unsigned int*)(bytecode))[header_data]);
+	printf(MARKER TOY_CC_NOTICE "Subs Size:\t\t%u" TOY_CC_RESET "\n", MARKER_VALUE(header_subs, unsigned int), ((unsigned int*)(bytecode))[header_subs]);
+
+	printf("\n---\n");
+
+	//some addresses may be absent
+	unsigned int addr_pc = 4;
+	unsigned int code_addr = 0;
+	unsigned int jumps_addr = 0;
+	unsigned int param_addr = 0;
+	unsigned int data_addr = 0;
+	unsigned int subs_addr = 0;
+
+
+	//header addresses
+	if (true) {
+		addr_pc++;
+		printf(MARKER TOY_CC_NOTICE "Code Address:\t\t%u" TOY_CC_RESET "\n", MARKER_VALUE(addr_pc, unsigned int), ((unsigned int*)(bytecode))[addr_pc]);
+		code_addr = ((unsigned int*)(bytecode))[addr_pc];
+	}
+
+	if (((unsigned int*)(bytecode))[header_jumps] > 0) {
+		addr_pc++;
+		printf(MARKER TOY_CC_NOTICE "Jumps Address:\t\t%u" TOY_CC_RESET "\n", MARKER_VALUE(addr_pc, unsigned int), ((unsigned int*)(bytecode))[addr_pc]);
+		jumps_addr = ((unsigned int*)(bytecode))[addr_pc];
+	}
+
+	if (((unsigned int*)(bytecode))[header_param] > 0) {
+		addr_pc++;
+		printf(MARKER TOY_CC_NOTICE "Param Address:\t\t%u" TOY_CC_RESET "\n", MARKER_VALUE(addr_pc, unsigned int), ((unsigned int*)(bytecode))[addr_pc]);
+		param_addr = ((unsigned int*)(bytecode))[addr_pc];
+	}
+
+	if (((unsigned int*)(bytecode))[header_data] > 0) {
+		addr_pc++;
+		printf(MARKER TOY_CC_NOTICE "Data Address:\t\t%u" TOY_CC_RESET "\n", MARKER_VALUE(addr_pc, unsigned int), ((unsigned int*)(bytecode))[addr_pc]);
+		data_addr = ((unsigned int*)(bytecode))[addr_pc];
+	}
+
+	if (((unsigned int*)(bytecode))[header_subs] > 0) {
+		addr_pc++;
+		printf(MARKER TOY_CC_NOTICE "Subs Address:\t\t%u" TOY_CC_RESET "\n", MARKER_VALUE(addr_pc, unsigned int), ((unsigned int*)(bytecode))[addr_pc]);
+		subs_addr = ((unsigned int*)(bytecode))[addr_pc];
+	}
+
+	printf("\n---\n");
+
+	//read the code instructions one-by-one
+	unsigned int pc = code_addr;
+	while(pc < jumps_addr) {
+		pc += inspect_instruction(bytecode, pc, jumps_addr, data_addr);
+	}
+
+	(void)jumps_addr;
+	(void)param_addr;
+	(void)data_addr;
+	(void)subs_addr;
+}
+
+int inspect_instruction(unsigned char* bytecode, unsigned int pc, unsigned int jumps_addr, unsigned int data_addr) {
+	//read and print the opcode instruction at 'pc'
+
+	Toy_OpcodeType opcode = bytecode[pc];
+
+	switch(opcode) {
+		case TOY_OPCODE_READ:
+			return inspect_read(bytecode, pc, jumps_addr, data_addr);
+
+		case TOY_OPCODE_RETURN:
+			printf(MARKER "Keyword RETURN (%u)\n", MARKER_VALUE(pc, unsigned char), bytecode[pc + 1]);
+			return 4;
+
+		case TOY_OPCODE_PRINT:
+			printf(MARKER "Keyword PRINT\n", MARKER_VALUE(pc, unsigned char));
+			return 4;
+
+		default:
+			printf(MARKER TOY_CC_WARN "Unknown Word: [%u, %u, %u, %u]" TOY_CC_RESET "\n", MARKER_VALUE(pc, unsigned char), bytecode[pc], bytecode[pc+1], bytecode[pc+2], bytecode[pc+3]);
+			return 4;
+	}
+}
+
+int inspect_read(unsigned char* bytecode, unsigned int pc, unsigned int jumps_addr, unsigned int data_addr) {
+	Toy_ValueType type = bytecode[pc + 1];
+
+	switch(type) {
+		case TOY_VALUE_NULL: {
+			printf(MARKER "READ NULL\n", MARKER_VALUE(pc, unsigned char));
+			return 4;
+		}
+
+		case TOY_VALUE_BOOLEAN: {
+			if (bytecode[pc + 2]) {
+				printf(MARKER "READ BOOL true\n", MARKER_VALUE(pc, unsigned char));
+			}
+			else {
+				
+			}
+			return 4;
+		}
+
+		case TOY_VALUE_INTEGER: {
+			int i = *(int*)(bytecode + pc + 4);
+			printf(MARKER "READ INTEGER %d\n", MARKER_VALUE(pc, unsigned char), i);
+			return 8;
+		}
+
+		case TOY_VALUE_FLOAT: {
+			float i = *(float*)(bytecode + pc + 4);
+			printf(MARKER "READ FLOAT %f\n", MARKER_VALUE(pc, unsigned char), i);
+			return 8;
+		}
+
+		case TOY_VALUE_STRING: {
+			Toy_StringType stringType = (Toy_StringType)(*(bytecode + pc + 2)); //TODO: not needed?
+			int len = bytecode[pc + 3]; //only used for names?
+
+			(void)stringType;
+
+			unsigned int indexValue = *((unsigned int*)(bytecode + pc + 4));
+			unsigned int jumpValue = *((unsigned int*)(bytecode + jumps_addr + indexValue));
+			char* cstr = ((char*)(bytecode + data_addr + jumpValue));
+
+			printf(MARKER "READ STRING (%d) %s\n", MARKER_VALUE(pc, unsigned char), len, cstr);
+
+			return 8;
+		}
+
+		case TOY_VALUE_ARRAY:
+		case TOY_VALUE_TABLE: 
+		case TOY_VALUE_FUNCTION:
+		case TOY_VALUE_OPAQUE:
+		case TOY_VALUE_ANY:
+		case TOY_VALUE_UNKNOWN:
+		default: {
+			printf(MARKER "READ ???\n", MARKER_VALUE(pc, unsigned char));
+			return 4;
+		}
+	}
+}

repl/bytecode_inspector.h

@@ -0,0 +1,3 @@
+#pragma once
+
+void inspect_bytecode(unsigned char* bytecode);

repl/main.c

@@ -1,3 +1,5 @@
+#include "bytecode_inspector.h"
+
 #include "toy_console_colors.h"
 
 #include "toy_lexer.h"
@@ -127,7 +129,7 @@ typedef struct CmdLine {
 	bool silentPrint;
 	bool silentAssert;
 	bool removeAssert;
-	bool verboseDebugPrint;
+	bool verbose;
 } CmdLine;
 
 void usageCmdLine(int argc, const char* argv[]) {
@@ -189,7 +191,7 @@ CmdLine parseCmdLine(int argc, const char* argv[]) {
 		.silentPrint = false,
 		.silentAssert = false,
 		.removeAssert = false,
-		.verboseDebugPrint = false,
+		.verbose = false,
 	};
 
 	for (int i = 1; i < argc; i++) {
@@ -239,7 +241,7 @@ CmdLine parseCmdLine(int argc, const char* argv[]) {
 		}
 
 		else if (!strcmp(argv[i], "-d") || !strcmp(argv[i], "--verbose")) {
-			cmd.verboseDebugPrint = true;
+			cmd.verbose = true;
 		}
 
 		else {
@@ -250,80 +252,6 @@ CmdLine parseCmdLine(int argc, const char* argv[]) {
 	return cmd;
 }
 
-//repl function
-int repl(const char* filepath) {
-	//output options
-	Toy_setPrintCallback(printCallback);
-	Toy_setErrorCallback(errorAndContinueCallback);
-	Toy_setAssertFailureCallback(assertFailureAndContinueCallback);
-
-	//vars to use
-	char prompt[256];
-	getFileName(prompt, filepath);
-	unsigned int INPUT_BUFFER_SIZE = 4096;
-	char inputBuffer[INPUT_BUFFER_SIZE];
-	memset(inputBuffer, 0, INPUT_BUFFER_SIZE);
-
-	Toy_Bucket* bucket = Toy_allocateBucket(TOY_BUCKET_IDEAL);
-
-	Toy_VM vm;
-	Toy_initVM(&vm);
-
-	printf("%s> ", prompt); //shows the terminal prompt and begin
-
-	unsigned int runCount = 0; //used for initial preserveScope
-
-	//read from the terminal
-	while(fgets(inputBuffer, INPUT_BUFFER_SIZE, stdin)) {
-		//work around fgets() adding a newline
-		unsigned int length = strlen(inputBuffer);
-		if (inputBuffer[length - 1] == '\n') {
-			inputBuffer[--length] = '\0';
-		}
-
-		if (length == 0 || !inputBuffer[ strspn(inputBuffer, " \r\n\t") ]) {
-			printf("%s> ", prompt); //shows the terminal prompt and restart
-			continue;
-		}
-
-		//end
-		if (strlen(inputBuffer) == 4 && (strncmp(inputBuffer, "exit", 4) == 0 || strncmp(inputBuffer, "quit", 4) == 0)) {
-			break;
-		}
-
-		//parse the input, prep the VM for execution
-		Toy_Lexer lexer;
-		Toy_bindLexer(&lexer, inputBuffer);
-		Toy_Parser parser;
-		Toy_bindParser(&parser, &lexer);
-		Toy_Ast* ast = Toy_scanParser(&bucket, &parser); //Ast is in the bucket, so it doesn't need to be freed
-
-		//parsing error, retry
-		if (parser.error) {
-			printf("%s> ", prompt); //shows the terminal prompt
-			continue;
-		}
-
-		unsigned char* bytecode = Toy_compileToBytecode(ast);
-		Toy_bindVM(&vm, bytecode, runCount++ > 0);
-
-		//run
-		Toy_runVM(&vm);
-
-		//free the memory, and leave the VM ready for the next loop
-		Toy_resetVM(&vm, true);
-		free(bytecode);
-
-		printf("%s> ", prompt); //shows the terminal prompt
-	}
-
-	//cleanup all memory
-	Toy_freeVM(&vm);
-	Toy_freeBucket(&bucket);
-
-	return 0;
-}
-
 //debugging
 static void debugStackPrint(Toy_Stack* stack) {
 	//DEBUG: if there's anything on the stack, print it
@@ -385,6 +313,87 @@ static void debugScopePrint(Toy_Scope* scope, int depth) {
 	}
 }
 
+//repl function
+int repl(const char* filepath, bool verbose) {
+	//output options
+	Toy_setPrintCallback(printCallback);
+	Toy_setErrorCallback(errorAndContinueCallback);
+	Toy_setAssertFailureCallback(assertFailureAndContinueCallback);
+
+	//vars to use
+	char prompt[256];
+	getFileName(prompt, filepath);
+	unsigned int INPUT_BUFFER_SIZE = 4096;
+	char inputBuffer[INPUT_BUFFER_SIZE];
+	memset(inputBuffer, 0, INPUT_BUFFER_SIZE);
+
+	Toy_Bucket* bucket = Toy_allocateBucket(TOY_BUCKET_IDEAL);
+
+	Toy_VM vm;
+	Toy_initVM(&vm);
+
+	printf("%s> ", prompt); //shows the terminal prompt and begin
+
+	unsigned int runCount = 0; //used for initial preserveScope
+
+	//read from the terminal
+	while(fgets(inputBuffer, INPUT_BUFFER_SIZE, stdin)) {
+		//work around fgets() adding a newline
+		unsigned int length = strlen(inputBuffer);
+		if (inputBuffer[length - 1] == '\n') {
+			inputBuffer[--length] = '\0';
+		}
+
+		if (length == 0 || !inputBuffer[ strspn(inputBuffer, " \r\n\t") ]) {
+			printf("%s> ", prompt); //shows the terminal prompt and restart
+			continue;
+		}
+
+		//end
+		if (strlen(inputBuffer) == 4 && (strncmp(inputBuffer, "exit", 4) == 0 || strncmp(inputBuffer, "quit", 4) == 0)) {
+			break;
+		}
+
+		//parse the input, prep the VM for execution
+		Toy_Lexer lexer;
+		Toy_bindLexer(&lexer, inputBuffer);
+		Toy_Parser parser;
+		Toy_bindParser(&parser, &lexer);
+		Toy_Ast* ast = Toy_scanParser(&bucket, &parser); //Ast is in the bucket, so it doesn't need to be freed
+
+		//parsing error, retry
+		if (parser.error) {
+			printf("%s> ", prompt); //shows the terminal prompt
+			continue;
+		}
+
+		unsigned char* bytecode = Toy_compileToBytecode(ast);
+		Toy_bindVM(&vm, bytecode, runCount++ > 0);
+
+		//run
+		Toy_runVM(&vm);
+
+		//print the debug info
+		if (verbose) {
+			debugStackPrint(vm.stack);
+			debugScopePrint(vm.scope, 0);
+			inspect_bytecode(bytecode);
+		}
+
+		//free the memory, and leave the VM ready for the next loop
+		Toy_resetVM(&vm, true);
+		free(bytecode);
+
+		printf("%s> ", prompt); //shows the terminal prompt
+	}
+
+	//cleanup all memory
+	Toy_freeVM(&vm);
+	Toy_freeBucket(&bucket);
+
+	return 0;
+}
+
 //main file
 int main(int argc, const char* argv[]) {
 	Toy_setPrintCallback(printCallback);
@@ -462,9 +471,10 @@ int main(int argc, const char* argv[]) {
 		Toy_runVM(&vm);
 
 		//print the debug info
-		if (cmd.verboseDebugPrint) { //URGENT: 'verbose' option is mainly for the WIP elements, like decompiler
+		if (cmd.verbose) {
 			debugStackPrint(vm.stack);
 			debugScopePrint(vm.scope, 0);
+			inspect_bytecode(bytecode);
 		}
 
 		//cleanup
@@ -472,7 +482,7 @@ int main(int argc, const char* argv[]) {
 		free(bytecode);
 	}
 	else {
-		repl(argv[0]);
+		repl(argv[0], cmd.verbose);
 	}
 
 	return 0;

scripts/hello_world.toy

@@ -1,20 +1,5 @@
 
 
-/*
-{
-	fn hello() {
-		print "Hello world";
-	}
-	hello();
-}
-*/
-
-{
-	fn hello(arg) {
-		print arg;
-	}
-
-	hello("world");
-}
-
+print "hello world";
 
+print [];

source/toy_vm.c

@@ -62,7 +62,7 @@ static void processRead(Toy_VM* vm) {
 		}
 
 		case TOY_VALUE_STRING: {
-			enum Toy_StringType stringType = READ_BYTE(vm);
+			Toy_StringType stringType = READ_BYTE(vm);
 			int len = (int)READ_BYTE(vm); //WARN: only used for name strings
 			(void)len;