From 42fab467f1847cce32a3d3524384c7fdf0a9641e Mon Sep 17 00:00:00 2001
From: Maxwell Jeffress <maxwelljeffress@proton.me>
Date: Thu, 16 Jul 2026 13:29:53 +1000
Subject: [PATCH] more assembler stuff

---
 meson.build              |   3 +-
 src/assembler/assemble.c | 163 ++++++++++++++++++++++++++++++++++++++-
 src/assembler/assemble.h |  14 ++++
 tests/add.asm            |   2 +
 4 files changed, 180 insertions(+), 2 deletions(-)

diff --git a/meson.build b/meson.build
index 23328d7..fb0eeb3 100644
--- a/meson.build
+++ b/meson.build
@@ -9,7 +9,8 @@ sources = files(
 )
 
 asmSources = files(
-    'src/assembler/main.c'
+    'src/assembler/main.c',
+    'src/assembler/assemble.c'
 )
 
 args = ['-Wall', '-Wextra']
diff --git a/src/assembler/assemble.c b/src/assembler/assemble.c
index 99d58e5..90bb02e 100644
--- a/src/assembler/assemble.c
+++ b/src/assembler/assemble.c
@@ -1,9 +1,170 @@
 #include "assemble.h"
 
 #include <stdlib.h>
+#include <stdio.h>
+#include <stdbool.h>
+#include <string.h>
 
 #include "../instruction.h"
 
-void assemble(const char* input, const char* outputFileName) {
+// Table of strings to instructions
+struct _stringToInstructionOpcodeTable {char* str; InstructionOpcode opcode;} stringToInstructionOpcodeTable[] = {
+    {"mov", INST_MOV},
+    {"mvi", INST_MVI},
+    {"add", INST_ADD},
+    {"sub", INST_SUB},
+    {"load", INST_LOAD},
+    {"store", INST_STORE},
+    {"and", INST_AND},
+    {"or", INST_OR},
+    {"xor", INST_XOR},
+    {"shf", INST_SHF},
+    {"jmp", INST_JMP},
+    {"jiz", INST_JIZ},
+    {"call", INST_CALL},
+    {"ret", INST_RET},
+    {"sys", INST_SYS},
+    {"hlt", INST_HLT},
+};
 
+InstructionOpcode stringToInstructionOpcode(const char* string) {
+    for (size_t i = 0; i < sizeof(stringToInstructionOpcodeTable) / sizeof(struct _stringToInstructionOpcodeTable)); i++) {
+        if (strcmp(stringToInstructionOpcodeTable[i].str, string) == 0) {
+            return stringToInstructionOpcodeTable[i].opcode;
+        }
+    }
+
+    fprintf(stderr, "Invalid instruction string %s\n", string);
+    exit(1);
+}
+
+Program newProgram(void) {
+    Program program = {
+        .at = malloc(sizeof(Instruction) * 16),
+        .len = 0,
+        .capacity = 16
+    };
+
+    if (program.at == NULL) {
+        fprintf(stderr, "malloc failed while creating new program\n");
+        exit(1);
+    }
+    return program;
+}
+
+void addInstructionToProgram(Program* program, Instruction inst) {
+    if (program->at == NULL) {
+        fprintf(stderr, "unexpected NULL value when adding instruction to program\n");
+        exit(1);
+    }
+
+    if (program->len + 1 >= program->capacity) {
+        Instruction* tmp = malloc(sizeof(Instruction) * program->capacity * 2);
+        if (tmp == NULL) {
+            fprintf(stderr, "malloc failed while expanding program\n");
+            exit(1);
+        }
+        program->at = tmp;
+        program->capacity *= 2;
+    }
+
+    program->at[program->len] = inst;
+    program->len++;
+}
+
+void freeProgram(Program* program) {
+    if (program->at != NULL) {
+        free(program->at);
+    }
+    program->capacity = 0;
+    program->len = 0;
+}
+
+/**
+ * Essentially a lexer for the assembly language.
+ * @param [input] The text to create a program from. NOT the file name
+ * @return A new program, with all instructions from the input
+ */
+Program createProgramFromText(const char* input) {
+    Program program = newProgram();
+
+    char buf[16] = {'\0'};
+    size_t bufSize = 0;
+    size_t programLen = strlen(input);
+    size_t current = 0;
+
+    bool processingInstWord = true;
+    uint8_t processingArgNum = 0;
+
+    Instruction currentInst = {};
+
+    for (;;) {
+        // scan through until we hit a space, colon or new line
+        char c = input[current];
+        switch (c) {
+            case ' ': {
+                if (bufSize == 0) {
+                    // Ignore extra spaces
+                    break;
+                }
+
+                if (processingInstWord) {
+                    currentInst.opcode = stringToInstructionOpcode(buf);
+                } else {
+                    if (processingArgNum >= 3) {
+                        fprintf(stderr, "expecting new line after instruction args, got %s\n", buf);
+                        exit(1);
+                    }
+                    OperandType type = INSTRUCTION_OPERAND_TYPES[currentInst.opcode][processingArgNum];
+                    if (type == OP_NONE) {
+                        fprintf(stderr, "expecting new line after instruction args, got %s\n", buf);
+                        exit(1);
+                    }
+                }
+                
+                buf[0] = '\0';
+                bufSize = 0;
+                break;
+            }
+            case '\n': {
+                if (processingInstWord && bufSize == 0) {
+                    // Ignore extra new lines
+                    break;
+                }
+                break;
+            }
+            case ':': {
+                if (!processingInstWord) {
+                    fprintf(stderr, "label must be first word on a new line");
+                    exit(1);
+                }
+                break;
+            }
+            case '\t': {
+                // Ignore tabs
+                break;
+            }
+            case '\r': {
+                // Ignore Windows specific formatting
+                break;
+            }
+            default: {
+                buf[bufSize] = c;
+                buf[bufSize + 1] = '\0';
+                bufSize++;
+                break;
+            }
+        }
+
+        current++;
+        if (current >= programLen) {
+            break;
+        }
+    }
+
+    return program;
+}
+
+void assemble(const char* input, const char* outputFileName) {
+    Program program = createProgramFromText(input);
 }
diff --git a/src/assembler/assemble.h b/src/assembler/assemble.h
index cee9c2a..a17c380 100644
--- a/src/assembler/assemble.h
+++ b/src/assembler/assemble.h
@@ -1,3 +1,17 @@
 #pragma once
 
+#include "../instruction.h"
+
+typedef struct {
+    Instruction* at;
+    size_t len;
+    size_t capacity;
+} Program;
+
+Program newProgram(void);
+void addInstructionToProgram(Program* program, Instruction inst);
+void freeProgram(Program* program);
+
+Program createProgramFromText(const char* input);
+
 void assemble(const char* input, const char* outputFileName);
diff --git a/tests/add.asm b/tests/add.asm
index c0ee31a..8f6a875 100644
--- a/tests/add.asm
+++ b/tests/add.asm
@@ -1,3 +1,5 @@
+dingus:
+
 mvi a 2
 mvi b 2