From dc210e409cca65b2efdc22fe03a3095ff10323b6 Mon Sep 17 00:00:00 2001
From: omicron <omicron.me@protonmail.com>
Date: Mon, 7 Apr 2025 12:50:39 +0200
Subject: [PATCH 01/14] fix parse_immediate to accept label_reference instead
 of identifier

---
 src/parser/parser.c | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/src/parser/parser.c b/src/parser/parser.c
index f3db142..7e60c81 100644
--- a/src/parser/parser.c
+++ b/src/parser/parser.c
@@ -83,7 +83,7 @@ parse_result_t parse_register_expression(tokenlist_entry_t *current) {
 }
 
 parse_result_t parse_immediate(tokenlist_entry_t *current) {
-    parser_t parsers[] = {parse_number, parse_identifier, nullptr};
+    parser_t parsers[] = {parse_number, parse_label_reference, nullptr};
     parse_result_t result = parse_any(current, parsers);
     return parse_result_wrap(NODE_IMMEDIATE, result);
 }
-- 
2.47.2


From d3d69b82d513d0e32866a12106942594e0748b2d Mon Sep 17 00:00:00 2001
From: omicron <omicron.me@protonmail.com>
Date: Mon, 7 Apr 2025 10:49:57 +0200
Subject: [PATCH 02/14] Add .import and .export directive to the grammar and
 parser

---
 doc/parser_grammar.txt  |  6 +++++-
 src/ast.c               |  8 ++++++++
 src/ast.h               |  4 ++++
 src/parser/parser.c     | 18 +++++++++++++++++-
 src/parser/primitives.c | 16 ++++++++++++++++
 src/parser/primitives.h |  2 ++
 6 files changed, 52 insertions(+), 2 deletions(-)

diff --git a/doc/parser_grammar.txt b/doc/parser_grammar.txt
index 9a651b9..00c7275 100644
--- a/doc/parser_grammar.txt
+++ b/doc/parser_grammar.txt
@@ -3,10 +3,14 @@
 
 <label> ::= <identifier> <colon>
 
-<directive> ::= <dot> <section_directive> <newline>
+<directive> ::= <dot> (<section_directive> | <export_directive> | <import_directive> ) <newline>
 
 <section_directive> ::= "section" <identifier>
 
+<export_directive> ::= "export" <identifier>
+
+<import_directive> ::= "import" <identifier>
+
 <instruction> ::= <identifier> <operands> <newline>
 
 <operands> ::= <operand> ( <comma> <operand> )*
diff --git a/src/ast.c b/src/ast.c
index e47c949..d9275f7 100644
--- a/src/ast.c
+++ b/src/ast.c
@@ -123,6 +123,10 @@ const char *ast_node_id_to_cstr(node_id_t id) {
         return "NODE_PLUS_OR_MINUS";
     case NODE_SECTION_DIRECTIVE:
         return "NODE_SECTION_DIRECTIVE";
+    case NODE_IMPORT_DIRECTIVE:
+        return "NODE_IMPORT_DIRECTIVE";
+    case NODE_EXPORT_DIRECTIVE:
+        return "NODE_EXPORT_DIRECTIVE";
     case NODE_REGISTER:
         return "NODE_REGISTER";
     case NODE_SECTION:
@@ -159,6 +163,10 @@ const char *ast_node_id_to_cstr(node_id_t id) {
         return "NODE_DOT";
     case NODE_NEWLINE:
         return "NODE_NEWLINE";
+    case NODE_IMPORT:
+        return "NODE_IMPORT";
+    case NODE_EXPORT:
+        return "NODE_EXPORT";
     }
     assert(!"Unreachable, weird node id" && id);
     __builtin_unreachable();
diff --git a/src/ast.h b/src/ast.h
index 038d88e..0751b80 100644
--- a/src/ast.h
+++ b/src/ast.h
@@ -29,10 +29,14 @@ typedef enum node_id {
     NODE_REGISTER_OFFSET,
     NODE_PLUS_OR_MINUS,
     NODE_SECTION_DIRECTIVE,
+    NODE_IMPORT_DIRECTIVE,
+    NODE_EXPORT_DIRECTIVE,
 
     // Validated primitives
     NODE_REGISTER,
     NODE_SECTION,
+    NODE_IMPORT,
+    NODE_EXPORT,
 
     // Primitive nodes
     NODE_IDENTIFIER,
diff --git a/src/parser/parser.c b/src/parser/parser.c
index 7e60c81..05dcbf0 100644
--- a/src/parser/parser.c
+++ b/src/parser/parser.c
@@ -119,8 +119,24 @@ parse_result_t parse_section_directive(tokenlist_entry_t *current) {
     return parse_consecutive(current, NODE_SECTION_DIRECTIVE, parsers);
 }
 
+parse_result_t parse_import_directive(tokenlist_entry_t *current) {
+    parser_t parsers[] = {parse_import, parse_identifier, nullptr};
+    return parse_consecutive(current, NODE_IMPORT_DIRECTIVE, parsers);
+}
+
+parse_result_t parse_export_directive(tokenlist_entry_t *current) {
+    parser_t parsers[] = {parse_export, parse_identifier, nullptr};
+    return parse_consecutive(current, NODE_EXPORT_DIRECTIVE, parsers);
+}
+
+parse_result_t parse_directive_options(tokenlist_entry_t *current) {
+    parser_t parsers[] = {parse_section_directive, parse_import_directive,
+                          parse_export_directive, nullptr};
+    return parse_any(current, parsers);
+}
+
 parse_result_t parse_directive(tokenlist_entry_t *current) {
-    parser_t parsers[] = {parse_dot, parse_section_directive, parse_newline,
+    parser_t parsers[] = {parse_dot, parse_directive_options, parse_newline,
                           nullptr};
     return parse_consecutive(current, NODE_DIRECTIVE, parsers);
 }
diff --git a/src/parser/primitives.c b/src/parser/primitives.c
index c93231e..a8acf56 100644
--- a/src/parser/primitives.c
+++ b/src/parser/primitives.c
@@ -105,3 +105,19 @@ parse_result_t parse_section(tokenlist_entry_t *current) {
     return parse_token(current, TOKEN_IDENTIFIER, NODE_SECTION,
                        is_section_token);
 }
+
+bool is_import_token(lexer_token_t *token) {
+    return strcmp(token->value, "import") == 0;
+}
+
+parse_result_t parse_import(tokenlist_entry_t *current) {
+    return parse_token(current, TOKEN_IDENTIFIER, NODE_IMPORT, is_import_token);
+}
+
+bool is_export_token(lexer_token_t *token) {
+    return strcmp(token->value, "export") == 0;
+}
+
+parse_result_t parse_export(tokenlist_entry_t *current) {
+    return parse_token(current, TOKEN_IDENTIFIER, NODE_EXPORT, is_export_token);
+}
diff --git a/src/parser/primitives.h b/src/parser/primitives.h
index 46ed4ef..6c5d04a 100644
--- a/src/parser/primitives.h
+++ b/src/parser/primitives.h
@@ -27,5 +27,7 @@ parse_result_t parse_label_reference(tokenlist_entry_t *current);
  */
 parse_result_t parse_register(tokenlist_entry_t *current);
 parse_result_t parse_section(tokenlist_entry_t *current);
+parse_result_t parse_import(tokenlist_entry_t *current);
+parse_result_t parse_export(tokenlist_entry_t *current);
 
 #endif // INCLUDE_PARSER_PRIMITIVES_H_
-- 
2.47.2


From 8c0e9926c5fd7bfee0c6c1b9e3b40636a818c674 Mon Sep 17 00:00:00 2001
From: omicron <omicron.me@protonmail.com>
Date: Mon, 7 Apr 2025 10:50:57 +0200
Subject: [PATCH 03/14] Make main properly return with failure on parsing
 errors

---
 src/main.c | 43 +++++++++++++++++++++++++------------------
 1 file changed, 25 insertions(+), 18 deletions(-)

diff --git a/src/main.c b/src/main.c
index 7b09662..6825254 100644
--- a/src/main.c
+++ b/src/main.c
@@ -32,21 +32,22 @@ void print_text(tokenlist_t *list) {
     }
 }
 
-void print_ast(tokenlist_t *list) {
+error_t *print_ast(tokenlist_t *list) {
     parse_result_t result = parse(list->head);
-    if (result.err) {
-        puts(result.err->message);
-        error_free(result.err);
-        return;
-    }
+    if (result.err)
+        return result.err;
+
     ast_node_print(result.node);
 
     if (result.next != nullptr) {
         puts("First unparsed token:");
         lexer_token_print(&result.next->token);
     }
-
     ast_node_free(result.node);
+    if (result.next != nullptr) {
+        return errorf("did not parse entire input token stream");
+    }
+    return nullptr;
 }
 
 int get_execution_mode(int argc, char *argv[]) {
@@ -63,6 +64,20 @@ int get_execution_mode(int argc, char *argv[]) {
     return MODE_AST;
 }
 
+error_t *do_action(mode_t mode, tokenlist_t *list) {
+    switch (mode) {
+    case MODE_TOKENS:
+        print_tokens(list);
+        return nullptr;
+    case MODE_TEXT:
+        print_text(list);
+        return nullptr;
+    case MODE_AST:
+        return print_ast(list);
+    }
+    __builtin_unreachable();
+}
+
 int main(int argc, char *argv[]) {
     mode_t mode = get_execution_mode(argc, argv);
     char *filename = argv[2];
@@ -81,17 +96,9 @@ int main(int argc, char *argv[]) {
     if (err)
         goto cleanup_tokens;
 
-    switch (mode) {
-    case MODE_TOKENS:
-        print_tokens(list);
-        break;
-    case MODE_TEXT:
-        print_text(list);
-        break;
-    case MODE_AST:
-        print_ast(list);
-        break;
-    }
+    err = do_action(mode, list);
+    if (err)
+        goto cleanup_tokens;
 
     tokenlist_free(list);
     error_free(err);
-- 
2.47.2


From ef22c0b6200d1de8c58f1bffec448bdae59eb4b2 Mon Sep 17 00:00:00 2001
From: omicron <omicron.me@protonmail.com>
Date: Mon, 7 Apr 2025 10:52:49 +0200
Subject: [PATCH 04/14] Add .import and .export to the input test file

---
 tests/input/valid.asm | 5 +++++
 1 file changed, 5 insertions(+)

diff --git a/tests/input/valid.asm b/tests/input/valid.asm
index 274b656..d56c48d 100644
--- a/tests/input/valid.asm
+++ b/tests/input/valid.asm
@@ -2,6 +2,9 @@
 
 ; Small valid code snippet that should contain all different AST nodes
 
+.export _start
+.import exit
+
 _start:
     mov eax, ebx
     lea eax, [eax + ebx * 4 + 8]
@@ -19,3 +22,5 @@ _start:
     push 0xffff:64
     push 0o777:16
     push 0b0001:16
+    mov rax, 0
+    call exit
-- 
2.47.2


From 2a7bb479ac6f6f22566378322c34f93172adfcb9 Mon Sep 17 00:00:00 2001
From: omicron <omicron.me@protonmail.com>
Date: Thu, 3 Apr 2025 03:43:56 +0200
Subject: [PATCH 05/14] initial symbol table implementation

---
 src/encoder/symbols.c | 159 ++++++++++++++++++++++++++++++++++++++++++
 src/encoder/symbols.h |  46 ++++++++++++
 2 files changed, 205 insertions(+)
 create mode 100644 src/encoder/symbols.c
 create mode 100644 src/encoder/symbols.h

diff --git a/src/encoder/symbols.c b/src/encoder/symbols.c
new file mode 100644
index 0000000..29f5330
--- /dev/null
+++ b/src/encoder/symbols.c
@@ -0,0 +1,159 @@
+#include "symbols.h"
+#include "../error.h"
+#include <assert.h>
+#include <stdlib.h>
+#include <string.h>
+
+constexpr size_t symbol_table_default_cap = 64;
+constexpr size_t symbol_table_max_cap = 1 << 16;
+
+error_t *const err_symbol_table_invalid_node = &(error_t){
+    .message = "Unexpected node id when adding symbol to symbol table"};
+error_t *const err_symbol_table_max_cap = &(error_t){
+    .message = "Failed to increase symbol table length, max capacity reached"};
+error_t *const err_symbol_table_incompatible_symbols =
+    &(error_t){.message = "Failed to update symbol with incompatible kind"};
+
+error_t *symbol_table_alloc(symbol_table_t **output) {
+    *output = nullptr;
+
+    symbol_table_t *table = calloc(1, sizeof(symbol_table_t));
+    if (table == nullptr)
+        return err_allocation_failed;
+
+    table->symbols = calloc(symbol_table_default_cap, sizeof(symbol_t));
+    if (table->symbols == nullptr) {
+        free(table);
+        return err_allocation_failed;
+    }
+
+    table->cap = symbol_table_default_cap;
+    table->len = 0;
+
+    *output = table;
+    return nullptr;
+}
+
+void symbol_table_free(symbol_table_t *table) {
+    free(table->symbols);
+    free(table);
+}
+
+error_t *symbol_table_grow_cap(symbol_table_t *table) {
+    if (table->cap >= symbol_table_max_cap)
+        return err_symbol_table_max_cap;
+
+    size_t new_cap = table->cap * 2;
+    symbol_t *new_symbols = realloc(table->symbols, new_cap * sizeof(symbol_t));
+    if (new_symbols == nullptr)
+        return err_allocation_failed;
+
+    table->symbols = new_symbols;
+    table->cap = new_cap;
+
+    return nullptr;
+}
+
+error_t *symbol_table_get_node_info(ast_node_t *node, symbol_kind_t *kind,
+                                    char **name) {
+    switch (node->id) {
+    case NODE_LABEL:
+        *kind = SYMBOL_LOCAL;
+        *name = node->children[0]->token_entry->token.value;
+        return nullptr;
+    case NODE_LABEL_REFERENCE:
+        *kind = SYMBOL_REFERENCE;
+        *name = node->token_entry->token.value;
+        return nullptr;
+    case NODE_IMPORT_DIRECTIVE:
+        *kind = SYMBOL_IMPORT;
+        *name = node->children[1]->token_entry->token.value;
+        return nullptr;
+    case NODE_EXPORT_DIRECTIVE:
+        *kind = SYMBOL_EXPORT;
+        *name = node->children[1]->token_entry->token.value;
+        return nullptr;
+    default:
+        return err_symbol_table_invalid_node;
+    }
+    __builtin_unreachable();
+}
+
+/*
+old  \  new  | REFERENCE | LOCAL    | IMPORT   | EXPORT   |
+-------------|-----------|----------|----------|----------|
+REFERENCE    |           | replace  | replace  | replace  |
+-------------|-----------|----------|----------|----------|
+LOCAL        |           |          |   ERR    | replace  |
+-------------|-----------|----------|----------|----------|
+IMPORT       |           |          |          |   ERR    |
+-------------|-----------|----------|----------|----------|
+EXPORT       |           |          |   ERR    |          |
+-------------|-----------|----------|----------|----------|
+*/
+
+bool symbol_table_should_update(symbol_kind_t old, symbol_kind_t new) {
+    if (old == SYMBOL_REFERENCE)
+        return new != SYMBOL_REFERENCE;
+    if (old == SYMBOL_LOCAL)
+        return new == SYMBOL_EXPORT;
+    return false;
+}
+
+bool symbol_table_should_error(symbol_kind_t old, symbol_kind_t new) {
+    if (new == SYMBOL_IMPORT)
+        return old == SYMBOL_LOCAL || old == SYMBOL_EXPORT;
+    if (new == SYMBOL_EXPORT)
+        return old == SYMBOL_IMPORT;
+    return false;
+}
+
+/**
+ * @pre The symbol _must not_ already be in the table.
+ */
+error_t *symbol_table_add(symbol_table_t *table, char *name, symbol_kind_t kind,
+                          ast_node_t *node) {
+    if (table->len >= table->cap) {
+        error_t *err = symbol_table_grow_cap(table);
+        if (err)
+            return err;
+    }
+
+    table->symbols[table->len] = (symbol_t){
+        .name = name,
+        .kind = kind,
+        .node = node,
+    };
+
+    table->len += 1;
+
+    return nullptr;
+}
+
+error_t *symbol_table_update(symbol_table_t *table, ast_node_t *node) {
+    char *name;
+    symbol_kind_t kind;
+    error_t *err = symbol_table_get_node_info(node, &kind, &name);
+    if (err)
+        return err;
+
+    symbol_t *symbol = symbol_table_lookup(table, name);
+    if (!symbol)
+        return symbol_table_add(table, name, kind, node);
+    if (symbol_table_should_error(symbol->kind, kind))
+        return err_symbol_table_incompatible_symbols;
+    if (symbol_table_should_update(symbol->kind, kind)) {
+        symbol->name = name;
+        symbol->kind = kind;
+        symbol->node = node;
+    }
+    return nullptr;
+}
+
+symbol_t *symbol_table_lookup(symbol_table_t *table, const char *name) {
+    for (size_t i = 0; i < table->len; ++i) {
+        if (strcmp(table->symbols[i].name, name) == 0)
+            return &table->symbols[i];
+    }
+    return nullptr;
+}
diff --git a/src/encoder/symbols.h b/src/encoder/symbols.h
new file mode 100644
index 0000000..9c4e7f7
--- /dev/null
+++ b/src/encoder/symbols.h
@@ -0,0 +1,46 @@
+#ifndef INCLUDE_ENCODER_SYMBOLS_H_
+#define INCLUDE_ENCODER_SYMBOLS_H_
+
+#include "../ast.h"
+
+extern error_t *const err_symbol_table_invalid_node;
+extern error_t *const err_symbol_table_max_cap;
+extern error_t *const err_symbol_table_incompatible_symbols;
+
+typedef enum symbol_kind {
+    SYMBOL_REFERENCE,
+    SYMBOL_LOCAL,
+    SYMBOL_EXPORT,
+    SYMBOL_IMPORT,
+} symbol_kind_t;
+
+/**
+ * Represent a symbol in the program
+ *
+ * Symbols with the same name can only be in the table once. IMPORT or EXPORT
+ * symbols take precedence over REFERENCE symbols. If any reference symbols
+ * remain after the first encoding pass this indicates an error. Trying to add
+ * an IMPORT or EXPORT symbol if the same name already exists as the other kind
+ * is an error.
+ *
+ * This symbol table never taken ownership of the name string, it's lifted
+ * straight from the node->token.value.
+ */
+typedef struct symbol {
+    char *name;
+    symbol_kind_t kind;
+    ast_node_t *node;
+} symbol_t;
+
+typedef struct symbol_table {
+    size_t cap;
+    size_t len;
+    symbol_t *symbols;
+} symbol_table_t;
+
+error_t *symbol_table_alloc(symbol_table_t **table);
+void symbol_table_free(symbol_table_t *table);
+error_t *symbol_table_update(symbol_table_t *table, ast_node_t *node);
+symbol_t *symbol_table_lookup(symbol_table_t *table, const char *name);
+
+#endif // INCLUDE_ENCODER_SYMBOLS_H_
-- 
2.47.2


From ac14925a0ad5656efbf724794790e32a7bd36e07 Mon Sep 17 00:00:00 2001
From: omicron <omicron.me@protonmail.com>
Date: Tue, 8 Apr 2025 20:38:08 +0200
Subject: [PATCH 06/14] Add symbols tests

---
 tests/input/manysymbols.asm |  65 +++++++
 tests/input/symbols.asm     |  12 ++
 tests/main.c                |   2 +
 tests/symbols.c             | 351 ++++++++++++++++++++++++++++++++++++
 4 files changed, 430 insertions(+)
 create mode 100644 tests/input/manysymbols.asm
 create mode 100644 tests/input/symbols.asm
 create mode 100644 tests/symbols.c

diff --git a/tests/input/manysymbols.asm b/tests/input/manysymbols.asm
new file mode 100644
index 0000000..76a0ba0
--- /dev/null
+++ b/tests/input/manysymbols.asm
@@ -0,0 +1,65 @@
+lbl_0:  ; 65 symbols used for testing growing the symbols table
+lbl_1:
+lbl_2:
+lbl_3:
+lbl_4:
+lbl_5:
+lbl_6:
+lbl_7:
+lbl_8:
+lbl_9:
+lbl_10:
+lbl_11:
+lbl_12:
+lbl_13:
+lbl_14:
+lbl_15:
+lbl_16:
+lbl_17:
+lbl_18:
+lbl_19:
+lbl_20:
+lbl_21:
+lbl_22:
+lbl_23:
+lbl_24:
+lbl_25:
+lbl_26:
+lbl_27:
+lbl_28:
+lbl_29:
+lbl_30:
+lbl_31:
+lbl_32:
+lbl_33:
+lbl_34:
+lbl_35:
+lbl_36:
+lbl_37:
+lbl_38:
+lbl_39:
+lbl_40:
+lbl_41:
+lbl_42:
+lbl_43:
+lbl_44:
+lbl_45:
+lbl_46:
+lbl_47:
+lbl_48:
+lbl_49:
+lbl_50:
+lbl_51:
+lbl_52:
+lbl_53:
+lbl_54:
+lbl_55:
+lbl_56:
+lbl_57:
+lbl_58:
+lbl_59:
+lbl_60:
+lbl_61:
+lbl_62:
+lbl_63:
+lbl_64:
diff --git a/tests/input/symbols.asm b/tests/input/symbols.asm
new file mode 100644
index 0000000..921d90a
--- /dev/null
+++ b/tests/input/symbols.asm
@@ -0,0 +1,12 @@
+.import test
+.export test
+test:
+    call test
+.import more
+.export more
+more:
+    call more
+.import other
+.export other
+other:
+    call other
diff --git a/tests/main.c b/tests/main.c
index c94d54d..cf7162f 100644
--- a/tests/main.c
+++ b/tests/main.c
@@ -3,12 +3,14 @@
 extern MunitTest ast_tests[];
 extern MunitTest lexer_tests[];
 extern MunitTest regression_tests[];
+extern MunitTest symbols_tests[];
 
 int main(int argc, char *argv[MUNIT_ARRAY_PARAM(argc + 1)]) {
     MunitSuite suites[] = {
         {"/regression", regression_tests, nullptr, 1, MUNIT_SUITE_OPTION_NONE},
         {"/ast",        ast_tests,        nullptr, 1, MUNIT_SUITE_OPTION_NONE},
         {"/lexer",      lexer_tests,      nullptr, 1, MUNIT_SUITE_OPTION_NONE},
+        {"/symbols",    symbols_tests,    nullptr, 1, MUNIT_SUITE_OPTION_NONE},
         {nullptr,       nullptr,          nullptr, 0, MUNIT_SUITE_OPTION_NONE},
     };
 
diff --git a/tests/symbols.c b/tests/symbols.c
new file mode 100644
index 0000000..f844e34
--- /dev/null
+++ b/tests/symbols.c
@@ -0,0 +1,351 @@
+#include "../src/encoder/symbols.h"
+#include "../src/ast.h"
+#include "../src/error.h"
+#include "../src/lexer.h"
+#include "../src/parser/parser.h"
+#include "munit.h"
+#include <string.h>
+
+void symbols_setup_test(ast_node_t **node, tokenlist_t **list, char *path) {
+    lexer_t *lex = &(lexer_t){};
+    lexer_open(lex, path);
+    tokenlist_alloc(list);
+    tokenlist_fill(*list, lex);
+    parse_result_t result = parse((*list)->head);
+    lexer_close(lex);
+
+    *node = result.node;
+}
+
+MunitResult test_symbol_table_alloc(const MunitParameter params[], void *data) {
+    (void)params;
+    (void)data;
+
+    symbol_table_t *table = nullptr;
+    error_t *err = symbol_table_alloc(&table);
+
+    munit_assert_ptr_not_null(table);
+    munit_assert_ptr_null(err);
+    munit_assert_size(table->cap, ==, 64); // Default capacity
+    munit_assert_size(table->len, ==, 0);
+    munit_assert_ptr_not_null(table->symbols);
+
+    symbol_table_free(table);
+    return MUNIT_OK;
+}
+
+MunitResult test_symbol_table_lookup_empty(const MunitParameter params[], void *data) {
+    (void)params;
+    (void)data;
+
+    symbol_table_t *table = nullptr;
+    symbol_table_alloc(&table);
+
+    symbol_t *symbol = symbol_table_lookup(table, "nonexistent");
+    munit_assert_ptr_null(symbol);
+
+    symbol_table_free(table);
+    return MUNIT_OK;
+}
+
+MunitResult test_symbol_add_reference(const MunitParameter params[], void *data) {
+    (void)params;
+    (void)data;
+    ast_node_t *root;
+    tokenlist_t *list;
+    symbol_table_t *table = nullptr;
+    symbols_setup_test(&root, &list, "tests/input/symbols.asm");
+    symbol_table_alloc(&table);
+
+    ast_node_t *reference = root->children[3]->children[1]->children[0]->children[0];
+    munit_assert_int(reference->id, ==, NODE_LABEL_REFERENCE);
+    munit_assert_size(table->len, ==, 0);
+
+    error_t *err = symbol_table_update(table, reference);
+    munit_assert_null(err);
+    munit_assert_size(table->len, ==, 1);
+
+    symbol_t *symbol = symbol_table_lookup(table, "test");
+    munit_assert_not_null(symbol);
+    munit_assert_int(SYMBOL_REFERENCE, ==, symbol->kind);
+    munit_assert_ptr_equal(reference, symbol->node);
+    munit_assert_string_equal(symbol->name, "test");
+
+    symbol_table_free(table);
+    ast_node_free(root);
+    tokenlist_free(list);
+    return MUNIT_OK;
+}
+
+MunitResult test_symbol_add_label(const MunitParameter params[], void *data) {
+    (void)params;
+    (void)data;
+    ast_node_t *root;
+    tokenlist_t *list;
+    symbol_table_t *table = nullptr;
+    symbols_setup_test(&root, &list, "tests/input/symbols.asm");
+    symbol_table_alloc(&table);
+
+    ast_node_t *label = root->children[2];
+    munit_assert_int(label->id, ==, NODE_LABEL);
+    munit_assert_size(table->len, ==, 0);
+
+    error_t *err = symbol_table_update(table, label);
+    munit_assert_null(err);
+    munit_assert_size(table->len, ==, 1);
+
+    symbol_t *symbol = symbol_table_lookup(table, "test");
+    munit_assert_not_null(symbol);
+    munit_assert_int(SYMBOL_LOCAL, ==, symbol->kind);
+    munit_assert_ptr_equal(label, symbol->node);
+    munit_assert_string_equal(symbol->name, "test");
+
+    symbol_table_free(table);
+    ast_node_free(root);
+    tokenlist_free(list);
+    return MUNIT_OK;
+}
+
+MunitResult test_symbol_add_import(const MunitParameter params[], void *data) {
+    (void)params;
+    (void)data;
+    ast_node_t *root;
+    tokenlist_t *list;
+    symbol_table_t *table = nullptr;
+    symbols_setup_test(&root, &list, "tests/input/symbols.asm");
+    symbol_table_alloc(&table);
+
+    ast_node_t *import_directive = root->children[0]->children[1];
+    munit_assert_int(import_directive->id, ==, NODE_IMPORT_DIRECTIVE);
+    munit_assert_size(table->len, ==, 0);
+
+    error_t *err = symbol_table_update(table, import_directive);
+    munit_assert_null(err);
+    munit_assert_size(table->len, ==, 1);
+
+    symbol_t *symbol = symbol_table_lookup(table, "test");
+    munit_assert_not_null(symbol);
+    munit_assert_int(SYMBOL_IMPORT, ==, symbol->kind);
+    munit_assert_ptr_equal(import_directive, symbol->node);
+    munit_assert_string_equal(symbol->name, "test");
+
+    symbol_table_free(table);
+    ast_node_free(root);
+    tokenlist_free(list);
+    return MUNIT_OK;
+}
+
+void test_symbol_update(const char *name, ast_node_t *first, symbol_kind_t first_kind, ast_node_t *second,
+                        symbol_kind_t second_kind, bool should_succeed, bool should_update) {
+    symbol_table_t *table = nullptr;
+    symbol_table_alloc(&table);
+
+    munit_assert_size(table->len, ==, 0);
+    error_t *err = symbol_table_update(table, first);
+    munit_assert_null(err);
+    munit_assert_size(table->len, ==, 1);
+
+    symbol_t *symbol = symbol_table_lookup(table, name);
+    munit_assert_not_null(symbol);
+    munit_assert_int(first_kind, ==, symbol->kind);
+    munit_assert_ptr_equal(first, symbol->node);
+    munit_assert_string_equal(symbol->name, name);
+
+    err = symbol_table_update(table, second);
+    if (should_succeed)
+        munit_assert_null(err);
+    else
+        munit_assert_ptr_equal(err, err_symbol_table_incompatible_symbols);
+    munit_assert_size(table->len, ==, 1);
+
+    symbol = symbol_table_lookup(table, name);
+    if (should_update) {
+        munit_assert_not_null(symbol);
+        munit_assert_int(second_kind, ==, symbol->kind);
+        munit_assert_ptr_equal(second, symbol->node);
+        munit_assert_string_equal(symbol->name, name);
+    } else {
+        munit_assert_not_null(symbol);
+        munit_assert_int(first_kind, ==, symbol->kind);
+        munit_assert_ptr_equal(first, symbol->node);
+        munit_assert_string_equal(symbol->name, name);
+    }
+
+    symbol_table_free(table);
+}
+
+MunitResult test_symbol_upgrade_valid(const MunitParameter params[], void *data) {
+    ast_node_t *root;
+    tokenlist_t *list;
+
+    symbols_setup_test(&root, &list, "tests/input/symbols.asm");
+
+    ast_node_t *reference = root->children[3]->children[1]->children[0]->children[0];
+    ast_node_t *label = root->children[2];
+    ast_node_t *import_directive = root->children[0]->children[1];
+    ast_node_t *export_directive = root->children[1]->children[1];
+
+    // real upgrades
+    test_symbol_update("test", reference, SYMBOL_REFERENCE, label, SYMBOL_LOCAL, true, true);
+    test_symbol_update("test", reference, SYMBOL_REFERENCE, import_directive, SYMBOL_IMPORT, true, true);
+    test_symbol_update("test", reference, SYMBOL_REFERENCE, export_directive, SYMBOL_EXPORT, true, true);
+    test_symbol_update("test", label, SYMBOL_LOCAL, export_directive, SYMBOL_EXPORT, true, true);
+
+    // identity upgrades
+    test_symbol_update("test", reference, SYMBOL_REFERENCE, reference, SYMBOL_REFERENCE, true, false);
+    test_symbol_update("test", label, SYMBOL_LOCAL, label, SYMBOL_LOCAL, true, false);
+    test_symbol_update("test", import_directive, SYMBOL_IMPORT, import_directive, SYMBOL_IMPORT, true, false);
+    test_symbol_update("test", export_directive, SYMBOL_EXPORT, export_directive, SYMBOL_EXPORT, true, false);
+
+    // downgrades that are allowed and ignored
+    test_symbol_update("test", label, SYMBOL_LOCAL, reference, SYMBOL_REFERENCE, true, false);
+    test_symbol_update("test", import_directive, SYMBOL_IMPORT, reference, SYMBOL_REFERENCE, true, false);
+    test_symbol_update("test", export_directive, SYMBOL_EXPORT, reference, SYMBOL_REFERENCE, true, false);
+    test_symbol_update("test", export_directive, SYMBOL_EXPORT, label, SYMBOL_LOCAL, true, false);
+    test_symbol_update("test", import_directive, SYMBOL_IMPORT, label, SYMBOL_LOCAL, true, false);
+
+    ast_node_free(root);
+    tokenlist_free(list);
+    return MUNIT_OK;
+}
+
+MunitResult test_symbol_upgrade_invalid(const MunitParameter params[], void *data) {
+    ast_node_t *root;
+    tokenlist_t *list;
+
+    symbols_setup_test(&root, &list, "tests/input/symbols.asm");
+
+    ast_node_t *reference = root->children[3]->children[1]->children[0]->children[0];
+    ast_node_t *label = root->children[2];
+    ast_node_t *import_directive = root->children[0]->children[1];
+    ast_node_t *export_directive = root->children[1]->children[1];
+
+    // invalid upgrades
+    test_symbol_update("test", label, SYMBOL_LOCAL, import_directive, SYMBOL_IMPORT, false, false);
+    test_symbol_update("test", export_directive, SYMBOL_EXPORT, import_directive, SYMBOL_IMPORT, false, false);
+    test_symbol_update("test", import_directive, SYMBOL_IMPORT, export_directive, SYMBOL_EXPORT, false, false);
+
+    ast_node_free(root);
+    tokenlist_free(list);
+    return MUNIT_OK;
+}
+
+MunitResult test_symbol_add_export(const MunitParameter params[], void *data) {
+    (void)params;
+    (void)data;
+    ast_node_t *root;
+    tokenlist_t *list;
+    symbol_table_t *table = nullptr;
+    symbols_setup_test(&root, &list, "tests/input/symbols.asm");
+    symbol_table_alloc(&table);
+
+    ast_node_t *export_directive = root->children[1]->children[1];
+    munit_assert_int(export_directive->id, ==, NODE_EXPORT_DIRECTIVE);
+    munit_assert_size(table->len, ==, 0);
+
+    error_t *err = symbol_table_update(table, export_directive);
+    munit_assert_null(err);
+    munit_assert_size(table->len, ==, 1);
+
+    symbol_t *symbol = symbol_table_lookup(table, "test");
+    munit_assert_not_null(symbol);
+    munit_assert_int(SYMBOL_EXPORT, ==, symbol->kind);
+    munit_assert_ptr_equal(export_directive, symbol->node);
+    munit_assert_string_equal(symbol->name, "test");
+
+    symbol_table_free(table);
+    ast_node_free(root);
+    tokenlist_free(list);
+    return MUNIT_OK;
+}
+
+MunitResult test_symbol_table_growth(const MunitParameter params[], void *data) {
+    (void)params;
+    (void)data;
+    ast_node_t *root;
+    tokenlist_t *list;
+    symbol_table_t *table = nullptr;
+
+    // Set up with our manysymbols.asm file
+    symbols_setup_test(&root, &list, "tests/input/manysymbols.asm");
+    symbol_table_alloc(&table);
+
+    // Initial capacity should be the default (64)
+    munit_assert_size(table->cap, ==, 64);
+    munit_assert_size(table->len, ==, 0);
+
+    // Add the first 64 labels (indices 0-63)
+    size_t initial_cap = table->cap;
+    for (size_t i = 0; i < 64; i++) {
+        ast_node_t *label = root->children[i];
+        munit_assert_int(label->id, ==, NODE_LABEL);
+
+        error_t *err = symbol_table_update(table, label);
+        munit_assert_null(err);
+        munit_assert_size(table->len, ==, i + 1);
+
+        // Capacity should remain the same for the first 64 labels
+        munit_assert_size(table->cap, ==, initial_cap);
+    }
+
+    // Now add the 65th label (index 64), which should trigger growth
+    ast_node_t *final_label = root->children[64];
+    munit_assert_int(final_label->id, ==, NODE_LABEL);
+
+    error_t *err = symbol_table_update(table, final_label);
+    munit_assert_null(err);
+    munit_assert_size(table->len, ==, 65);
+
+    // Capacity should have doubled
+    munit_assert_size(table->cap, ==, initial_cap * 2);
+
+    // Validate we can look up all the symbols
+    for (size_t i = 0; i <= 64; i++) {
+        char name[10];
+        sprintf(name, "lbl_%zu", i);
+
+        symbol_t *symbol = symbol_table_lookup(table, name);
+        munit_assert_not_null(symbol);
+        munit_assert_int(SYMBOL_LOCAL, ==, symbol->kind);
+        munit_assert_string_equal(symbol->name, name);
+    }
+
+    symbol_table_free(table);
+    ast_node_free(root);
+    tokenlist_free(list);
+    return MUNIT_OK;
+}
+
+MunitResult test_symbol_invalid_node(const MunitParameter params[], void *data) {
+    (void)params;
+    (void)data;
+    ast_node_t *root;
+    tokenlist_t *list;
+    symbol_table_t *table = nullptr;
+    symbols_setup_test(&root, &list, "tests/input/symbols.asm");
+    symbol_table_alloc(&table);
+
+    munit_assert_size(table->len, ==, 0);
+    error_t *err = symbol_table_update(table, root);
+    munit_assert_ptr_equal(err, err_symbol_table_invalid_node);
+    munit_assert_size(table->len, ==, 0);
+
+    symbol_table_free(table);
+    ast_node_free(root);
+    tokenlist_free(list);
+    return MUNIT_OK;
+}
+
+MunitTest symbols_tests[] = {
+    {"/table_alloc",        test_symbol_table_alloc,        nullptr, nullptr, MUNIT_TEST_OPTION_NONE, nullptr},
+    {"/table_lookup_empty", test_symbol_table_lookup_empty, nullptr, nullptr, MUNIT_TEST_OPTION_NONE, nullptr},
+    {"/add_reference",      test_symbol_add_reference,      nullptr, nullptr, MUNIT_TEST_OPTION_NONE, nullptr},
+    {"/add_label",          test_symbol_add_label,          nullptr, nullptr, MUNIT_TEST_OPTION_NONE, nullptr},
+    {"/add_import",         test_symbol_add_import,         nullptr, nullptr, MUNIT_TEST_OPTION_NONE, nullptr},
+    {"/add_export",         test_symbol_add_export,         nullptr, nullptr, MUNIT_TEST_OPTION_NONE, nullptr},
+    {"/upgrade_valid",      test_symbol_upgrade_valid,      nullptr, nullptr, MUNIT_TEST_OPTION_NONE, nullptr},
+    {"/upgrade_invalid",    test_symbol_upgrade_invalid,    nullptr, nullptr, MUNIT_TEST_OPTION_NONE, nullptr},
+    {"/table_growth",       test_symbol_table_growth,       nullptr, nullptr, MUNIT_TEST_OPTION_NONE, nullptr},
+    {"/invalid_node",       test_symbol_invalid_node,       nullptr, nullptr, MUNIT_TEST_OPTION_NONE, nullptr},
+    {nullptr,               nullptr,                        nullptr, nullptr, MUNIT_TEST_OPTION_NONE, nullptr}
+};
-- 
2.47.2


From d59559d327cf093080774cb113a2d9026b3ae4ad Mon Sep 17 00:00:00 2001
From: omicron <omicron.me@protonmail.com>
Date: Sun, 13 Apr 2025 15:17:08 +0200
Subject: [PATCH 07/14] Add registers data table

Change the validated primitive parse_register so that it uses the data
table instead
---
 src/data/registers.c    | 92 +++++++++++++++++++++++++++++++++++++++++
 src/data/registers.h    | 82 ++++++++++++++++++++++++++++++++++++
 src/parser/primitives.c | 17 +-------
 3 files changed, 176 insertions(+), 15 deletions(-)
 create mode 100644 src/data/registers.c
 create mode 100644 src/data/registers.h

diff --git a/src/data/registers.c b/src/data/registers.c
new file mode 100644
index 0000000..075c1ed
--- /dev/null
+++ b/src/data/registers.c
@@ -0,0 +1,92 @@
+#include "registers.h"
+
+register_data_t *const registers[] = {
+    // Instruction pointer
+    &(register_data_t){"rip",  REG_RIP, OPERAND_SIZE_64},
+    &(register_data_t){"eip",  REG_RIP, OPERAND_SIZE_32},
+    &(register_data_t){"ip",   REG_RIP, OPERAND_SIZE_16},
+
+    // 64-bit general purpose registers
+    &(register_data_t){"rax",  REG_A,   OPERAND_SIZE_64},
+    &(register_data_t){"rcx",  REG_C,   OPERAND_SIZE_64},
+    &(register_data_t){"rdx",  REG_D,   OPERAND_SIZE_64},
+    &(register_data_t){"rbx",  REG_B,   OPERAND_SIZE_64},
+    &(register_data_t){"rsp",  REG_SP,  OPERAND_SIZE_64},
+    &(register_data_t){"rbp",  REG_BP,  OPERAND_SIZE_64},
+    &(register_data_t){"rsi",  REG_SI,  OPERAND_SIZE_64},
+    &(register_data_t){"rdi",  REG_DI,  OPERAND_SIZE_64},
+    &(register_data_t){"r8",   REG_8,   OPERAND_SIZE_64},
+    &(register_data_t){"r9",   REG_9,   OPERAND_SIZE_64},
+    &(register_data_t){"r10",  REG_10,  OPERAND_SIZE_64},
+    &(register_data_t){"r11",  REG_11,  OPERAND_SIZE_64},
+    &(register_data_t){"r12",  REG_12,  OPERAND_SIZE_64},
+    &(register_data_t){"r13",  REG_13,  OPERAND_SIZE_64},
+    &(register_data_t){"r14",  REG_14,  OPERAND_SIZE_64},
+    &(register_data_t){"r15",  REG_15,  OPERAND_SIZE_64},
+
+    // 32-bit general purpose registers
+    &(register_data_t){"eax",  REG_A,   OPERAND_SIZE_32},
+    &(register_data_t){"ecx",  REG_C,   OPERAND_SIZE_32},
+    &(register_data_t){"edx",  REG_D,   OPERAND_SIZE_32},
+    &(register_data_t){"ebx",  REG_B,   OPERAND_SIZE_32},
+    &(register_data_t){"esp",  REG_SP,  OPERAND_SIZE_32},
+    &(register_data_t){"ebp",  REG_BP,  OPERAND_SIZE_32},
+    &(register_data_t){"esi",  REG_SI,  OPERAND_SIZE_32},
+    &(register_data_t){"edi",  REG_DI,  OPERAND_SIZE_32},
+    &(register_data_t){"r8d",  REG_8,   OPERAND_SIZE_32},
+    &(register_data_t){"r9d",  REG_9,   OPERAND_SIZE_32},
+    &(register_data_t){"r10d", REG_10,  OPERAND_SIZE_32},
+    &(register_data_t){"r11d", REG_11,  OPERAND_SIZE_32},
+    &(register_data_t){"r12d", REG_12,  OPERAND_SIZE_32},
+    &(register_data_t){"r13d", REG_13,  OPERAND_SIZE_32},
+    &(register_data_t){"r14d", REG_14,  OPERAND_SIZE_32},
+    &(register_data_t){"r15d", REG_15,  OPERAND_SIZE_32},
+
+    // 16-bit general purpose registers
+    &(register_data_t){"ax",   REG_A,   OPERAND_SIZE_16},
+    &(register_data_t){"cx",   REG_C,   OPERAND_SIZE_16},
+    &(register_data_t){"dx",   REG_D,   OPERAND_SIZE_16},
+    &(register_data_t){"bx",   REG_B,   OPERAND_SIZE_16},
+    &(register_data_t){"sp",   REG_SP,  OPERAND_SIZE_16},
+    &(register_data_t){"bp",   REG_BP,  OPERAND_SIZE_16},
+    &(register_data_t){"si",   REG_SI,  OPERAND_SIZE_16},
+    &(register_data_t){"di",   REG_DI,  OPERAND_SIZE_16},
+    &(register_data_t){"r8w",  REG_8,   OPERAND_SIZE_16},
+    &(register_data_t){"r9w",  REG_9,   OPERAND_SIZE_16},
+    &(register_data_t){"r10w", REG_10,  OPERAND_SIZE_16},
+    &(register_data_t){"r11w", REG_11,  OPERAND_SIZE_16},
+    &(register_data_t){"r12w", REG_12,  OPERAND_SIZE_16},
+    &(register_data_t){"r13w", REG_13,  OPERAND_SIZE_16},
+    &(register_data_t){"r14w", REG_14,  OPERAND_SIZE_16},
+    &(register_data_t){"r15w", REG_15,  OPERAND_SIZE_16},
+
+    // 8-bit general purpose registers (low byte)
+    &(register_data_t){"al",   REG_A,   OPERAND_SIZE_8 },
+    &(register_data_t){"cl",   REG_C,   OPERAND_SIZE_8 },
+    &(register_data_t){"dl",   REG_D,   OPERAND_SIZE_8 },
+    &(register_data_t){"bl",   REG_B,   OPERAND_SIZE_8 },
+    &(register_data_t){"spl",  REG_SP,  OPERAND_SIZE_8 },
+    &(register_data_t){"bpl",  REG_BP,  OPERAND_SIZE_8 },
+    &(register_data_t){"sil",  REG_SI,  OPERAND_SIZE_8 },
+    &(register_data_t){"dil",  REG_DI,  OPERAND_SIZE_8 },
+    &(register_data_t){"r8b",  REG_8,   OPERAND_SIZE_8 },
+    &(register_data_t){"r9b",  REG_9,   OPERAND_SIZE_8 },
+    &(register_data_t){"r10b", REG_10,  OPERAND_SIZE_8 },
+    &(register_data_t){"r11b", REG_11,  OPERAND_SIZE_8 },
+    &(register_data_t){"r12b", REG_12,  OPERAND_SIZE_8 },
+    &(register_data_t){"r13b", REG_13,  OPERAND_SIZE_8 },
+    &(register_data_t){"r14b", REG_14,  OPERAND_SIZE_8 },
+    &(register_data_t){"r15b", REG_15,  OPERAND_SIZE_8 },
+
+    // x87 floating point registers
+    &(register_data_t){"st0",  REG_ST0, OPERAND_SIZE_80},
+    &(register_data_t){"st1",  REG_ST1, OPERAND_SIZE_80},
+    &(register_data_t){"st2",  REG_ST2, OPERAND_SIZE_80},
+    &(register_data_t){"st3",  REG_ST3, OPERAND_SIZE_80},
+    &(register_data_t){"st4",  REG_ST4, OPERAND_SIZE_80},
+    &(register_data_t){"st5",  REG_ST5, OPERAND_SIZE_80},
+    &(register_data_t){"st6",  REG_ST6, OPERAND_SIZE_80},
+    &(register_data_t){"st7",  REG_ST7, OPERAND_SIZE_80},
+
+    nullptr,
+};
diff --git a/src/data/registers.h b/src/data/registers.h
new file mode 100644
index 0000000..29bb90f
--- /dev/null
+++ b/src/data/registers.h
@@ -0,0 +1,82 @@
+#ifndef INCLUDE_DATA_REGISTERS_H_
+#define INCLUDE_DATA_REGISTERS_H_
+
+typedef enum operand_size {
+    OPERAND_SIZE_INVALID = 0,
+
+    OPERAND_SIZE_8 = 1 << 0,
+    OPERAND_SIZE_16 = 1 << 1,
+    OPERAND_SIZE_32 = 1 << 2,
+    OPERAND_SIZE_64 = 1 << 3,
+
+    OPERAND_SIZE_80 = 1 << 4,
+    OPERAND_SIZE_128 = 1 << 5,
+    OPERAND_SIZE_256 = 1 << 6,
+    OPERAND_SIZE_512 = 1 << 7,
+} operand_size_t;
+
+static inline operand_size_t bits_to_operand_size(int bits) {
+    switch (bits) {
+    case 8:
+        return OPERAND_SIZE_8;
+    case 16:
+        return OPERAND_SIZE_16;
+    case 32:
+        return OPERAND_SIZE_32;
+    case 64:
+        return OPERAND_SIZE_64;
+    case 80:
+        return OPERAND_SIZE_80;
+    case 128:
+        return OPERAND_SIZE_128;
+    case 256:
+        return OPERAND_SIZE_256;
+    case 512:
+        return OPERAND_SIZE_512;
+    default:
+        return OPERAND_SIZE_INVALID;
+    }
+}
+
+typedef enum register_id {
+    // Special registers
+    REG_RIP = -1,
+
+    // General purpose registers
+    REG_A = 0x0000,
+    REG_C,
+    REG_D,
+    REG_B,
+    REG_SP,
+    REG_BP,
+    REG_SI,
+    REG_DI,
+
+    REG_8,
+    REG_9,
+    REG_10,
+    REG_11,
+    REG_12,
+    REG_13,
+    REG_14,
+    REG_15,
+
+    REG_ST0 = 0x1000,
+    REG_ST1,
+    REG_ST2,
+    REG_ST3,
+    REG_ST4,
+    REG_ST5,
+    REG_ST6,
+    REG_ST7,
+} register_id_t;
+
+typedef struct register_data {
+    const char *name;
+    register_id_t id;
+    operand_size_t size;
+} register_data_t;
+
+extern register_data_t *const registers[];
+
+#endif // INCLUDE_DATA_REGISTERS_H_
diff --git a/src/parser/primitives.c b/src/parser/primitives.c
index a8acf56..166c89a 100644
--- a/src/parser/primitives.c
+++ b/src/parser/primitives.c
@@ -1,5 +1,6 @@
 #include "primitives.h"
 #include "../ast.h"
+#include "../data/registers.h"
 #include <string.h>
 
 parse_result_t parse_identifier(tokenlist_entry_t *current) {
@@ -71,23 +72,9 @@ parse_result_t parse_label_reference(tokenlist_entry_t *current) {
                        nullptr);
 }
 
-const char *registers[] = {
-    // 64-bit registers
-    "rax", "rcx", "rdx", "rbx", "rsp", "rbp", "rsi", "rdi", "r8", "r9", "r10",
-    "r11", "r12", "r13", "r14", "r15",
-    // 32-bit registers
-    "eax", "ecx", "edx", "ebx", "esp", "ebp", "esi", "edi", "r8d", "r9d",
-    "r10d", "r11d", "r12d", "r13d", "r14d", "r15d",
-    // 16-bit registers
-    "ax", "cx", "dx", "bx", "sp", "bp", "si", "di", "r8w", "r9w", "r10w",
-    "r11w", "r12w", "r13w", "r14w", "r15w",
-    // 8-bit low registers
-    "al", "cl", "dl", "bl", "spl", "bpl", "sil", "dil", "r8b", "r9b", "r10b",
-    "r11b", "r12b", "r13b", "r14b", "r15b", nullptr};
-
 bool is_register_token(lexer_token_t *token) {
     for (size_t i = 0; registers[i] != nullptr; ++i)
-        if (strcmp(token->value, registers[i]) == 0)
+        if (strcmp(token->value, registers[i]->name) == 0)
             return true;
     return false;
 }
-- 
2.47.2


From 2cf69f5e18ef2fc57e838fdffa1a1f3fcfc67378 Mon Sep 17 00:00:00 2001
From: omicron <omicron.me@protonmail.com>
Date: Sun, 13 Apr 2025 20:35:25 +0200
Subject: [PATCH 08/14] Add initial limited opcode data

---
 src/data/opcodes.c | 145 +++++++++++++++++++++++++++++++++++++++++++++
 src/data/opcodes.h |  56 +++++++++++++++++
 2 files changed, 201 insertions(+)
 create mode 100644 src/data/opcodes.c
 create mode 100644 src/data/opcodes.h

diff --git a/src/data/opcodes.c b/src/data/opcodes.c
new file mode 100644
index 0000000..f74f68a
--- /dev/null
+++ b/src/data/opcodes.c
@@ -0,0 +1,145 @@
+#include "opcodes.h"
+
+// clang-format off
+opcode_data_t *const opcodes[] = {
+    // RET
+    &(opcode_data_t) {
+        .mnemonic = "ret",
+        .opcode = 0xC3,
+        .opcode_extension = opcode_extension_none,
+        .operand_count = 0,
+    },
+    // RET imm16
+    &(opcode_data_t) {
+        .mnemonic = "ret",
+        .opcode = 0xC2,
+        .opcode_extension = opcode_extension_none,
+        .operand_count = 1,
+        .operands = {
+            { .kind = OPERAND_IMMEDIATE, .size = OPERAND_SIZE_16 },
+        },
+    },
+    // PUSH imm8
+    &(opcode_data_t) {
+        .mnemonic = "push",
+        .opcode = 0x6A,
+        .opcode_extension = opcode_extension_none,
+        .operand_count = 1,
+        .operands = {
+            { .kind = OPERAND_IMMEDIATE, .size = OPERAND_SIZE_8},
+        },
+    },
+    // PUSH imm16
+    &(opcode_data_t) {
+        .mnemonic = "push",
+        .opcode = 0x68,
+        .opcode_extension = opcode_extension_none,
+        .operand_size_prefix = true,
+        .operand_count = 1,
+        .operands = {
+            { .kind = OPERAND_IMMEDIATE, .size = OPERAND_SIZE_16},
+        },
+    },
+    // PUSH imm32
+    &(opcode_data_t) {
+        .mnemonic = "push",
+        .opcode = 0x68,
+        .opcode_extension = opcode_extension_none,
+        .operand_size_prefix = false,
+        .operand_count = 1,
+        .operands = {
+            { .kind = OPERAND_IMMEDIATE, .size = OPERAND_SIZE_32},
+        },
+    },
+    // PUSH reg16, 
+    &(opcode_data_t) {
+        .mnemonic = "push",
+        .opcode = 0x50,
+        .opcode_extension = opcode_extension_none,
+        .encoding_class = ENCODING_OPCODE_REGISTER,
+        .operand_count = 1,
+        .operands = {
+            { .kind = OPERAND_REGISTER, .size = OPERAND_SIZE_16 },
+        },
+    },
+    // PUSH reg64
+    &(opcode_data_t) {
+        .mnemonic = "push",
+        .opcode = 0x50,
+        .opcode_extension = opcode_extension_none,
+        .encoding_class = ENCODING_OPCODE_REGISTER,
+        .operand_count = 1,
+        .operands = {
+            { .kind = OPERAND_REGISTER, .size = OPERAND_SIZE_64 },
+        },
+    },
+    // NOT reg16
+    &(opcode_data_t) {
+        .mnemonic = "not",
+        .opcode = 0xF7,
+        .opcode_extension = 2,
+        .operand_size_prefix = true,
+        .operand_count = 1,
+        .operands = {
+            { .kind = OPERAND_REGISTER, .size = OPERAND_SIZE_16 },
+        },
+    },
+    // NOT reg32
+    &(opcode_data_t) {
+        .mnemonic = "not",
+        .opcode = 0xF7,
+        .opcode_extension = 2,
+        .operand_count = 1,
+        .operands = {
+            { .kind = OPERAND_REGISTER, .size = OPERAND_SIZE_32 },
+        },
+    },
+    // NOT reg64
+    &(opcode_data_t) {
+        .mnemonic = "not",
+        .opcode = 0xF7,
+        .opcode_extension = 2,
+        .rex_w_prefix = true,
+        .operand_count = 1,
+        .operands = {
+            { .kind = OPERAND_REGISTER, .size = OPERAND_SIZE_64 },
+        },
+    },
+
+    // NEG reg16
+    &(opcode_data_t) {
+        .mnemonic = "neg",
+        .opcode = 0xF7,
+        .opcode_extension = 3,
+        .operand_size_prefix = true,
+        .operand_count = 1,
+        .operands = {
+            { .kind = OPERAND_REGISTER, .size = OPERAND_SIZE_16 },
+        },
+    },
+    // NEG reg32
+    &(opcode_data_t) {
+        .mnemonic = "neg",
+        .opcode = 0xF7,
+        .opcode_extension = 3,
+        .operand_count = 1,
+        .operands = {
+            { .kind = OPERAND_REGISTER, .size = OPERAND_SIZE_32 },
+        },
+    },
+    // NEG reg64
+    &(opcode_data_t) {
+        .mnemonic = "neg",
+        .opcode = 0xF7,
+        .opcode_extension = 3,
+        .rex_w_prefix = true,
+        .operand_count = 1,
+        .operands = {
+            { .kind = OPERAND_REGISTER, .size = OPERAND_SIZE_64 },
+        },
+    },
+
+
+    nullptr,
+};
+
diff --git a/src/data/opcodes.h b/src/data/opcodes.h
new file mode 100644
index 0000000..14c842b
--- /dev/null
+++ b/src/data/opcodes.h
@@ -0,0 +1,56 @@
+#ifndef INCLUDE_DATA_OPCODES_H_
+#define INCLUDE_DATA_OPCODES_H_
+
+#include "../data/registers.h"
+#include <stddef.h>
+#include <stdint.h>
+
+constexpr uint8_t rex_prefix = 0x40;
+constexpr uint8_t rex_prefix_w = 0x48;
+constexpr uint8_t rex_prefix_r = 0x44;
+constexpr uint8_t rex_prefix_x = 0x42;
+constexpr uint8_t rex_prefix_b = 0x41;
+
+constexpr uint8_t operand_size_prefix = 0x66;
+constexpr uint8_t memory_size_prefix = 0x67;
+constexpr uint8_t lock_prefix = 0xF0;
+constexpr uint8_t repne_prefix = 0xF2;
+constexpr uint8_t rep_prefix = 0xF3;
+
+typedef enum encoding_class {
+    ENCODING_DEFAULT,         // use modrm+sib for registers and memory, append
+                              // immediates
+    ENCODING_OPCODE_REGISTER, // encode the register in the last 3 bits of the
+                              // opcode
+} encoding_class_t;
+
+typedef enum operand_kind {
+    OPERAND_REGISTER,
+    OPERAND_MEMORY,
+    OPERAND_IMMEDIATE,
+} operand_kind_t;
+
+typedef struct operand_info {
+    operand_kind_t kind;
+    operand_size_t size;
+} operand_info_t;
+
+constexpr uint8_t opcode_extension_none = 0xFF;
+
+typedef struct opcode_data {
+    const char *mnemonic;
+
+    uint16_t opcode;
+    uint8_t opcode_extension; // 3 bits for the opcode extension in the reg
+                              // field of a modr/m byte
+    encoding_class_t encoding_class;
+    bool operand_size_prefix;
+    bool address_size_prefix;
+    bool rex_w_prefix;
+    size_t operand_count;
+    operand_info_t operands[3];
+} opcode_data_t;
+
+extern opcode_data_t *const opcodes[];
+
+#endif // INCLUDE_DATA_OPCODES_H_
-- 
2.47.2


From dcf90b72e058e24150cecefa36afb96a5a444a1e Mon Sep 17 00:00:00 2001
From: omicron <omicron.me@protonmail.com>
Date: Mon, 14 Apr 2025 23:55:09 +0200
Subject: [PATCH 09/14] Add register and number values to AST nodes

---
 src/ast.h | 18 +++++++++++++-----
 1 file changed, 13 insertions(+), 5 deletions(-)

diff --git a/src/ast.h b/src/ast.h
index 0751b80..e5cfd04 100644
--- a/src/ast.h
+++ b/src/ast.h
@@ -1,6 +1,7 @@
 #ifndef INCLUDE_SRC_AST_H_
 #define INCLUDE_SRC_AST_H_
 
+#include "data/registers.h"
 #include "error.h"
 #include "lexer.h"
 #include "tokenlist.h"
@@ -63,6 +64,16 @@ constexpr size_t node_default_children_cap = 8;
 /* 65K ought to be enough for anybody */
 constexpr size_t node_max_children_cap = 1 << 16;
 
+typedef struct number {
+    uint64_t value;
+    operand_size_t size;
+} number_t;
+
+typedef struct register_ {
+    register_id_t id;
+    operand_size_t size;
+} register_t;
+
 struct ast_node {
     node_id_t id;
     tokenlist_entry_t *token_entry;
@@ -71,11 +82,8 @@ struct ast_node {
     ast_node_t **children;
 
     union {
-        struct {
-            uint64_t value;
-            int size;
-        } integer;
-        char *name;
+        register_t reg;
+        number_t number;
     } value;
 };
 
-- 
2.47.2


From 68dcd9dccece6f410d1593dfa96127fb121372b2 Mon Sep 17 00:00:00 2001
From: omicron <omicron.me@protonmail.com>
Date: Tue, 15 Apr 2025 00:01:40 +0200
Subject: [PATCH 10/14] Add first encoding pass

First pass collects all the symbols and interprets number and register
tokens into usable data for the later passes.
---
 src/encoder/encoder.c | 211 ++++++++++++++++++++++++++++++++++++++++++
 src/encoder/encoder.h |  20 ++++
 2 files changed, 231 insertions(+)
 create mode 100644 src/encoder/encoder.c
 create mode 100644 src/encoder/encoder.h

diff --git a/src/encoder/encoder.c b/src/encoder/encoder.c
new file mode 100644
index 0000000..a43a23b
--- /dev/null
+++ b/src/encoder/encoder.c
@@ -0,0 +1,211 @@
+#include "encoder.h"
+#include "../data/opcodes.h"
+#include "symbols.h"
+#include <assert.h>
+#include <errno.h>
+#include <string.h>
+
+error_t *const err_encoder_invalid_register =
+    &(error_t){.message = "Invalid register"};
+error_t *const err_encoder_number_overflow =
+    &(error_t){.message = "Number overflows the storage"};
+error_t *const err_encoder_invalid_number_format =
+    &(error_t){.message = "Invalid number format"};
+error_t *const err_encoder_invalid_size_suffix =
+    &(error_t){.message = "Invalid number size suffix"};
+error_t *const err_encoder_unknown_symbol_reference =
+    &(error_t){.message = "Referenced an unknown symbol"};
+
+error_t *encoder_alloc(encoder_t **output) {
+    *output = nullptr;
+    encoder_t *encoder = calloc(1, sizeof(encoder_t));
+
+    if (encoder == nullptr)
+        return err_allocation_failed;
+
+    error_t *err = symbol_table_alloc(&encoder->symbols);
+    if (err) {
+        free(encoder);
+        return err;
+    }
+
+    *output = encoder;
+    return nullptr;
+}
+
+void encoder_free(encoder_t *encoder) {
+    if (encoder == nullptr)
+        return;
+    symbol_table_free(encoder->symbols);
+    free(encoder);
+}
+
+bool encoder_is_symbols_node(ast_node_t *node) {
+    switch (node->id) {
+    case NODE_LABEL:
+    case NODE_LABEL_REFERENCE:
+    case NODE_EXPORT_DIRECTIVE:
+    case NODE_IMPORT_DIRECTIVE:
+        return true;
+    default:
+        return false;
+    }
+}
+
+int encoder_get_number_base(ast_node_t *number) {
+    switch (number->children[0]->id) {
+    case NODE_BINARY:
+        return 2;
+    case NODE_OCTAL:
+        return 8;
+    case NODE_DECIMAL:
+        return 10;
+    case NODE_HEXADECIMAL:
+        return 16;
+    default:
+        assert(false);
+    }
+    __builtin_unreachable();
+}
+
+bool is_valid_size_suffix(int bits) {
+    switch (bits) {
+    case 0:
+    case 8:
+    case 16:
+    case 32:
+    case 64:
+        return true;
+    default:
+        return false;
+    }
+}
+
+bool is_overflow(uint64_t value, int bits) {
+    if (bits == 0 || bits >= 64)
+        return false;
+
+    uint64_t max_value = (1ULL << bits) - 1;
+    return value > max_value;
+}
+
+operand_size_t encoder_get_size_mask(uint64_t value, int bits) {
+    if (bits != 0)
+        return bits_to_operand_size(bits);
+
+    operand_size_t mask = OPERAND_SIZE_64;
+    if (value < (1ULL << 8))
+        mask |= OPERAND_SIZE_8;
+    if (value < (1ULL << 16))
+        mask |= OPERAND_SIZE_16;
+    if (value < (1ULL << 32))
+        mask |= OPERAND_SIZE_32;
+    return mask;
+}
+
+error_t *encoder_set_number_value(ast_node_t *node) {
+    assert(node->id == NODE_NUMBER);
+    assert(node->children[0]);
+    const char *number = node->children[0]->token_entry->token.value;
+    int base = encoder_get_number_base(node);
+
+    if (base != 10)
+        number += 2; // all except base 10 use a 0x, 0o or 0b prefix
+
+    char *endptr;
+    errno = 0;
+    uint64_t value = strtoull(number, &endptr, base);
+
+    if (errno == ERANGE)
+        return err_encoder_number_overflow;
+
+    if (endptr == number)
+        return err_encoder_invalid_number_format;
+
+    int bits = 0;
+    if (*endptr == ':') {
+        const char *suffix = endptr + 1;
+
+        bits = strtol(suffix, &endptr, 10);
+
+        if (endptr == suffix)
+            return err_encoder_invalid_number_format;
+    }
+
+    if (*endptr != '\0')
+        return err_encoder_invalid_number_format;
+
+    if (!is_valid_size_suffix(bits))
+        return err_encoder_invalid_size_suffix;
+
+    if (is_overflow(value, bits))
+        return err_encoder_number_overflow;
+
+    node->value.number.value = value;
+    node->value.number.size = encoder_get_size_mask(value, bits);
+
+    return nullptr;
+}
+
+error_t *encoder_set_register_value(ast_node_t *node) {
+    assert(node->id == NODE_REGISTER);
+
+    const char *value = node->token_entry->token.value;
+
+    for (size_t i = 0; registers[i] != nullptr; ++i) {
+        if (strcmp(value, registers[i]->name) == 0) {
+            node->value.reg.id = registers[i]->id;
+            node->value.reg.size = registers[i]->size;
+            return nullptr;
+        }
+    }
+    return err_encoder_invalid_register;
+}
+
+/**
+ * Perform the initial pass over the AST. Records all symbols and sets the
+ * values of registers and numbers.
+ */
+error_t *encoder_first_pass(encoder_t *encoder, ast_node_t *node) {
+    error_t *err = nullptr;
+
+    if (encoder_is_symbols_node(node))
+        err = symbol_table_update(encoder->symbols, node);
+    else if (node->id == NODE_NUMBER)
+        err = encoder_set_number_value(node);
+    else if (node->id == NODE_REGISTER)
+        err = encoder_set_register_value(node);
+    if (err)
+        return err;
+
+    for (size_t i = 0; i < node->len; ++i) {
+        error_t *err = encoder_first_pass(encoder, node->children[i]);
+        if (err)
+            return err;
+    }
+
+    return nullptr;
+}
+
+opcode_data_t *encoder_find_opcode(ast_node_t *instruction) {
+    for (size_t i = 0; opcodes[i] != nullptr; ++i) {
+    }
+    return nullptr;
+}
+
+error_t *encoder_check_symbols(encoder_t *encoder) {
+    for (size_t i = 0; i < encoder->symbols->len; ++i)
+        if (encoder->symbols->symbols[i].kind == SYMBOL_REFERENCE)
+            return err_encoder_unknown_symbol_reference;
+    return nullptr;
+}
+
+error_t *encoder_encode(encoder_t *encoder, ast_node_t *ast) {
+    error_t *err = encoder_first_pass(encoder, ast);
+    if (err)
+        return err;
+    err = encoder_check_symbols(encoder);
+    if (err)
+        return err;
+    return nullptr;
+}
diff --git a/src/encoder/encoder.h b/src/encoder/encoder.h
new file mode 100644
index 0000000..f10966a
--- /dev/null
+++ b/src/encoder/encoder.h
@@ -0,0 +1,20 @@
+#ifndef INCLUDE_ENCODER_ENCODER_H_
+#define INCLUDE_ENCODER_ENCODER_H_
+
+#include "symbols.h"
+
+typedef struct encoder {
+    symbol_table_t *symbols;
+} encoder_t;
+
+error_t *encoder_alloc(encoder_t **encoder);
+error_t *encoder_encode(encoder_t *encoder, ast_node_t *ast);
+void encoder_free(encoder_t *encoder);
+
+extern error_t *const err_encoder_invalid_register;
+extern error_t *const err_encoder_number_overflow;
+extern error_t *const err_encoder_invalid_number_format;
+extern error_t *const err_encoder_invalid_size_suffix;
+extern error_t *const err_encoder_unknown_symbol_reference;
+
+#endif // INCLUDE_ENCODER_ENCODER_H_
-- 
2.47.2


From d8ae126e9ad73aca736b4da5163f04098b2be1a1 Mon Sep 17 00:00:00 2001
From: omicron <omicron.me@protonmail.com>
Date: Wed, 16 Apr 2025 00:20:35 +0200
Subject: [PATCH 11/14] Add opcode encoding value for NODE_INSTRUCTION entries
 in the AST

---
 src/ast.h | 6 ++++++
 1 file changed, 6 insertions(+)

diff --git a/src/ast.h b/src/ast.h
index e5cfd04..12c9b79 100644
--- a/src/ast.h
+++ b/src/ast.h
@@ -74,6 +74,11 @@ typedef struct register_ {
     operand_size_t size;
 } register_t;
 
+typedef struct opcode_encoding {
+    uint8_t encoding[32];
+    size_t len;
+} opcode_encoding_t;
+
 struct ast_node {
     node_id_t id;
     tokenlist_entry_t *token_entry;
@@ -84,6 +89,7 @@ struct ast_node {
     union {
         register_t reg;
         number_t number;
+        opcode_encoding_t encoding;
     } value;
 };
 
-- 
2.47.2


From 7e9c1bfda2cd2ab5f7375a49e213dad34f6109a1 Mon Sep 17 00:00:00 2001
From: omicron <omicron.me@protonmail.com>
Date: Wed, 16 Apr 2025 21:34:43 +0200
Subject: [PATCH 12/14] Add bytes type and tests

bytes_t is a local (automatic) allocation array that carries the length
and capacity with it.
---
 src/bytes.c   |   6 ++
 src/bytes.h   |  60 ++++++++++++++++++
 tests/bytes.c | 164 ++++++++++++++++++++++++++++++++++++++++++++++++++
 tests/main.c  |   2 +
 4 files changed, 232 insertions(+)
 create mode 100644 src/bytes.c
 create mode 100644 src/bytes.h
 create mode 100644 tests/bytes.c

diff --git a/src/bytes.c b/src/bytes.c
new file mode 100644
index 0000000..471f3d6
--- /dev/null
+++ b/src/bytes.c
@@ -0,0 +1,6 @@
+#include "bytes.h"
+#include "error.h"
+
+error_t *const err_bytes_no_capacity = &(error_t){
+    .message = "Not enough capacity in bytes buffer",
+};
diff --git a/src/bytes.h b/src/bytes.h
new file mode 100644
index 0000000..5041320
--- /dev/null
+++ b/src/bytes.h
@@ -0,0 +1,60 @@
+#ifndef INCLUDE_SRC_BYTES_H_
+#define INCLUDE_SRC_BYTES_H_
+
+#include "error.h"
+#include <stddef.h>
+#include <stdint.h>
+#include <string.h>
+
+extern error_t *const err_bytes_no_capacity;
+
+typedef struct bytes {
+    size_t len;
+    size_t cap;
+    uint8_t buffer[];
+} bytes_t;
+
+#define LOCAL_BYTES_ANONYMOUS(N)                                               \
+    &(struct {                                                                 \
+        size_t len;                                                            \
+        size_t cap;                                                            \
+        uint8_t buffer[(N)];                                                   \
+    }) {                                                                       \
+        0, (N), {}                                                             \
+    }
+
+#define LOCAL_BYTES(N) (bytes_t *)LOCAL_BYTES_ANONYMOUS(N);
+
+static inline error_t *bytes_append_uint8(bytes_t *bytes, uint8_t value) {
+    if (bytes->len >= bytes->cap)
+        return err_bytes_no_capacity;
+    bytes->buffer[bytes->len++] = value;
+    return nullptr;
+}
+
+static inline error_t *bytes_append_array(bytes_t *dst, size_t n,
+                                          uint8_t buffer[static n]) {
+    if (dst->len + n >= dst->cap)
+        return err_bytes_no_capacity;
+    memcpy(dst->buffer + dst->len, buffer, n);
+    dst->len += n;
+    return nullptr;
+}
+
+static inline error_t *bytes_append_bytes(bytes_t *dst, bytes_t *src) {
+    return bytes_append_array(dst, src->len, src->buffer);
+}
+
+static inline error_t *bytes_append_uint16(bytes_t *dst, uint16_t value) {
+    return bytes_append_array(dst, sizeof(value), (uint8_t *)&value);
+}
+
+static inline error_t *bytes_append_uint32(bytes_t *dst, uint32_t value) {
+    return bytes_append_array(dst, sizeof(value), (uint8_t *)&value);
+}
+
+static inline error_t *bytes_append_uint64(bytes_t *dst, uint64_t value) {
+    return bytes_append_array(dst, sizeof(value), (uint8_t *)&value);
+}
+
+#endif // INCLUDE_SRC_BYTES_H_
diff --git a/tests/bytes.c b/tests/bytes.c
new file mode 100644
index 0000000..3c98388
--- /dev/null
+++ b/tests/bytes.c
@@ -0,0 +1,164 @@
+#include "../src/bytes.h"
+#include "munit.h"
+
+MunitResult test_bytes_initializer(const MunitParameter params[], void *data) {
+    (void)params;
+    (void)data;
+    bytes_t *bytes = LOCAL_BYTES(16);
+    munit_assert_size(bytes->len, ==, 0);
+    munit_assert_size(bytes->cap, ==, 16);
+    for (size_t i = 0; i < 16; ++i)
+        munit_assert_uint8(bytes->buffer[i], ==, 0);
+    return MUNIT_OK;
+}
+
+MunitResult test_bytes_append_uint8(const MunitParameter params[], void *data) {
+    (void)params;
+    (void)data;
+    bytes_t *bytes = LOCAL_BYTES(16);
+    munit_assert_size(bytes->len, ==, 0);
+    munit_assert_size(bytes->cap, ==, 16);
+    for (size_t i = 0; i < 16; ++i) {
+        error_t *err = bytes_append_uint8(bytes, (uint8_t)i);
+        munit_assert_null(err);
+        munit_assert_uint8(bytes->buffer[i], ==, (uint8_t)i);
+    }
+
+    error_t *err = bytes_append_uint8(bytes, 0xFF);
+    munit_assert_ptr(err, ==, err_bytes_no_capacity);
+
+    return MUNIT_OK;
+}
+
+MunitResult test_bytes_append_array(const MunitParameter params[], void *data) {
+    (void)params;
+    (void)data;
+
+    bytes_t *bytes = LOCAL_BYTES(16);
+    munit_assert_size(bytes->len, ==, 0);
+    munit_assert_size(bytes->cap, ==, 16);
+
+    uint8_t test_array[] = {0x01, 0x02, 0x03, 0x04, 0x05};
+    size_t array_len = sizeof(test_array) / sizeof(test_array[0]);
+    error_t *err = bytes_append_array(bytes, array_len, test_array);
+    munit_assert_null(err);
+    munit_assert_size(bytes->len, ==, array_len);
+
+    for (size_t i = 0; i < array_len; ++i) {
+        munit_assert_uint8(bytes->buffer[i], ==, test_array[i]);
+    }
+
+    uint8_t second_array[] = {0x06, 0x07, 0x08};
+    size_t second_len = sizeof(second_array) / sizeof(second_array[0]);
+    err = bytes_append_array(bytes, second_len, second_array);
+    munit_assert_null(err);
+    munit_assert_size(bytes->len, ==, array_len + second_len);
+    for (size_t i = 0; i < second_len; ++i) {
+        munit_assert_uint8(bytes->buffer[array_len + i], ==, second_array[i]);
+    }
+
+    uint8_t overflow_array[10] = {0}; // Array that would exceed capacity
+    err = bytes_append_array(bytes, sizeof(overflow_array), overflow_array);
+    munit_assert_ptr(err, ==, err_bytes_no_capacity);
+    munit_assert_size(bytes->len, ==, array_len + second_len);
+
+    return MUNIT_OK;
+}
+
+MunitResult test_bytes_append_bytes(const MunitParameter params[], void *data) {
+    (void)params;
+    (void)data;
+
+    bytes_t *src = LOCAL_BYTES(8);
+    bytes_t *dst = LOCAL_BYTES(16);
+
+    // Fill source bytes with test data
+    for (uint8_t i = 0; i < 5; ++i) {
+        error_t *err = bytes_append_uint8(src, i + 1);
+        munit_assert_null(err);
+    }
+    munit_assert_size(src->len, ==, 5);
+
+    // Append source to destination
+    error_t *err = bytes_append_bytes(dst, src);
+    munit_assert_null(err);
+    munit_assert_size(dst->len, ==, src->len);
+
+    // Verify destination contents match source
+    for (size_t i = 0; i < src->len; ++i) {
+        munit_assert_uint8(dst->buffer[i], ==, src->buffer[i]);
+    }
+
+    // Fill source with more data and append again
+    for (uint8_t i = 0; i < 3; ++i) {
+        err = bytes_append_uint8(src, i + 6);
+        munit_assert_null(err);
+    }
+    munit_assert_size(src->len, ==, 8);
+
+    // Append updated source
+    err = bytes_append_bytes(dst, src);
+    munit_assert_null(err);
+    munit_assert_size(dst->len, ==, 13); // 5 + 8
+
+    // Test capacity boundary
+    src->len = 4; // manually set length to barely not fit
+    err = bytes_append_bytes(dst, src);
+    munit_assert_ptr(err, ==, err_bytes_no_capacity);
+    munit_assert_size(dst->len, ==, 13); // Length unchanged after error
+
+    return MUNIT_OK;
+}
+MunitResult test_bytes_append_uint16(const MunitParameter params[], void *data) {
+    bytes_t *bytes = LOCAL_BYTES(16);
+    munit_assert_size(bytes->len, ==, 0);
+    munit_assert_size(bytes->cap, ==, 16);
+
+    bytes_append_uint16(bytes, 0xFFAA);
+    munit_assert_size(bytes->len, ==, 2);
+    munit_assert_uint8(bytes->buffer[0], ==, 0xAA);
+    munit_assert_uint8(bytes->buffer[1], ==, 0xFF);
+
+    return MUNIT_OK;
+}
+MunitResult test_bytes_append_uint32(const MunitParameter params[], void *data) {
+    bytes_t *bytes = LOCAL_BYTES(16);
+    munit_assert_size(bytes->len, ==, 0);
+    munit_assert_size(bytes->cap, ==, 16);
+
+    bytes_append_uint32(bytes, 0xAABBCCDD);
+    munit_assert_size(bytes->len, ==, 4);
+    munit_assert_uint8(bytes->buffer[0], ==, 0xDD);
+    munit_assert_uint8(bytes->buffer[1], ==, 0xCC);
+    munit_assert_uint8(bytes->buffer[2], ==, 0xBB);
+    munit_assert_uint8(bytes->buffer[3], ==, 0xAA);
+    return MUNIT_OK;
+}
+MunitResult test_bytes_append_uint64(const MunitParameter params[], void *data) {
+    bytes_t *bytes = LOCAL_BYTES(16);
+    munit_assert_size(bytes->len, ==, 0);
+    munit_assert_size(bytes->cap, ==, 16);
+
+    bytes_append_uint64(bytes, 0xAABBCCDDEEFF9988);
+    munit_assert_size(bytes->len, ==, 8);
+    munit_assert_uint8(bytes->buffer[0], ==, 0x88);
+    munit_assert_uint8(bytes->buffer[1], ==, 0x99);
+    munit_assert_uint8(bytes->buffer[2], ==, 0xFF);
+    munit_assert_uint8(bytes->buffer[3], ==, 0xEE);
+    munit_assert_uint8(bytes->buffer[4], ==, 0xDD);
+    munit_assert_uint8(bytes->buffer[5], ==, 0xCC);
+    munit_assert_uint8(bytes->buffer[6], ==, 0xBB);
+    munit_assert_uint8(bytes->buffer[7], ==, 0xAA);
+    return MUNIT_OK;
+}
+
+MunitTest bytes_tests[] = {
+    {"/initializer",   test_bytes_initializer,   nullptr, nullptr, MUNIT_TEST_OPTION_NONE, nullptr},
+    {"/append_uint8",  test_bytes_append_uint8,  nullptr, nullptr, MUNIT_TEST_OPTION_NONE, nullptr},
+    {"/append_array",  test_bytes_append_array,  nullptr, nullptr, MUNIT_TEST_OPTION_NONE, nullptr},
+    {"/append_bytes",  test_bytes_append_bytes,  nullptr, nullptr, MUNIT_TEST_OPTION_NONE, nullptr},
+    {"/append_uint16", test_bytes_append_uint16, nullptr, nullptr, MUNIT_TEST_OPTION_NONE, nullptr},
+    {"/append_uint32", test_bytes_append_uint32, nullptr, nullptr, MUNIT_TEST_OPTION_NONE, nullptr},
+    {"/append_uint64", test_bytes_append_uint64, nullptr, nullptr, MUNIT_TEST_OPTION_NONE, nullptr},
+    {nullptr,          nullptr,                  nullptr, nullptr, MUNIT_TEST_OPTION_NONE, nullptr}
+};
diff --git a/tests/main.c b/tests/main.c
index cf7162f..be92b14 100644
--- a/tests/main.c
+++ b/tests/main.c
@@ -4,6 +4,7 @@ extern MunitTest ast_tests[];
 extern MunitTest lexer_tests[];
 extern MunitTest regression_tests[];
 extern MunitTest symbols_tests[];
+extern MunitTest bytes_tests[];
 
 int main(int argc, char *argv[MUNIT_ARRAY_PARAM(argc + 1)]) {
     MunitSuite suites[] = {
@@ -11,6 +12,7 @@ int main(int argc, char *argv[MUNIT_ARRAY_PARAM(argc + 1)]) {
         {"/ast",        ast_tests,        nullptr, 1, MUNIT_SUITE_OPTION_NONE},
         {"/lexer",      lexer_tests,      nullptr, 1, MUNIT_SUITE_OPTION_NONE},
         {"/symbols",    symbols_tests,    nullptr, 1, MUNIT_SUITE_OPTION_NONE},
+        {"/bytes",      bytes_tests,      nullptr, 1, MUNIT_SUITE_OPTION_NONE},
         {nullptr,       nullptr,          nullptr, 0, MUNIT_SUITE_OPTION_NONE},
     };
 
-- 
2.47.2


From 99c9dcd98578c98b94028f40a10a41fc3ec0828f Mon Sep 17 00:00:00 2001
From: omicron <omicron.me@protonmail.com>
Date: Wed, 16 Apr 2025 00:21:41 +0200
Subject: [PATCH 13/14] Incomplete second pass encoding

---
 src/encoder/encoder.c | 317 +++++++++++++++++++++++++++++++++++++++++-
 src/encoder/encoder.h |  12 ++
 2 files changed, 328 insertions(+), 1 deletion(-)

diff --git a/src/encoder/encoder.c b/src/encoder/encoder.c
index a43a23b..d3b999a 100644
--- a/src/encoder/encoder.c
+++ b/src/encoder/encoder.c
@@ -1,4 +1,5 @@
 #include "encoder.h"
+#include "../bytes.h"
 #include "../data/opcodes.h"
 #include "symbols.h"
 #include <assert.h>
@@ -15,6 +16,12 @@ error_t *const err_encoder_invalid_size_suffix =
     &(error_t){.message = "Invalid number size suffix"};
 error_t *const err_encoder_unknown_symbol_reference =
     &(error_t){.message = "Referenced an unknown symbol"};
+error_t *const err_encoder_no_encoding_found =
+    &(error_t){.message = "No encoding found for instruction"};
+error_t *const err_encoder_not_implemented =
+    &(error_t){.message = "Implementation for this opcode is missing"};
+error_t *const err_encoder_unexpected_length =
+    &(error_t){.message = "Unexpectedly long encoding"};
 
 error_t *encoder_alloc(encoder_t **output) {
     *output = nullptr;
@@ -162,6 +169,50 @@ error_t *encoder_set_register_value(ast_node_t *node) {
     return err_encoder_invalid_register;
 }
 
+/**
+ * Set the opcode extension in the modrm field
+ */
+static inline uint8_t modrm_extension(uint8_t modrm, uint8_t extension) {
+    assert(extension != opcode_extension_none);
+    assert((extension & 0b111) == extension);
+    return (modrm & ~modrm_reg_mask) | extension << 3;
+}
+
+/**
+ * Return the rex bit for reg field in modrm
+ */
+static inline uint8_t modrm_reg_rex(uint8_t rex, register_id_t id) {
+    if (id & 0b1000)
+        rex |= rex_prefix_r;
+    return rex;
+}
+
+/**
+ * update modrm reg field with the given register, must be used alongside
+ * modrm_reg_rex
+ */
+static inline uint8_t modrm_reg(uint8_t modrm, register_id_t id) {
+    return (modrm & ~modrm_reg_mask) | (id & 0b111) << 3;
+}
+
+/**
+ * Return the rex bit for rm field in modrm
+ */
+static inline uint8_t modrm_rm_rex(uint8_t rex, register_id_t id) {
+    if (id & 0b1000)
+        rex |= rex_prefix_b;
+    return rex;
+}
+
+/**
+ * update modrm rm field with the given register, must be used alongside
+ * modrm_rm_rex
+ */
+static inline uint8_t modrm_rm(uint8_t modrm, register_id_t id) {
+    assert((modrm & modrm_mod_mask) == modrm_mod_register);
+    return (modrm & ~modrm_rm_mask) | (id & 0b111);
+}
+
 /**
  * Perform the initial pass over the AST. Records all symbols and sets the
  * values of registers and numbers.
@@ -187,8 +238,272 @@ error_t *encoder_first_pass(encoder_t *encoder, ast_node_t *node) {
     return nullptr;
 }
 
+bool is_operand_match(operand_info_t *info, ast_node_t *operand) {
+    switch (info->kind) {
+    case OPERAND_REGISTER:
+        return operand->id == NODE_REGISTER &&
+               operand->value.reg.size == info->size;
+    case OPERAND_MEMORY:
+        return operand->id == NODE_MEMORY;
+    case OPERAND_IMMEDIATE: {
+        if (operand->id != NODE_IMMEDIATE)
+            return false;
+        ast_node_t *child = operand->children[0];
+
+        if (child->id == NODE_NUMBER)
+            return (child->value.number.size & info->size) > 0;
+        else if (child->id == NODE_LABEL_REFERENCE)
+            return info->size == OPERAND_SIZE_32;
+        // FIXME: first pass should give us information about the distance of
+        // the label reference so we can pick a size more appropriately instead
+        // of just defaulting to 32 bits
+        break;
+    } // end OPERAND_IMMEDIATE case
+    }
+    assert(false && "unreachable");
+    __builtin_unreachable();
+}
+
+bool is_opcode_match(opcode_data_t *opcode, const char *mnemonic,
+                     ast_node_t *operands) {
+    if (strcmp(opcode->mnemonic, mnemonic) != 0)
+        return false;
+
+    if (opcode->operand_count != operands->len)
+        return false;
+
+    for (size_t i = 0; i < operands->len; ++i) {
+        if (!is_operand_match(&opcode->operands[i], operands->children[i]))
+            return false;
+    }
+
+    return true;
+}
+
+error_t *encoder_get_opcode_data(ast_node_t *instruction, ast_node_t *operands,
+                                 opcode_data_t **opcode_out) {
+    const char *mnemonic = instruction->children[0]->token_entry->token.value;
+
+    for (size_t i = 0; opcodes[i]; ++i) {
+        opcode_data_t *opcode = opcodes[i];
+        if (is_opcode_match(opcode, mnemonic, operands)) {
+            *opcode_out = opcode;
+            return nullptr;
+        }
+    }
+    return err_encoder_no_encoding_found;
+}
+
+error_t *encode_two_operand(encoder_t *encoder, opcode_data_t *opcode,
+                            ast_node_t *operands, bytes_t *encoding,
+                            uint8_t *rex) {
+    (void)encoder;
+    (void)opcode;
+    (void)operands;
+    (void)encoding;
+    (void)rex;
+    assert(encoding->len >= 1 && "must have 1+ opcode byte in buffer already");
+    return err_encoder_not_implemented;
+}
+
+error_t *encode_one_register_in_opcode(encoder_t *encoder,
+                                       opcode_data_t *opcode,
+                                       ast_node_t *operands, bytes_t *encoding,
+                                       uint8_t *rex) {
+    (void)encoder;
+    (void)opcode;
+
+    register_id_t id = operands->children[0]->value.reg.id;
+    encoding->buffer[encoding->len - 1] |= id & 0b111;
+    if ((id & 0b1000) > 0) {
+        *rex |= rex_prefix_r;
+    }
+    return nullptr;
+}
+
+error_t *encode_one_register(encoder_t *encoder, opcode_data_t *opcode,
+                             ast_node_t *operands, bytes_t *encoding,
+                             uint8_t *rex) {
+    (void)encoder;
+    assert(operands->len == 1);
+    assert(operands->children[0]->id == NODE_REGISTER);
+
+    register_id_t id = operands->children[0]->value.reg.id;
+
+    uint8_t modrm = modrm_mod_register;
+
+    if (opcode->opcode_extension != opcode_extension_none) {
+        // register goes in rm field, extension goes in mod field
+        modrm = modrm_extension(modrm, opcode->opcode_extension);
+        modrm = modrm_rm(modrm, id);
+        *rex = modrm_rm_rex(*rex, id);
+    } else {
+        // register goes in reg field
+        // NOTE:
+        // it's actually likely this case just doesn't exist at all and all
+        // opcodes that take one register in modr/m _all_ have extended opcdes
+        modrm = modrm_reg(modrm, id);
+        *rex = modrm_reg_rex(*rex, id);
+    }
+
+    return bytes_append_uint8(encoding, modrm);
+}
+
+error_t *encode_one_immediate(encoder_t *encoder, opcode_data_t *opcode,
+                              ast_node_t *operands, bytes_t *encoding,
+                              uint8_t *rex) {
+    (void)encoder;
+    (void)opcode;
+    (void)rex;
+    assert(operands->len == 1);
+    assert(operands->children[0]->id == NODE_IMMEDIATE);
+    assert(operands->children[0]->len == 1);
+    ast_node_t *immediate = operands->children[0]->children[0];
+    assert(immediate->id == NODE_NUMBER ||
+           immediate->id == NODE_LABEL_REFERENCE);
+
+    if (immediate->id == NODE_NUMBER) {
+        uint64_t value = immediate->value.number.value;
+        operand_size_t size = opcode->operands[0].size;
+        error_t *err = nullptr;
+        switch (size) {
+        case OPERAND_SIZE_8:
+            err = bytes_append_uint8(encoding, value);
+            break;
+        case OPERAND_SIZE_16:
+            err = bytes_append_uint16(encoding, value);
+            break;
+        case OPERAND_SIZE_32:
+            err = bytes_append_uint32(encoding, value);
+            break;
+        case OPERAND_SIZE_64:
+            err = bytes_append_uint64(encoding, value);
+            break;
+        default:
+            assert(false && "intentionally unhandled");
+        }
+        return err;
+    } else {
+        // FIXME: this still assumes references are always 32 bit
+        uint32_t value = 0xDEADBEEF;
+        return bytes_append_uint32(encoding, value);
+    }
+}
+
+error_t *encode_one_memory(encoder_t *encoder, opcode_data_t *opcode,
+                           ast_node_t *operands, bytes_t *encoding,
+                           uint8_t *rex) {
+    (void)encoder;
+    (void)opcode;
+    (void)operands;
+    (void)encoding;
+    (void)rex;
+    return err_encoder_not_implemented;
+}
+
+error_t *encode_one_operand(encoder_t *encoder, opcode_data_t *opcode,
+                            ast_node_t *operands, bytes_t *encoding,
+                            uint8_t *rex) {
+    switch (opcode->operands[0].kind) {
+    case OPERAND_REGISTER:
+        if (opcode->encoding_class == ENCODING_OPCODE_REGISTER)
+            return encode_one_register_in_opcode(encoder, opcode, operands,
+                                                 encoding, rex);
+        else
+            return encode_one_register(encoder, opcode, operands, encoding,
+                                       rex);
+    case OPERAND_MEMORY:
+        return encode_one_memory(encoder, opcode, operands, encoding, rex);
+    case OPERAND_IMMEDIATE:
+        return encode_one_immediate(encoder, opcode, operands, encoding, rex);
+    }
+}
+
+error_t *encoder_encode_instruction(encoder_t *encoder,
+                                    ast_node_t *instruction) {
+    ast_node_t *operands = instruction->children[1];
+
+    opcode_data_t *opcode = nullptr;
+    error_t *err = encoder_get_opcode_data(instruction, operands, &opcode);
+    if (err)
+        return err;
+
+    uint8_t rex = 0;
+    bytes_t *encoding = LOCAL_BYTES(32);
+
+    if (opcode->opcode > 0xFF &&
+        (err = bytes_append_uint8(encoding, opcode->opcode >> 8)))
+        return err;
+    if ((err = bytes_append_uint8(encoding, opcode->opcode & 0xFF)))
+        return err;
+
+    // NOTE:operand encoders all expect the opcode to be in the buffer already.
+    // Some of them rely on this to encode the register value in the opcode
+    // byte.
+    switch (opcode->operand_count) {
+    case 0:
+        break;
+    case 1:
+        err = encode_one_operand(encoder, opcode, operands, encoding, &rex);
+        break;
+    case 2:
+        err = encode_two_operand(encoder, opcode, operands, encoding, &rex);
+        break;
+    default:
+        err = err_encoder_not_implemented;
+    }
+    if (err)
+        return err;
+
+    // produce the actual encoding output in the NODE_INSTRUCTION value
+    uint8_t *output = instruction->value.encoding.encoding;
+    size_t output_len = 0;
+
+    // Handle prefixes
+    if (opcode->rex_w_prefix)
+        rex = rex_prefix_w;
+    if (opcode->address_size_prefix)
+        output[output_len++] = memory_size_prefix;
+    if (opcode->operand_size_prefix)
+        output[output_len++] = operand_size_prefix;
+    if (rex > 0)
+        output[output_len++] = rex;
+
+    // copy the encoded opcode and operands
+    if (encoding->len > 20)
+        return err_encoder_unexpected_length;
+    memcpy(output + output_len, encoding->buffer, encoding->len);
+    output_len += encoding->len;
+
+    instruction->value.encoding.len = output_len;
+
+    return nullptr;
+}
+
+/**
+ * Perform the second pass that performs actual encoding. Will use
+ * placeholder values for label references because instruction size has not
+ * yet been determined.
+ */
+error_t *encoder_encoding_pass(encoder_t *encoder, ast_node_t *root) {
+    for (size_t i = 0; i < root->len; ++i) {
+        if (root->children[i]->id != NODE_INSTRUCTION)
+            continue;
+        ast_node_t *instruction = root->children[i];
+        error_t *err = encoder_encode_instruction(encoder, instruction);
+        if (err)
+            return err;
+    }
+    return nullptr;
+}
+
 opcode_data_t *encoder_find_opcode(ast_node_t *instruction) {
     for (size_t i = 0; opcodes[i] != nullptr; ++i) {
+        const char *mnemonic =
+            instruction->children[0]->token_entry->token.value;
+        ast_node_t *operands = instruction->children[1];
+        if (is_opcode_match(opcodes[i], mnemonic, operands))
+            return opcodes[i];
     }
     return nullptr;
 }
@@ -207,5 +522,5 @@ error_t *encoder_encode(encoder_t *encoder, ast_node_t *ast) {
     err = encoder_check_symbols(encoder);
     if (err)
         return err;
-    return nullptr;
+    return encoder_encoding_pass(encoder, ast);
 }
diff --git a/src/encoder/encoder.h b/src/encoder/encoder.h
index f10966a..45d34d7 100644
--- a/src/encoder/encoder.h
+++ b/src/encoder/encoder.h
@@ -7,6 +7,15 @@ typedef struct encoder {
     symbol_table_t *symbols;
 } encoder_t;
 
+constexpr uint8_t modrm_mod_memory = 0b00'000'000;
+constexpr uint8_t modrm_mod_memory_displacement8 = 0b01'000'000;
+constexpr uint8_t modrm_mod_memory_displacement32 = 0b10'000'000;
+constexpr uint8_t modrm_mod_register = 0b11'000'000;
+
+constexpr uint8_t modrm_reg_mask = 0b00'111'000;
+constexpr uint8_t modrm_rm_mask = 0b00'000'111;
+constexpr uint8_t modrm_mod_mask = 0b11'000'000;
+
 error_t *encoder_alloc(encoder_t **encoder);
 error_t *encoder_encode(encoder_t *encoder, ast_node_t *ast);
 void encoder_free(encoder_t *encoder);
@@ -16,5 +25,8 @@ extern error_t *const err_encoder_number_overflow;
 extern error_t *const err_encoder_invalid_number_format;
 extern error_t *const err_encoder_invalid_size_suffix;
 extern error_t *const err_encoder_unknown_symbol_reference;
+extern error_t *const err_encoder_no_encoding_found;
+extern error_t *const err_encoder_not_implemented;
+extern error_t *const err_encoder_unexpected_length;
 
 #endif // INCLUDE_ENCODER_ENCODER_H_
-- 
2.47.2


From d97cfb97be7444df6664bc571df8430415a0e955 Mon Sep 17 00:00:00 2001
From: omicron <omicron.me@protonmail.com>
Date: Wed, 16 Apr 2025 14:31:09 +0200
Subject: [PATCH 14/14] Implement printing the encoding in main

---
 src/main.c | 73 ++++++++++++++++++++++++++++++++++++++++++++++++------
 1 file changed, 66 insertions(+), 7 deletions(-)

diff --git a/src/main.c b/src/main.c
index 6825254..fb00342 100644
--- a/src/main.c
+++ b/src/main.c
@@ -1,3 +1,5 @@
+#include "ast.h"
+#include "encoder/encoder.h"
 #include "error.h"
 #include "lexer.h"
 #include "parser/parser.h"
@@ -8,7 +10,13 @@
 #include <stdlib.h>
 #include <string.h>
 
-typedef enum mode { MODE_AST, MODE_TEXT, MODE_TOKENS } mode_t;
+typedef enum mode {
+    MODE_INVALID = -1,
+    MODE_AST,
+    MODE_TEXT,
+    MODE_TOKENS,
+    MODE_ENCODING,
+} mode_t;
 
 void print_tokens(tokenlist_t *list) {
     for (auto entry = list->head; entry; entry = entry->next) {
@@ -50,18 +58,61 @@ error_t *print_ast(tokenlist_t *list) {
     return nullptr;
 }
 
-int get_execution_mode(int argc, char *argv[]) {
-    if (argc != 3 || (strcmp(argv[1], "tokens") != 0 &&
-                      strcmp(argv[1], "text") != 0 && strcmp(argv[1], "ast"))) {
-        puts("Usage: oas [tokens|text|ast] <filename>");
-        exit(1);
+void print_hex(size_t len, uint8_t bytes[static len]) {
+    for (size_t i = 0; i < len; i++) {
+        printf("%02x", bytes[i]);
+        if (i < len - 1) {
+            printf(" ");
+        }
     }
+    printf("\n");
+}
+
+error_t *print_encoding(tokenlist_t *list) {
+    parse_result_t result = parse(list->head);
+    if (result.err)
+        return result.err;
+
+    encoder_t *encoder;
+    error_t *err = encoder_alloc(&encoder);
+    if (err)
+        goto cleanup_ast;
+
+    err = encoder_encode(encoder, result.node);
+    if (err)
+        goto cleanup_ast;
+
+    ast_node_t *root = result.node;
+    for (size_t i = 0; i < root->len; ++i) {
+        ast_node_t *node = root->children[i];
+        if (node->id != NODE_INSTRUCTION)
+            continue;
+
+        print_hex(node->value.encoding.len, node->value.encoding.encoding);
+    }
+
+    encoder_free(encoder);
+    ast_node_free(result.node);
+    return nullptr;
+
+cleanup_ast:
+    ast_node_free(result.node);
+    return err;
+}
+
+int get_execution_mode(int argc, char *argv[]) {
+    if (argc != 3)
+        return MODE_INVALID;
 
     if (strcmp(argv[1], "tokens") == 0)
         return MODE_TOKENS;
     if (strcmp(argv[1], "text") == 0)
         return MODE_TEXT;
-    return MODE_AST;
+    if (strcmp(argv[1], "ast") == 0)
+        return MODE_AST;
+    if (strcmp(argv[1], "encoding") == 0)
+        return MODE_ENCODING;
+    return MODE_INVALID;
 }
 
 error_t *do_action(mode_t mode, tokenlist_t *list) {
@@ -74,12 +125,20 @@ error_t *do_action(mode_t mode, tokenlist_t *list) {
         return nullptr;
     case MODE_AST:
         return print_ast(list);
+    case MODE_ENCODING:
+        return print_encoding(list);
+    case MODE_INVALID:
+        /* can't happen */
     }
     __builtin_unreachable();
 }
 
 int main(int argc, char *argv[]) {
     mode_t mode = get_execution_mode(argc, argv);
+    if (mode == MODE_INVALID) {
+        puts("Usage: oas [tokens|text|ast|encoding] <filename>");
+        exit(1);
+    }
     char *filename = argv[2];
 
     lexer_t *lex = &(lexer_t){};
-- 
2.47.2