Add .import and .export to the input test file

doc/parser_grammar.txt

@@ -3,10 +3,14 @@
 
 <label> ::= <identifier> <colon>
 
-<directive> ::= <dot> <section_directive>
+<directive> ::= <dot> (<section_directive> | <export_directive> | <import_directive> )
 
 <section_directive> ::= "section" <identifier>
 
+<export_directive> ::= "export" <identifier>
+
+<import_directive> ::= "import" <identifier>
+
 <instruction> ::= <identifier> <operands>
 
 <operands> ::= <operand> ( <comma> <operand> )*

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:
@@ -157,6 +161,10 @@ const char *ast_node_id_to_cstr(node_id_t id) {
         return "NODE_ASTERISK";
     case NODE_DOT:
         return "NODE_DOT";
+    case NODE_IMPORT:
+        return "NODE_IMPORT";
+    case NODE_EXPORT:
+        return "NODE_EXPORT";
     }
     assert(!"Unreachable, weird node id" && id);
     __builtin_unreachable();

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,

src/encoder/symbols.c

@@ -0,0 +1,149 @@
+#include "symbols.h"
+#include "../error.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 *err_symbol_table_invalid_node = &(error_t){
+    .message = "Unexpected node id when adding symbol to symbol table"};
+error_t *err_symbol_table_max_cap = &(error_t){
+    .message = "Failed to increase symbol table length, max capacity reached"};
+error_t *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;
+        break;
+    case NODE_LABEL_REFERENCE:
+        *kind = SYMBOL_REFERENCE;
+        *name = node->token_entry->token.value;
+        break;
+    // TODO: when .import and .export directives are created add them here
+    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,
+    };
+
+    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;
+}

src/encoder/symbols.h

@@ -0,0 +1,42 @@
+#ifndef INCLUDE_ENCODER_SYMBOLS_H_
+#define INCLUDE_ENCODER_SYMBOLS_H_
+
+#include "../ast.h"
+
+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_

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);

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, nullptr};
+    parser_t parsers[] = {parse_dot, parse_directive_options, nullptr};
     return parse_consecutive(current, NODE_DIRECTIVE, parsers);
 }
 

src/parser/primitives.c

@@ -101,3 +101,19 @@ 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);
+}

src/parser/primitives.h

@@ -26,5 +26,7 @@ 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_

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