feat: Include directive (#61)

* fix: Fix castings

* feat: Implement include directive

lexer.c

@@ -4,6 +4,7 @@
 #include <string.h>
 #include <stdio.h>
 #include <stdlib.h>
+#include <sys/stat.h>
 #include "lexer.h"
 
 char is_num(char c)
@@ -569,3 +570,37 @@ typed_token *tokenize(char *inp)
     }
     return first;
 }
+
+typed_token *tokenize_file(char *path)
+{
+    FILE *fp = NULL;
+
+    fp = fopen(path, "rb");
+    if (fp == NULL)
+    {
+        fprintf(stderr, "Error opening file: %s\n", path);
+        exit(1);
+    }
+
+    char *data = NULL;
+    struct stat st;
+
+    if (fstat(fileno(fp), &st) == -1)
+        goto ret;
+
+    data = calloc(st.st_size + 1, sizeof(char));
+    if (!data)
+        goto ret;
+
+    int rd = fread(data, sizeof(char), st.st_size, fp);
+    if (rd != st.st_size)
+    {
+        data = NULL;
+        goto ret;
+    }
+    data[st.st_size] = '\0';
+
+    return tokenize(data);
+ret:
+    return NULL;
+}

lexer.h

@@ -11,6 +11,7 @@ typedef struct typed_token_
 
 typed_token *new_tkn(int tkn_id, void *data, void (*debug)(typed_token *));
 void str_tkn_debug(typed_token *tkn);
+typed_token *tokenize_file(char *path);
 typed_token *tokenize(char *inp);
 
 #define TKN_EOF 0

main.c

@@ -7,73 +7,35 @@
 #include "linked_list.h"
 #include "preprocess/preprocess.h"
 
-char *read_source_file(FILE *fp);
-
-#define MAX_INCLUDE_DEPTH 8
-/**
- * free() ignores NULL pointer by default. set each pointer to NULL
- * after calling free() to avoid double free error.
- */
-#define xfree(p)    \
-    do              \
-    {               \
-        free((p));  \
-        (p) = NULL; \
-    } while (0)
-
-typed_token *process(const char *filename, int depth, int log_lex, int log_prep)
+typed_token *process(const char *filename, int log_lex, int log_prep)
 {
-    typed_token *tkn = NULL;
-    char *content = NULL;
-    FILE *fp = NULL;
-
-    if (depth > MAX_INCLUDE_DEPTH)
-    {
-        fprintf(stderr, "exceeded maximum include depth (%d)\n", MAX_INCLUDE_DEPTH);
-        goto cleanup;
-    }
-
-    fp = fopen(filename, "rb");
-    if (fp == NULL)
-    {
-        fprintf(stderr, "Error opening file: %s\n", filename);
-        goto cleanup;
-    }
-    content = read_source_file(fp);
-
-    if (!content)
-    {
-        fprintf(stderr, "failed to read source file: read_source_file returned NULL\n");
-        goto cleanup;
-    }
+    prep_ctx *ctx = (prep_ctx *)malloc(sizeof(prep_ctx));
+    ctx->curr_path = NULL;
+    ctx->defs = new_linked_list();
 
-    tkn = tokenize(content);
+    typed_token *lexed = tokenize_file(filename);
+    typed_token *prep = preprocess(ctx, filename);
     if (log_lex)
     {
-        typed_token *t = tkn;
+        typed_token *t = lexed;
         while (t)
         {
             t->debug(t);
             t = t->next;
         }
     }
-    tkn = preprocess(tkn);
+
     if (log_prep)
     {
-        typed_token *t = tkn;
+        typed_token *t = prep;
         while (t)
         {
             t->debug(t);
             t = t->next;
         }
     }
 
-cleanup:
-    if (fp)
-        fclose(fp);
-    if (content)
-        xfree(content);
-    return tkn;
+    return prep;
 }
 
 int main(int argc, char *argv[])
@@ -84,15 +46,16 @@ int main(int argc, char *argv[])
         return 1;
     }
 
-    if (strcmp(argv[2], "--lex") != 0 && strcmp(argv[2], "--asm") != 0 && strcmp(argv[2], "--tree") != 0&& strcmp(argv[2], "--prep") != 0) {
+    if (strcmp(argv[2], "--lex") != 0 && strcmp(argv[2], "--asm") != 0 && strcmp(argv[2], "--tree") != 0 && strcmp(argv[2], "--prep") != 0)
+    {
         fprintf(stderr, "Unknown argument %s", argv[2]);
         return 1;
     }
 
     int log_lex = !strcmp(argv[2], "--lex");
     int log_prep = !strcmp(argv[2], "--prep");
 
-    typed_token *tkn = process(argv[1], 0, log_lex, log_prep);
+    typed_token *tkn = process(argv[1], log_lex, log_prep);
     if (tkn == NULL)
     {
         return 1;
@@ -135,27 +98,3 @@ int main(int argc, char *argv[])
 
     return 0;
 }
-
-char *read_source_file(FILE *fp)
-{
-    char *data = NULL;
-    struct stat st;
-
-    if (fstat(fileno(fp), &st) == -1)
-        goto ret;
-
-    data = calloc(st.st_size + 1, sizeof(char));
-    if (!data)
-        goto ret;
-
-    int rd = fread(data, sizeof(char), st.st_size, fp);
-    if (rd != st.st_size)
-    {
-        data = NULL;
-        goto ret;
-    }
-    data[st.st_size] = '\0';
-
-ret:
-    return data;
-}

preprocess/macro_include.c

@@ -0,0 +1,76 @@
+#include <stdlib.h>
+#include <stdio.h>
+#include "macro_include.h"
+#include "preprocess.h"
+#include <string.h>
+
+char *get_path(char *curr_path, char *header_path)
+{
+    char *new_path = (char *)malloc(128);
+    strcpy(new_path, curr_path);
+
+    char *last_slash = new_path;
+    char *ptr = new_path;
+    while (*ptr)
+    {
+        if (*ptr == '/')
+        {
+            last_slash = ptr + 1;
+        }
+        ptr++;
+    }
+    strcpy(last_slash, header_path);
+    return new_path;
+}
+
+linked_list *seg_include_gen(seg *self, prep_ctx *ctx)
+{
+    seg_include *inc = (seg_include *)self->data;
+    linked_list *ret = new_linked_list();
+
+    typed_token *t = preprocess(ctx, get_path(ctx->curr_path, inc->path));
+    while (t->type_id != TKN_EOF)
+    {
+        add_to_list(ret, t);
+        t = t->next;
+    }
+    return ret;
+}
+
+seg *parse_include(prep_ctx *ctx, typed_token **tkns_ptr)
+{
+    typed_token *tkn = *tkns_ptr;
+    if (tkn->type_id == TKN_DIRECTIVE)
+    {
+        typed_token *inner_tkn = (typed_token *)tkn->data;
+        tkn = tkn->next;
+        if (inner_tkn->type_id == TKN_ID)
+        {
+            if (strcmp((char *)inner_tkn->data, "include") == 0)
+            {
+                inner_tkn = inner_tkn->next;
+                if (inner_tkn->type_id == TKN_LIT_STR)
+                {
+                    char *path = (char *)inner_tkn->data;
+                    inner_tkn = inner_tkn->next;
+
+                    if (inner_tkn->type_id == TKN_EOF)
+                    {
+                        *tkns_ptr = tkn;
+                        seg *ret = (seg *)malloc(sizeof(seg));
+                        ret->data = malloc(sizeof(seg_include));
+                        ret->to_code = seg_include_gen;
+                        ret->stringified = NULL;
+                        ((seg_include *)ret->data)->path = path;
+                        return ret;
+                    }
+                    else
+                    {
+                        return NULL;
+                    }
+                }
+            }
+        }
+    }
+    return NULL;
+}

preprocess/macro_include.h

@@ -0,0 +1,14 @@
+#ifndef MACRO_INCLUDE_H
+#define MACRO_INCLUDE_H
+
+#include "../linked_list.h"
+#include "preprocess.h"
+
+typedef struct
+{
+    char *path;
+} seg_include;
+
+seg *parse_include(prep_ctx *ctx, typed_token **tkns_ptr);
+
+#endif

preprocess/preprocess.c

@@ -8,6 +8,7 @@
 #include "preprocess.h"
 #include "macro_define.h"
 #include "macro_call.h"
+#include "macro_include.h"
 #include "token.h"
 
 typed_token *chain_tokens(linked_list *tkns)
@@ -27,18 +28,22 @@ typed_token *chain_tokens(linked_list *tkns)
     return res_first;
 }
 
-typed_token *preprocess(typed_token *tkn)
+typed_token *preprocess(prep_ctx *ctx, char *path)
 {
+    char *prev_path = ctx->curr_path;
+    ctx->curr_path = path;
+    typed_token *tkn = tokenize_file(path);
     linked_list *ret = new_linked_list();
-    prep_ctx *ctx = (prep_ctx *)malloc(sizeof(prep_ctx));
-    ctx->defs = new_linked_list();
+
     while (tkn->type_id != TKN_EOF)
     {
         seg *s = NULL;
         if (!s)
             s = parse_token(ctx, &tkn);
         if (!s)
             s = parse_define(ctx, &tkn);
+        if (!s)
+            s = parse_include(ctx, &tkn);
         if (!s)
             s = parse_macro_call(ctx, &tkn);
 
@@ -55,5 +60,7 @@ typed_token *preprocess(typed_token *tkn)
         }
     }
 
+    ctx->curr_path = prev_path;
+
     return chain_tokens(ret);
 }

preprocess/preprocess.h

@@ -5,6 +5,7 @@
 
 typedef struct
 {
+    char *curr_path;
     linked_list *defs;
 } prep_ctx;
 
@@ -15,6 +16,6 @@ typedef struct seg_
     linked_list *(*to_code)(struct seg_ *self, prep_ctx *ctx);
 } seg;
 
-typed_token *preprocess(typed_token *tkn);
+typed_token *preprocess(prep_ctx *ctx, char *path);
 
 #endif

tests/output/inp_include.c_asm_output.asm

@@ -1,19 +1,8 @@
-TKN_INT
-TKN_ID(function2)
-TKN_L_PAREN
-TKN_R_PAREN
-TKN_L_BRACE
-TKN_RETURN
-TKN_LIT_INT(13)
-TKN_SEMICOLON
-TKN_R_BRACE
-TKN_EOF
 section .data
 __function2_size: equ 16
 __function_size: equ 16
-__printf_size: equ 16
-__temp_str_3 db `function: %d\n`, 0
-__temp_str_4 db `function2: %d\n`, 0
+__temp_str_2 db `function: %d\n`, 0
+__temp_str_3 db `function2: %d\n`, 0
 __main_size: equ 96
 section .text
 global function2
@@ -46,7 +35,7 @@ main:
 push rbp
 mov rbp, rsp
 sub rsp, __main_size
-mov rax, __temp_str_3
+mov rax, __temp_str_2
 mov [rsp+0], rax
 mov rax, rsp
 add rax, 0
@@ -61,7 +50,7 @@ mov rax, rsp
 add rax, 0
 mov [rsp+32], rax
 call printf
-mov rax, __temp_str_4
+mov rax, __temp_str_3
 mov [rsp+40], rax
 mov rax, rsp
 add rax, 0

tests/output/inp_include.c_lex_output.txt

@@ -1,3 +1,7 @@
+TKN_DIRECTIVE:
+  TKN_ID(include)
+  TKN_LIT_STR(header1.h)
+  TKN_EOF
 TKN_VOID
 TKN_ID(printf)
 TKN_L_PAREN