From b83187f66175d93a0dba45f6d110ed94badac7c5 Mon Sep 17 00:00:00 2001
From: Mistivia <i@mistivia.com>
Date: Sun, 16 Mar 2025 20:01:42 +0800
Subject: refactor using allocator pattern

---
 src/as_main.c      |   7 ++--
 src/as_op.c        |   7 ++--
 src/as_op.h        |   3 +-
 src/as_parser.c    | 103 +++++++++++++++++++++++++++++------------------------
 src/as_parser.h    |  39 ++++++++++----------
 src/as_tokenizer.c |  49 +++++++++++++------------
 src/as_tokenizer.h |  39 +++++++++++---------
 src/utils.c        |  37 +++++++++++++++++++
 src/utils.h        |  14 ++++++++
 9 files changed, 185 insertions(+), 113 deletions(-)
 create mode 100644 src/utils.c
 create mode 100644 src/utils.h

diff --git a/src/as_main.c b/src/as_main.c
index eb5de20..2809c59 100644
--- a/src/as_main.c
+++ b/src/as_main.c
@@ -1,15 +1,18 @@
 #include <stdio.h>
 
 #include "as_tokenizer.h"
+#include "utils.h"
 
 int main(int argc, char** argv) {
     if (argc != 2) {
         fprintf(stderr, "Usage: fvm-as <inputfile>\n");
         return 1;
     }
-
+    Allocator alct = newAllocator();
     FILE *fp = fopen(argv[1], "r");
-    TokenStream* ts = makeTokenStream(fp);
+    TokenStream ts = makeTokenStream(alct, fp);
+
+    deleteAllocator(alct);
     return 0;
 }
 
diff --git a/src/as_op.c b/src/as_op.c
index 847214d..88884b8 100644
--- a/src/as_op.c
+++ b/src/as_op.c
@@ -3,12 +3,11 @@
 #include <string.h>
 
 struct opTableEntry{
-    Op op;
+    enum op op;
     const char* name;
 };
-typedef struct opTableEntry OpTableEntry;
 
-OpTableEntry opTable[] = {
+struct opTableEntry opTable[] = {
     {ADD, "add"},
     {SUB, "sub"},
     {MUL, "mul"},
@@ -18,7 +17,7 @@ OpTableEntry opTable[] = {
     {OPEND, NULL}
 };
 
-Op str2op(const char* str) {
+enum op str2op(const char* str) {
     for (int i = 0; opTable[i].name != NULL; i++) {
         if (strcmp(opTable[i].name, str) == 0) {
             return opTable[i].op;
diff --git a/src/as_op.h b/src/as_op.h
index ec88670..dcd77e0 100644
--- a/src/as_op.h
+++ b/src/as_op.h
@@ -6,9 +6,8 @@ enum op {
     // place holder for the end of the table
     OPEND
 };
-typedef enum op Op;
 
-Op str2op(const char *str);
+enum op str2op(const char *str);
 
 #endif
 
diff --git a/src/as_parser.c b/src/as_parser.c
index b419be5..f848730 100644
--- a/src/as_parser.c
+++ b/src/as_parser.c
@@ -1,36 +1,46 @@
 #include "as_parser.h"
+#include "as_tokenizer.h"
+#include "utils.h"
 
+#include <stdio.h>
 #include <stdlib.h>
 
 // BNF
 // ===
 //
 // <prog> ::= <stmts>
-// <stmts> ::= <stmt> <newline> | <stmt> <stmts>
+// <stmts> ::= <stmt> newline | <stmt> newline <stmts>
 // <stmt> ::= <label> <instr> | <instr> | <label>
-// <instr> ::= <op> | <op> <arg> | <op> <tag>
-// <label> ::= <tag> :
-// <op> ::= add | sub | mul | div | mod | eq
+// <instr> ::= <op> | <op> arg | <op> tag
+// <label> ::= tag ":"
+// <op> ::= "add" | "sub" | "mul" | "div" | "mod" | "eq"
 
 
-Prog* parseProg(TokenStream *ts) {
-    Prog *p = malloc(sizeof(Prog));
-    p->stmts = parseStmts(ts);
+Prog parseProg(Allocator alct, TokenStream ts) {
+    Prog p = allocate(alct, sizeof(struct prog));
+    p->stmts = parseStmts(alct, ts);
     return p;
 }
 
-Stmts* parseStmts(TokenStream *ts) {
-    Stmts *ss = malloc(sizeof(Stmts));
-    ss->stmts = malloc(sizeof(Stmt*));
+Stmts parseStmts(Allocator alct, TokenStream ts) {
+    Stmts ss = allocate(alct, sizeof(struct stmts));
+    ss->stmts = allocate(alct, sizeof(Stmt));
     ss->stmts[0] = NULL;
     int capacity = 0;
     int len = 0;
-    while (peekToken(ts)->type != ENDOFFILE) {
-        Stmt *s = parseStmt(ts);
+    while (peekToken(alct, ts)->type != ENDOFFILE) {
+        Stmt s = parseStmt(alct, ts);
+        if (s == NULL) continue;
         if (len == capacity) {
             capacity = capacity * 2 + 1;
             ss->stmts = realloc(ss->stmts, sizeof(Stmt*) * capacity);
         }
+        // expect newline
+        if (peekToken(alct, ts)->type == NEWLINE) {
+            nextToken(alct, ts);
+        } else {
+            fprintf(stderr, "%d:%d Expect NEWLINE.\n", peekToken(alct, ts)->line, peekToken(alct, ts)->col);
+        }
         ss->stmts[len] = s;
         len++;
     }
@@ -38,15 +48,15 @@ Stmts* parseStmts(TokenStream *ts) {
     return ss;
 }
 
-Label* parseLabel(TokenStream *ts) {
-    Token *t = nextToken(ts);
+Label parseLabel(Allocator alct, TokenStream ts) {
+    Token t = nextToken(alct, ts);
     if (t->type != TAG) {
         fprintf(stderr, "%d:%d Expect LABEL.\n", t->line, t->col);
         exit(-1);
     }
-    Label *l = malloc(sizeof(Label));
+    Label l = allocate(alct, sizeof(Label));
     l->name = t->sval;
-    t = nextToken(ts);
+    t = nextToken(alct, ts);
     if (t->type != COLON) {
         fprintf(stderr, "%d:%d Expect COLON.\n", t->line, t->col);
         exit(-1);
@@ -54,34 +64,33 @@ Label* parseLabel(TokenStream *ts) {
     return l;
 }
 
-Stmt* parseStmt(TokenStream *ts) {
-    Token *t = peekToken(ts);
+Stmt parseStmt(Allocator alct, TokenStream ts) {
+    Token t = peekToken(alct, ts);
+    Stmt stmt = allocate(alct, sizeof(struct stmt));
+    stmt->label = NULL;
+    stmt->instr = NULL;
     if (t->type == TAG) {
-        Label *l = parseLabel(ts);
-        t = peekToken(ts);
-        if (t->type == COLON) {
-            nextToken(ts);
-            return (Stmt*)l;
-        } else {
-            Instr *i = parseInstr(ts);
-            Stmt *s = malloc(sizeof(Stmt));
-            s->label = l;
-            s->instr = i;
-            return s;
+        stmt->label = parseLabel(alct, ts);
+        if (peekToken(alct, ts)->type == NEWLINE) {
+            return stmt;
         }
-    } else {
-        Instr *i = parseInstr(ts);
-        Stmt *s = malloc(sizeof(Stmt));
-        s->label = NULL;
-        s->instr = i;
-        return s;
     }
+    if (t->type == OP) {
+        stmt->instr = parseInstr(alct, ts);
+        if (peekToken(alct, ts)->type == NEWLINE) {
+            return stmt;
+        }
+    }
+    if (t->type == NEWLINE) {
+        return NULL;
+    }
+    fprintf(stderr, "%d:%d Expect lable + instruction, lable, or instruction.", t->line, t->col);
+    exit(-1);
 }
 
-// parse op
-Op parseOp(TokenStream *ts) {
-    Token *t = nextToken(ts);
-    Op op;
+enum op parseOp(Allocator alct, TokenStream ts) {
+    Token t = nextToken(alct, ts);
+    enum op op;
     if (t->type == OP) {
         op = str2op(t->sval);
         if (op == OPEND) {
@@ -95,22 +104,22 @@ Op parseOp(TokenStream *ts) {
     return op;
 }
 
-Instr* parseInstr(TokenStream *ts) {
-    Token *t = nextToken(ts);
-    Instr *i = malloc(sizeof(Instr));
+Instr parseInstr(Allocator alct, TokenStream ts) {
+    Token t = nextToken(alct, ts);
+    Instr i = allocate(alct, sizeof(struct instr));
     i->tagName = NULL;
     if (t->type == OP) {
-        i->op = parseOp(ts);
-        t = peekToken(ts);
+        i->op = parseOp(alct, ts);
+        t = peekToken(alct, ts);
         if (t->type == ARG) {
-            Arg *a = malloc(sizeof(Arg));
+            Arg a = allocate(alct, sizeof(struct arg));
             a->ival = t->ival;
             a->fval = t->fval;
             i->arg = a;
-            nextToken(ts);
+            nextToken(alct, ts);
         } else if (t->type == TAG) {
             i->tagName = t->sval;
-            nextToken(ts);
+            nextToken(alct, ts);
         }
     }
     return i;
diff --git a/src/as_parser.h b/src/as_parser.h
index 3306909..a3db024 100644
--- a/src/as_parser.h
+++ b/src/as_parser.h
@@ -2,6 +2,7 @@
 #define FVM_AS_PARSER_H_
 
 #include "as_tokenizer.h"
+#include "utils.h"
 
 #include "as_op.h"
 
@@ -9,41 +10,41 @@ struct arg {
     int64_t ival;
     double fval;
 };
-typedef struct arg Arg;
+typedef struct arg * Arg;
 
 struct instr {
-    Op op;
-    Arg* arg;
+    enum op op;
+    Arg arg;
     const char* tagName;
 };
-typedef struct instr Instr;
+typedef struct instr *Instr;
 
 struct label {
     const char* name;
 };
-typedef struct label Label;
+typedef struct label *Label;
 
 struct stmt {
-    Label* label;
-    Instr* instr;
+    Label label;
+    Instr instr;
 };
-typedef struct stmt Stmt;
+typedef struct stmt *Stmt;
 
 struct stmts {
-    Stmt** stmts;
+    Stmt* stmts;
 };
-typedef struct stmts Stmts;
+typedef struct stmts *Stmts;
 
 struct prog {
-    Stmts *stmts;
+    Stmts stmts;
 };
-typedef struct prog Prog;
-
-Prog* parseProg(TokenStream *ts);
-Stmt* parseStmt(TokenStream *ts);
-Stmts* parseStmts(TokenStream *ts);
-Instr* parseInstr(TokenStream *ts);
-Label* parseLabel(TokenStream *ts);
-Op parseOp(TokenStream *ts);
+typedef struct prog *Prog;
+
+Prog parseProg(Allocator alct, TokenStream ts);
+Stmt parseStmt(Allocator alct, TokenStream ts);
+Stmts parseStmts(Allocator alct, TokenStream ts);
+Instr parseInstr(Allocator alct, TokenStream ts);
+Label parseLabel(Allocator alct, TokenStream ts);
+enum op parseOp(Allocator alct, TokenStream ts);
 
 #endif
diff --git a/src/as_tokenizer.c b/src/as_tokenizer.c
index 23c0ce1..4cb695e 100644
--- a/src/as_tokenizer.c
+++ b/src/as_tokenizer.c
@@ -3,8 +3,9 @@
 #include <string.h>
 
 #include "as_tokenizer.h"
+#include "utils.h"
 
-int InputStream_nextChar(InputStream *s) {
+int InputStream_nextChar(InputStream s) {
     if (s->cursor == -1) {
         return EOF;
     }
@@ -28,7 +29,7 @@ int InputStream_nextChar(InputStream *s) {
     return c;
 }
 
-int InputStream_peekChar(InputStream *s) {
+int InputStream_peekChar(InputStream s) {
     if (s->cursor == -1) {
         return EOF;
     }
@@ -79,7 +80,8 @@ int isPartOfIndentifier(int c) {
     return 0;
 }
 
-Token nextTokenImpl(InputStream *s) {
+Token nextTokenImpl(Allocator alct, InputStream s) {
+    Token t = allocate(alct, sizeof(struct token));
     int c;
     while (1) {
         c = InputStream_peekChar(s);
@@ -88,12 +90,13 @@ Token nextTokenImpl(InputStream *s) {
         }
         if (c == '\n') {
             InputStream_nextChar(s);
-            Token t = (Token){.type = NEWLINE, .line = s->line, .col = s->col};
+            *t = (struct token){.type = NEWLINE, .line = s->line, .col = s->col};
             return t;
         }
         if (c == ':') {
             InputStream_nextChar(s);
-            return (Token){.type = COLON, .line = s->line, .col = s->col};
+            *t = (struct token){.type = COLON, .line = s->line, .col = s->col};
+            return t;
         }
         if (c == ' ' || c == '\t') {
             InputStream_nextChar(s);
@@ -109,12 +112,13 @@ Token nextTokenImpl(InputStream *s) {
                     break;
                 }
             } 
-            return (Token){.type = ARG, .ival = ival, .line = s->line, .col = s->col};
+            *t = (struct token){.type = ARG, .ival = ival, .line = s->line, .col = s->col};
+            return t;
         }
         if (isStartOfIndentifier(c)) {
             size_t line = s->line;
             size_t col = s->col;
-            char *sval = malloc(256);
+            char *sval = allocate(alct, 256);
             size_t i = 0;
             while(1) {
                 if (i >= 255) {
@@ -130,37 +134,38 @@ Token nextTokenImpl(InputStream *s) {
             }
             sval[i] = '\0';
             if (isOp(sval)) {
-                return (Token){.type = OP, .sval = sval, .line = line, .col = col};
+                *t = (struct token){.type = OP, .sval = sval, .line = line, .col = col};
+                return t;
             }
-            return (Token){.type = TAG, .sval = sval, .line = line, .col = col};
+            *t = (struct token){.type = TAG, .sval = sval, .line = line, .col = col};
+            return t;
         }
         fprintf(stderr, "error: invalid character %c at line %d, col %d\n", c, s->line, s->col);
     }
     // end of file
-    return (Token){.type = ENDOFFILE};
+    *t = (struct token){.type = ENDOFFILE};
+    return t;
 }
 
-Token *nextToken(TokenStream *ts) {
+Token nextToken(Allocator alct, TokenStream ts) {
     if (ts->buf != NULL) {
-        Token *t = ts->buf;
+        Token t = ts->buf;
         ts->buf = NULL;
         return t;
     }
-    Token *t = malloc(sizeof(Token));
-    *t = nextTokenImpl(ts->s);
+    Token t = nextTokenImpl(alct, ts->s);
     return t;
 }
 
-Token *peekToken(TokenStream *ts) {
+Token peekToken(Allocator alct, TokenStream ts) {
     if (ts->buf != NULL) {
         return ts->buf;
     }
-    ts->buf = malloc(sizeof(Token));
-    *(ts->buf) = nextTokenImpl(ts->s);
+    ts->buf = nextTokenImpl(alct, ts->s);
     return ts->buf;
 }
 
-void printToken(Token *t) {
+void printToken(Token t) {
     switch (t->type) {
         case OP:
             printf("OP: %s, line: %d, col: %d\n", t->sval, t->line, t->col);
@@ -183,15 +188,15 @@ void printToken(Token *t) {
     }
 }
 
-TokenStream* makeTokenStream(FILE* fp) {
-    InputStream *s = malloc(sizeof(InputStream));
+TokenStream makeTokenStream(Allocator alct, FILE* fp) {
+    InputStream s = allocate(alct, sizeof(struct inputStream));
     s->fp = fp;
-    s->buf = malloc(INPUT_STREAM_BUF_SIZE);
+    s->buf = allocate(alct, INPUT_STREAM_BUF_SIZE);
     s->buf_pos = 0;
     s->cursor = 0;
     s->line = 1;
     s->col = 1;
-    TokenStream *ts = malloc(sizeof(TokenStream));
+    TokenStream ts = allocate(alct, sizeof(struct tokenStream));
     ts->s = s;
     ts->buf = NULL;
     return ts;
diff --git a/src/as_tokenizer.h b/src/as_tokenizer.h
index 377aca0..e9025ba 100644
--- a/src/as_tokenizer.h
+++ b/src/as_tokenizer.h
@@ -4,38 +4,43 @@
 #include <stdint.h>
 #include <stdio.h>
 
-typedef enum {
+#include "utils.h"
+
+enum tokenType {
     OP, ARG, TAG, COLON, NEWLINE, ENDOFFILE
-} TokenType;
+};
 
-typedef struct {
-    TokenType type;
+struct token {
+    enum tokenType type;
     int line;
     int col;
     char *sval;
     int64_t ival;
     double fval;
-} Token;
+};
+typedef struct token *Token;
 
 #define INPUT_STREAM_BUF_SIZE 1024
 
-typedef struct {
+struct inputStream {
     FILE *fp;
     char *buf;
     int buf_pos;
     int cursor;
     int line;
     int col;
-} InputStream;
-
-typedef struct {
-    Token* buf;
-    InputStream *s;
-} TokenStream;
-
-Token *nextToken(TokenStream *ts);
-Token *peekToken(TokenStream *ts);
-void printToken(Token *t);
-TokenStream* makeTokenStream(FILE* fp);
+}; 
+typedef struct inputStream *InputStream;
+
+struct tokenStream {
+    Token buf;
+    InputStream s;
+};
+typedef struct tokenStream *TokenStream;
+
+Token nextToken(Allocator alct, TokenStream ts);
+Token peekToken(Allocator alct, TokenStream ts);
+void printToken(Token t);
+TokenStream makeTokenStream(Allocator alct, FILE* fp);
 
 #endif // FMV_AS_TOKENIZER_H_
diff --git a/src/utils.c b/src/utils.c
new file mode 100644
index 0000000..354159b
--- /dev/null
+++ b/src/utils.c
@@ -0,0 +1,37 @@
+#include "utils.h"
+
+struct allocator {
+    void** bufs;
+    size_t cap;
+    size_t len;
+};
+
+Allocator newAllocator() {
+    Allocator alct = malloc(sizeof(struct allocator));
+    alct->bufs = malloc(sizeof(void*) * 16);
+    alct->cap = 16;
+    alct->len = 0;
+    alct->bufs[0] = NULL;
+    return alct;
+}
+
+void deleteAllocator(Allocator alct) {
+    for (size_t i = 0; i < alct->len; i++) {
+        free(alct->bufs[i]);
+    }
+    free(alct);
+}
+
+void * allocate(Allocator alct, size_t size) {
+    if (alct->len >= alct->cap) {
+        alct->cap = alct->cap * 2; // Doubling the capacity
+        alct->bufs = realloc(alct->bufs, sizeof(void*) * alct->cap);
+    }
+
+    void* ptr = malloc(size); // Allocate requested size
+    alct->bufs[alct->len] = ptr; // Store pointer in array
+    alct->len++;
+    return ptr;
+}
+
+
diff --git a/src/utils.h b/src/utils.h
new file mode 100644
index 0000000..af5e820
--- /dev/null
+++ b/src/utils.h
@@ -0,0 +1,14 @@
+#ifndef FVM_UTILS_H_
+#define FVM_UTILS_H_
+
+#include <stdlib.h>
+
+struct allocator;
+typedef struct allocator *Allocator;
+
+Allocator newAllocator();
+void deleteAllocator(Allocator allocator);
+
+void* allocate(Allocator allocator, size_t size);
+
+#endif // FVM_UTILS_H_
-- 
cgit v1.0