From 1ce0d45242097a07b7a4ee539a074ec812851a58 Mon Sep 17 00:00:00 2001
From: Mistivia <i@mistivia.com>
Date: Wed, 5 Mar 2025 18:31:55 +0800
Subject: finish parser

---
 src/as_op.c        | 28 ++++++++++++++++++++++++++++
 src/as_op.h        | 14 ++++++++++++++
 src/as_parser.c    | 52 ++++++++++++++++++++++++++++++++++++++++++----------
 src/as_parser.h    | 24 +++++++-----------------
 src/as_tokenizer.c |  6 +++---
 src/as_tokenizer.h |  7 ++++++-
 6 files changed, 100 insertions(+), 31 deletions(-)
 create mode 100644 src/as_op.c
 create mode 100644 src/as_op.h

diff --git a/src/as_op.c b/src/as_op.c
new file mode 100644
index 0000000..847214d
--- /dev/null
+++ b/src/as_op.c
@@ -0,0 +1,28 @@
+#include "as_op.h"
+
+#include <string.h>
+
+struct opTableEntry{
+    Op op;
+    const char* name;
+};
+typedef struct opTableEntry OpTableEntry;
+
+OpTableEntry opTable[] = {
+    {ADD, "add"},
+    {SUB, "sub"},
+    {MUL, "mul"},
+    {DIV, "div"},
+    {MOD, "mod"},
+    {EQ, "eq"},
+    {OPEND, NULL}
+};
+
+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;
+        }
+    }
+    return OPEND;
+}
diff --git a/src/as_op.h b/src/as_op.h
new file mode 100644
index 0000000..ec88670
--- /dev/null
+++ b/src/as_op.h
@@ -0,0 +1,14 @@
+#ifndef FVM_AS_OP_H_
+#define FVM_AS_OP_H_
+
+enum op {
+    ADD, SUB, MUL, DIV, MOD, EQ,
+    // place holder for the end of the table
+    OPEND
+};
+typedef enum op Op;
+
+Op str2op(const char *str);
+
+#endif
+
diff --git a/src/as_parser.c b/src/as_parser.c
index d80cdd9..b419be5 100644
--- a/src/as_parser.c
+++ b/src/as_parser.c
@@ -7,9 +7,9 @@
 //
 // <prog> ::= <stmts>
 // <stmts> ::= <stmt> <newline> | <stmt> <stmts>
-// <stmt> ::= <tag> <instr> | <instr> | <tag>
-// <instr> ::= <op> | <op> <arg> | <op> <label>
-// <tag> ::= <label> :
+// <stmt> ::= <label> <instr> | <instr> | <label>
+// <instr> ::= <op> | <op> <arg> | <op> <tag>
+// <label> ::= <tag> :
 // <op> ::= add | sub | mul | div | mod | eq
 
 
@@ -38,9 +38,25 @@ Stmts* parseStmts(TokenStream *ts) {
     return ss;
 }
 
+Label* parseLabel(TokenStream *ts) {
+    Token *t = nextToken(ts);
+    if (t->type != TAG) {
+        fprintf(stderr, "%d:%d Expect LABEL.\n", t->line, t->col);
+        exit(-1);
+    }
+    Label *l = malloc(sizeof(Label));
+    l->name = t->sval;
+    t = nextToken(ts);
+    if (t->type != COLON) {
+        fprintf(stderr, "%d:%d Expect COLON.\n", t->line, t->col);
+        exit(-1);
+    }
+    return l;
+}
+
 Stmt* parseStmt(TokenStream *ts) {
     Token *t = peekToken(ts);
-    if (t->type == LABEL) {
+    if (t->type == TAG) {
         Label *l = parseLabel(ts);
         t = peekToken(ts);
         if (t->type == COLON) {
@@ -49,23 +65,40 @@ Stmt* parseStmt(TokenStream *ts) {
         } else {
             Instr *i = parseInstr(ts);
             Stmt *s = malloc(sizeof(Stmt));
-            s->tag = l;
+            s->label = l;
             s->instr = i;
             return s;
         }
     } else {
         Instr *i = parseInstr(ts);
         Stmt *s = malloc(sizeof(Stmt));
-        s->tag = NULL;
+        s->label = NULL;
         s->instr = i;
         return s;
     }
 }
 
+// parse op
+Op parseOp(TokenStream *ts) {
+    Token *t = nextToken(ts);
+    Op op;
+    if (t->type == OP) {
+        op = str2op(t->sval);
+        if (op == OPEND) {
+            fprintf(stderr, "%d:%d Invalid OP.\n", t->line, t->col);
+            exit(-1);
+        }
+    } else {
+        fprintf(stderr, "%d:%d Expect OP.\n", t->line, t->col);
+        exit(-1);
+    }
+    return op;
+}
+
 Instr* parseInstr(TokenStream *ts) {
     Token *t = nextToken(ts);
     Instr *i = malloc(sizeof(Instr));
-    i->labelName = NULL;
+    i->tagName = NULL;
     if (t->type == OP) {
         i->op = parseOp(ts);
         t = peekToken(ts);
@@ -73,11 +106,10 @@ Instr* parseInstr(TokenStream *ts) {
             Arg *a = malloc(sizeof(Arg));
             a->ival = t->ival;
             a->fval = t->fval;
-            a->sval = t->sval;
             i->arg = a;
             nextToken(ts);
-        } else if (t->type == LABEL) {
-            i->labelName = t->sval;
+        } else if (t->type == TAG) {
+            i->tagName = t->sval;
             nextToken(ts);
         }
     }
diff --git a/src/as_parser.h b/src/as_parser.h
index 268f45a..3306909 100644
--- a/src/as_parser.h
+++ b/src/as_parser.h
@@ -1,31 +1,20 @@
+#ifndef FVM_AS_PARSER_H_
+#define FVM_AS_PARSER_H_
+
 #include "as_tokenizer.h"
 
-// BNF
-// ===
-//
-// <prog> ::= <stmts>
-// <stmts> ::= <stmt> <newline> | <stmt> <stmts>
-// <stmt> ::= <tag> <instr> | <instr> | <tag>
-// <instr> ::= <op> | <op> <arg> | <op> <label>
-// <tag> ::= <label> :
-// <op> ::= add | sub | mul | div | mod | eq
-
-enum op {
-    ADD, SUB, MUL, DIV, MOD, EQ
-};
-typedef enum op Op;
+#include "as_op.h"
 
 struct arg {
     int64_t ival;
     double fval;
-    const char *sval;
 };
 typedef struct arg Arg;
 
 struct instr {
     Op op;
     Arg* arg;
-    const char* labelName;
+    const char* tagName;
 };
 typedef struct instr Instr;
 
@@ -35,7 +24,7 @@ struct label {
 typedef struct label Label;
 
 struct stmt {
-    Label* tag;
+    Label* label;
     Instr* instr;
 };
 typedef struct stmt Stmt;
@@ -57,3 +46,4 @@ Instr* parseInstr(TokenStream *ts);
 Label* parseLabel(TokenStream *ts);
 Op parseOp(TokenStream *ts);
 
+#endif
diff --git a/src/as_tokenizer.c b/src/as_tokenizer.c
index 7cf244a..23c0ce1 100644
--- a/src/as_tokenizer.c
+++ b/src/as_tokenizer.c
@@ -132,7 +132,7 @@ Token nextTokenImpl(InputStream *s) {
             if (isOp(sval)) {
                 return (Token){.type = OP, .sval = sval, .line = line, .col = col};
             }
-            return (Token){.type = LABEL, .sval = sval, .line = line, .col = col};
+            return (Token){.type = TAG, .sval = sval, .line = line, .col = col};
         }
         fprintf(stderr, "error: invalid character %c at line %d, col %d\n", c, s->line, s->col);
     }
@@ -164,11 +164,11 @@ void printToken(Token *t) {
     switch (t->type) {
         case OP:
             printf("OP: %s, line: %d, col: %d\n", t->sval, t->line, t->col);
- 狗太厉害了           break;
+            break;
         case ARG:
             printf("ARG: %ld, line: %d, col: %d\n", t->ival, t->line, t->col);
             break;
-        case LABEL:
+        case TAG:
             printf("LABEL: %s, line: %d, col: %d\n", t->sval, t->line, t->col);
             break;
         case COLON:
diff --git a/src/as_tokenizer.h b/src/as_tokenizer.h
index fef8625..377aca0 100644
--- a/src/as_tokenizer.h
+++ b/src/as_tokenizer.h
@@ -1,8 +1,11 @@
+#ifndef FMV_AS_TOKENIZER_H_
+#define FMV_AS_TOKENIZER_H_
+
 #include <stdint.h>
 #include <stdio.h>
 
 typedef enum {
-    OP, ARG, LABEL, COLON, NEWLINE, ENDOFFILE
+    OP, ARG, TAG, COLON, NEWLINE, ENDOFFILE
 } TokenType;
 
 typedef struct {
@@ -34,3 +37,5 @@ Token *nextToken(TokenStream *ts);
 Token *peekToken(TokenStream *ts);
 void printToken(Token *t);
 TokenStream* makeTokenStream(FILE* fp);
+
+#endif // FMV_AS_TOKENIZER_H_
-- 
cgit v1.0