From 869420ef7ab2bfa8bc04c44584e55bb337d83f32 Mon Sep 17 00:00:00 2001
From: andromeda <andromeda@lenovo>
Date: Mon, 9 Mar 2026 11:00:59 +0100
Subject: [PATCH] fix some bugs, work on assembler

---
 twasm/asm/main.asm  | 72 +++++++++++++++++++++++++++++++++++++--------
 twasm/asm/tests.asm | 10 +++++++
 2 files changed, 69 insertions(+), 13 deletions(-)

diff --git a/twasm/asm/main.asm b/twasm/asm/main.asm
index 7128a7f..57af625 100644
--- a/twasm/asm/main.asm
+++ b/twasm/asm/main.asm
@@ -37,8 +37,11 @@ start:
   mov rdi, program      ; -> program
   mov rsi, [program.size] ; = size of program
   call tokenise
-
+  ; rax = number of tokens processed
+  mov rdi, rax
+  push rdi
   call clear_output_arena
+  pop rdi
   call assemble
 
   jmp halt
@@ -49,30 +52,63 @@ start:
 
 ; ------------------------------------------------------------------------------
 ; assemble
-; TODO write tests
+; TODO write testsr
 ; TODO make it work :/ putting the cart before the horse
 ;
 ; description:
 ; assembles the program from tokens located at TOKEN_TABLE_ADDR into a flat
 ; binary located at OUTPUT_ADDR. It's probably desirable to clear the output
 ; arena before calling this function.
+;
+; parameters:
+; rdi = number of tokens in the token table
 ; ------------------------------------------------------------------------------
 
 assemble:
   xor rax, rax ; number of tokens processed
   .loop:
-    cmp rax, TOKEN_TABLE_SIZE / TOKEN_TABLE_ENTRY_SIZE ; check incrementer
-                                                       ; against the number of
-                                                       ; entries in the token
-                                                       ; table
-    jg .break                                          ; if overflown, break
+    cmp rax, rdi ; check incrementer against the number of tokens in the token
+    jge .break    ; table. If overflown, break
 
-    mov rdi, [rax * TOKEN_TABLE_ENTRY_SIZE + TOKEN_TABLE_ADDR] ; rdi -> next
+    push rdi
+    xor edi, edi
+    mov di, [rax * TOKEN_TABLE_ENTRY_SIZE + TOKEN_TABLE_ADDR] ; rdi = next tte
+    push rax
+    xor eax, eax
+    call get_tte_type
+
+    cmp ax, 0x01 ; check if it's an operator
+    je .operator
+    jne .continue_operator
+
+  .operator
+    push rsi
+    mov rsi, .msg_found_operator
+    call print
+    pop rsi
+
+  .continue_operator
+    cmp ax, 0x02 ; check if it's a register
+    je .register
+    jne .continue_register
+
+  .register
+    push rsi
+    mov rsi, .msg_found_register
+    call print
+    pop rsi
+
+  .continue_register
+    pop rax ; incrementer
+    pop rdi ; total number of tokens
 
     inc rax ; move to next token
     jmp .loop
+
   .break:
     ret
+  .msg_found_operator db "found operator", 0x0A, 0x00
+  .msg_found_register db "found register", 0x0A, 0x00
 
 ; ------------------------------------------------------------------------------
 ; get_tte_type
@@ -85,12 +121,13 @@ assemble:
 ; di = token table entry
 ;
 ; returned:
-; al = type of token, or UNRECOGNISED_ID_TYPE. The upper 4 bytes of al are
+; al = type of token, or UNRECOGNISED_ID_TYPE. The upper 4 bits of al are
 ;      zeroed; the rest of rax is zeroed.
 ; ------------------------------------------------------------------------------
 
 get_tte_type:
   and rdi, 0xFFFF ; mask input so it behaves as expected
+  xor eax, eax
 
   .loop:
     cmp rax, (tokens.by_id_end - tokens.by_id) / 4 ; make sure it's still in range
@@ -104,11 +141,12 @@ get_tte_type:
     inc rax
     jmp .loop
   .not_found:
-    xor rax, rax
     mov al, UNRECOGNISED_ID_TYPE
+    and ax, 0xF ; mask as expected
     ret
   .found:
     mov al, [2 + tokens.by_id + rax * 4]
+    and ax, 0xF ; mask as expected
     ret
 
 ; ------------------------------------------------------------------------------
@@ -128,6 +166,7 @@ get_tte_type:
 
 get_tte_typed_metadata:
   and rdi, 0xFFFF ; mask input so it behaves as expected
+  xor eax, eax
 
   .loop:
     cmp rax, (tokens.by_id_end - tokens.by_id) / 4 ; make sure it's still in range
@@ -141,7 +180,7 @@ get_tte_typed_metadata:
     inc rax
     jmp .loop
   .not_found:
-    xor rax, rax
+    xor eax, eax
     mov al, UNRECOGNISED_ID_METADATA
     ret
   .found:
@@ -163,11 +202,14 @@ get_tte_typed_metadata:
 ; parameters:
 ; rdi -> first byte of program
 ; rsi = size of program in bytes
+;
+; returned:
+; rax = number of tokens processed
 ; ------------------------------------------------------------------------------
 
 tokenise:
   add rsi, rdi ; last byte of program
-  xor rcx, rcx ; number of tokens processed
+  xor ecx, ecx ; number of tokens processed
   .loop:
     cmp rdi, rsi ; if current byte greater than last byte
     jg .break    ; then break
@@ -218,16 +260,20 @@ tokenise:
 
   .close_bracket:
     ; rewrite open bracket token entry with a filled out one
+    push rcx
+
     mov dl, [.data_open_bracket]
     sub cl, dl
     mov byte [TOKEN_TABLE_ADDR + rdx * TOKEN_TABLE_ENTRY_SIZE], cl
     mov byte [1 + TOKEN_TABLE_ADDR + rdx * TOKEN_TABLE_ENTRY_SIZE], 0x10
-    add cl, dl
+
+    pop rcx
 
   .continue_close_bracket:
     inc rcx ; +1 token processed
     jmp .loop
   .break:
+    mov rax, rcx
     ret
 
   .data_open_bracket db 0x00 ; represents the token # of the latest open bracket
diff --git a/twasm/asm/tests.asm b/twasm/asm/tests.asm
index d5341f0..f48d8bf 100644
--- a/twasm/asm/tests.asm
+++ b/twasm/asm/tests.asm
@@ -403,6 +403,11 @@ test_get_tte_type:
   cmp al, 0x02      ; register
   jne .fail
 
+  mov di, 0x0056    ; mov
+  call get_tte_type
+  cmp al, 0x01      ; operator
+  jne .fail
+
   mov di, 0xFFFF    ; unrecognised token
   call get_tte_type
   cmp al, UNRECOGNISED_ID_TYPE
@@ -439,6 +444,11 @@ test_get_tte_typed_metadata:
   cmp al, 0x03                ; width: 64 bits
   jne .fail
 
+  mov di, 0x0056              ; mov
+  call get_tte_typed_metadata
+  cmp al, 0x02                ; # operands
+  jne .fail
+
   mov di, 0xFFFF              ; unrecognised token
   call get_tte_typed_metadata
   cmp al, UNRECOGNISED_ID_METADATA