From 2e222c304abd45dc33c252da757119e588652100 Mon Sep 17 00:00:00 2001
From: andromeda <andromeda@lenovo>
Date: Wed, 18 Mar 2026 20:26:30 +0100
Subject: add evaluate_operand, fix operator tokenising

---
 twasm/asm/main.asm | 102 ++++++++++++++++++++++++++++++++++++++---------------
 1 file changed, 74 insertions(+), 28 deletions(-)

(limited to 'twasm/asm/main.asm')

diff --git a/twasm/asm/main.asm b/twasm/asm/main.asm
index 4b23759..8b2d7ef 100644
--- a/twasm/asm/main.asm
+++ b/twasm/asm/main.asm
@@ -578,8 +578,8 @@ tokenise:
     pop rsi
 
     push rax
-    mov dword [.pending_operator], 0 ; zero pending operator
-    xor eax, eax                     ; eax = number of bytes in operator
+    xor eax, eax                 ; eax = number of bytes in operator
+    mov [.pending_operator], eax ; zero pending operator
 
   .operator_loop:
     ; TODO give this its own error
@@ -606,35 +606,19 @@ tokenise:
     jmp .operator_loop ; and loop
 
   .operator_break:
-    pop rax
-
-    push rcx
-    mov rcx, tokens.operators ; rcx -> entry in lookup table
-
-  .operator_id_loop:
-    cmp rcx, tokens.operators_end ; check if index still in range
-    ; TODO give own error
-    jg .unexpected_operator       ; if not, error
-
-    ; TODO use something other than r8 and r9
-    mov r8d, [rcx]
-    mov r9d, [.pending_operator]
-    cmp r8d, r9d
-    je .found_id
-
-    add rcx, 6 ; next entry
-
-    jmp .operator_id_loop
+    ; rax already pushed from .operator
+    push rdi
 
-  .found_id:
-    push rdx
-    mov dx, [rcx + 4] ; dx = token id
+    mov edi, [.pending_operator] ; edi = operator to be searched
+    call identify_operator
+               ; ax = operator's token ID
+    mov cx, ax ; cx = operator's token ID for safe keeping
 
-    mov [TOKEN_TABLE_ADDR + rax * TOKEN_TABLE_ENTRY_SIZE], dx ; write to token
-    inc rax                                                   ; table
+    pop rdi ; rdi = byte counter
+    pop rax ; rax = tokens processed
 
-    pop rdx
-    pop rcx
+    mov [TOKEN_TABLE_ADDR + rax * TOKEN_TABLE_ENTRY_SIZE], cx
+    inc rax ; plus 1 token processed
 
     mov byte [.expecting], E_COMMENT | E_NEWLINE | E_WHITESPACE | E_OPERAND
     jmp .loop
@@ -650,6 +634,7 @@ tokenise:
 
     test byte [.expecting], E_OPERAND ; make sure an operand was expected
     jz .unexpected_operand            ; if not, error
+
   .operand_loop:
     mov dl, [rdi]
 
@@ -720,6 +705,67 @@ tokenise:
   .msg_operand db "operand.", 0x0A, 0x00
   .pending_operator dd 0 ; the operator token that is pending processing
 
+; ------------------------------------------------------------------------------
+; evaluate_operand
+;
+; description:
+; takes the location and length of an operand and evaluates it into binary data
+; and a return code to interpret the binary data.
+;
+; | code | rsi contents         | notes |
+; |------|----------------------|-------|
+; | 0x00 | token ID of register |       |
+; | 0xFF | -                    | error |
+;
+; parameters:
+; rdi -> first byte of operand
+; rsi = size of operand in bytes
+;
+; returned:
+; rax = binary data corresponding to the operand
+; dl = return code
+; ------------------------------------------------------------------------------
+
+evaluate_operand:
+  cmp rsi, 4
+  jg .unrecognised
+
+  .register:
+    push rdi
+    mov edi, [rdi] ; edi = register to be searched
+
+    ; TODO figure out how to mask elegantly :/
+    ; mask edi for lower rsi bits
+    cmp rsi, 4
+    je .register4
+    cmp rsi, 3
+    je .register3
+    cmp rsi, 2
+    je .register2
+    cmp rsi, 1
+    je .register1
+  .register1:
+    and edi, 0xFF
+  .register2:
+    and edi, 0xFFFF
+  .register3:
+    and edi, 0xFFFFFF
+  .register4:
+
+    call identify_register
+    ; ax = register's token ID or UNRECOGNISED_TOKEN_ID
+    pop rdi
+
+    cmp ax, UNRECOGNISED_TOKEN_ID
+    je .unrecognised
+
+    mov dl, 0x00
+    ret
+
+  .unrecognised:
+    mov dl, 0xFF
+    ret
+
 ; ------------------------------------------------------------------------------
 ; evaluate_constant
 ;
-- 
cgit v1.3.1