From 4a3350fe4ec27bd31acbae3454449dbbf9d34584 Mon Sep 17 00:00:00 2001
From: andromeda <andromeda@lenovo>
Date: Fri, 13 Mar 2026 22:49:15 +0100
Subject: [PATCH] get token id of operators

---
 twasm/asm/main.asm | 71 ++++++++++++++++++++++++++++++++++++++++++----
 1 file changed, 65 insertions(+), 6 deletions(-)

diff --git a/twasm/asm/main.asm b/twasm/asm/main.asm
index 22a1084..8fc7b5b 100644
--- a/twasm/asm/main.asm
+++ b/twasm/asm/main.asm
@@ -487,8 +487,9 @@ tokenise:
     push rdi
     push rax
     push rdx
-    mov rsi, whitespace_2 ; rsi -> list of whitespace (ignored) bytes
-    mov rdi,            2 ; rdi = size of list in bytes
+    ; TODO probably should not ignore null bytes
+    mov rsi, whitespace_3 ; rsi -> list of whitespace (ignored) bytes
+    mov rdi,            3 ; rdi = size of list in bytes
                           ; dl = current byte
     call elemb
     ; al = 0 if not whitespace, 1 if whitespace
@@ -568,47 +569,100 @@ tokenise:
     jmp .loop
 
   .operator:
+    ; debug message
     push rsi
     mov rsi, .found
     call print.debug
     mov rsi, .msg_operator
     call print
     pop rsi
+
+    push rax
+    mov dword [.pending_operator], 0 ; zero pending operator
+    xor eax, eax                     ; eax = number of bytes in operator
+
   .operator_loop:
+    ; TODO give this its own error
+    ; TODO make this pop rax
+    cmp eax, 4              ; check that operator is short enough
+    jg .unexpected_operator ; if not, error
+
     mov dl, [rdi] ; next byte
 
+    ; TODO have better check for operator end
     cmp dl, " "
     je .operator_break
     cmp dl, 0x0A
     je .operator_break
+    cmp dl, 0x00
+    je .operator_break
     cmp dl, ";"
     je .operator_break
 
-    inc rdi            ; inc byte counter
+    mov [.pending_operator + eax], dl
+
+    inc rax            ; inc byte counter
+    inc rdi            ; inc byte pointer
     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
+
+  .found_id
+    push rdx
+    mov dx, [rcx + 4] ; dx = token id
+
+    mov [TOKEN_TABLE_ADDR + rax * TOKEN_TABLE_ENTRY_SIZE], dx ; write to token
+    inc rax                                                   ; table
+
+    pop rdx
+    pop rcx
+
     mov byte [.expecting], E_COMMENT | E_NEWLINE | E_WHITESPACE | E_OPERAND
     jmp .loop
 
   .operand:
+    ; debug message
     push rsi
     mov rsi, .found
     call print.debug
     mov rsi, .msg_operand
     call print
     pop rsi
+
     test byte [.expecting], E_OPERAND ; make sure an operand was expected
     jz .unexpected_operand            ; if not, error
   .operand_loop:
     mov dl, [rdi]
+
     cmp dl, ","
     je .operand_break
     cmp dl, 0x0A
     je .operand_break
     cmp dl, 0x00
     je .operand_break
+
     inc rdi
     jmp .operand_loop
+
   .operand_break:
     mov byte [.expecting], E_COMMENT | E_NEWLINE | E_WHITESPACE | E_COMMA
     jmp .loop
@@ -650,6 +704,12 @@ tokenise:
     mov rsi, .msg_operand
     call print
     jmp halt
+  .unexpected_operator:
+    mov rsi, .err_unexpected
+    call print.error
+    mov rsi, .msg_operator
+    call print
+    jmp halt
   .err_unexpected db "unexpected ", 0x00
   .found db "found ", 0x00
   .msg_whitespace db "whitespace.", 0x0A, 0x00
@@ -658,6 +718,7 @@ tokenise:
   .msg_comma db "comma.", 0x0A, 0x00
   .msg_operator db "operator.", 0x0A, 0x00
   .msg_operand db "operand.", 0x0A, 0x00
+  .pending_operator dd 0 ; the operator token that is pending processing
 
 ; ------------------------------------------------------------------------------
 ; utilities
@@ -825,8 +886,6 @@ clear_output_arena:
 ; data
 ; ------------------------------------------------------------------------------
 
-
-
 tokens:
   .by_id:
     dw 0x0010    ; eax
@@ -1081,7 +1140,7 @@ msg_halt db "halted.", 0x0A, 0x00
 
 token_terminator_8 db 0x00, " ", 0x0A, 0x0D, ",", 0x00, 0x00, 0x00
 
-whitespace_2 db " ", 0x0D
+whitespace_3 db " ", 0x0D, 0x00
 
 ; test program
 program: