identify some tokens

twasm/asm/main.asm

@@ -912,6 +912,71 @@ evaluate_constant:
     mov rdx, 0xFF ; unrecognised type
     ret
 
+; ------------------------------------------------------------------------------
+; identify_register
+;
+; description:
+; takes a register in ascii-encoded text and returns its token ID or
+; UNRECOGNISED_TOKEN_ID if not recognised
+;
+; parameters:
+; edi = register to be searched
+;
+; returned:
+; ax = register's token ID or UNRECOGNISED_TOKEN_ID
+; ------------------------------------------------------------------------------
+
+identify_register:
+  xor eax, eax ; tokens.registers + eax -> entry in tokens.registers
+  .loop:
+    cmp eax, (tokens.registers_end - tokens.registers)
+    jge .not_found
+
+    cmp edi, [tokens.registers + eax]
+    je .found
+
+    add eax, 6
+    jmp .loop
+  .found:
+    mov ax, [tokens.registers + eax + 4]
+    ret
+  .not_found:
+    mov ax, UNRECOGNISED_TOKEN_ID
+    ret
+
+; ------------------------------------------------------------------------------
+; identify_operator
+; TODO combine with identify_register
+;
+; description:
+; takes an operator in ascii-encoded text and returns its token ID or
+; UNRECOGNISED_TOKEN_ID if not recognised
+;
+; parameters:
+; edi = operator to be searched
+;
+; returned:
+; ax = operator's token ID or UNRECOGNISED_TOKEN_ID
+; ------------------------------------------------------------------------------
+
+identify_operator:
+  xor eax, eax ; tokens.operators + eax -> entry in tokens.operators
+  .loop:
+    cmp eax, (tokens.operators_end - tokens.operators)
+    jge .not_found
+
+    cmp edi, [tokens.operators + eax]
+    je .found
+
+    add eax, 6
+    jmp .loop
+  .found:
+    mov ax, [tokens.operators + eax + 4]
+    ret
+  .not_found:
+    mov ax, UNRECOGNISED_TOKEN_ID
+    ret
+
 ; ------------------------------------------------------------------------------
 ; utilities
 ; ------------------------------------------------------------------------------

twasm/asm/tests.asm

@@ -34,6 +34,12 @@ run_tests:
   call clear_test_arena
   call test_evaluate_constant
 
+  call clear_test_arena
+  call test_identify_register
+
+  call clear_test_arena
+  call test_identify_operator
+
   ret
   .msg db "running test suite...", 0x0A, 0x00
 
@@ -405,6 +411,7 @@ test_evaluate_constant:
   .case0c_solution dq "char"
   .case1c db '"chars"'
   .case1c_solution dq "chars"
+  ; " wow my editor really doesn't like highlighting quotes correctly
 
   .case0q db "0q31103755242102"
   .case0q_solution dq 0q31103755242102
@@ -412,6 +419,84 @@ test_evaluate_constant:
   .case0b db "0b0110011001101001011100100111001101110100001000000011011000110100"
   .case0b_solution dq 0b0110011001101001011100100111001101110100001000000011011000110100
 
+
+; ------------------------------------------------------------------------------
+; test_identify_register
+;
+; description:
+; tests identify_register described funtionality
+; ------------------------------------------------------------------------------
+
+test_identify_register:
+  mov rsi, .msg
+  call print.test
+
+  mov edi, "rcx"
+  call identify_register
+  cmp ax, 0x0002
+  jne .fail
+
+  mov edi, "RaNd"
+  call identify_register
+  cmp ax, UNRECOGNISED_TOKEN_ID
+  jne .fail
+
+  mov edi, ""
+  call identify_register
+  cmp ax, UNRECOGNISED_TOKEN_ID
+  jne .fail
+
+  .pass:
+    mov rsi, msg_pass
+    call print
+    ret
+  .fail:
+    mov rsi, msg_fail
+    call print
+    ret
+  .msg db "test_identify_register...", 0x00
+
+; ------------------------------------------------------------------------------
+; test_identify_operator
+;
+; description:
+; tests identify_operator described funtionality
+; ------------------------------------------------------------------------------
+
+test_identify_operator:
+  mov rsi, .msg
+  call print.test
+
+  mov edi, "xor"
+  call identify_operator
+  cmp ax, 0x0053
+  jne .fail
+
+  mov edi, [tokens.operators_end]
+  call identify_operator
+  cmp ax, UNRECOGNISED_TOKEN_ID
+  jne .fail
+
+  mov edi, "RaNd"
+  call identify_operator
+  cmp ax, UNRECOGNISED_TOKEN_ID
+  jne .fail
+
+  mov edi, ""
+  call identify_operator
+  cmp ax, UNRECOGNISED_TOKEN_ID
+  jne .fail
+
+  .pass:
+    mov rsi, msg_pass
+    call print
+    ret
+  .fail:
+    mov rsi, msg_fail
+    call print
+    ret
+  .msg db "test_identify_operator...", 0x00
+
 msg_pass:
   db 0x0A
   times (TEST_LINE_LENGTH + .start - .end) db " ", ; right align