rearrange test case

twasm/asm/main.asm

@@ -732,8 +732,7 @@ tokenise:
 ; | 0x00 | 0x | hexidecimal  |
 ; | 0x01 | 0q | octal        |
 ; | 0x02 | 0b | binary       |
-; | 0x03 | '' | ascii char   |
+; | 0x03 | "  | char         |
-; | 0x04 | "" | ascii string |
 ; | 0xFF |    | unrecognised |
 ;
 ; where `p.` is the prefix or otherwise indicator
@@ -763,19 +762,12 @@ evaluate_constant:
   cmp dl, '0'
   je .numeric
 
-  ; char case
+  ; chr case
   mov rcx, 0x03
   push rcx
-  cmp dl, "'"
-  je .chr
-  pop rcx
-
-  ; str case
-  mov rcx, 0x04
-  push rcx
   xor ecx, ecx ; rcx = number of times right-rolled
   cmp dl, '"'
-  je .str
+  je .chr
   pop rcx
 
   jmp .unrecognised
@@ -874,26 +866,10 @@ evaluate_constant:
     jmp .bin_loop
 
   .chr:
-    mov al, [rdi]
+    cmp rcx, 4       ; ensure char is only 4 bytes long
-
-    ; bound check byte as printable char
-    cmp al, 0x20
-    jl .unrecognised
-    cmp al, 0x7E
     jg .unrecognised
-
-    dec rsi
-    inc rdi
-    mov dl, [rdi]
-    cmp dl, "'"
-    jne .unrecognised
-
-    jmp .break
-
-  .str:
-    ; TODO range check rcx / return string longer as a register
     cmp rsi, 1 ; range check
-    je .str_break
+    je .chr_break
 
     ror rax, 8
     inc rcx
@@ -911,17 +887,17 @@ evaluate_constant:
     dec rsi
     inc rdi
 
-    jmp .str
+    jmp .chr
 
-  .str_break:
+  .chr_break:
     cmp rcx, 1             ; for each [1..rcx]
-    jle .str_break_for_good
+    jle .chr_break_for_good
     rol rax, 8             ; roll left to make up for the roll right earlier
     dec rcx
-    jmp .str_break
+    jmp .chr_break
 
-  .str_break_for_good:
+  .chr_break_for_good:
-    mov dl, [rdi]
+    mov dl, [rdi] ; make sure the chr is closed
     cmp dl, '"'
     jne .unrecognised
 
@@ -932,9 +908,75 @@ evaluate_constant:
     ret
 
   .unrecognised:
+    pop rdx
     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
 
@@ -50,31 +56,31 @@ test_elemb:
 
   ; [0]
   mov rdi, 5
-  mov rsi, test_elemb_5
+  mov rsi, .case1
-  mov dl, [test_elemb_5]
+  mov dl, [.case1]
   call elemb
   cmp al, 1
   jne .fail
 
   ; [n - 1]
   mov rdi, 5
-  mov rsi, test_elemb_5
+  mov rsi, .case1
-  mov dl, [test_elemb_5 + 4]
+  mov dl, [.case1 + 4]
   call elemb
   cmp al, 1
   jne .fail
 
   ; [1]
   mov rdi, 5
-  mov rsi, test_elemb_5
+  mov rsi, .case1
-  mov dl, [test_elemb_5 + 1]
+  mov dl, [.case1 + 1]
   call elemb
   cmp al, 1
   jne .fail
 
   ; not present
   mov rdi, 5
-  mov rsi, test_elemb_5
+  mov rsi, .case1
   mov dl, 0xDA
   call elemb
   cmp al, 0
@@ -82,7 +88,7 @@ test_elemb:
 
   ; 0 length list
   mov rdi, 0
-  mov rsi, test_elemb_0
+  mov rsi, .case0
   mov dl, 0x34
   call elemb
   cmp al, 0
@@ -96,6 +102,8 @@ test_elemb:
     mov rsi, msg_fail
     call print
     ret
+  .case0: ; [This Page Intentionally Left Blank]
+  .case1 db 0x54, 0x00, 0x21, 0x20, 0x34
   .msg db "test_elemb...", 0x00
 
 ; ------------------------------------------------------------------------------
@@ -370,20 +378,19 @@ test_evaluate_constant:
 
   ; char
   mov rdi, .case0c
-  mov rsi, 3
+  mov rsi, 6
   call evaluate_constant
   cmp rax, [.case0c_solution]
   jne .fail
   cmp rdx, 0x03
   jne .fail
 
-  ; str
+  ; oversized char
-  mov rdi, .case0s
+  mov rdi, .case1c
-  mov rsi, 5
+  mov rsi, 7
   call evaluate_constant
-  cmp rax, [.case0s_solution]
+  cmp rdx, 0xFF
   jne .fail
-  cmp rdx, 0x04
 
   .pass:
     mov rsi, msg_pass
@@ -402,10 +409,11 @@ test_evaluate_constant:
   .case2h_solution dq 0x1234567890
   .case3h db "0x243F6A8885A308D3"
   .case3h_solution dq 0x243F6A8885A308D3
-  .case0c db "' '"
+  .case0c db '"char"'
-  .case0c_solution dq ' '
+  .case0c_solution dq "char"
-  .case0s db '"str"'
+  .case1c db '"chars"'
-  .case0s_solution dq "str"
+  .case1c_solution dq "chars"
+  ; " wow my editor really doesn't like highlighting quotes correctly
 
   .case0q db "0q31103755242102"
   .case0q_solution dq 0q31103755242102
@@ -413,6 +421,84 @@ 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
@@ -423,6 +509,3 @@ msg_fail:
   times (TEST_LINE_LENGTH + .start - .end) db " ",
   .start db "failed."
   .end db 0x0A, 0x00
-
-test_elemb_0: ; [This Page Intentionally Left Blank]
-test_elemb_5 db 0x54, 0x00, 0x21, 0x20, 0x34