add rough char/str parsing

2 files changed, 120 insertions, 21 deletions

diff --git a/twasm/asm/main.asm b/twasm/asm/main.asm
index c792eb9..2c28dbf 100644
--- a/twasm/asm/main.asm
+++ b/twasm/asm/main.asm
@@ -732,9 +732,11 @@ tokenise:
 ; | 0x00 | 0x | hexidecimal  |
 ; | 0x01 | 0q | octal        |
 ; | 0x02 | 0b | binary       |
+; | 0x03 | '' | ascii char   |
+; | 0x04 | "" | ascii string |
 ; | 0xFF |    | unrecognised |
 ;
-; where `p.` is the prefix
+; where `p.` is the prefix or otherwise indicator
 ;
 ; parameters:
 ; rdi -> first byte of constant
@@ -750,40 +752,62 @@ evaluate_constant:
                 ; rdi -> current byte of constant
   xor eax, eax  ; rax = value of constant
 
-  ; TODO fix this cheap trick xD
-  mov dl, [rdi]
-  dec rsi ; one fewer byte left
-  inc rdi ; point to next byte
-  cmp dl, '0'
-  jne .unrecognised
+  ; each case pushes the return value of dl into `rcx`, which is popped into dl
+  ; to return
 
   mov dl, [rdi]
   dec rsi ; one fewer byte left
   inc rdi ; point to next byte
 
-  ; hex case
-  mov rcx, 0x00
-  push rcx
-  cmp dl, 'x'
-  je .hex_loop
-  pop rcx
+  ; all numeric prefixes further handled in .numeric
+  cmp dl, '0'
+  je .numeric
 
-  ; octal case
-  mov rcx, 0x01
+  ; char case
+  mov rcx, 0x03
   push rcx
-  cmp dl, 'q'
-  je .oct_loop
+  cmp dl, "'"
+  je .chr
   pop rcx
 
-  ; binary case
-  mov rcx, 0x02
+  ; str case
+  mov rcx, 0x04
   push rcx
-  cmp dl, 'b'
-  je .bin_loop
+  xor ecx, ecx ; rcx = number of times right-rolled
+  cmp dl, '"'
+  je .str
   pop rcx
 
   jmp .unrecognised
 
+  .numeric:
+    mov dl, [rdi]
+    dec rsi ; one fewer byte left
+    inc rdi ; point to next byte
+
+    ; hex case
+    mov rcx, 0x00
+    push rcx
+    cmp dl, 'x'
+    je .hex_loop
+    pop rcx
+
+    ; octal case
+    mov rcx, 0x01
+    push rcx
+    cmp dl, 'q'
+    je .oct_loop
+    pop rcx
+
+    ; binary case
+    mov rcx, 0x02
+    push rcx
+    cmp dl, 'b'
+    je .bin_loop
+    pop rcx
+
+    jmp .unrecognised
+
   .hex_loop:
     cmp rsi, 0 ; make sure we're in range
     je .break  ; if not, break
@@ -849,6 +873,60 @@ evaluate_constant:
     inc rdi
     jmp .bin_loop
 
+  .chr:
+    mov al, [rdi]
+
+    ; 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
+
+    ror rax, 8
+    inc rcx
+
+    mov dl, [rdi]
+
+    ; bound check byte as printable char
+    cmp dl, 0x20
+    jl .unrecognised
+    cmp dl, 0x7E
+    jg .unrecognised
+
+    or al, dl
+
+    dec rsi
+    inc rdi
+
+    jmp .str
+
+  .str_break:
+    cmp rcx, 1             ; for each [1..rcx]
+    jle .str_break_for_good
+    rol rax, 8             ; roll left to make up for the roll right earlier
+    dec rcx
+    jmp .str_break
+
+  .str_break_for_good:
+    mov dl, [rdi]
+    cmp dl, '"'
+    jne .unrecognised
+
+    jmp .break
+
   .break:
     pop rdx
     ret
diff --git a/twasm/asm/tests.asm b/twasm/asm/tests.asm
index bc2ad84..4d0ddc8 100644
--- a/twasm/asm/tests.asm
+++ b/twasm/asm/tests.asm
@@ -368,6 +368,23 @@ test_evaluate_constant:
   cmp rdx, 0x02
   jne .fail
 
+  ; char
+  mov rdi, .case0c
+  mov rsi, 3
+  call evaluate_constant
+  cmp rax, [.case0c_solution]
+  jne .fail
+  cmp rdx, 0x03
+  jne .fail
+
+  ; str
+  mov rdi, .case0s
+  mov rsi, 5
+  call evaluate_constant
+  cmp rax, [.case0s_solution]
+  jne .fail
+  cmp rdx, 0x04
+
   .pass:
     mov rsi, msg_pass
     call print
@@ -385,6 +402,10 @@ test_evaluate_constant:
   .case2h_solution dq 0x1234567890
   .case3h db "0x243F6A8885A308D3"
   .case3h_solution dq 0x243F6A8885A308D3
+  .case0c db "' '"
+  .case0c_solution dq ' '
+  .case0s db '"str"'
+  .case0s_solution dq "str"
 
   .case0q db "0q31103755242102"
   .case0q_solution dq 0q31103755242102