From f18199c5d0eb133d49eee3b40c795fc716631145 Mon Sep 17 00:00:00 2001 From: PleegWat Date: Thu, 3 Sep 2026 16:45:04 +0200 Subject: [PATCH] Only allow 0o prefix for octal, not just 0 --- strint.asm | 53 ++++++++++++++++++++++++------------------------- strint_test.asm | 13 ++++++------ 2 files changed, 33 insertions(+), 33 deletions(-) diff --git a/strint.asm b/strint.asm index 84bebbc..a807bf2 100644 --- a/strint.asm +++ b/strint.asm @@ -1,4 +1,4 @@ -; Internal convention: +; Internal register assignments: ; r1 - Partially parsed output integer ; r2 - Parsing position ; r3 - Character being parsed @@ -13,6 +13,7 @@ ; Clobbers: ; flags ; r3 - last character read +; r4 - Negative marker ; Note: Unless the input is "0", tail-calls into a base-specific specialization. pub atoi: mov r2, r1 @@ -23,26 +24,20 @@ cmp r3, 0x2D ; '-' jne atoi_positive sub r4, zr, 1 ; Set r4 to -1 add r2, r2, 1 -load_8 r3, [r2] atoi_positive: -cmp r3, 0x30 ; '0' -jne atoi_dec_loop ; No prefix, parse decimal -add r2, r2, 1 -load_8 r3, [r2] ; From "boxBOX", or an octal digit -cmp r3, 0x30 ; '0' -jl atoi_done -cmp r3, 0x37 ; '7' -jle atoi_oct_loop ; Octal with just '0' prefix -add r2, r2, 1 ; 2-byte prefix -and r3, r3, 0x5F ; to uppercase -cmp r3, 0x42 ; 'B' +load_16 r3, [r2] ; 2-byte prefix "0b", "0o", "0x", etc. +or r3, r3, 0x20 ; 2nd char to lower case +add r2, r2, 2 +cmp r3, 0x3062 ; "0b" je atoi_bin_loop -cmp r3, 0x4F ; 'O' +cmp r3, 0x306F ; "0o" je atoi_oct_loop -cmp r3, 0x58 ; 'X' +cmp r3, 0x3078 ; "0x" je atoi_hex_loop +sub r2, r2, 2 ; no matching prefix, move pointer back +jmp atoi_dec_loop -; FALLTHROUGH: not a valid prefix, but we did see a zero +; Epilogue atoi_done: add r1, r1, r4 ; If r4 is -1, negate r1. Else it's 0 and no effect. xor r1, r1, r4 @@ -58,6 +53,7 @@ jmp r13 ; Clobbers: ; flags ; r3 - last character read +; r4 - Negative marker pub atoi_dec: mov r2, r1 mov r1, 0 @@ -76,7 +72,7 @@ cmp r3, 0x39 ; '9' jg atoi_done add r2, r2, 1 sub r3, r3, 0x30 ; '0' -lsl flags, r1, 2 +lsl flags, r1, 2 ; Use flags to help multiply by 10 add r1, r1, flags lsl r1, r1, 1 add r1, r1, r3 @@ -91,6 +87,7 @@ jmp atoi_dec_loop ; Clobbers: ; flags ; r3 - last character read +; r4 - Negative marker pub atoi_bin: mov r2, r1 mov r1, 0 @@ -102,8 +99,8 @@ sub r4, zr, 1 ; Set r4 to -1 add r2, r2, 1 atoi_bin_positive: load_16 r3, [r2] ; check for prefix -and r3, r3, 0xFF5F ; to uppercase -cmp r3, 0x3042 ; "0B" +or r3, r3, 0x20 ; 2nd char to lower case +cmp r3, 0x3062 ; "0b" jne atoi_bin_loop add r2, r2, 2 atoi_bin_loop: @@ -127,6 +124,7 @@ jmp atoi_bin_loop ; Clobbers: ; flags ; r3 - last character read +; r4 - Negative marker pub atoi_oct: mov r2, r1 mov r1, 0 @@ -138,8 +136,8 @@ sub r4, zr, 1 ; Set r4 to -1 add r2, r2, 1 atoi_oct_positive: load_16 r3, [r2] ; check for prefix -and r3, r3, 0xFF5F ; to uppercase -cmp r3, 0x304F ; "0O" +or r3, r3, 0x20 ; 2nd char to lower case +cmp r3, 0x306F ; "0o" jne atoi_oct_loop add r2, r2, 2 atoi_oct_loop: @@ -163,6 +161,7 @@ jmp atoi_oct_loop ; Clobbers: ; flags ; r3 - last character read +; r4 - Negative marker pub atoi_hex: mov r2, r1 mov r1, 0 @@ -174,8 +173,8 @@ sub r4, zr, 1 ; Set r4 to -1 add r2, r2, 1 atoi_hex_positive: load_16 r3, [r2] ; check for prefix -and r3, r3, 0xFF5F ; to uppercase -cmp r3, 0x3058 ; "0X" +or r3, r3, 0x20 ; 2nd char to lower case +cmp r3, 0x3078 ; "0x" jne atoi_oct_loop add r2, r2, 2 atoi_hex_loop: @@ -191,13 +190,13 @@ add r1, r1, r3 jmp atoi_hex_loop atoi_hex_letter: -and r3, r3, 0x5F ; to uppercase -cmp r3, 0x41 ; 'A' +or r3, r3, 0x20 ; to lower case +cmp r3, 0x61 ; 'a' jl atoi_done -cmp r3, 0x46 ; 'F' +cmp r3, 0x66 ; 'f' jg atoi_done add r2, r2, 1 -sub r3, r3, 0x37 ; 'A' - 10 +sub r3, r3, 0x57 ; 'a' - 10 lsl r1, r1, 4 add r1, r1, r3 jmp atoi_hex_loop diff --git a/strint_test.asm b/strint_test.asm index 407443b..e44c051 100644 --- a/strint_test.asm +++ b/strint_test.asm @@ -46,9 +46,10 @@ tests: @0x1040 U32 42 U8 2 "42\0" @0x1050 U32 67 U8 2 "67lol\0" @0x1060 U32 0x69a U8 5 "0x69a\0" -@0x1070 U32 0o23 U8 3 "0239\0" -@0x1080 U32 0b1011 U8 6 "0b1011\0" -@0x1090 U32 0b10001 U8 7 "0b100012\0" -@0x10a0 U32 0xFFFFFFEB U8 3 "-21\0" ; Yuk -@0x10b0 -end_of_tests: \ No newline at end of file +@0x1070 U32 0x4B4 U8 5 "0x4B4\0" +@0x1080 U32 0o23 U8 4 "0o239\0" +@0x1090 U32 0b1011 U8 6 "0b1011\0" +@0x10a0 U32 0b10001 U8 7 "0b100012\0" +@0x10b0 U32 0xFFFFFFEB U8 3 "-21\0" ; Yuk +@0x10c0 +end_of_tests: