language-servers/roc
1
0
Fork 0
mirror of https://github.com/roc-lang/roc.git synced 2025-09-26 13:29:12 +00:00
Code Issues Projects Releases Packages Wiki Activity Actions 6

fix accessing bug

Browse source
This commit is contained in:
Brendan Hansknecht 2023-03-15 09:16:19 -07:00
parent 88a69a23a8
commit 0f708d7577
No known key found for this signature in database
GPG key ID: 0EA784685083E75B
1 changed files with 7 additions and 8 deletions
Download patch file Download diff file Expand all files Collapse all files

15
crates/compiler/builtins/bitcode/src/str.zig
View file

@ -1811,7 +1811,7 @@ pub fn fromUtf8RangeC(
pub fn fromUtf8Range(arg: RocList, start: usize, count: usize, update_mode: UpdateMode) FromUtf8Result { pub fn fromUtf8Range(arg: RocList, start: usize, count: usize, update_mode: UpdateMode) FromUtf8Result {
const bytes = @ptrCast([*]const u8, arg.bytes)[start..count]; const bytes = @ptrCast([*]const u8, arg.bytes)[start..count];
if (isValidUnicode(@ptrCast([*]const u8, bytes), bytes.len)) { if (isValidUnicode(bytes)) {
// the output will be correct. Now we need to clone the input // the output will be correct. Now we need to clone the input
// TODO: rework this to properly take advantage fo seamless slices. // TODO: rework this to properly take advantage fo seamless slices.
@ -1877,9 +1877,7 @@ fn errorToProblem(bytes: [*]u8, length: usize) struct { index: usize, problem: U
unreachable; unreachable;
} }
pub fn isValidUnicode(ptr: [*]const u8, len: usize) callconv(.C) bool { pub fn isValidUnicode(buf: []const u8) bool {
const buf: []const u8 = ptr[0..len];
const size = @sizeOf(u64); const size = @sizeOf(u64);
// TODO: we should test changing the step on other platforms. // TODO: we should test changing the step on other platforms.
// The general tradeoff is making extremely large strings potentially much faster // The general tradeoff is making extremely large strings potentially much faster
@ -1887,7 +1885,7 @@ pub fn isValidUnicode(ptr: [*]const u8, len: usize) callconv(.C) bool {
const step = size; const step = size;
var i: usize = 0; var i: usize = 0;
while (i + step < buf.len) { while (i + step < buf.len) {
var bytes: u64 = 0; var bytes: u64 = undefined;
@memcpy(@ptrCast([*]u8, &bytes), @ptrCast([*]const u8, buf) + i, size); @memcpy(@ptrCast([*]u8, &bytes), @ptrCast([*]const u8, buf) + i, size);
const unicode_bytes = bytes & 0x8080_8080_8080_8080; const unicode_bytes = bytes & 0x8080_8080_8080_8080;
if (unicode_bytes == 0) { if (unicode_bytes == 0) {
@ -1899,21 +1897,22 @@ pub fn isValidUnicode(ptr: [*]const u8, len: usize) callconv(.C) bool {
while (buf[i] >= 0b1000_0000) { while (buf[i] >= 0b1000_0000) {
// This forces prefetching, otherwise the loop can run at about half speed. // This forces prefetching, otherwise the loop can run at about half speed.
var small_buf = [4]u8{ 0, 0, 0, 0 }; var small_buf: [4]u8 = undefined;
@memcpy(&small_buf, @ptrCast([*]const u8, buf) + i, size); @memcpy(&small_buf, @ptrCast([*]const u8, buf) + i, 4);
// TODO: Should we always inline these function calls below? // TODO: Should we always inline these function calls below?
if (std.unicode.utf8ByteSequenceLength(small_buf[0])) |cp_len| { if (std.unicode.utf8ByteSequenceLength(small_buf[0])) |cp_len| {
if (std.meta.isError(std.unicode.utf8Decode(small_buf[0..cp_len]))) { if (std.meta.isError(std.unicode.utf8Decode(small_buf[0..cp_len]))) {
return false; return false;
} }
i += cp_len; i += cp_len;
if (i == buf.len) return true; if (i + 4 >= buf.len) break;
} else |_| { } else |_| {
return false; return false;
} }
} }
} }
if (i == buf.len) return true;
while (buf[i] < 0b1000_0000) { while (buf[i] < 0b1000_0000) {
i += 1; i += 1;
if (i == buf.len) return true; if (i == buf.len) return true;