Enchancement: use CFStringGetCharacters over iteration and CFStringGetCharacterAtIndex

[jakewilliami]

Currently we use CFStringGetCharacterAtIndex and iterate over the CF String to construct the String. This is obviously inefficient, and we should use CFStringGetCharacters instead.

Some example code which I haven't quite got working yet:

mutable struct CFRange
    length::Int
    location::Int
end
function _cfstring_get_characters(cfstr::Cstring, range::CFRange)
    charbuf = Vector{UInt8}(undef, range.length)
    ccall(:CFStringGetCharacters, Ptr{UInt32}, (Cstring, Ptr{Cvoid}, Ptr{Cvoid}), cfstr, pointer_from_objref(range), charbuf)
    return charbuf
end
function _string_from_cfstring(cfstr::Cstring, encoding::Unsigned = K_CFSTRING_ENCODING_MACROMAN)
    strlen = _cfstring_get_length(cfstr)
    maxsz = _cfstring_get_maximum_size_for_encoding(strlen, encoding)
    return _cfstring_get_characters(cfstr, CFRange(maxsz, 0))
end

Then, we will simple be able to write

function _string_from_cf_string(cfstr::Cstring)
    charbuf = _string_from_cfstring(cfstr)
    return String(charbuf)
end

[jakewilliami]

NOTE: when I work on this optimisation, it would be nice to benchmark the results and see how much more efficient it is.

[jakewilliami]

Just using initial tests, you can see how much overhead this might add:

julia> using HiddenFiles, BenchmarkTools

julia> ishidden("/System/Applications/Utilities/Terminal.app/Contents/MacOS/Terminal")
true

julia> @btime ishidden("/System/Applications/Utilities/Terminal.app/Contents/MacOS/Terminal");
  2.010 ms (114 allocations: 7.26 KiB)

julia> @btime ishidden("/System/Applications/Utilities/Terminal.app/Contents/");
  807.882 μs (62 allocations: 4.38 KiB)

[jakewilliami]

Might be able to use CFRangeMake

[jakewilliami]

Still not working

julia> using HiddenFiles

julia> mutable struct CFRange
           length::Int
           location::Int
       end

julia> cfstr = HiddenFiles._cfstring_create_with_cstring("Project.toml")
Cstring(0x000060000031c6a0)

julia> r = CFRange(10, 1)
CFRange(10, 1)

julia> charbuf = Vector{UInt16}(undef, 12);

julia> ccall(:CFStringGetCharacters, Cstring, (Cstring, Ptr{CFRange}, Ptr{Cvoid}), cfstr, pointer_from_objref(r), charbuf)

signal (11): Segmentation fault: 11
in expression starting at REPL[6]:1
__CFStrConvertBytesToUnicode at /System/Library/Frameworks/CoreFoundation.framework/Versions/A/CoreFoundation (unknown line

[jakewilliami]

Tried with Ref but still not working

julia> using HiddenFiles

julia> mutable struct CFRange
                  length::Int
                  location::Int
              end

julia> cfstr = HiddenFiles._cfstring_create_with_cstring("Project.toml")
Cstring(0x0000600000d2cae0)

julia> r = CFRange(10, 1)
CFRange(10, 1)

julia> charbuf = Vector{UInt16}(undef, 12);

julia> ccall(:CFStringGetCharacters, Cstring, (Cstring, Ref{CFRange}, Ptr{Cvoid}), cfstr, Ref{CFRange}(r), charbuf)