Standardise string encoding with constant variable

Addresses #1, with #6 in mind
jakewilliami · Sep 1, 2022 · 1c1ba4b · 1c1ba4b
1 parent e38e925
commit 1c1ba4b
Show file tree

Hide file tree

Showing 2 changed files with 23 additions and 21 deletions.
diff --git a/src/HiddenFiles.jl b/src/HiddenFiles.jl
@@ -71,7 +71,7 @@ export ishidden
         # https://github.com/osquery/osquery/blob/598983db97459f858e7a9cc5c731409ffc089b48/osquery/tables/system/darwin/extended_attributes.cpp#L111-L144
         # https://github.com/objective-see/ProcInfo/blob/ec51090fcf741a9e045dd3e5119cb5cc8750efd3/procInfo/Binary.m#L121-L172
         # NOTE: this function will fail if you give it f as "/"
-        function _k_mditem_content_type_tree(f::AbstractString, str_encoding::Unsigned = K_CFSTRING_ENCODING_MACROMAN)
+        function _k_mditem_content_type_tree(f::AbstractString, str_encoding::Unsigned = CF_STRING_ENCODING)
             cfstr = _cfstring_create_with_cstring(f, str_encoding)
             mditem = _mditem_create(cfstr)
             mdattrs = _mditem_copy_attribute(mditem, K_MDITEM_CONTENT_TYPE_TREE)
@@ -81,7 +81,7 @@ export ishidden
             for i in 0:(cfarr_len - 1)
                 attr = _cfarray_get_value_at_index(mdattrs, i)
                 if attr != C_NULL #&& !iszero(_cfstring_get_length(attr))
-                    push!(content_types, _string_from_cf_string(attr))
+                    push!(content_types, _string_from_cf_string(attr, str_encoding))
                 end
             end
             return content_types

diff --git a/src/utils/darwin.jl b/src/utils/darwin.jl
@@ -1,24 +1,26 @@
 # https://opensource.apple.com/source/CF/CF-635/CFString.h.auto.html
 # https://developer.apple.com/documentation/corefoundation/cfstringbuiltinencodings
-const K_CFSTRING_ENCODING_MACROMAN = 0x0
-const K_CFSTRING_ENCODING_WINDOWSLATIN1 = 0x0500 # ANSI codepage 1252
-const K_CFSTRING_ENCODING_ISOLATIN1 = 0x0201     # ISO 8859-1
-const K_CFSTRING_ENCODING_NEXTSTEPLATIN = 0x0B01 # NextStep encoding
-const K_CFSTRING_ENCODING_ASCII = 0x0600         # 0..127 (in creating CFString, values greater than 0x7F are treated as corresponding Unicode value)
-const K_CFSTRING_ENCODING_UNICODE = 0x0100       # kTextEncodingUnicodeDefault  + kTextEncodingDefaultFormat (aka kUnicode16BitFormat)
-const K_CFSTRING_ENCODING_UTF8 = 0x08000100      # kTextEncodingUnicodeDefault + kUnicodeUTF8Format
-const K_CFSTRING_ENCODING_NONLOSSYASCII = 0x0BFF # 7bit Unicode variants used by Cocoa & Java
-const K_CFSTRING_ENCODING_UTF16 = 0x0100         # kTextEncodingUnicodeDefault + kUnicodeUTF16Format (alias of kCFStringEncodingUnicode)
-const K_CFSTRING_ENCODING_UTF16BE = 0x10000100   # kTextEncodingUnicodeDefault + kUnicodeUTF16BEFormat
-const K_CFSTRING_ENCODING_UTF16LE = 0x14000100   # kTextEncodingUnicodeDefault + kUnicodeUTF16LEFormat
-const K_CFSTRING_ENCODING_UTF32 = 0x0c000100     # kTextEncodingUnicodeDefault + kUnicodeUTF32Format
-const K_CFSTRING_ENCODING_UTF32BE = 0x18000100   # kTextEncodingUnicodeDefault + kUnicodeUTF32BEFormat
-const K_CFSTRING_ENCODING_UTF32LE = 0x1c000100   # kTextEncodingUnicodeDefault + kUnicodeUTF32LEFormat
-# https://opensource.apple.com/source/CF/CF-368/String.subproj/CFStringUtilities.c.auto.html
-const K_CF_STRING_ENCODING_UTF8 = UInt32(65001)
+const K_CF_STRING_ENCODING_MAC_ROMAN       = 0x0
+const K_CF_STRING_ENCODING_WINDOWS_LATIN_1 = 0x0500      # ANSI codepage 1252
+const K_CF_STRING_ENCODING_ISO_LATIN_1     = 0x0201      # ISO 8859-1
+const K_CF_STRING_ENCODING_NEXT_STEP_LATIN = 0x0B01      # NextStep encoding
+const K_CF_STRING_ENCODING_ASCII           = 0x0600      # 0..127 (in creating CFString, values greater than 0x7F are treated as corresponding Unicode value)
+const K_CF_STRING_ENCODING_UNICODE         = 0x0100      # kTextEncodingUnicodeDefault  + kTextEncodingDefaultFormat (aka kUnicode16BitFormat)
+const K_CF_STRING_ENCODING_UTF8            = 0x08000100  # kTextEncodingUnicodeDefault + kUnicodeUTF8Format
+const K_CF_STRING_ENCODING_NON_LOSSY_ASCII = 0x0BFF      # 7bit Unicode variants used by Cocoa & Java
+const K_CF_STRING_ENCODING_UTF16           = 0x0100      # kTextEncodingUnicodeDefault + kUnicodeUTF16Format (alias of kCFStringEncodingUnicode)
+const K_CF_STRING_ENCODING_UTF16BE         = 0x10000100  # kTextEncodingUnicodeDefault + kUnicodeUTF16BEFormat
+const K_CF_STRING_ENCODING_UTF16LE         = 0x14000100  # kTextEncodingUnicodeDefault + kUnicodeUTF16LEFormat
+const K_CF_STRING_ENCODING_UTF32           = 0x0c000100  # kTextEncodingUnicodeDefault + kUnicodeUTF32Format
+const K_CF_STRING_ENCODING_UTF32BE         = 0x18000100  # kTextEncodingUnicodeDefault + kUnicodeUTF32BEFormat
+const K_CF_STRING_ENCODING_UTF32LE         = 0x1c000100  # kTextEncodingUnicodeDefault + kUnicodeUTF32LEFormat
+const K_CF_STRING_ENCODING_UTF8            = 0x08000100  # https://opensource.apple.com/source/CF/CF-368/String.subproj/CFStringUtilities.c.auto.html
+
+# This will be out main/default string encoding
+const CF_STRING_ENCODING = K_CF_STRING_ENCODING_MAC_ROMAN # K_CF_STRING_ENCODING_UTF8 or UTF16 doesn't seem to work
 
 # https://developer.apple.com/documentation/corefoundation/1542942-cfstringcreatewithcstring        
-function _cfstring_create_with_cstring(s::AbstractString, encoding::Unsigned = K_CFSTRING_ENCODING_MACROMAN)
+function _cfstring_create_with_cstring(s::AbstractString, encoding::Unsigned = CF_STRING_ENCODING)
     cfstr = ccall(:CFStringCreateWithCString, Cstring, 
                   (Ptr{Cvoid}, Cstring, UInt32),
                   C_NULL, s, encoding)
@@ -56,7 +58,7 @@ function _cfstring_get_length(cfstr::Cstring)
 end
 
 # https://developer.apple.com/documentation/corefoundation/1542143-cfstringgetmaximumsizeforencodin
-function _cfstring_get_maximum_size_for_encoding(strlen::T, encoding::Unsigned = K_CFSTRING_ENCODING_MACROMAN) where {T <: Integer}
+function _cfstring_get_maximum_size_for_encoding(strlen::T, encoding::Unsigned = CF_STRING_ENCODING) where {T <: Integer}
     return ccall(:CFStringGetMaximumSizeForEncoding, Int32, (Int32, UInt32), strlen, encoding)
 end
 
@@ -67,7 +69,7 @@ end
 
 # https://github.com/vovkasm/input-source-switcher/blob/c5bab3de716db5e3dae3703ed3b72f2bf1cd51d3/utils.cpp#L9-L18
 # https://www.tabnine.com/code/java/methods/org.eclipse.swt.internal.webkit.WebKit_win32/CFStringGetCharactersPtr
-function _string_from_cf_string(cfstr::Cstring, encoding::Unsigned = K_CFSTRING_ENCODING_MACROMAN)
+function _string_from_cf_string(cfstr::Cstring, encoding::Unsigned = CF_STRING_ENCODING)
     strlen = _cfstring_get_length(cfstr)
     maxsz = _cfstring_get_maximum_size_for_encoding(strlen, encoding)
     cfio = IOBuffer()