Add appropriate cconvert/unsafe_convert methods to avoid String alloc… (#56)

* Add appropriate cconvert/unsafe_convert methods to avoid String allocation in Base

* Address review comments

Author: quinnj

src/InlineStrings.jl

@@ -73,6 +73,8 @@ for sz in (1, 4, 8, 16, 32, 64, 128, 256)
     end
 end
 
+# used to zero out n lower bytes of an inline string
+clear_n_bytes(s, n) = Base.shl_int(Base.lshr_int(s, 8 * n), 8 * n)
 _bswap(x::T) where {T <: InlineString} = T === InlineString1 ? x : Base.bswap_int(x)
 
 const InlineStringTypes = Union{InlineString1,
@@ -147,6 +149,26 @@ function Base.Symbol(x::T) where {T <: InlineString}
         (Ref{T}, Int), ref, sizeof(x))
 end
 
+Base.cconvert(::Type{Ptr{UInt8}}, x::InlineString1) = Base.cconvert(Ptr{UInt8}, InlineString3(x))
+Base.cconvert(::Type{Ptr{Int8}}, x::InlineString1) = Base.cconvert(Ptr{Int8}, InlineString3(x))
+Base.cconvert(::Type{Cstring}, x::InlineString1) = Base.cconvert(Cstring, InlineString3(x))
+Base.cconvert(::Type{Ptr{UInt8}}, x::T) where {T <: InlineString} =
+    Ref{T}(_bswap(clear_n_bytes(x, 1)))
+Base.cconvert(::Type{Ptr{Int8}}, x::T) where {T <: InlineString} =
+    Ref{T}(_bswap(clear_n_bytes(x, 1)))
+function Base.cconvert(::Type{Cstring}, x::T) where {T <: InlineString}
+    ref = Ref{T}(_bswap(clear_n_bytes(x, 1)))
+    Base.containsnul(Ptr{Int8}(pointer_from_objref(ref)), sizeof(x)) &&
+        throw(ArgumentError("embedded NULs are not allowed in C strings: $x"))
+    return ref
+end
+Base.unsafe_convert(::Type{Ptr{UInt8}}, x::Ref{T}) where {T <: InlineString} =
+    Ptr{UInt8}(pointer_from_objref(x))
+Base.unsafe_convert(::Type{Ptr{Int8}}, x::Ref{T}) where {T <: InlineString} =
+    Ptr{Int8}(pointer_from_objref(x))
+Base.unsafe_convert(::Type{Cstring}, s::Ref{T}) where {T <: InlineString} =
+    Cstring(Base.unsafe_convert(Ptr{Cchar}, s))
+
 Base.Vector{UInt8}(s::InlineString) = Vector{UInt8}(codeunits(s))
 Base.Array{UInt8}(s::InlineString) = Vector{UInt8}(codeunits(s))
 
@@ -506,9 +528,6 @@ function Base.rstrip(f, s::InlineString)
     return nc == 0 ? s : _chopsuffix(s, nc)
 end
 
-# used to zero out n lower bytes of an inline string
-clear_n_bytes(s, n) = Base.shl_int(Base.lshr_int(s, 8 * n), 8 * n)
-
 Base.chomp(s::InlineString1) = chomp(String3(s))
 function Base.chomp(s::InlineString)
     i = lastindex(s)

test/runtests.jl

@@ -254,6 +254,9 @@ for S in SUBTYPES
     end
 end
 
+# can't contain NUL when converting to Cstring
+@test_throws ArgumentError Base.cconvert(Cstring, InlineString("a\0c"))
+
 end # @testset
 
 const STRINGS = ["", "🍕", "a", "a"^3, "a"^7, "a"^15, "a"^31, "a"^63, "a"^127, "a"^255]
@@ -339,6 +342,10 @@ const INLINES = map(InlineString, STRINGS)
         @test last(x, sizeof(x)) == last(y, sizeof(y))
         y != "" && @test last(x, sizeof(x) - 1) == last(y, sizeof(y) - 1)
         @test last(x, sizeof(x) + 1) == last(y, sizeof(y) + 1)
+        # https://github.com/JuliaDatabases/SQLite.jl/issues/306
+        @test unsafe_string(Base.unsafe_convert(Ptr{UInt8}, Base.cconvert(Ptr{UInt8}, x))) == y
+        @test unsafe_string(Base.unsafe_convert(Ptr{Int8}, Base.cconvert(Ptr{Int8}, x))) == y
+        @test unsafe_string(Base.unsafe_convert(Cstring, Base.cconvert(Cstring, x))) == y
     end
 end