Added support for read_xport

src/C_interface.jl

@@ -12,7 +12,7 @@ function readstat_get_file_format_version(metadata::Ptr{Nothing})
 end
 
 function readstat_get_row_count(metadata::Ptr{Nothing})
-    return ccall((:readstat_get_row_count, libreadstat), UInt, (Ptr{Nothing},), metadata)
+    return ccall((:readstat_get_row_count, libreadstat), Int, (Ptr{Nothing},), metadata)
 end
 
 function readstat_get_var_count(metadata::Ptr{Nothing})
@@ -69,4 +69,8 @@ end
 
 function readstat_parse(filename::String, type::Val{:sas7bdat}, parser::Ptr{Nothing}, ds::ReadStatDataFrame)
     return ccall((:readstat_parse_sas7bdat, libreadstat), Int, (Ptr{Nothing}, Cstring, Any), parser, string(filename), ds)
-end
+end
+
+function readstat_parse(filename::String, type::Val{:xport}, parser::Ptr{Nothing}, ds::ReadStatDataFrame)
+    return ccall((:readstat_parse_xport, libreadstat), Int, (Ptr{Nothing}, Cstring, Any), parser, string(filename), ds)
+end

src/ReadStat.jl

@@ -18,7 +18,7 @@ include(depsjl_path)
 using DataValues: DataValueVector
 using Dates
 
-export ReadStatDataFrame, read_dta, read_sav, read_por, read_sas7bdat
+export ReadStatDataFrame, read_dta, read_sav, read_por, read_sas7bdat, read_xport
 
 ##############################################################################
 ##
@@ -83,6 +83,8 @@ end
 
 include("C_interface.jl")
 
+const READSTAT_NALLOC_ROWS = 100000
+
 ##############################################################################
 ##
 ## Julia functions
@@ -154,7 +156,13 @@ function handle_variable!(var_index::Cint, variable::Ptr{Nothing},
     push!(ds.formats, get_format(variable))
     jtype = get_type(variable)
     push!(ds.types, jtype)
-    push!(ds.data, DataValueVector{jtype}(ds.rows))
+
+    if ds.rows < 0
+        push!(ds.data, DataValueVector{jtype}(READSTAT_NALLOC_ROWS))
+    else
+        push!(ds.data, DataValueVector{jtype}(ds.rows))
+    end
+
     push!(ds.storagewidths, get_storagewidth(variable))
     push!(ds.measures, get_measure(variable))
     push!(ds.alignments, get_alignment(variable))
@@ -182,6 +190,16 @@ function handle_value!(obs_index::Cint, variable::Ptr{Nothing},
                        value::ReadStatValue, ds_ptr::Ptr{ReadStatDataFrame})
     ds = unsafe_pointer_to_objref(ds_ptr)
     var_index = readstat_variable_get_index(variable)
+
+    nl = length(ds.data[var_index + 1])
+    if (obs_index + 1) >= nl
+        resize!(ds.data[var_index + 1], nl + READSTAT_NALLOC_ROWS)
+    end
+
+    if obs_index > ds.rows
+        ds.rows = obs_index + 1
+    end
+
     if !readstat_value_is_missing(value, variable)
         readfield!(ds.data[var_index + 1], obs_index + 1, value)
     end
@@ -232,6 +250,10 @@ function read_data_file(filename::AbstractString, filetype::Val)
     parser = Parser()
     # parse
     parse_data_file!(ds, parser, filename, filetype)
+    # Resize
+    for i in 1:length(ds.data)
+        resize!(ds.data[i], ds.rows)
+    end
     # return dataframe instead of ReadStatDataFrame
     return ds
 end
@@ -260,5 +282,6 @@ read_dta(filename::AbstractString) = read_data_file(filename, Val(:dta))
 read_sav(filename::AbstractString) = read_data_file(filename, Val(:sav))
 read_por(filename::AbstractString) = read_data_file(filename, Val(:por))
 read_sas7bdat(filename::AbstractString) = read_data_file(filename, Val(:sas7bdat))
+read_xport(filename::AbstractString) = read_data_file(filename, Val(:xport))
 
 end #module ReadStat

test/runtests.jl

@@ -52,4 +52,20 @@ data = rsdf.data
 @test data[6] == DataValueArray{String}(["2", "7", ""])
 end
 
+@testset "XPT files" begin
+
+dtafile = joinpath(dirname(@__FILE__), "types.xpt")
+rsdf = read_xport(dtafile)
+data = rsdf.data
+
+@test length(data) == 6
+@test rsdf.headers == [:vfloat, :vdouble, :vlong, :vint, :vbyte, :vstring]
+@test data[1] == DataValueArray{Float32}([3.14, 7., NA])
+@test data[2] == DataValueArray{Float64}([3.14, 7., NA])
+@test data[3] == DataValueArray{Int32}([2, 7, NA])
+@test data[4] == DataValueArray{Int16}([2, 7, NA])
+@test data[5] == DataValueArray{Int8}([2, 7., NA])
+@test data[6] == DataValueArray{String}(["2", "7", ""])
+end
+
 end