Merge pull request #18 from CarloLucibello/cl/pil

fix for PIL imports + fix DatasetDict julia format

Author: CarloLucibello

.github/workflows/CI.yml

@@ -18,8 +18,8 @@ jobs:
       fail-fast: false
       matrix:
         version:
-          - '1.7'
-          - '1'
+          - '1.9'
+          # - '1' # add back when 1.10 is out
           - 'nightly'
         os:
           - ubuntu-latest

.github/workflows/TagBot.yml

@@ -30,4 +30,4 @@ jobs:
           token: ${{ secrets.GITHUB_TOKEN }}
           # Edit the following line to reflect the actual name of the GitHub Secret containing your private key
           ssh: ${{ secrets.DOCUMENTER_KEY }}
-          # ssh: ${{ secrets.NAME_OF_MY_SSH_PRIVATE_KEY_SECRET }}
+          # ssh: ${{ secrets.NAME_OF_MY_SSH_PRIVATE_KEY_SECRET }}

CondaPkg.toml

@@ -1,8 +1,10 @@
 channels = ["conda-forge"]
 
 [deps]
-h5py = ""
-pillow = ">=9.1, <10"
+# h5py = ""
+# pillow = ">=9.1, <10"
+# pyarrow = "==6.0.0"
+datasets = ">=2.12, <3"
 numpy = ">=1.20, <2"
-datasets = ">=2.7, <3"
-pyarrow = "==6.0.0"
+pillow = ""
+

Project.toml

@@ -1,7 +1,7 @@
 name = "HuggingFaceDatasets"
 uuid = "d94b9a45-fdf5-4270-b024-5cbb9ef7117d"
 authors = ["Carlo Lucibello"]
-version = "0.3.0"
+version = "0.3.1"
 
 [deps]
 CondaPkg = "992eb4ea-22a4-4c89-a5bb-47a3300528ab"
@@ -16,7 +16,7 @@ DLPack = "0.1"
 ImageCore = "0.9"
 MLUtils = "0.4.1"
 PythonCall = "0.9"
-julia = "1.7"
+julia = "1.9"
 
 [extras]
 Test = "8dfed614-e22c-5e08-85e1-65c5234f0b40"

src/HuggingFaceDatasets.jl

@@ -41,6 +41,8 @@ function __init__()
     # https://cjdoris.github.io/PythonCall.jl/dev/pythoncall-reference/#PythonCall.pycopy!
     PythonCall.pycopy!(datasets, pyimport("datasets"))
     PythonCall.pycopy!(PIL, pyimport("PIL"))
+    pyimport("PIL.PngImagePlugin")
+    pyimport("PIL.JpegImagePlugin")
     PythonCall.pycopy!(np, pyimport("numpy"))
     PythonCall.pycopy!(copy, pyimport("copy"))
 end

src/dataset.jl

@@ -107,7 +107,7 @@ version of [`with_format`](@ref).
 """
 function set_format!(ds::Dataset, format)
     if format == "julia"
-        # ds.pyds.set_format("numpy")
+        ds.pyds.reset_format() # or d.pyd.set_format("python")
         ds.jltransform = py2jl
     else
         ds.pyds.set_format(format)

src/datasetdict.jl

@@ -102,7 +102,7 @@ version of [`with_format`](@ref).
 """
 function set_format!(d::DatasetDict, format)
     if format == "julia"
-        d.pyd.set_format("numpy")
+        d.pyd.reset_format()
         d.jltransform = py2jl
     else
         d.pyd.set_format(format)

src/transforms.jl

@@ -21,6 +21,7 @@ function _pyconvert(x::Py)
     end
 end
 
+# Do nothing on a non-Py object.
 _pyconvert(x) = x
 
 """
@@ -30,6 +31,7 @@ Convert Python types to Julia types applying `pyconvert` recursively.
 """
 py2jl
 
+# py2jl recurses through pycanonicalize and converts through _pyconvert
 py2jl(x) = pycanonicalize(_pyconvert(x))
 
 pycanonicalize(x) = x

test/dataset.jl

@@ -74,7 +74,6 @@ end
     @test x isa Dict
     @test x["label"] == -1
     @test x["idx"] == 0
-    @show x["premise"] |> typeof
     @test x["premise"] isa AbstractString
     @test x["premise"] == "The cat sat on the mat."
     @test x["hypothesis"] isa AbstractString

test/datasetdict.jl

@@ -19,12 +19,12 @@ end
 @testset "with_format(julia)" begin
     d = with_format(mnist, "julia")
     ds = d["test"]
-    @test ds.format["type"] == "numpy"
+    @test ds.format["type"] == nothing
     x = ds[1]
     @test x isa Dict
     @test x["label"] isa Int
     @test x["label"] == 7
-    @test x["image"] isa Matrix{UInt8}
+    @test x["image"] isa AbstractMatrix{<:Gray}
     @test size(x["image"]) == (28, 28)
 end