Merge #231

231: enable S3Paths to be (de)serialized to/from Arrow r=omus a=jrevels

closes #184 

This adds ArrowTypes as a lightweight dependency, which is hopefully okay. 

Co-authored-by: Jarrett Revels <[email protected]>
Co-authored-by: Curtis Vogt <[email protected]>

Project.toml

@@ -4,6 +4,7 @@ version = "0.9.4"
 
 [deps]
 AWS = "fbe9abb3-538b-5e4e-ba9e-bc94f4f92ebc"
+ArrowTypes = "31f734f8-188a-4ce0-8406-c8a06bd891cd"
 Base64 = "2a0f44e3-6c83-55bd-87e4-b1978d98bd5f"
 Compat = "34da2185-b29b-5c13-b0c7-acf172513d20"
 Dates = "ade2ca70-3891-5945-98fb-dc099432e06a"
@@ -20,6 +21,7 @@ XMLDict = "228000da-037f-5747-90a9-8195ccbf91a5"
 
 [compat]
 AWS = "1.63.1"
+ArrowTypes = "1.2"
 Compat = "3.29.0"
 EzXML = "0.9, 1"
 FilePathsBase = "=0.9.11, =0.9.12, =0.9.13, =0.9.14, =0.9.15"
@@ -35,9 +37,10 @@ XMLDict = "0.3, 0.4"
 julia = "1.3"
 
 [extras]
+Arrow = "69666777-d1a9-59fb-9406-91d4454c9d45"
 JSON3 = "0f8b85d8-7281-11e9-16c2-39a750bddbf1"
 Minio = "4281f0d9-7ae0-406e-9172-b7277c1efa20"
 Test = "8dfed614-e22c-5e08-85e1-65c5234f0b40"
 
 [targets]
-test = ["Test", "JSON3", "Minio"]
+test = ["Test", "Arrow", "JSON3", "Minio"]

src/AWSS3.jl

@@ -38,6 +38,7 @@ export S3Path,
 
 using AWS
 using AWS.AWSServices: s3
+using ArrowTypes
 using FilePathsBase
 using FilePathsBase: /, join
 using HTTP

src/s3path.jl

@@ -640,3 +640,16 @@ function FilePathsBase.mktmpdir(parent::S3Path)
     fp = parent / string(uuid4(), "/")
     return mkdir(fp)
 end
+
+const S3PATH_ARROW_NAME = Symbol("JuliaLang.AWSS3.S3Path")
+ArrowTypes.arrowname(::Type{<:S3Path}) = S3PATH_ARROW_NAME
+ArrowTypes.ArrowType(::Type{<:S3Path}) = String
+ArrowTypes.JuliaType(::Val{S3PATH_ARROW_NAME}, ::Any) = S3Path
+ArrowTypes.fromarrow(::Type{<:S3Path}, uri_string) = S3Path(uri_string)
+
+function ArrowTypes.toarrow(path::S3Path)
+    if !isnothing(path.config)
+        throw(ArgumentError("`S3Path` config must be `nothing` to serialize to Arrow"))
+    end
+    return string(path)
+end

test/runtests.jl

@@ -2,6 +2,7 @@ using AWS
 using AWS.AWSExceptions: AWSException
 using AWS.AWSServices: s3
 using AWSS3
+using Arrow
 using Test
 using Dates
 using Retry

test/s3path.jl

@@ -433,6 +433,19 @@ function s3path_tests(config)
         rm(json_path)
     end
 
+    @testset "Arrow <-> S3Path (de)serialization" begin
+        ver = String('A':'Z') * String('0':'5')
+        paths = [
+            S3Path("s3://$(bucket_name)/a"),
+            S3Path("s3://$(bucket_name)/b?versionId=$ver"),
+            # format trick: using this comment to force use of multiple lines
+        ]
+        tbl = Arrow.Table(Arrow.tobuffer((; paths=paths)))
+        @test all(tbl.paths .== paths)
+        push!(paths, S3Path("s3://$(bucket_name)/c"; config=config))
+        @test_throws Arrow.tobuffer((; paths=paths)) ArgumentError
+    end
+
     @testset "tryparse" begin
         cfg = global_aws_config()
         ver = String('A':'Z') * String('0':'5')