Artifact rework

.gitignore

@@ -5,3 +5,4 @@
 /Manifest.toml
 /docs/Manifest.toml
 /docs/build/
+/Artifacts.toml

Project.toml

@@ -1,27 +1,27 @@
 name = "TimeZoneFinder"
 uuid = "3ccf6684-3f25-4581-8c58-114637dcab4a"
 authors = ["Tom Gillam <[email protected]>"]
-version = "0.5.2"
+version = "0.6.0"
 
 [deps]
+Downloads = "f43a241f-c20a-4ad4-852c-f6b1247861c6"
 JSON3 = "0f8b85d8-7281-11e9-16c2-39a750bddbf1"
-LazyArtifacts = "4af54fe1-eca0-43a8-85a7-787d91b784e3"
 Memoize = "c03570c3-d221-55d1-a50c-7939bbd78826"
 Meshes = "eacbb407-ea5a-433e-ab97-5258b1ca43fa"
 Pkg = "44cfe95a-1eb2-52ea-b672-e2afdf69b78f"
-PrecompileTools = "aea7be01-6a6a-4083-8856-8a6e6704d82a"
 Scratch = "6c6a2e73-6563-6170-7368-637461726353"
 Serialization = "9e88b42a-f829-5b0c-bbe9-9e923198166b"
 TimeZones = "f269a46b-ccf7-5d73-abea-4c690281aa53"
+ZipArchives = "49080126-0e18-4c2a-b176-c102e4b3760c"
 
 [compat]
+Downloads = "1"
 JSON3 = "1"
-LazyArtifacts = "1"
 Memoize = "0.4"
 Meshes = "0.32,0.33,0.34,0.35,0.36, 0.37, 0.38, 0.39"
-PrecompileTools = "1"
 Scratch = "1"
 TimeZones = "1.10"
+ZipArchives = "1"
 julia = "1.6"
 
 [extras]

src/TimeZoneFinder.jl

@@ -2,15 +2,16 @@ module TimeZoneFinder
 
 export timezone_at, timezones_at
 
+using Downloads: download
 using JSON3
-using LazyArtifacts
 using Memoize
 using Meshes
+using Pkg.Artifacts
 using Pkg.TOML
-using PrecompileTools
 using Scratch
 using Serialization
 using TimeZones
+using ZipArchives: ZipBufferReader, zip_names, zip_openentry
 
 """Get points that form a closed loop.
 
@@ -72,12 +73,60 @@ function Base.in(point::Point, bpa::BoundedPolyArea)
     return in(point, bpa.polyarea)
 end
 
+"""
+    _get_artifact_path(version) -> String
+
+Get the path to the artifact for `version`, e.g. "2023b".
+
+This will download the necessary data if it doesn't already exist.
+"""
+function _get_artifact_path(version::AbstractString)
+    artifacts_toml = joinpath(dirname(@__DIR__), "Artifacts.toml")
+    artifact_name = "timezone-boundary-builder-$version"
+    hash = artifact_hash(artifact_name, artifacts_toml)
+
+    if !isnothing(hash) && artifact_exists(hash)
+        # The artifact is known, and exists on-disk, we can use it.
+        return artifact_path(hash)
+    end
+
+    # We need to download and extract the dataset.
+    # We aren't going to keep the zip archive around, so download to memory only, then
+    # decompress
+    hash = create_artifact() do artifact_dir
+        url = (
+            "https://github.com/evansiroky/timezone-boundary-builder/releases/download/" *
+            "$version/timezones-with-oceans.geojson.zip"
+        )
+        reader = ZipBufferReader(take!(download(url, IOBuffer())))
+        # We expect this archive to contain a single file, which we will
+        # extract into `artifact_dir`.
+        filename = only(zip_names(reader))
+        # We use `basename` here, since sometimes the archive includes an additional
+        # level of indirection. e.g. 2018d contains:
+        #   dist/combined-with-oceans.json
+        # whereas more recent releases contain:
+        #   combined-with-oceans.json
+        output_path = joinpath(artifact_dir, basename(filename))
+        zip_openentry(reader, filename) do io
+            open(output_path, "w") do f
+                write(f, read(io))
+            end
+        end
+    end
+
+    # We are happy to overwrite any existing mapping; this means that we set
+    # `force` to be true. (Otherwise we would fail here if e.g. the artifacts
+    # directory had been emptied).
+    bind_artifact!(artifacts_toml, artifact_name, hash; force=true)
+    return artifact_path(hash)
+end
+
 """
 Generate the timezone map data from the artifact identified by `version`.
 """
 function generate_data(version::AbstractString)
-    artifact_name = "timezone-boundary-builder-$version"
-    dir = LazyArtifacts.@artifact_str(artifact_name)
+    dir = _get_artifact_path(version)
     obj = open(JSON3.read, joinpath(dir, "combined-with-oceans.json"))
 
     # Vectors that will be populated in the loop below.
@@ -147,16 +196,42 @@ Julia process.
     end
 end
 
+function _read_gh_api_paginated(url::AbstractString, per_page::Int64, page::Int64)
+    return JSON3.read(
+        take!(download("$(url)?per_page=$(per_page)&page=$(page)", IOBuffer()))
+    )
+end
+
+function _read_gh_api_paginated(url::AbstractString)
+    responses = []
+    # TODO: This is the maximum per-page limit, at least for the "releases" command
+    per_page = 100
+    page = 1
+    while isempty(responses) || length(responses[end]) > 0
+        response = _read_gh_api_paginated(url, per_page, page)
+        push!(responses, response)
+        page += 1
+    end
+    return reduce(vcat, responses)
+end
+
 """
     _get_boundary_builder_versions()
 
-Get a list of versions for we have boundary data. Will be e.g. `["2022a", "2023b"]`.
+Get a list of versions for we have boundary data. 
 
-The list will be sorted in order of increasing versions.
+Will be e.g. `["2022a", "2023b"]`. The list will be sorted in order of increasing versions.
 """
-function _get_boundary_builder_versions()
-    toml = TOML.parsefile(find_artifacts_toml(@__FILE__))
-    return sort!([last(split(name, "-")) for name in keys(toml)])
+@memoize function _get_boundary_builder_versions()
+    # TODO: There are some older versions than 2018d (back to 2016d), but these provide a differently named
+    #   zip file. We could aim to support these if there is demand.
+
+    # NOTE: we are doing this manually to avoid a moderately heavy dependency on GitHub.jl
+    release_data = _read_gh_api_paginated(
+        "https://api.github.com/repos/evansiroky/timezone-boundary-builder/releases"
+    )
+    all_tags = [x[:tag_name] for x in release_data]
+    return sort(filter(tag -> tag >= "2018d", all_tags))
 end
 
 """
@@ -170,7 +245,7 @@ the `TimeZones` package. The map from tzdata version -> boundary version is memo
 
 This is determined by the rules in the "note" in the docstring for [`timezone_at`](@ref).
 """
-@memoize function _timezone_boundary_builder_version(tzdata_version::AbstractString)
+function _timezone_boundary_builder_version(tzdata_version::AbstractString)
     boundary_builder_versions = _get_boundary_builder_versions()
 
     i = searchsortedlast(boundary_builder_versions, tzdata_version)
@@ -254,7 +329,4 @@ function timezone_at(latitude::Real, longitude::Real)
     return only(tzs)
 end
 
-# Precompile the primary API.
-@compile_workload timezone_at(1.0, 1.0)
-
 end

test/runtests.jl

@@ -1,7 +1,8 @@
 using Memoize
 using Test
 using TimeZoneFinder
-using TimeZoneFinder: _timezone_boundary_builder_version
+using TimeZoneFinder:
+    _get_artifact_path, _get_boundary_builder_versions, _timezone_boundary_builder_version
 using TimeZones
 
 """
@@ -197,7 +198,7 @@ const TEST_LOCATIONS =
 
     @testset "old tzdata versions" begin
         # Run for several tzdata versions that we should be able to support.
-        for version in ["2021c", "2022d", "2022f"]
+        for version in ["2018d", "2021c", "2022d", "2022f"]
             tzdata_context(version) do
                 @test timezone_at(52.5061, 13.358) == TimeZone("Europe/Berlin")
             end
@@ -211,4 +212,18 @@ const TEST_LOCATIONS =
             @test timezone_at(50.438114, 30.5179595) == TimeZone("Europe/Kyiv")
         end
     end
+
+    @testset "_get_artifact_path" begin
+        dir = _get_artifact_path("2023b")
+        @test isfile(joinpath(dir, "combined-with-oceans.json"))
+        dir2 = _get_artifact_path("2023b")
+        @test dir == dir2
+    end
+
+    @testset "_get_boundary_builder_versions" begin
+        versions = _get_boundary_builder_versions()
+        @test sort(versions) == versions
+        @test versions[1] == "2018d"
+        @test length(versions) >= 10
+    end
 end