.juliabundleignore should apply to top-level files, too (#99)

Co-authored-by: Morten Piibeleht <morten.piibeleht@juliahub.com>

Author: mbauman

CHANGELOG.md

@@ -2,6 +2,12 @@
 
 The format is based on [Keep a Changelog](https://keepachangelog.com/en/1.1.0/), and this project adheres to [Semantic Versioning](https://semver.org/spec/v2.0.0.html).
 
+## Unreleased
+
+### Fixed
+
+* Top-level `.juliabundleignore` files are now correctly handled when using `JuliaHub.appbundle`. ([#99], [#100])
+
 ## Version [v0.1.14] - 2025-06-11
 
 ### Added
@@ -186,3 +192,5 @@ Initial package release.
 [#92]: https://github.com/JuliaComputing/JuliaHub.jl/issues/92
 [#94]: https://github.com/JuliaComputing/JuliaHub.jl/issues/94
 [#96]: https://github.com/JuliaComputing/JuliaHub.jl/issues/96
+[#99]: https://github.com/JuliaComputing/JuliaHub.jl/issues/99
+[#100]: https://github.com/JuliaComputing/JuliaHub.jl/issues/100

src/PackageBundler/utils.jl

@@ -77,15 +77,15 @@ function get_bundleignore(file, top)
     dir = dirname(file)
     patterns = Set{Any}()
     try
-        while dir != top
+        while true
             if isfile(joinpath(dir, ".juliabundleignore"))
                 union!(
                     patterns,
                     Glob.FilenameMatch.(strip.(readlines(joinpath(dir, ".juliabundleignore")))),
                 )
                 return patterns, dir
             end
-            if dir == dirname(dir)
+            if dir == dirname(dir) || dir == top
                 break
             end
             dir = dirname(dir)

src/utils.jl

@@ -304,22 +304,55 @@ function Base.show(io::IO, filehash::FileHash)
 end
 
 # Estimates the size of the bundle directory
-function _max_appbundle_dir_size(dir; maxsize=100 * 1024 * 1024)
+function _max_appbundle_dir_size(dir::AbstractString; maxsize=100 * 1024 * 1024)
     sz = 0
+    _walk_appbundle_files(dir) do filepath
+        if sz > maxsize
+            return _WalkFilesReturnEarly()
+        end
+        sz += filesize(filepath)
+    end
+    return sz, sz <= maxsize
+end
+
+function _walk_appbundle_files(f::Base.Callable, dir::AbstractString)
+    # Note: even if if `path_filterer` says that directory `foo/bar`
+    # should not be included, it will still likely return `true` for
+    # any files in there, like `foo/bar/baz`. So we need to make sure
+    # we stop recursing into subdirectories.
     pred = _PackageBundler.path_filterer(dir)
-    for (root, _, files) in walkdir(dir)
-        for file in files
-            file = joinpath(root, file)
-            if !pred(file)
-                @debug "ignoring $file in dir size measurement"
-                continue
-            end
+    _walkfiles(dir; descend=pred) do filepath
+        if !pred(filepath)
+            @debug "ignoring file in _walk_appbundle_files: $(filepath)"
+            return nothing
+        end
+        return f(filepath)
+    end
+end
+
+struct _WalkFilesReturnEarly end
 
-            sz > maxsize && return sz, false
-            sz += filesize(file)
+# Calls `f` on any non-directory in `root`.
+# `descend` gets called on any directory, and if it returns false,
+function _walkfiles(f::Base.Callable, root::AbstractString; descend::Base.Callable)
+    if !isdir(root)
+        error("Not a directory: $(root)")
+    end
+    directories = String[root]
+    while !isempty(directories)
+        dir = popfirst!(directories)
+        for subpath in readdir(dir; join=true)
+            if isdir(subpath)
+                if descend(subpath)::Bool
+                    push!(directories, subpath)
+                end
+            else
+                if f(subpath) === _WalkFilesReturnEarly()
+                    break
+                end
+            end
         end
     end
-    return sz, sz < maxsize
 end
 
 function _json_get(d::Dict, key, ::Type{T}; var::AbstractString, parse=false) where {T}

test/fixtures/ignorefiles/Pkg3/src/foo

@@ -0,0 +1 @@
+000000000000000000000000000000000000000000000000

test/fixtures/ignorefiles/Pkg3/test/bar/test

@@ -0,0 +1 @@
+000000000000000000000000000000000000000000000000

test/fixtures/ignorefiles/Pkg3/test/foo/test

@@ -0,0 +1 @@
+000000000000000000000000000000000000000000000000

test/packagebundler.jl

@@ -225,3 +225,44 @@ end
     end
     @test isfile(out)
 end
+
+@testset "path_filterer" begin
+    @testset "subdirectory" begin
+        dir = joinpath(@__DIR__, "fixtures", "ignorefiles")
+        pred = JuliaHub._PackageBundler.path_filterer(dir)
+        @test pred(joinpath(dir, "Pkg3", "README.md"))
+
+        @test pred(joinpath(dir, "Pkg3", "src", "bar"))
+        @test !pred(joinpath(dir, "Pkg3", "src", "foo"))
+        @test pred(joinpath(dir, "Pkg3", "src", "fooo"))
+
+        @test !pred(joinpath(dir, "Pkg3", "test", "bar"))
+        @test !pred(joinpath(dir, "Pkg3", "test", "foo"))
+        @test pred(joinpath(dir, "Pkg3", "test", "fooo", "test"))
+
+        # Note: even though the test/foo and test/bar directories are
+        # excluded, the predicate function does not return false if you
+        # check for file within the directories.
+        @test pred(joinpath(dir, "Pkg3", "test", "bar", "test"))
+        @test pred(joinpath(dir, "Pkg3", "test", "foo", "test"))
+    end
+    @testset "toplevel" begin
+        dir = joinpath(@__DIR__, "fixtures", "ignorefiles", "Pkg3")
+        pred = JuliaHub._PackageBundler.path_filterer(dir)
+        @test pred(joinpath(dir, "README.md"))
+
+        @test pred(joinpath(dir, "src", "bar"))
+        @test !pred(joinpath(dir, "src", "foo"))
+        @test pred(joinpath(dir, "src", "fooo"))
+
+        @test !pred(joinpath(dir, "test", "bar"))
+        @test !pred(joinpath(dir, "test", "foo"))
+        @test pred(joinpath(dir, "test", "fooo", "test"))
+
+        # Note: even though the test/foo and test/bar directories are
+        # excluded, the predicate function does not return false if you
+        # check for file within the directories.
+        @test pred(joinpath(dir, "test", "bar", "test"))
+        @test pred(joinpath(dir, "test", "foo", "test"))
+    end
+end

test/utils.jl

@@ -101,3 +101,31 @@ end
         Dict("_id_missing" => "123e4567-e89b-12d3-a456-426614174000"), "id", UUIDs.UUID
     )
 end
+
+@testset "_max_appbundle_dir_size" begin
+    # We check here that the `.juliabundleignore` is honored by making
+    # sure that the calculated total file size of the Pkg3/ directory is
+    dir = joinpath(@__DIR__, "fixtures", "ignorefiles", "Pkg3")
+
+    appbundle_files = String[]
+    JuliaHub._walk_appbundle_files(dir) do filepath
+        push!(appbundle_files, relpath(filepath, dir))
+    end
+    @test sort(appbundle_files) == [
+        ".gitignore", ".juliabundleignore", "Project.toml", "README.md",
+        joinpath("src", "Pkg3.jl"), joinpath("src", "bar"), joinpath("src", "fooo"),
+        joinpath("test", "fooo", "test"), joinpath("test", "runtests.jl"),
+    ]
+
+    # The files that are not meant to be included in the /Pkg3/ bundle here are
+    # all 50 byte files. Should they should show up in the total size here.
+    #
+    # Note: on windows, the files may be checked out with different line endings,
+    # so the total file size may be slightly different.
+    sz, sz_is_low = JuliaHub._max_appbundle_dir_size(dir)
+    @test sz_is_low
+    @test sz in (Sys.iswindows() ? (405, 432) : (405,))
+
+    _, sz_low = JuliaHub._max_appbundle_dir_size(dir; maxsize=200)
+    @test !sz_low
+end