Create PackageCollection type for nicer show method

Project.toml

@@ -13,7 +13,9 @@ JSON3 = "0f8b85d8-7281-11e9-16c2-39a750bddbf1"
 JuliaSyntax = "70703baa-626e-46a2-a12c-08ffd08c73b4"
 Legolas = "741b9549-f6ed-4911-9fbf-4a1c0c97f0cd"
 LicenseCheck = "726dbf0d-6eb6-41af-b36c-cd770e0f00cc"
+OrderedCollections = "bac558e1-5e72-5ebc-8fee-abe8a469f55d"
 Pkg = "44cfe95a-1eb2-52ea-b672-e2afdf69b78f"
+PrecompileTools = "aea7be01-6a6a-4083-8856-8a6e6704d82a"
 Printf = "de0858da-6303-5e67-8744-51eddeeeb8d7"
 RegistryInstances = "2792f1a3-b283-48e8-9a74-f99dce5104f3"
 TOML = "fa267f1f-6049-4f14-aa54-33bafae1ed76"
@@ -26,10 +28,12 @@ AbstractTrees = "0.4"
 CodecZlib = "0.7"
 Git = "1.2.1"
 GitHub = "5.4"
-Legolas = "0.5"
 JSON3 = "1.5.1"
 JuliaSyntax = "0.4"
+Legolas = "0.5"
 LicenseCheck = "0.2"
+OrderedCollections = "1"
+PrecompileTools = "1.2"
 RegistryInstances = "0.1"
 julia = "1.6"
 

src/PackageAnalyzer.jl

@@ -16,6 +16,8 @@ using JuliaSyntax
 using JuliaSyntax: @K_str
 using Legolas
 using Legolas: @schema, @version
+using OrderedCollections
+using PrecompileTools
 
 # We wrap `registry_info` for thread-safety, so we don't want to pull into the namespace here
 using RegistryInstances: RegistryInstances, reachable_registries, PkgEntry
@@ -24,6 +26,7 @@ using RegistryInstances: RegistryInstances, reachable_registries, PkgEntry
 export find_package, find_packages, find_packages_in_manifest
 # Ways to analyze them
 export analyze, analyze_manifest, analyze_packages, LineCategories
+export PackageCollection
 
 ##
 # Borrowed from
@@ -163,6 +166,10 @@ function PackageV1(name, uuid, repo;
 end
 
 function Base.show(io::IO, p::PackageV1)
+    compact = get(io, :compact, false)::Bool
+    if compact
+        return print(io, PackageV1, "(\"", p.name, "\", …)")
+    end
     body = """
         PackageV1 $(p.name):
           * repo: $(p.repo)
@@ -355,6 +362,8 @@ function Base.show(io::IO, d::Trunk)
     print(io, ")")
 end
 
+include("package_collection.jl")
+
 # Provides methods to obtain a `PkgSource`
 include("find_packages.jl")
 
@@ -380,5 +389,10 @@ include("count_loc.jl")
 
 include("deprecated_schemas.jl")
 
+@compile_workload begin
+    p = analyze(PackageAnalyzer)
+    sprint(show, MIME"text/plain"(), p)
+    nothing
+end
 
 end # module

src/package_collection.jl

@@ -0,0 +1,69 @@
+struct PackageCollection <: AbstractVector{PackageV1}
+    pkgs::Vector{PackageV1}
+end
+
+Base.getindex(p::PackageCollection, i::Int) = p.pkgs[i]
+Base.size(p::PackageCollection) = size(p.pkgs)
+
+function _to_underscore(n::Integer)
+    x = Iterators.partition(digits(n), 3)
+    return join(reverse(join.(reverse.(x))), '_')
+end
+
+
+function Base.show(io::IO, ::MIME"text/plain", c::PackageCollection)
+    n = length(c)
+    summary(io, c)
+    println(io)
+    l_src = l_ext = l_test = l_docs = l_readme = l_src_docstring = 0
+
+    for p in c
+        l_src += sum_julia_loc(p, "src")
+        l_ext += sum_julia_loc(p, "ext")
+        l_test += sum_julia_loc(p, "test")
+        l_docs += sum_doc_lines(p)
+        l_readme += sum_readme_lines(p)
+        l_src_docstring += sum_docstrings(p, "src")
+    end
+
+    p_ext = @sprintf("%.1f", 100 * l_ext / (l_test + l_src + l_ext))
+    p_test = @sprintf("%.1f", 100 * l_test / (l_test + l_src + l_ext))
+    p_docs = @sprintf("%.1f", 100 * l_docs / (l_docs + l_src + l_ext))
+
+    p_ext = @sprintf("%.1f", 100 * l_ext / (l_test + l_src + l_ext))
+    p_test = @sprintf("%.1f", 100 * l_test / (l_test + l_src + l_ext))
+    p_docs = @sprintf("%.1f", 100 * l_docs / (l_docs + l_src + l_ext))
+
+    body = """
+        * Source code
+          * Total Julia source code: $(_to_underscore(l_src)) lines
+          * Total Julia extension code: $(_to_underscore(l_ext)) lines ($(p_ext)% of `test` + `src` + `ext`)
+          * Julia code test code: $(_to_underscore(l_test)) lines ($(p_test)% of `test` + `src` + `ext`)
+          * Lines of documentation: $(_to_underscore(l_docs)) lines ($(p_docs)% of `docs` + `src` + `ext`)
+        """
+
+    n = l_src_docstring + l_readme
+    p_docstrings = @sprintf("%.1f", 100 * n / (n + l_src))
+    body *= """
+          * Total lines of documentation in README & docstrings: $(n) lines ($(p_docstrings)% of README + `src`)
+        """
+    print(io, strip(body))
+
+    licenses = OrderedDict{String, Vector{String}}()
+    for p in c
+        for lic in p.license_files
+            for license in lic.licenses_found
+                v = get!(Vector{String}, licenses, license)
+                push!(v, p.name)
+            end
+        end
+    end
+    sort!(licenses; by=((k,v),) -> length(v))
+    println(io, "\n* Licenses")
+    for (k, v) in licenses
+        sort!(unique!(v))
+        print(io, "  * $k ($(length(v))): ")
+        join(io, v, ", ")
+        println(io)
+    end
+end

src/parallel.jl

@@ -1,9 +1,9 @@
 """
-    analyze_packages(pkg_entries; auth::GitHub.Authorization=github_auth(), sleep=0, root=mktempdir()) -> Vector{PackageV1}
+    analyze_packages(pkg_entries; auth::GitHub.Authorization=github_auth(), sleep=0, root=mktempdir()) -> PackageCollection
 
 
 Analyze all packages in the iterable `pkg_entries`, using threads, storing their code in `root`
-if it needs to be downloaded.  Returns a `Vector{PackageV1}`.
+if it needs to be downloaded.  Returns a [`PackageCollection`](@ref), which can be used like a `Vector{PackageV1}`.
 
 Each element of `pkg_entries` should be a valid input to [`analyze`](@ref).
 
@@ -24,5 +24,5 @@ function analyze_packages(pkg_entries; auth::GitHub.Authorization=github_auth(),
         put!(outputs, (i, analyze(pkg; auth, sleep, root)))
     end
     close(outputs)
-    return last.(sort!(collect(outputs); by=first))
+    return PackageCollection(last.(sort!(collect(outputs); by=first)))
 end