A quick and dirty script to find broken links in the Julia package registry, and then find any dependencies on those repos

PackageErrors.jl

import Pkg
using GitHub
using CSV
using DataFrames
using Dates

# Don't commit this once you add your auth token!!
GITHUB_AUTH = "" # INSERT your Github auth token
myauth = GitHub.authenticate(GITHUB_AUTH)
path = "" # INSERT the local path to your clone of the JuliaRegistries/General repo

# Build the registry as a dict with the following fields
# name
# path
# repo
# lastCommit
# stars
# description
function buildRegistry()
  df = DataFrame(uuid = String[],
                  name=String[],
                  path=String[],
                  repo=String[],
                  lastCommit=DateTime[],
                  stars=Int[],
                  description=String[])

  registry = Pkg.TOML.parsefile(joinpath(path, "Registry.toml"))
  packages = registry["packages"]

  errors = []
  for (i, (uuid, dict)) in collect(enumerate(packages))
    packageFile = Pkg.TOML.parsefile(joinpath(path, dict["path"], "Package.toml"))
    repoURL = packageFile["repo"]
    repo = replace(repoURL, "https://github.com/" => "")[1:end-4]

    dict = copy(dict)
    dict["uuid"] = uuid
    dict["name"] = dict["name"]
    dict["path"] = dict["path"]
    dict["repo"] = repo

    try
      repo = GitHub.repo(repo, auth=myauth)
      dict["stars"] = repo.stargazers_count
      dict["lastCommit"] = repo.pushed_at
      dict["description"] = repo.description == nothing ? "" : repo.description

      push!(df, dict)
    catch e
      push!(errors, dict)
      println(repo, e)
    end

    # Deal with Github rate limiting
    if i % 1000 == 0
      println("At ", i)
      sleep(15*60)
    end
  end

  return df, errors
end

df, errors = buildRegistry()
CSV.write("registry.csv", df) # Save in case we need to reload later

gitErrors = filter(d -> !contains(d["repo"], "gitlab"), (errors))
for e in gitErrors
  println("https://github.com/" * e["repo"])
end

# Find dependencies
function findDependents(uuid, df)
  filter(df) do row
    dependencyFile = joinpath(path, row.path, "Deps.toml")
    if !isfile(dependencyFile)
      return false
    end

    dependencies = Pkg.TOML.parsefile(dependencyFile)
    return any([uuid in values(version) for version in values(dependencies)])
  end
end

for e in gitErrors
  dependents = findDependents(e["uuid"], df)
  println(e["name"])
  println(dependents)
end