Merge pull request #10677 from sjackman/sj/pr-upload

pr-upload: Upload bottles to Archive.org

Library/Homebrew/archive.rb

@@ -0,0 +1,177 @@
+# typed: false
+# frozen_string_literal: true
+
+require "digest/md5"
+require "utils/curl"
+
+# The Internet Archive API client.
+#
+# @api private
+class Archive
+  extend T::Sig
+
+  include Context
+  include Utils::Curl
+
+  class Error < RuntimeError
+  end
+
+  sig { returns(String) }
+  def inspect
+    "#<Archive: item=#{@archive_item}>"
+  end
+
+  sig { params(item: T.nilable(String)).void }
+  def initialize(item: "homebrew")
+    raise UsageError, "Must set the Archive item!" unless item
+
+    @archive_item = item
+  end
+
+  def open_api(url, *args, auth: true)
+    if auth
+      key = Homebrew::EnvConfig.internet_archive_key
+      raise UsageError, "HOMEBREW_INTERNET_ARCHIVE_KEY is unset." if key.blank?
+
+      if key.exclude?(":")
+        raise UsageError, "Use HOMEBREW_INTERNET_ARCHIVE_KEY=access:secret. See https://archive.org/account/s3.php"
+      end
+
+      args += ["--header", "Authorization: AWS #{key}"]
+    end
+
+    curl(*args, url, print_stdout: false, secrets: key)
+  end
+
+  sig {
+    params(local_file:    String,
+           directory:     String,
+           remote_file:   String,
+           warn_on_error: T.nilable(T::Boolean)).void
+  }
+  def upload(local_file, directory:, remote_file:, warn_on_error: false)
+    local_file = Pathname.new(local_file)
+    unless local_file.exist?
+      msg = "#{local_file} for upload doesn't exist!"
+      raise Error, msg unless warn_on_error
+
+      # Warn and return early here since we know this upload is going to fail.
+      opoo msg
+      return
+    end
+
+    md5_base64 = Digest::MD5.base64digest(local_file.read)
+    url = "https://#{@archive_item}.s3.us.archive.org/#{directory}/#{remote_file}"
+    args = ["--upload-file", local_file, "--header", "Content-MD5: #{md5_base64}"]
+    args << "--fail" unless warn_on_error
+    result = T.unsafe(self).open_api(url, *args)
+    return if result.success? && result.stdout.exclude?("Error")
+
+    msg = "Bottle upload failed: #{result.stdout}"
+    raise msg unless warn_on_error
+
+    opoo msg
+  end
+
+  sig {
+    params(formula:       Formula,
+           directory:     String,
+           warn_on_error: T::Boolean).returns(String)
+  }
+  def mirror_formula(formula, directory: "mirror", warn_on_error: false)
+    formula.downloader.fetch
+
+    filename = ERB::Util.url_encode(formula.downloader.basename)
+    destination_url = "https://archive.org/download/#{@archive_item}/#{directory}/#{filename}"
+
+    odebug "Uploading to #{destination_url}"
+
+    upload(
+      formula.downloader.cached_location,
+      directory:     directory,
+      remote_file:   filename,
+      warn_on_error: warn_on_error,
+    )
+
+    destination_url
+  end
+
+  # Gets the MD5 hash of the specified remote file.
+  #
+  # @return the hash, the empty string (if the file doesn't have a hash), nil (if the file doesn't exist)
+  sig { params(directory: String, remote_file: String).returns(T.nilable(String)) }
+  def remote_md5(directory:, remote_file:)
+    url = "https://#{@archive_item}.s3.us.archive.org/#{directory}/#{remote_file}"
+    result = curl_output "--fail", "--silent", "--head", "--location", url
+    if result.success?
+      result.stdout.match(/^ETag: "(\h{32})"/)&.values_at(1)&.first || ""
+    else
+      raise Error if result.status.exitstatus != 22 && result.stderr.exclude?("404 Not Found")
+
+      nil
+    end
+  end
+
+  sig { params(directory: String, filename: String).returns(String) }
+  def file_delete_instructions(directory, filename)
+    <<~EOS
+      Run:
+        curl -X DELETE -H "Authorization: AWS $HOMEBREW_INTERNET_ARCHIVE_KEY" https://#{@archive_item}.s3.us.archive.org/#{directory}/#{filename}
+      Or run:
+        ia delete #{@archive_item} #{directory}/#{filename}
+    EOS
+  end
+
+  sig {
+    params(bottles_hash:  T::Hash[String, T.untyped],
+           warn_on_error: T.nilable(T::Boolean)).void
+  }
+  def upload_bottles(bottles_hash, warn_on_error: false)
+    bottles_hash.each do |_formula_name, bottle_hash|
+      directory = bottle_hash["bintray"]["repository"]
+      bottle_count = bottle_hash["bottle"]["tags"].length
+
+      bottle_hash["bottle"]["tags"].each_value do |tag_hash|
+        filename = tag_hash["filename"] # URL encoded in Bottle::Filename#archive
+        delete_instructions = file_delete_instructions(directory, filename)
+
+        local_filename = tag_hash["local_filename"]
+        md5 = Digest::MD5.hexdigest(File.read(local_filename))
+
+        odebug "Checking remote file #{@archive_item}/#{directory}/#{filename}"
+        result = remote_md5(directory: directory, remote_file: filename)
+        case result
+        when nil
+          # File doesn't exist.
+          odebug "Uploading #{@archive_item}/#{directory}/#{filename}"
+          upload(local_filename,
+                 directory:     directory,
+                 remote_file:   filename,
+                 warn_on_error: warn_on_error)
+        when md5
+          # File exists, hash matches.
+          odebug "#{filename} is already published with matching hash."
+          bottle_count -= 1
+        when ""
+          # File exists, but can't find hash
+          failed_message = "#{filename} is already published!"
+          raise Error, "#{failed_message}\n#{delete_instructions}" unless warn_on_error
+
+          opoo failed_message
+        else
+          # File exists, but hash either doesn't exist or is mismatched.
+          failed_message = <<~EOS
+            #{filename} is already published with a mismatched hash!
+              Expected: #{md5}
+              Actual:   #{result}
+          EOS
+          raise Error, "#{failed_message}#{delete_instructions}" unless warn_on_error
+
+          opoo failed_message
+        end
+      end
+
+      odebug "Uploaded #{bottle_count} bottles"
+    end
+  end
+end

Library/Homebrew/dev-cmd/pr-pull.rb

@@ -49,6 +49,8 @@ module Homebrew
              description: "Message to include when autosquashing revision bumps, deletions, and rebuilds."
       flag   "--artifact=",
              description: "Download artifacts with the specified name (default: `bottles`)."
+      flag   "--archive-item=",
+             description: "Upload to the specified Internet Archive item (default: `homebrew`)."
       flag   "--bintray-org=",
              description: "Upload to the specified Bintray organisation (default: `homebrew`)."
       flag   "--tap=",
@@ -65,6 +67,7 @@ module Homebrew
                   description: "Comma-separated list of workflows which can be ignored if they have not been run."
 
       conflicts "--clean", "--autosquash"
+      conflicts "--archive-item", "--bintray-org"
 
       named_args :pull_request, min: 1
     end
@@ -357,6 +360,7 @@ module Homebrew
 
     workflows = args.workflows.presence || ["tests.yml"]
     artifact = args.artifact || "bottles"
+    archive_item = args.archive_item
     bintray_org = args.bintray_org || "homebrew"
     mirror_repo = args.bintray_mirror || "mirror"
     tap = Tap.fetch(args.tap || CoreTap.instance.name)
@@ -424,7 +428,11 @@ module Homebrew
           upload_args << "--keep-old" if args.keep_old?
           upload_args << "--warn-on-upload-failure" if args.warn_on_upload_failure?
           upload_args << "--root-url=#{args.root_url}" if args.root_url
-          upload_args << "--bintray-org=#{bintray_org}"
+          upload_args << if archive_item.present?
+            "--archive-item=#{archive_item}"
+          else
+            "--bintray-org=#{bintray_org}"
+          end
           safe_system HOMEBREW_BREW_FILE, *upload_args
         end
       end

Library/Homebrew/dev-cmd/pr-upload.rb

@@ -2,6 +2,7 @@
 # frozen_string_literal: true
 
 require "cli/parser"
+require "archive"
 require "bintray"
 
 module Homebrew
@@ -27,6 +28,8 @@ module Homebrew
       switch "--warn-on-upload-failure",
              description: "Warn instead of raising an error if the bottle upload fails. "\
                           "Useful for repairing bottle uploads that previously failed."
+      flag   "--archive-item=",
+             description: "Upload to the specified Internet Archive item (default: `homebrew`)."
       flag   "--bintray-org=",
              description: "Upload to the specified Bintray organisation (default: `homebrew`)."
       flag   "--root-url=",
@@ -47,6 +50,18 @@ module Homebrew
     end
   end
 
+  def archive?(bottles_hash)
+    @archive ||= bottles_hash.values.all? do |bottle_hash|
+      bottle_hash["bottle"]["root_url"].start_with? "https://archive.com/"
+    end
+  end
+
+  def bintray?(bottles_hash)
+    @bintray ||= bottles_hash.values.all? do |bottle_hash|
+      bottle_hash["bottle"]["root_url"].match? %r{^https://[\w-]+\.bintray\.com/}
+    end
+  end
+
   def github_releases?(bottles_hash)
     @github_releases ||= bottles_hash.values.all? do |bottle_hash|
       root_url = bottle_hash["bottle"]["root_url"]
@@ -76,11 +91,16 @@ module Homebrew
     bottle_args += json_files
 
     if args.dry_run?
-      service = if github_releases?(bottles_hash)
-        "GitHub Releases"
-      else
-        "Bintray"
-      end
+      service =
+        if archive?(bottles_hash)
+          "Internet Archive"
+        elsif bintray?(bottles_hash)
+          "Bintray"
+        elsif github_releases?(bottles_hash)
+          "GitHub Releases"
+        else
+          odie "Service specified by root_url is not recognized"
+        end
       puts <<~EOS
         brew #{bottle_args.join " "}
         Upload bottles described by these JSON files to #{service}:
@@ -102,7 +122,20 @@ module Homebrew
       safe_system HOMEBREW_BREW_FILE, *audit_args
     end
 
-    if github_releases?(bottles_hash)
+    if archive?(bottles_hash)
+      # Handle uploading to the Internet Archive.
+      archive_item = args.archive_item || "homebrew"
+      archive = Archive.new(item: archive_item)
+      archive.upload_bottles(bottles_hash,
+                             warn_on_error: args.warn_on_upload_failure?)
+    elsif bintray?(bottles_hash)
+      # Handle uploading to Bintray.
+      bintray_org = args.bintray_org || "homebrew"
+      bintray = Bintray.new(org: bintray_org)
+      bintray.upload_bottles(bottles_hash,
+                             publish_package: !args.no_publish?,
+                             warn_on_error:   args.warn_on_upload_failure?)
+    elsif github_releases?(bottles_hash)
       # Handle uploading to GitHub Releases.
       bottles_hash.each_value do |bottle_hash|
         root_url = bottle_hash["bottle"]["root_url"]
@@ -128,12 +161,7 @@ module Homebrew
         end
       end
     else
-      # Handle uploading to Bintray.
-      bintray_org = args.bintray_org || "homebrew"
-      bintray = Bintray.new(org: bintray_org)
-      bintray.upload_bottles(bottles_hash,
-                             publish_package: !args.no_publish?,
-                             warn_on_error:   args.warn_on_upload_failure?)
+      odie "Service specified by root_url is not recognized"
     end
   end
 end

Library/Homebrew/env_config.rb

@@ -179,6 +179,10 @@ module Homebrew
         default_text: 'The "Beer Mug" emoji.',
         default:      "🍺",
       },
+      HOMEBREW_INTERNET_ARCHIVE_KEY:          {
+        description: "Use this API key when accessing the Internet Archive S3 API, where bottles are stored. " \
+                     "The format is access:secret. See https://archive.org/account/s3.php",
+      },
       HOMEBREW_LIVECHECK_WATCHLIST:           {
         description: "Consult this file for the list of formulae to check by default when no formula argument " \
                      "is passed to `brew livecheck`.",

Library/Homebrew/test/archive_spec.rb

@@ -0,0 +1,20 @@
+# typed: false
+# frozen_string_literal: true
+
+require "archive"
+
+describe Archive, :needs_network do
+  subject(:archive) { described_class.new(item: "homebrew") }
+
+  describe "::remote_checksum" do
+    it "detects a published file" do
+      hash = archive.remote_md5(directory: ".", remote_file: "cmake-3.1.2.yosemite.bottle.tar.gz")
+      expect(hash).to eq("c6e525d472124670b0b635800488f438")
+    end
+
+    it "fails on a non-existent file" do
+      hash = archive.remote_md5(directory: "bottles", remote_file: "my-fake-bottle-1.0.snow_hyena.tar.gz")
+      expect(hash).to be nil
+    end
+  end
+end

completions/bash/brew

@@ -1475,6 +1475,7 @@ _brew_pr_pull() {
   case "$cur" in
     -*)
       __brewcomp "
+      --archive-item
       --artifact
       --autosquash
       --bintray-mirror
@@ -1507,6 +1508,7 @@ _brew_pr_upload() {
   case "$cur" in
     -*)
       __brewcomp "
+      --archive-item
       --bintray-org
       --debug
       --dry-run

completions/fish/brew.fish

@@ -1053,6 +1053,7 @@ __fish_brew_complete_arg 'pr-publish' -l workflow -d 'Target workflow filename (
 
 
 __fish_brew_complete_cmd 'pr-pull' 'Download and publish bottles, and apply the bottle commit from a pull request with artifacts generated by GitHub Actions'
+__fish_brew_complete_arg 'pr-pull' -l archive-item -d 'Upload to the specified Internet Archive item (default: `homebrew`)'
 __fish_brew_complete_arg 'pr-pull' -l artifact -d 'Download artifacts with the specified name (default: `bottles`)'
 __fish_brew_complete_arg 'pr-pull' -l autosquash -d 'Automatically reformat and reword commits in the pull request to our preferred format'
 __fish_brew_complete_arg 'pr-pull' -l bintray-mirror -d 'Use the specified Bintray repository to automatically mirror stable URLs defined in the formulae (default: `mirror`)'
@@ -1077,6 +1078,7 @@ __fish_brew_complete_arg 'pr-pull' -l workflows -d 'Retrieve artifacts from the
 
 
 __fish_brew_complete_cmd 'pr-upload' 'Apply the bottle commit and publish bottles to Bintray or GitHub Releases'
+__fish_brew_complete_arg 'pr-upload' -l archive-item -d 'Upload to the specified Internet Archive item (default: `homebrew`)'
 __fish_brew_complete_arg 'pr-upload' -l bintray-org -d 'Upload to the specified Bintray organisation (default: `homebrew`)'
 __fish_brew_complete_arg 'pr-upload' -l debug -d 'Display any debugging information'
 __fish_brew_complete_arg 'pr-upload' -l dry-run -d 'Print what would be done rather than doing it'

completions/zsh/_brew

@@ -1229,10 +1229,11 @@ _brew_pr_publish() {
 # brew pr-pull
 _brew_pr_pull() {
   _arguments \
+    '(--bintray-org)--archive-item[Upload to the specified Internet Archive item (default: `homebrew`)]' \
     '--artifact[Download artifacts with the specified name (default: `bottles`)]' \
     '(--clean)--autosquash[Automatically reformat and reword commits in the pull request to our preferred format]' \
     '--bintray-mirror[Use the specified Bintray repository to automatically mirror stable URLs defined in the formulae (default: `mirror`)]' \
-    '--bintray-org[Upload to the specified Bintray organisation (default: `homebrew`)]' \
+    '(--archive-item)--bintray-org[Upload to the specified Bintray organisation (default: `homebrew`)]' \
     '--branch-okay[Do not warn if pulling to a branch besides the repository default (useful for testing)]' \
     '(--autosquash)--clean[Do not amend the commits from pull requests]' \
     '--debug[Display any debugging information]' \
@@ -1255,6 +1256,7 @@ _brew_pr_pull() {
 # brew pr-upload
 _brew_pr_upload() {
   _arguments \
+    '--archive-item[Upload to the specified Internet Archive item (default: `homebrew`)]' \
     '--bintray-org[Upload to the specified Bintray organisation (default: `homebrew`)]' \
     '--debug[Display any debugging information]' \
     '--dry-run[Print what would be done rather than doing it]' \

docs/Manpage.md

@@ -1181,6 +1181,8 @@ Requires write access to the repository.
   Message to include when autosquashing revision bumps, deletions, and rebuilds.
 * `--artifact`:
   Download artifacts with the specified name (default: `bottles`).
+* `--archive-item`:
+  Upload to the specified Internet Archive item (default: `homebrew`).
 * `--bintray-org`:
   Upload to the specified Bintray organisation (default: `homebrew`).
 * `--tap`:
@@ -1208,6 +1210,8 @@ Apply the bottle commit and publish bottles to Bintray or GitHub Releases.
   Do not generate a new commit before uploading.
 * `--warn-on-upload-failure`:
   Warn instead of raising an error if the bottle upload fails. Useful for repairing bottle uploads that previously failed.
+* `--archive-item`:
+  Upload to the specified Internet Archive item (default: `homebrew`).
 * `--bintray-org`:
   Upload to the specified Bintray organisation (default: `homebrew`).
 * `--root-url`:
@@ -1834,6 +1838,9 @@ example, run `export HOMEBREW_NO_INSECURE_REDIRECT=1` rather than just
 
   *Default:* The "Beer Mug" emoji.
 
+- `HOMEBREW_INTERNET_ARCHIVE_KEY`
+  <br>Use this API key when accessing the Internet Archive S3 API, where bottles are stored. The format is access:secret. See https://archive.org/account/s3.php
+
 - `HOMEBREW_LIVECHECK_WATCHLIST`
   <br>Consult this file for the list of formulae to check by default when no formula argument is passed to `brew livecheck`.
 

manpages/brew.1

@@ -1645,6 +1645,10 @@ Message to include when autosquashing revision bumps, deletions, and rebuilds\.
 Download artifacts with the specified name (default: \fBbottles\fR)\.
 .
 .TP
+\fB\-\-archive\-item\fR
+Upload to the specified Internet Archive item (default: \fBhomebrew\fR)\.
+.
+.TP
 \fB\-\-bintray\-org\fR
 Upload to the specified Bintray organisation (default: \fBhomebrew\fR)\.
 .
@@ -1692,6 +1696,10 @@ Do not generate a new commit before uploading\.
 Warn instead of raising an error if the bottle upload fails\. Useful for repairing bottle uploads that previously failed\.
 .
 .TP
+\fB\-\-archive\-item\fR
+Upload to the specified Internet Archive item (default: \fBhomebrew\fR)\.
+.
+.TP
 \fB\-\-bintray\-org\fR
 Upload to the specified Bintray organisation (default: \fBhomebrew\fR)\.
 .
@@ -2608,6 +2616,12 @@ Print this text before the installation summary of each successful build\.
 \fIDefault:\fR The "Beer Mug" emoji\.
 .
 .TP
+\fBHOMEBREW_INTERNET_ARCHIVE_KEY\fR
+.
+.br
+Use this API key when accessing the Internet Archive S3 API, where bottles are stored\. The format is access:secret\. See https://archive\.org/account/s3\.php
+.
+.TP
 \fBHOMEBREW_LIVECHECK_WATCHLIST\fR
 .
 .br