Merge pull request #14583 from SMillerDev/fix/analytics/improve_keys

fix: add better keys and fuller values to influxDB analytics

Library/Homebrew/.rubocop.yml

@@ -41,7 +41,7 @@ Metrics/PerceivedComplexity:
 Metrics/MethodLength:
   Max: 232
 Metrics/ModuleLength:
-  Max: 475
+  Max: 480
   Exclude:
     # TODO: extract more of the bottling logic
     - "dev-cmd/bottle.rb"

Library/Homebrew/brew.sh

@@ -651,6 +651,7 @@ export HOMEBREW_GIT
 export HOMEBREW_GIT_WARNING
 export HOMEBREW_MINIMUM_GIT_VERSION
 export HOMEBREW_LINUX_MINIMUM_GLIBC_VERSION
+export HOMEBREW_PHYSICAL_PROCESSOR
 export HOMEBREW_PROCESSOR
 export HOMEBREW_PRODUCT
 export HOMEBREW_OS_VERSION

Library/Homebrew/cask/installer.rb

@@ -111,7 +111,7 @@ module Cask
       install_artifacts
 
       if @cask.tap&.should_report_analytics?
-        ::Utils::Analytics.report_event("cask_install", @cask.token, on_request: true)
+        ::Utils::Analytics.report_event(:cask_install, @cask.token, on_request: true)
       end
 
       purge_backed_up_versioned_files

Library/Homebrew/env_config.rb

@@ -297,6 +297,11 @@ module Homebrew
         description: "If set, do not print any hints about changing Homebrew's behaviour with environment variables.",
         boolean:     true,
       },
+      HOMEBREW_NO_GOOGLE_ANALYTICS:              {
+        description: "If set, do not send analytics to Google Analytics but allow sending to Homebrew's InfluxDB " \
+                     "analytics server. For more information, see: <https://docs.brew.sh/Analytics>",
+        boolean:     true,
+      },
       HOMEBREW_NO_GITHUB_API:                    {
         description: "If set, do not use the GitHub API, e.g. for searches or fetching relevant issues " \
                      "after a failed install.",

Library/Homebrew/extend/os/mac/utils/analytics.rb

@@ -6,22 +6,18 @@ module Utils
     class << self
       extend T::Sig
 
-      sig { params(verbose: T::Boolean).returns(String) }
-      def custom_prefix_label(verbose: false)
-        return generic_custom_prefix_label if Hardware::CPU.arm?
+      sig { returns(String) }
+      def custom_prefix_label_google
+        return generic_custom_prefix_label_google if Hardware::CPU.arm?
 
         "non-/usr/local"
       end
 
-      sig { params(verbose: T::Boolean).returns(String) }
-      def arch_label(verbose: false)
-        if Hardware::CPU.arm?
-          "ARM"
-        elsif Hardware::CPU.in_rosetta2?
-          "Rosetta"
-        else
-          ""
-        end
+      sig { returns(String) }
+      def arch_label_google
+        return "Rosetta" if Hardware::CPU.in_rosetta2?
+
+        generic_arch_label_google
       end
     end
   end

Library/Homebrew/formula_installer.rb

@@ -416,7 +416,7 @@ class FormulaInstaller
 
     if formula.tap&.should_report_analytics?
       action = "#{formula.full_name} #{options}".strip
-      Utils::Analytics.report_event("install", action, on_request: installed_on_request?)
+      Utils::Analytics.report_event(:formula_install, action, on_request: installed_on_request?)
     end
 
     self.class.attempted << formula

Library/Homebrew/global.rb

@@ -48,6 +48,7 @@ HOMEBREW_WWW = "https://brew.sh"
 HOMEBREW_DOCS_WWW = "https://docs.brew.sh"
 HOMEBREW_SYSTEM = ENV.fetch("HOMEBREW_SYSTEM").freeze
 HOMEBREW_PROCESSOR = ENV.fetch("HOMEBREW_PROCESSOR").freeze
+HOMEBREW_PHYSICAL_PROCESSOR = ENV.fetch("HOMEBREW_PHYSICAL_PROCESSOR").freeze
 
 HOMEBREW_BREWED_CURL_PATH = Pathname(ENV.fetch("HOMEBREW_BREWED_CURL_PATH")).freeze
 HOMEBREW_USER_AGENT_CURL = ENV.fetch("HOMEBREW_USER_AGENT_CURL").freeze

Library/Homebrew/test/utils/analytics_spec.rb

@@ -5,49 +5,93 @@ require "utils/analytics"
 require "formula_installer"
 
 describe Utils::Analytics do
-  describe "::os_arch_prefix_ci" do
-    context "when os_arch_prefix_ci is not set" do
-      before do
-        described_class.clear_os_arch_prefix_ci
-      end
+  describe "::label_google" do
+    before do
+      described_class.clear_additional_tags_cache
+    end
 
-      let(:ci) { ", CI" if ENV["CI"] }
+    let(:ci) { ", CI" if ENV["CI"] }
 
-      it "returns OS_VERSION and prefix when HOMEBREW_PREFIX is a custom prefix on intel" do
-        allow(Hardware::CPU).to receive(:type).and_return(:intel)
-        allow(Hardware::CPU).to receive(:in_rosetta2?).and_return(false)
-        allow(Homebrew).to receive(:default_prefix?).and_return(false)
-        expect(described_class.os_arch_prefix_ci).to have_key(:prefix)
-        expect(described_class.os_arch_prefix_ci[:prefix]).to eq described_class.custom_prefix_label
-      end
+    it "returns OS_VERSION and prefix when HOMEBREW_PREFIX is a custom prefix on intel" do
+      allow(Hardware::CPU).to receive(:type).and_return(:intel)
+      allow(Hardware::CPU).to receive(:in_rosetta2?).and_return(false)
+      allow(Homebrew).to receive(:default_prefix?).and_return(false)
+      expect(described_class.label_google).to include described_class.custom_prefix_label_google
+    end
 
-      it "returns OS_VERSION, ARM and prefix when HOMEBREW_PREFIX is a custom prefix on arm" do
-        allow(Hardware::CPU).to receive(:type).and_return(:arm)
-        allow(Hardware::CPU).to receive(:in_rosetta2?).and_return(false)
-        allow(Homebrew).to receive(:default_prefix?).and_return(false)
-        expect(described_class.os_arch_prefix_ci).to have_key(:arch)
-        expect(described_class.os_arch_prefix_ci[:arch]).to eq described_class.arch_label
-        expect(described_class.os_arch_prefix_ci).to have_key(:prefix)
-        expect(described_class.os_arch_prefix_ci[:prefix]).to eq described_class.custom_prefix_label
-      end
+    it "returns OS_VERSION, ARM and prefix when HOMEBREW_PREFIX is a custom prefix on arm" do
+      allow(Hardware::CPU).to receive(:type).and_return(:arm)
+      allow(Hardware::CPU).to receive(:in_rosetta2?).and_return(false)
+      allow(Homebrew).to receive(:default_prefix?).and_return(false)
+      expect(described_class.label_google).to include described_class.arch_label_google
+      expect(described_class.label_google).to include described_class.custom_prefix_label_google
+    end
 
-      it "returns OS_VERSION, Rosetta and prefix when HOMEBREW_PREFIX is a custom prefix on Rosetta", :needs_macos do
-        allow(Hardware::CPU).to receive(:type).and_return(:intel)
-        allow(Hardware::CPU).to receive(:in_rosetta2?).and_return(true)
-        allow(Homebrew).to receive(:default_prefix?).and_return(false)
-        expect(described_class.os_arch_prefix_ci).to have_key(:prefix)
-        expect(described_class.os_arch_prefix_ci[:prefix]).to eq described_class.custom_prefix_label
-      end
+    it "returns OS_VERSION, Rosetta and prefix when HOMEBREW_PREFIX is a custom prefix on Rosetta", :needs_macos do
+      allow(Hardware::CPU).to receive(:type).and_return(:intel)
+      allow(Hardware::CPU).to receive(:in_rosetta2?).and_return(true)
+      allow(Homebrew).to receive(:default_prefix?).and_return(false)
+      expect(described_class.label_google).to include described_class.custom_prefix_label_google
+    end
 
-      it "does not include prefix when HOMEBREW_PREFIX is the default prefix" do
-        allow(Homebrew).to receive(:default_prefix?).and_return(true)
-        expect(described_class.os_arch_prefix_ci).not_to have_key(:prefix)
-      end
+    it "does not include prefix when HOMEBREW_PREFIX is the default prefix" do
+      allow(Homebrew).to receive(:default_prefix?).and_return(true)
+      expect(described_class.label_google).not_to include HOMEBREW_PREFIX.to_s
+    end
 
-      it "includes CI when ENV['CI'] is set" do
-        ENV["CI"] = "true"
-        expect(described_class.os_arch_prefix_ci).to have_key(:ci)
-      end
+    it "includes CI when ENV['CI'] is set" do
+      ENV["CI"] = "1"
+      expect(described_class.label_google).to include "CI"
+    end
+  end
+
+  describe "::additional_tags_influx" do
+    before do
+      described_class.clear_additional_tags_cache
+    end
+
+    let(:ci) { ", CI" if ENV["CI"] }
+
+    it "returns OS_VERSION and prefix when HOMEBREW_PREFIX is a custom prefix on intel" do
+      allow(Hardware::CPU).to receive(:type).and_return(:intel)
+      allow(Hardware::CPU).to receive(:in_rosetta2?).and_return(false)
+      allow(Homebrew).to receive(:default_prefix?).and_return(false)
+      expect(described_class.additional_tags_influx).to have_key(:prefix)
+      expect(described_class.additional_tags_influx[:prefix]).to eq "custom-prefix"
+    end
+
+    it "returns OS_VERSION, ARM and prefix when HOMEBREW_PREFIX is a custom prefix on arm" do
+      allow(Hardware::CPU).to receive(:type).and_return(:arm)
+      allow(Hardware::CPU).to receive(:in_rosetta2?).and_return(false)
+      allow(Homebrew).to receive(:default_prefix?).and_return(false)
+      expect(described_class.additional_tags_influx).to have_key(:arch)
+      expect(described_class.additional_tags_influx[:arch]).to eq HOMEBREW_PHYSICAL_PROCESSOR
+      expect(described_class.additional_tags_influx).to have_key(:prefix)
+      expect(described_class.additional_tags_influx[:prefix]).to eq "custom-prefix"
+    end
+
+    it "returns OS_VERSION, Rosetta and prefix when HOMEBREW_PREFIX is a custom prefix on Rosetta", :needs_macos do
+      allow(Hardware::CPU).to receive(:type).and_return(:intel)
+      allow(Hardware::CPU).to receive(:in_rosetta2?).and_return(true)
+      allow(Homebrew).to receive(:default_prefix?).and_return(false)
+      expect(described_class.additional_tags_influx).to have_key(:prefix)
+      expect(described_class.additional_tags_influx[:prefix]).to eq "custom-prefix"
+    end
+
+    it "does not include prefix when HOMEBREW_PREFIX is the default prefix" do
+      allow(Homebrew).to receive(:default_prefix?).and_return(true)
+      expect(described_class.additional_tags_influx).to have_key(:prefix)
+      expect(described_class.additional_tags_influx[:prefix]).to eq HOMEBREW_PREFIX.to_s
+    end
+
+    it "includes CI when ENV['CI'] is set" do
+      ENV["CI"] = "1"
+      expect(described_class.additional_tags_influx).to have_key(:ci)
+    end
+
+    it "includes developer when ENV['HOMEBREW_DEVELOPER'] is set" do
+      ENV["HOMEBREW_DEVELOPER"] = "1"
+      expect(described_class.additional_tags_influx).to have_key(:developer)
     end
   end
 
@@ -61,14 +105,14 @@ describe Utils::Analytics do
         ENV["HOMEBREW_NO_ANALYTICS"] = "true"
         expect(described_class).not_to receive(:report_google)
         expect(described_class).not_to receive(:report_influx)
-        described_class.report_event("install", action)
+        described_class.report_event(:install, action)
       end
 
       it "returns nil when HOMEBREW_NO_ANALYTICS_THIS_RUN is true" do
         ENV["HOMEBREW_NO_ANALYTICS_THIS_RUN"] = "true"
         expect(described_class).not_to receive(:report_google)
         expect(described_class).not_to receive(:report_influx)
-        described_class.report_event("install", action)
+        described_class.report_event(:install, action)
       end
 
       it "returns nil when HOMEBREW_ANALYTICS_DEBUG is true" do
@@ -78,13 +122,14 @@ describe Utils::Analytics do
         expect(described_class).to receive(:report_google)
         expect(described_class).to receive(:report_influx)
 
-        described_class.report_event("install", action)
+        described_class.report_event(:install, action)
       end
     end
 
     it "passes to the influxdb and google methods" do
       ENV.delete("HOMEBREW_NO_ANALYTICS_THIS_RUN")
       ENV.delete("HOMEBREW_NO_ANALYTICS")
+      ENV.delete("HOMEBREW_DEVELOPER")
       ENV["HOMEBREW_ANALYTICS_DEBUG"] = "true"
       expect(described_class).to receive(:report_google).with(:event, hash_including(ea: action)).once
       expect(described_class).to receive(:report_influx).with(:install, "formula_name --head", false,
@@ -95,6 +140,7 @@ describe Utils::Analytics do
     it "sends to google twice on request" do
       ENV.delete("HOMEBREW_NO_ANALYTICS_THIS_RUN")
       ENV.delete("HOMEBREW_NO_ANALYTICS")
+      ENV.delete("HOMEBREW_DEVELOPER")
       ENV["HOMEBREW_ANALYTICS_DEBUG"] = "true"
       expect(described_class).to receive(:report_google).with(:event, hash_including(ea: action, ec: :install)).once
       expect(described_class).to receive(:report_google).with(:event,
@@ -117,6 +163,7 @@ describe Utils::Analytics do
       ENV.delete("HOMEBREW_NO_ANALYTICS")
       ENV["HOMEBREW_ANALYTICS_DEBUG"] = "true"
       ENV["HOMEBREW_ANALYTICS_ENABLE_INFLUX"] = "true"
+      ENV["HOMEBREW_DEVELOPER"] = "1"
       expect(described_class).to receive(:deferred_curl).once
       described_class.report_influx(:install, action, true, developer: true, CI: true)
     end

Library/Homebrew/utils/analytics.rb

@@ -20,15 +20,8 @@ module Utils
       INFLUX_TOKEN = "y2JZsgE7glWT9V-S-nElETLp8oyH9PGh9JVa-kCdOdp7mEHIOws4BtdjsIe3HHpCBty7IQHLnmh0prqK2vBj9A=="
       INFLUX_HOST = "europe-west1-1.gcp.cloud2.influxdata.com"
 
-      sig { params(type: T.any(String, Symbol), metadata: T::Hash[Symbol, T.untyped]).void }
+      sig { params(type: Symbol, metadata: T::Hash[Symbol, T.untyped]).void }
       def report_google(type, metadata = {})
-        os = metadata[:el][:os]
-        arch = ", #{metadata[:el][:arch]}" if metadata[:el][:arch].present?
-        prefix = ", #{metadata[:el][:prefix]}" if metadata[:el][:prefix].present?
-        ci = ", CI" if metadata[:el][:CI] == true
-
-        metadata[:el] = "#{os}#{arch}#{prefix}#{ci}"
-
         analytics_ids = ENV.fetch("HOMEBREW_ANALYTICS_IDS", "").split(",")
         analytics_ids.each do |analytics_id|
           args = []
@@ -79,14 +72,14 @@ module Utils
       end
 
       sig {
-        params(category: T.any(String, Symbol), action: T.any(String, Symbol), on_request: T::Boolean,
+        params(measurement: Symbol, package_and_options: String, on_request: T::Boolean,
                additional_tags: T::Hash[Symbol, T.untyped]).void
       }
-      def report_influx(category, action, on_request, additional_tags = {})
+      def report_influx(measurement, package_and_options, on_request, additional_tags = {})
         return unless ENV["HOMEBREW_ANALYTICS_ENABLE_INFLUX"]
 
         # Append general information to device information
-        tags = additional_tags.merge(action: action, on_request: !on_request.nil?)
+        tags = additional_tags.merge(package_and_options: package_and_options, on_request: !on_request.nil?)
                               .compact_blank
                               .map { |k, v| "#{k}=#{v.to_s.sub(" ", "\\ ")}" } # convert to key/value parameters
                               .join(",")
@@ -96,7 +89,7 @@ module Utils
           "--header", "Content-Type: text/plain; charset=utf-8",
           "--header", "Accept: application/json",
           "--header", "Authorization: Token #{INFLUX_TOKEN}",
-          "--data-raw", "#{category},#{tags} count=1i #{Time.now.to_i}"
+          "--data-raw", "#{measurement},#{tags} count=1i #{Time.now.to_i}"
         ]
 
         url = "https://#{INFLUX_HOST}/api/v2/write?bucket=#{INFLUX_BUCKET}&precision=s"
@@ -117,40 +110,76 @@ module Utils
         end
       end
 
-      sig { params(category: T.any(String, Symbol), action: String, on_request: T::Boolean).void }
-      def report_event(category, action, on_request: false)
-        return if not_this_run?
-        return if disabled?
+      sig { params(measurement: Symbol, package_and_options: String, on_request: T::Boolean).void }
+      def report_event(measurement, package_and_options, on_request: false)
+        report_google_event(measurement, package_and_options, on_request: on_request)
+        report_influx_event(measurement, package_and_options, on_request: on_request)
+      end
 
-        google_label = os_arch_prefix_ci(verbose: false)
+      sig { params(category: Symbol, action: String, on_request: T::Boolean).void }
+      def report_google_event(category, action, on_request: false)
+        return if not_this_run? || disabled? || Homebrew::EnvConfig.no_google_analytics?
+
+        category = "install" if category == :formula_install
 
         report_google(:event,
                       ec: category,
                       ea: action,
-                      el: google_label,
+                      el: label_google,
                       ev: nil)
 
-        if on_request
-          report_google(:event,
-                        ec: :install_on_request,
-                        ea: action,
-                        el: google_label,
-                        ev: nil)
-        end
+        return unless on_request
 
-        influx_additional_data = os_arch_prefix_ci(verbose: true)
-        report_influx(category, action, on_request, influx_additional_data)
+        report_google(:event,
+                      ec: :install_on_request,
+                      ea: action,
+                      el: label_google,
+                      ev: nil)
       end
 
+      sig { params(measurement: Symbol, package_and_options: String, on_request: T::Boolean).void }
+      def report_influx_event(measurement, package_and_options, on_request: false)
+        return if not_this_run? || disabled?
+
+        report_influx(measurement, package_and_options, on_request, additional_tags_influx)
+      end
+
+      sig { params(exception: Exception).void }
       def report_build_error(exception)
+        report_google_build_error(exception)
+        report_influx_error(exception)
+      end
+
+      sig { params(exception: Exception).void }
+      def report_google_build_error(exception)
+        return if not_this_run? || disabled?
+
         return unless exception.formula.tap
         return unless exception.formula.tap.should_report_analytics?
 
-        action = exception.formula.full_name
-        if (options = exception.options.to_a.map(&:to_s).join(" ").presence)
-          action = "#{action} #{options}".strip
+        formula_full_name = exception.formula.full_name
+        package_and_options = if (options = exception.options.to_a.map(&:to_s).join(" ").presence)
+          "#{formula_full_name} #{options}".strip
+        else
+          formula_full_name
         end
-        report_event("BuildError", action)
+        report_event("BuildError", package_and_options)
+      end
+
+      sig { params(exception: Exception).void }
+      def report_influx_error(exception)
+        return if not_this_run? || disabled?
+
+        return unless exception.formula.tap
+        return unless exception.formula.tap.should_report_analytics?
+
+        formula_full_name = exception.formula.full_name
+        package_and_options = if (options = exception.options.to_a.map(&:to_s).join(" ").presence)
+          "#{formula_full_name} #{options}".strip
+        else
+          formula_full_name
+        end
+        report_event(:build_error, package_and_options)
       end
 
       def messages_displayed?
@@ -280,45 +309,55 @@ module Utils
         nil
       end
 
-      sig { params(verbose: T::Boolean).returns(String) }
-      def custom_prefix_label(verbose: false)
+      sig { returns(String) }
+      def custom_prefix_label_google
         "custom-prefix"
       end
-      alias generic_custom_prefix_label custom_prefix_label
+      alias generic_custom_prefix_label_google custom_prefix_label_google
 
-      sig { params(verbose: T::Boolean).returns(String) }
-      def arch_label(verbose: false)
+      sig { returns(String) }
+      def arch_label_google
         if Hardware::CPU.arm?
           "ARM"
         else
           ""
         end
       end
+      alias generic_arch_label_google arch_label_google
 
-      def clear_os_arch_prefix_ci
-        return unless instance_variable_defined?(:@os_arch_prefix_ci)
-
-        remove_instance_variable(:@os_arch_prefix_ci)
+      def clear_additional_tags_cache
+        remove_instance_variable(:@label_google) if instance_variable_defined?(:@label_google)
+        remove_instance_variable(:@additional_tags_influx) if instance_variable_defined?(:@additional_tags_influx)
       end
 
-      sig { params(verbose: T::Boolean).returns(T::Hash[Symbol, String]) }
-      def os_arch_prefix_ci(verbose: false)
-        @os_arch_prefix_ci ||= begin
-          data = {
-            os:        OS_VERSION,
-            developer: Homebrew::EnvConfig.developer?,
-            version:   HOMEBREW_VERSION,
-            system:    HOMEBREW_SYSTEM,
-            ci:        ENV["CI"].present?,
-            arch:      arch_label(verbose: verbose),
-            prefix:    custom_prefix_label(verbose: verbose),
-          }
-          unless verbose
-            data.delete(:arch) if data[:arch].blank?
-            data.delete(:prefix) if Homebrew.default_prefix?
-          end
+      sig { returns(String) }
+      def label_google
+        @label_google ||= begin
+          os = OS_VERSION
+          arch = ", #{arch_label_google}" if arch_label_google.present?
+          prefix = ", #{custom_prefix_label_google}" unless Homebrew.default_prefix?
+          ci = ", CI" if ENV["CI"]
+          "#{os}#{arch}#{prefix}#{ci}"
+        end
+      end
 
-          data
+      sig { returns(T::Hash[Symbol, String]) }
+      def additional_tags_influx
+        @additional_tags_influx ||= begin
+          version = HOMEBREW_VERSION.match(/^[\d.]+/)[0]
+          version = "#{version}-dev" if HOMEBREW_VERSION.include?("-")
+          prefix = Homebrew.default_prefix? ? HOMEBREW_PREFIX.to_s : "custom-prefix"
+
+          {
+            version:             version,
+            prefix:              prefix,
+            default_prefix:      Homebrew.default_prefix?,
+            ci:                  ENV["CI"].present?,
+            developer:           Homebrew::EnvConfig.developer?,
+            arch:                HOMEBREW_PHYSICAL_PROCESSOR,
+            os:                  HOMEBREW_SYSTEM,
+            os_name_and_version: OS_VERSION,
+          }
         end
       end
 

docs/Analytics.md

@@ -1,6 +1,6 @@
 # Anonymous Aggregate User Behaviour Analytics
 
-Homebrew gathers anonymous aggregate user behaviour analytics using Google Analytics. You will be notified the first time you run `brew update` or install Homebrew. Analytics are not enabled until after this notice is shown, to ensure that you can [opt out](Analytics.md#opting-out) without ever sending analytics data.
+Homebrew gathers anonymous aggregate user behaviour analytics using Google Analytics (until our in-progress migration to our own InfluxDB). You will be notified the first time you run `brew update` or install Homebrew. Analytics are not enabled until after this notice is shown, to ensure that you can [opt out](Analytics.md#opting-out) without ever sending analytics data.
 
 ## Why?
 
@@ -61,6 +61,12 @@ Homebrew analytics helps us maintainers and leaving it on is appreciated. Howeve
 export HOMEBREW_NO_ANALYTICS=1
 ```
 
+If you are fine with analytics being sent to Homebrew's InfluxDB but not to Google Analytics, you can set:
+
+```sh
+export HOMEBREW_NO_GOOGLE_ANALYTICS=1
+```
+
 Alternatively, this will prevent analytics from ever being sent:
 
 ```sh