greenhat 0.4.0 → 0.6.1

data/lib/greenhat/shell/process.rb

@@ -30,8 +30,10 @@ module GreenHat
         puts
       end
 
-      def self.filter_help
-        ShellHelper::Filter.help
+      def self.filter_help(raw = {})
+        args, flags, _args = Args.parse(raw)
+
+        ShellHelper.show(ShellHelper::Filter.help(args.first), flags)
       end
 
       def self.ps(raw = {})
@@ -45,6 +47,10 @@ module GreenHat
         ShellHelper.file_output(files, flags)
       end
 
+      def self.default(raw_list)
+        filter(raw_list)
+      end
+
       def self.filter(raw = {})
         # Argument Parsing
         files, flags, args = Args.parse(raw)
@@ -52,14 +58,14 @@ module GreenHat
         # Prepare Log List
         file_list = ShellHelper.prepare_list(files, GreenHat::Ps.things)
 
-        results = ShellHelper.filter_start(file_list, flags, args)
+        results = Query.start(file_list, flags, args)
 
         # Check Search Results
         if results.instance_of?(Hash) && results.values.flatten.empty?
           puts 'No results'.pastel(:red)
         else
-          # This causes the key 'colorized' output to also be included
-          ShellHelper.show(results.to_a.compact.flatten, flags)
+
+          ShellHelper.show(results, flags)
         end
       end
     end

data/lib/greenhat/shell/query.rb

@@ -0,0 +1,534 @@
+# NameSpace
+module GreenHat
+  # Query Handlers
+  # rubocop:disable Metrics/ModuleLength
+  module Query
+    # Main Entry Point for Filtering
+    def self.start(files, flags = {}, args = {})
+      # Convert to Things / Thing.list == processed files
+      files = ShellHelper.find_things(files, flags, Thing.list).select(&:query?)
+
+      # Breakdown by Interval
+      return process_interval(files, flags, args) if flags.key? :interval
+
+      # Ignore Archive/Host Dividers
+      if flags[:combine]
+        # Add SRC Field to Combined Entries
+        results = combine_src_add(files)
+        Query.filter(results.flatten.compact, flags, args)
+      else
+        file_filter(files, flags, args)
+      end
+    end
+
+    def self.combine_src_add(files)
+      files.reject(&:blank?).map do |file|
+        src = file.archive&.friendly_name || file&.name
+        file.data.map do |entry|
+          bit = entry.clone
+          bit[:src] = src
+          bit
+        end
+      end
+    end
+
+    # Simplify Loop for Query.start
+    def self.file_filter(files, flags = {}, args = {})
+      # Iterate and Preserve Archive/Host Index
+      files.each_with_object([]) do |file, obj|
+        # Ignore Empty Results / No Thing
+        next if file&.blank?
+
+        # Include Total Count in Name
+        results = Query.filter(file.data, flags, args)
+
+        next if results.count.zero? # Skip if there are no results
+
+        duration = calculate_duration(results)
+
+        # Create Title
+        title = create_title(file, flags, results)
+
+        # Append Duration
+        title += " #{duration.pastel(:cyan, :dim)}" unless duration.blank?
+
+        # Add Title
+        obj.push(title)
+
+        # Add Results
+        obj.concat(results)
+
+        obj
+      end
+    end
+
+    # Helper to simplify Title Creation
+    def self.create_title(file, flags, results)
+      title = file.friendly_name
+      # Ignore for Total Results
+      title += " #{results.count}".pastel(:bright_black) unless flags.key?(:total)
+      title
+    end
+
+    # Helper to transform fields from one to another (timestamp => time)
+    def self.process_transform(transform, results)
+      # TODO: Eventually provide more splat options
+      from, to, *splat = transform
+      results.each do |entry|
+        entry[to.to_sym] = case splat
+                           when [:to_i]
+                             entry[from.to_sym].to_i
+                           else
+                             entry[from.to_sym]
+                           end
+      end
+    end
+
+    # Organize Results by Interval
+    # TODO Simplify
+    # rubocop:disable Metrics/PerceivedComplexity, Metrics/CyclomaticComplexity
+    def self.process_interval(files, flags, args)
+      data = files.reject(&:blank?).map(&:data).flatten.compact
+
+      # Handle Transform
+      process_transform(flags[:transform], data) if flags.key? :transform
+
+      interval = ChronicDuration.parse(flags[:interval]) || 5.minutes
+      index = build_time_index(data, interval)
+      index.transform_values! { |_y| [] }
+
+      data.each do |r|
+        index[r.time.floor(interval)] << r
+      end
+
+      output = index.each_with_object({}) do |(time, entries), obj|
+        results = Query.filter(entries, flags, args)
+
+        next if results.count.zero? # Skip if No Results
+
+        duration = calculate_duration(results)
+
+        # Timzone Manipulation
+        time_title = flags.key?(:time_zone) ? time.in_time_zone(flags[:time_zone]) : time
+
+        title = time_title.to_s.pastel(:bright_blue, :bold)
+        title += " #{results.count}".pastel(:bright_black) unless flags.key?(:total)
+
+        # Append Duration
+        title += " #{duration.pastel(:cyan, :dim)}" unless duration.blank?
+
+        # Add Results
+        obj[title] = results
+
+        obj
+      end
+
+      # Remove Empty Intervals / There may be empty here due to results
+      output.reject! { |_k, v| v.flatten.empty? }
+
+      output.to_a.flatten(2)
+    end
+
+    def self.calculate_duration(results)
+      # Skip for Pluck
+      only_with_time = results.select { |x| x.instance_of?(Hash) && x.key?(:time) && x[:time].instance_of?(Time) }
+
+      # If slice is used ignore
+      return nil if only_with_time.empty?
+
+      sorted = only_with_time.map(&:time).sort
+      humanize_time(sorted.first, sorted.last)
+    end
+
+    # Replace TimeDifference with https://stackoverflow.com/a/4136485/1678507
+    def self.humanize_time(time_start, time_end, increments = 2)
+      miliseconds = (time_end - time_start) * 1000
+
+      list = [[1000, :ms], [60, :s], [60, :m], [24, :h]].map do |count, name|
+        next unless miliseconds.positive?
+
+        miliseconds, n = miliseconds.divmod(count)
+
+        "#{n.to_i}#{name}" unless n.to_i.zero?
+      end
+
+      list.compact.reverse[0..increments - 1].join(' ')
+    end
+
+    # Filter Logic
+    # TODO: Simplify
+    def self.filter(data, flags = {}, args = {})
+      results = data.clone
+
+      # Handle Transform
+      process_transform(flags[:transform], results) if flags.key? :transform
+
+      results.select! do |row|
+        args.send(flags.logic || :all?) do |arg|
+          filter_row_key(row, arg, flags)
+        end
+      end
+
+      # Ensure presecense of a specific field
+      results = filter_exists(results, flags[:exists]) if flags.key?(:exists)
+
+      # Time Zone
+      results = filter_modify_timezone(results, flags[:time_zone]) if flags.key?(:time_zone)
+
+      # Time Filtering
+      results = filter_time(results, flags) if flags.key?(:start) || flags.key?(:end)
+
+      # Strip Results if Slice is defined
+      results = filter_slice(results, flags[:slice]) if flags.key?(:slice)
+
+      # Strip Results if Except is defined
+      results = filter_except(results, flags[:except]) if flags.key?(:except)
+
+      # Remove Blank from either slice or except
+      results.reject!(&:empty?)
+
+      # Sort / Reverse by default
+      if flags.key?(:sort)
+        results.sort_by! { |x| x.slice(*flags[:sort]).values }
+        results.reverse!
+      end
+
+      # JSON Formatting
+      results = results.map { |x| Oj.dump(x) } if flags.key?(:json)
+
+      # Show Unique Only
+      results = filter_uniq(results, flags[:uniq]) if flags.key?(:uniq)
+
+      # Reverse
+      results.reverse! if flags[:reverse]
+
+      # Count occurrences / Skip Results
+      return filter_stats(results, flags) if flags.key?(:stats)
+
+      # Percentile Breakdown
+      return filter_percentile(results, flags) if flags.key?(:percentile)
+
+      # Limit before Pluck / Flattening
+      results = filter_limit(results, flags[:limit]) if flags.key?(:limit)
+
+      # Pluck
+      results = filter_pluck(results, flags[:pluck]) if flags.key?(:pluck)
+
+      # Total Counter
+      return [ShellHelper.total_count(results, flags)] if flags.key?(:total)
+
+      results
+    end
+    # rubocop:enable Metrics/CyclomaticComplexity, Metrics/PerceivedComplexity
+
+    # Create list of Times
+    def self.build_time_index(results, interval = 5.minutes)
+      index = {}
+      start = results.min_by(&:time).time.floor(interval)
+      finish = results.max_by(&:time).time.floor(interval)
+
+      loop do
+        index[start] = 0
+        start += interval
+        break if start > finish
+      end
+
+      index
+    rescue StandardError => e
+      LogBot.fatal('Index Error', e.message)
+    ensure
+      {}
+    end
+
+    # Limit / Ensure Exists and Valid Number
+    def self.filter_limit(results, limit)
+      return results unless limit.integer? && limit.positive?
+
+      results.shift limit
+    end
+
+    def self.filter_modify_timezone(results, time_zone)
+      results.map do |x|
+        next unless x.key? :time
+
+        x = x.clone # Prevent Top Level Modification
+        x[:time] = x[:time].in_time_zone time_zone
+
+        x
+      end
+    end
+
+    # Filter Start and End Times
+    # TODO: This is a bit icky, simplify/dry
+    def self.filter_time(results, flags)
+      if flags.key?(:start)
+        begin
+          time_start = Time.parse(flags[:start])
+
+          results.select! do |x|
+            if x.time
+              time_start < x.time
+            else
+              true
+            end
+          end
+        rescue StandardError
+          puts 'Unable to Process Start Time Filter'.pastel(:red)
+        end
+      end
+
+      if flags.key?(:end)
+        begin
+          time_start = Time.parse(flags[:end])
+
+          results.select! do |x|
+            if x.time
+              time_start > x.time
+            else
+              true
+            end
+          end
+        rescue StandardError
+          puts 'Unable to Process End Time Filter'.pastel(:red)
+        end
+      end
+
+      results
+    end
+
+    def self.filter_except(results, except)
+      # Avoid Empty Results
+      if except.empty?
+        filter_empty_arg('except')
+        return results
+      end
+
+      results.map { |row| row.except(*except) }
+    end
+
+    def self.filter_exists(results, exists)
+      # Avoid Empty Results
+      if exists.empty?
+        filter_empty_arg('exists')
+        return results
+      end
+
+      results.select { |row| (exists - row.keys).empty? }
+    end
+
+    def self.filter_slice(results, slice)
+      # Avoid Empty Results
+      if slice.empty?
+        filter_empty_arg('slice')
+        return results
+      end
+
+      results.compact.map { |row| row.slice(*slice) }
+    end
+
+    def self.filter_pluck(results, pluck)
+      # Avoid Empty Results
+      if pluck.empty?
+        filter_empty_arg('pluck')
+        return results
+      end
+
+      results.map { |x| x.slice(*pluck).values }.flatten
+    end
+
+    def self.filter_uniq(results, unique)
+      # Avoid Empty Results
+      if unique.empty?
+        filter_empty_arg('uniq')
+        return results
+      end
+
+      unique.map do |field|
+        results.uniq { |x| x[field] }
+      end.inject(:&)
+    end
+
+    # TODO: Make better
+    # rubocop:disable Metrics/PerceivedComplexity, Metrics/CyclomaticComplexity
+    def self.filter_percentile(results, flags)
+      output = {}
+
+      results.each do |entry|
+        entry.each do |key, value|
+          # Numbers only
+          next unless value.instance_of?(Integer) || value.instance_of?(Float)
+
+          output[key] ||= []
+          output[key].push value
+        end
+      end
+
+      data = output.map do |key, values|
+        # Support Rounding
+        l99 = values.percentile(0.99)
+        l99 = l99.round(flags.round) if flags.round
+
+        l95 = values.percentile(0.99)
+        l95 = l95.round(flags.round) if flags.round
+
+        {
+          key: key,
+          '99' => l99,
+          '95' => l95,
+          mean: flags.round ? values.mean.round(flags.round) : values.mean,
+          min: flags.round ? values.min.round(flags.round) : values.min,
+          max: flags.round ? values.max.round(flags.round) : values.max,
+          count: values.count
+        }
+      end
+
+      headers = data.flat_map(&:keys).uniq
+      [[headers, data.map(&:values)]] # Multiple Arrays for Results Flatten
+    end
+    # rubocop:enable Metrics/PerceivedComplexity, Metrics/CyclomaticComplexity
+
+    def self.filter_stats(results, flags)
+      stats = flags[:stats]
+
+      # Avoid Empty Results
+      if stats.empty?
+        filter_empty_arg('stats')
+        return results
+      end
+
+      # Loop through Stats, Separate Hash/Tables
+      stats.map do |field|
+        occurrences = filter_count_occurrences(results, field, flags)
+
+        # Use Truncate For Long Keys
+        occurrences.transform_keys! { |key| key.to_s[0..flags[:truncate]] } if flags[:truncate]
+
+        # Total Occurences
+        total = occurrences.values.sum
+
+        # Percs
+        occurrences.transform_values! do |count|
+          [
+            count,
+            " #{percent(count, total)}%".pastel(:bright_black)
+          ]
+        end
+
+        # Sort by total occurances / New Variable for Total
+        output = occurrences.sort_by(&:last).to_h.transform_values!(&:join).to_a
+
+        # Append Header / Total with field name
+        output.unshift([field.to_s.pastel(:bright_black), total])
+
+        # Format
+        output.to_h
+      end
+    end
+
+    # Percent Helper
+    def self.percent(value, total)
+      ((value / total.to_f) * 100).round
+    end
+
+    # Helper to Count occurrences
+    def self.filter_count_occurrences(results, field, flags = {})
+      results.each_with_object(Hash.new(0)) do |entry, counts|
+        if entry.key? field
+          # Rounding in pagination breaks stats
+          key = if flags.key?(:round) && entry[field].numeric?
+                  entry[field].to_f.round(flags.round)
+                else
+                  entry[field]
+                end
+
+          counts[key] += 1
+        else
+          counts['None'.pastel(:bright_black)] += 1
+        end
+
+        counts
+      end
+    end
+
+    def self.filter_empty_arg(arg)
+      puts [
+        'Ignoring'.pastel(:bright_yellow),
+        "--#{arg}".pastel(:cyan),
+        'it requires an argument'.pastel(:red)
+      ].join(' ')
+    end
+
+    # Break out filter row logic into separate method
+    def self.filter_row_key(row, arg, flags)
+      return false if row.nil? # Nothing to filter if row empty
+
+      # Ignore Other Logic if Field isn't even included / Full Text Searching
+      return false unless row.key?(arg[:field]) || arg[:field] == :text
+
+      # Sensitivity Check / Check for Match / Full Text Searching
+      search_data = arg[:field] == :text ? row : row[arg.field]
+      match = filter_row_entry(search_data.to_s, arg, flags)
+
+      # Pivot of off include vs exclude
+      if arg.bang
+        !match
+      else
+        match
+      end
+    end
+
+    # Field Partial / Case / Exact search
+    def self.filter_row_entry(entry, arg, flags)
+      # Exact Matching / Unless doing full text search
+      return entry == arg.value.to_s if flags.key?(:exact) && arg.field != :text
+
+      # Cast to String/Integer Helper
+      entry, value = filter_entry_cast(entry, arg, flags)
+
+      entry.send(arg.logic, value)
+    end
+
+    # Handle casting to strings or integers
+    def self.filter_entry_cast(entry, arg, flags)
+      # Cast to String
+      value = arg.value.to_s
+
+      # No Logic on Empty Entries
+      return [entry, value] if entry.empty?
+
+      case arg.logic
+      when :include?
+
+        # Exact Case argument
+        unless flags.key?(:case)
+          entry = entry.downcase
+          value = value.downcase
+        end
+      when :>=, :<=
+        entry = entry.to_i if entry.numeric?
+        value = value.to_i if value&.numeric?
+      end
+
+      [entry, value]
+    end
+
+    def self.filter_entry_logic(entry, arg)
+      entry.send(arg.logic, arg.value)
+    end
+  end
+
+  # Internal Query Helper
+  # query = 'gitlab-rails/application_json.log --message!="Cannot obtain an exclusive lease" --severity=error'
+  # Query.filter_internal(query)
+  def self.filter_internal(search = '')
+    files, flags, args = Args.parse(Shellwords.split(search))
+    flags[:combine] = true
+
+    # Default to everything
+    files = Thing.all.map(&:name) if files.empty?
+
+    Query.start(files, flags, args)
+  end
+
+  # rubocop:enable Metrics/ModuleLength
+end