btrc-hub/backend/app/services/counts_table_export.rb

require 'csv'

# Export uses the exact same population and grouped cells as the interactive table.
class CountsTableExport
  EXTRA = ['format', 'columns', 'headers', 'labels', 'children', 'child_sort',
           'order', 'nested_json', 'summary_columns', 'basis', 'axis', 'latest'].freeze
  attr_reader :format, :resource

  def initialize(params)
    raise ArgumentError if (params.keys - CountsTableQuery::KEYS - EXTRA).any?
    raise ArgumentError unless params.values.all? { _1.is_a?(String) }

    @params = params
    @labels = JSON.parse(params.fetch('labels', '{}'))
    raise ArgumentError unless @labels.is_a?(Hash)
    @selected = params.key?('children') ? params['children'].split(',') : nil
    @order = params.fetch('order', '').split(',')
    @child_sort = params.fetch('child_sort', '').split(',').map { _1.split(':', -1) }
    @child_sort.each do |path, direction, *extra|
      raise ArgumentError if extra.any? || !['asc', 'desc'].include?(direction)
      raise ArgumentError unless path&.match?(/\A[a-zA-Z_]+(?:\.[a-zA-Z]+)*\/[a-zA-Z]+\z/)
    end
    @format = params.fetch('format', 'json')
    @resource = params.fetch('resource', 'posts')
    raise ArgumentError unless ['json', 'csv'].include?(format)

    @query = CountsTableQuery.new(params.slice(*CountsTableQuery::KEYS)
        .merge('page' => '1', 'limit' => '100'))
    @summary = @query.mode == 'summary'
    @columns = params.fetch('columns', '').split(',')
    @headers = JSON.parse(params.fetch('headers', '[]'))
    raise ArgumentError unless @headers.is_a?(Array) && @headers.all? { _1.is_a?(String) }

    unless @summary
      if ['posts', 'tags'].include?(resource)
        @projection = CountsExportProjection.new(params, resource:, summary: false, tags: false)
      else
        allowed = @query.population.definition.fields.keys + ['is_latest']
        raise ArgumentError if (@columns - allowed).any?
        raise ArgumentError unless @columns.length == @headers.length && @columns.uniq == @columns
      end
    end
  end

  def body
    Enumerator.new do |output|
      ApplicationRecord.transaction(isolation: :repeatable_read) do
        @language_names = Language.pluck(:code, :name).to_h
        started = false
        first = true
        @query.each_page do |page|
          unless started
            metadata = page.except(:versions, :cells, :axis_labels, :column_values).merge(
                resource:, filters: @params.slice(*CountsTableQuery::KEYS),
                fields: @query.population.definition.metadata,
                columns: @columns, column_labels: @headers,
                order: @params['order'], children: @params['children'],
                sort: @params.fetch('sort', ''), child_sort: @params['child_sort'])
            output << (format == 'json' ? "{\"metadata\":#{ metadata.to_json },\"data\":[" : "\uFEFF")
            if format == 'csv'
              headers = @summary ? summary_headers : @headers
              output << CSV.generate_line(headers.map { safe(_1) }, row_sep: "\r\n")
            end
            started = true
          end
          (page[:cells] || page[:versions]).each do |row|
            record = project(row)
            if format == 'json'
              output << ',' unless first
              output << record.to_json
            else
              values = if @summary
                         row[:values].each_with_index.map do |value, index|
                           page[:axis_labels].fetch("a#{ index }", {}).fetch(value.to_s, value)
                         end + [row[:count]]
                       elsif @projection
                         @projection.csv_record(record, row)
                       else
                         record.map { |key, value| csv_logical(key, value) }
                       end
              output << CSV.generate_line(values.map { safe(_1) }, row_sep: "\r\n")
            end
            first = false
          end
        end
        output << ']}' if format == 'json'
      end
    end
  end

  private

  def summary_headers
    labels = @labels
    @query.axes.all.map { labels.fetch(_1['field'], _1['field']) } +
        [labels.fetch('count', 'count')]
  end

  def project(row)
    return { row: row[:values].first(@query.axes.rows.length),
             column: row[:values].drop(@query.axes.rows.length), count: row[:count] } if @summary
    return @projection.project(row) if @projection

    @columns.to_h do |column|
      key = column == 'user' ? 'created_by_user' : column
      value = row[key]
      value = project_nested(value, key) if ['members', 'placements', 'export_paths_json'].include?(key)
      [key, value]
    end
  end

  def camel(key)
    key.gsub(/_([a-z])/) { Regexp.last_match(1).upcase }
  end

  def project_nested(value, path)
    rows = path == 'export_paths_json' ? (value || {}).map { |profile, target|
      { 'profile' => profile, 'path' => target }
    } : Array(value)
    sorts = @child_sort.filter_map do |key, direction|
      parent, field = key.split('/', 2)
      [field, direction] if parent == path
    end
    rows = rows.each_with_index.sort do |(first, a), (last, b)|
      compared = 0
      sorts.each do |field, direction|
        left = first.find { |key, _| camel(key) == field }&.last
        right = last.find { |key, _| camel(key) == field }&.last
        compared = if left.nil? || right.nil?
                     left.nil? == right.nil? ? 0 : (left.nil? ? 1 : -1)
                   else
                     (left.to_s <=> right.to_s) * (direction == 'asc' ? 1 : -1)
                   end
        break unless compared.zero?
      end
      compared.zero? ? a <=> b : compared
    end.map(&:first)
    rows.map do |row|
      fields = row.keys.select { !@selected || @selected.include?("#{ path }.#{ camel(_1) }") }
      ordered = (@order + fields.map { "#{ path }.#{ camel(_1) }" }).uniq
          .filter_map { |item| fields.find { "#{ path }.#{ camel(_1) }" == item } }
      ordered.to_h do |key|
        item = row[key]
        item = project_nested(item, "#{ path }.#{ camel(key) }") if item.is_a?(Array)
        [key, item]
      end
    end
  end

  def csv_logical(key, value)
    return value ? @labels.fetch('latest', '') : '' if key == 'is_latest'
    if value.is_a?(Time) || value.is_a?(ActiveSupport::TimeWithZone)
      return @query.population.timezone.utc_to_local(value.utc).iso8601
    end
    if key == 'language_code' || key == 'source_language_code'
      return @language_names[value]
    end
    if value.is_a?(Hash)
      return value.to_h { |field, item| [@labels.fetch(camel(field.to_s), ''), csv_logical(field, item)] }
    end
    return value.map { csv_logical(key, _1) } if value.is_a?(Array)

    @labels.fetch("#{ key }:#{ value }", value)
  end

  def safe(value)
    return value if value.is_a?(Numeric)

    text = value.is_a?(Hash) || value.is_a?(Array) ? JSON.generate(value) : value.to_s
    text.match?(/\A\s*[=+@-]/) ? "'#{ text }" : text
  end
end