# encoding: UTF-8
module Scrivito
  # Provides an enumerator for iterating over obj search results and retrieving obj instances.
  # This is done using the {http://ruby-doc.org/core-1.8.7/Enumerable.html <code>Enumerable</code> mixin},
  # which provides methods such as <code>map</code>, <code>select</code> or <code>take</code>.
  #
  # This enumerator is lazy. If for example you are looking for {BasicObj Obj}s with the ObjClass "Publication",
  # and there are 93 objs in total, than <code>enum.take(10)</code> will fetch the first 10 objs only,
  # ignoring the other 83 objs.
  # This also means, that iterating multiple times over this enumerator causes the search results and objs to be fetched again.
  # If you want to get all objs at once, use <code>enum.to_a</code>.
  #
  # To start searching use one of the {BasicObj Obj} methods that returns an {ObjSearchEnumerator}. The preferred way is to start with {BasicObj.where Obj.where}.
  #
  # == Currently available fields and their values
  #
  # [+:*+]              Searches all fields.
  #                     This is only possible with the operators +contais+ or +starts_with+.
  # [+:id+]             Id of an {BasicObj Obj}. This is a +string+ field.
  # [+:_path+]          Path of an {BasicObj Obj}. This is a +string+ field.
  # [+:_name+]          Name of an {BasicObj Obj}. This is a +string+ field.
  # [+:title+]          Title of an {BasicObj Obj}. This is a +string+ field.
  # [+:body+]           Body of an {BasicObj Obj}. This is an +html+ field. Thus, only the +contains+ and
  #                     +contains_prefix+ operators can be applied to this field.
  # [+:_obj_class+]     ObjClass of an {BasicObj Obj}. This is a +string+ field.
  # [+:_permalink+]     Permalink of an {BasicObj Obj}. This is a +string+ field.
  # [+:_last_changed+]  Date of last change of an {BasicObj Obj}.
  # [every <em><code>:custom_attribute</code></em>] Custom attribute of an {BasicObj Obj}. Note that
  #                                                 depending on the attribute type (e.g. an
  #                                                 +html+ field), some operators can not be applied.
  #
  # All values are stored as strings.
  #
  # Date values are stored in the format YYYYMMDDHHMMSS in UTC. For example, 2000-01-01 00:00:00 UTC is stored as "<code>20000101000000</code>".
  # This is relevant for string comparisons using the +is_less_than+ and +is_greater_than+ operators.
  #
  # == Currently available operators
  #
  # For +:contains+ and +:contains_prefix+, the examples are based on the following field value:
  # "Behind every cloud is another cloud."
  #
  # [+:contains+]         Searches for one or more whole words. Each word needs to be present.
  #
  #                       Example subquery values:
  #
  #                       ✔ "behind cloud" (case insensitive)
  #
  #                       ✘ "behi clo" (not whole words)
  #
  #                       ✘ "behind everything" (second word does not match)
  # [+:contains_prefix+]  Searches for one prefix. A whole word is also a prefix.
  #
  #                       Example subquery values:
  #
  #                       ✔ "Clou" (case insensitive)
  #
  #                       ✔ "Every" (case insensitive)
  #
  # For +:equals+ and +:starts_with+, the examples are based on the following field value:
  # "Some content."
  #
  # [+:equals+]           The +field+ value needs to be identical to the +value+ of this subquery.
  #
  #                       Only applicable to +string+, +enum+, +multienum+ and +date+ fields.
  #
  #                       Example subquery values:
  #
  #                       ✔ "Some content." (exact value)
  #
  #                       ✘ "Some" (not exact value)
  #
  # [+:starts_with+]      The +field+ value needs to start exactly with the +value+ of this subquery.
  #
  #                       Only applicable to +string+, +enum+, +multienum+ and +date+ fields.
  #
  #                       Example subquery values:
  #
  #                       ✔ "Som" (prefix of the value)
  #
  #                       ✘ "som" (incorrect case of prefix)
  #
  #                       ✘ "content" (not prefix of the whole value)
  #
  # For +:is_less_than+ and +:is_greater_than+, the examples are based on the following field value (date string):
  # "20000101000000"
  #
  # [+:is_less_than+]     Matches if the field string value is less than the subquery string value.
  #
  #                       Only applicable to +string+, +enum+, +multienum+ and +date+ fields.
  #
  #                       Example subquery values:
  #
  #                       ✔ "19991231235959" (is less than "20000101000000")
  #
  #                       ✘ "20000101000000" (equal, not less than)
  #
  # [+:is_greater_than+]  Matches if the field string value is greater than the subquery string value.
  #
  #                       Only applicable to +string+, +enum+, +multienum+ and +date+ fields.
  #
  #                       Example subquery values:
  #
  #                       ✔ "20000101000001" (is greater than "20000101000000")
  #
  #                       ✘ "20000101000000" (equal, not greater than)
  #
  # @api public
  class ObjSearchEnumerator
    include Enumerable

    attr_reader :workspace

    attr_reader :query
    def initialize(workspace)
      @workspace = workspace
      @options = {}
    end

    # @group Chainable methods

    # Adds this additional AND subquery to this {ObjSearchEnumerator}.
    #
    # Compares the +field(s)+ with the +value(s)+ using the +operator+ of this subquery.
    # All objs to which this criterion applies remain in the result set.
    #
    # @param [Symbol, String, Array<Symbol, String>] field Name(s) of the field(s) to be searched.
    #   For arrays, the subquery matches, if one or more of these fields meet this criterion.
    # @param [Symbol, String] operator See "Currently available operators" at the top.
    # @param [String, Array<String>] value The value(s) with which the field value(s) are compared using the +operator+ of this subquery.
    #   For arrays, the subquery matches, if the condition is met for one or more of the array elements.
    # @param [Hash] boost A hash where the keys are field names and their values are boosting factors.
    #   Boosting factors must be in the range from 1 to 10.
    #   Boosting can only be applied to subqueries in which the +contains+ or +contains_prefix+ operator is used.
    # @return [ObjSearchEnumerator]
    # @api public
    def and(field, operator, value, boost = nil)
      real_operator = operator_mapping(operator)
      subquery = {:field => field, :operator => real_operator, :value => convert_value(value)}
      if boost.present?
        valid_boost_operators = [:contains, :contains_prefix]
        if valid_boost_operators.include?(operator.to_sym)
          subquery[:boost] = boost
        else
          raise "Boost is not allowed with operator '#{operator}'. " +
              "Valid operators are: #{valid_boost_operators.join(', ')}"
        end
      end
      @size = nil
      @query = (query || []) + [subquery]

      self
    end

    # Adds this additional negated AND subquery to this {ObjSearchEnumerator}.
    #
    # Compares the +field(s)+ with the +value(s)+ using the negated +operator+ of this subquery.
    # All objs to which this criterion applies are removed from the result set.
    #
    # @param [Symbol, String, Array<Symbol, String>] field Name(s) of the field(s) to be searched.
    #   For arrays, the subquery matches, if one or more of these fields meet this criterion.
    # @param [Symbol, String] operator Only applicable to subqueries in which the +equals+,
    #   +starts_with+, +is_greater_than+ and +is_less_than+ operator is used
    #   (See "Currently available operators" at the top).
    # @param [String, Array<String>] value The value(s) with which the field value(s) are compared using the +operator+ of this subquery.
    #   For arrays, the subquery matches, if the condition is met for one or more of the array elements.
    # @return [ObjSearchEnumerator]
    # @api public
    def and_not(field, operator, value)
      real_operator = operator_mapping(operator)
      valid_negated_operators = [:equals, :starts_with, :is_greater_than, :is_less_than]
      unless valid_negated_operators.include?(operator.to_sym)
        raise "Negating operator '#{operator}' is not valid."
      end
      subquery = {:field => field, :operator => real_operator, :value => convert_value(value),
          :negate => true}
      @size = nil
      @query = (query || []) + [subquery]

      self
    end

    # Orders the results by +field_name+.
    # @param [Symbol, String] field_name This parameter determines by which field the hits are sorted (e.g. +:_path+).
    # @return [ObjSearchEnumerator]
    # @api public
    def order(field_name)
      options[:sort_by] = field_name

      self
    end

    # Reverses the order of the results. Requires {#order} to be specified before.
    # @return [ObjSearchEnumerator]
    # @api public
    def reverse_order
      options[:sort_by].present? or raise "A search order has to be specified"\
          " before reverse_order can be applied."
      @reverse_order = !@reverse_order

      self
    end

    # Number of search results to be returned by each of the internal search requests.
    # The default is +10+. The server may limit a large batch size to a reasonable value.
    # @param [Integer] size A value in the range from +1+ to +100+.
    # @return [ObjSearchEnumerator]
    # @api public
    def batch_size(size)
      options[:size] = size

      self
    end


    # Omits the first +amount+ number of {BasicObj Obj}s from the results. The default is +0+.
    # @param [Integer] amount
    # @return [ObjSearchEnumerator]
    # @api public
    def offset(amount)
      options[:offset] ||= 0
      options[:offset] += amount

      self
    end

    def include_deleted
      @include_deleted = true

      self
    end

    # @!endgroup

    # Iterates over the search result, yielding {BasicObj Obj}.
    # @yield [Obj]
    # @return [void]
    # @api public
    def each
      offset = options[:offset] || 0
      current_batch, total = fetch_next_batch(offset)
      loop do
        if current_batch.size == 0
          if offset < total
            current_batch, total = fetch_next_batch(offset)
          else
            raise StopIteration
          end
        end

        offset += 1
        hit = current_batch.shift
        yield hit
      end
    end

    # The total number of hits.
    # @return [Integer]
    # @api public
    def size
      return @size if @size

      size_query = {
        query: query,
        size: 0
      }
      if @include_deleted
        size_query[:options] = {
          include_deleted: true
        }
      end

      @size ||= CmsBackend.instance.search_objs(workspace, size_query)['total'].to_i
    end

    # load a single batch of search results from the backend.
    # returns an array of Objs.
    # will usually return `batch_size` results if available,
    # but may occasionally return fewer than `batch_size` results (due to rate limit, for example).
    # @api public
    def load_batch
      next_batch = fetch_next_batch(options[:offset] || 0)

      next_batch.first
    end

    # @api public
    alias_method :length, :size

    # @api public
    alias_method :count, :size

    private

    attr_reader :options

    def convert_value(value)
      if value.kind_of?(Array)
        value.map{ |v| convert_single_value(v) }
      else
        convert_single_value(value)
      end
    end

    def convert_single_value(value)
      if value.is_a?(Time) || value.is_a?(Date)
        CmsRestApi::AttributeSerializer.convert_time(value)
      else
        value
      end
    end

    def operator_mapping(operator)
      case operator.to_sym
      when :contains
        :search
      when :contains_prefix
        :prefix_search
      when :equals
        :equal
      when :starts_with
        :prefix
      when :is_greater_than
        :greater_than
      when :is_less_than
        :less_than
      else
        raise "Operator '#{operator}'' is not valid!"
      end
    end

    def fetch_next_batch(offset)
      request_result = CmsBackend.instance.search_objs(workspace, search_dsl(offset))

      obj_ids = request_result['results'].map { |result| result['id'] || result['_id'] }
      objs = workspace.objs.find_including_deleted(obj_ids)

      @size = request_result['total'].to_i

      [objs, @size]
    end

    def search_dsl(offset)
      patches = {
        offset: offset,
        query: query,
      }
      if @reverse_order
        patches[:sort_order] = options[:sort_by].present? ? :desc : :asc
      end

      if @include_deleted
        patches[:options] = {
          include_deleted: true
        }
      end

      options.merge(patches)
    end

  end
end