mappings.yaml 

schemas:
  #uniform
  http_schema:
  - version: 2.1
    columns:
      - name: status_code
        required: false
      - name: tag_url
        required: false
        data_type:
          type: custom
          schema: url_struct
          version: 2.1
      - name: cookie
        data_type: map
      - name: ip_info
        required: false
        data_type:
          type: custom
          schema: ip_struct
          version: 2.0
      - name: user_agent_info
        data_type: map
      - name: headers
        data_type: map
  ip_struct:
  - version: 2.0
    columns:
      - name: ip_type
        required: false
      - name: ip_domain
        required: false
      - name: ip
        required: false
  url_struct:
  - version: 2.1
    columns:
      - name: scheme
        required: false
      - name: host
        required: false
      - name: path
        required: false
      - name: query
        data_type: map
  entity_struct:
  - version: 2.1
    columns:
      - name: entity_type
        required: false
      - name: entity_domain
        required: false
      - name: entity_id
        required: false
  #clicks
  gcm_clicks:
    columns:
      - name: "Event Time"
        data_type: string
      - name: "User ID"
        required: false
      - name: "Advertiser ID"
        required: false
      - name: "Campaign ID"
        required: false
      - name: "Ad ID"
        required: false
      - name: "Rendering ID"
        required: false
      - name: "Creative Version"
        required: false
      - name: "Site ID (CM360)"
        required: false
      - name: "Placement ID"
        required: false
      - name: "Country Code"
        required: false
      - name: "State/Region"
        required: false
      - name: "Browser/Platform ID"
        required: false
      - name: "Browser/Platform Version"
        required: false
      - name: "Operating System ID"
        required: false
      - name: "Designated Market Area (DMA) ID"
        required: false
      - name: "City ID"
        required: false
      - name: "ZIP/Postal Code"
        required: false
      - name: "U Value"
        required: false
      - name: "Segment Value 1"
        required: false
      - name: "Impression ID"
        required: false
      - name: "Landing Page URL ID"
        required: false
      - name: "Partner1 ID"
        required: false
  aqfer_click_subevent:
  - version: 2.0
    columns:
      - entity_id
      - name: event_timestamp
        data_type: long
      - event_type
      - event_id
      - name: source
        required: false
      - name: country_code
        required: false
      - name: region_code
        required: false
      - name: parent_event_id
        required: false
      - name: agency_id
        required: false
      - name: advertiser_id
        required: false
      - name: page_url
        required: false
        data_type:
          type: custom
          schema: url_struct
          version: 2.1
      - name: ad_url
        required: false
        data_type:
          type: custom
          schema: url_struct
          version: 2.1
      - name: page_referer_url
        required: false
        data_type:
          type: custom
          schema: url_struct
          version: 2.1
      - name: destination_url
        required: false
        data_type:
          type: custom
          schema: url_struct
          version: 2.1
      - name: campaign_id
        required: false
      - name: ad_group_id
        required: false
      - name: creative_id
        required: false
      - name: placement_id
        required: false
      - name: other_marketing_program_levels
        data_type: map
      - name: media_partner
        required: false
      - name: inventory_partner
        required: false
      - name: supply_vendor_publisher_id
        required: false
      - name: site_id
        required: false
      - name: other_media_group_levels
        data_type: map
      - name: has_click_conversion
        data_type: boolean
        required: false
      - name: cpm_currency
        required: false
      - name: cpm_paid
        required: false
        data_type: int
      - name: cpm_cost
        required: false
        data_type: int
      - name: cpc_paid
        required: false
        data_type: int
      - name: cpc_cost
        required: false
        data_type: int
      - name: keywords
        required: false
      - name: search_terms
        required: false
      - name: search_phrase
        required: false
      - name: other_entity_ids
        data_type:
          type: list
          element:
            type: custom
            schema: entity_struct
            version: 2.1
      - name: other_event_ids
        data_type: map
      - name: http_info
        data_type:
          type: custom
          schema: http_schema
          version: 2.1
      - name: geo
        data_type: map
      - name: metrics
        data_type: map
      - name: others
        data_type: map
    partitions:
      - entity_type
      - entity_domain
      #- event_date
      #- event_hour
      - batch_id
  #activity = conversion
  gcm_activity:
    columns:
      - name: "Event Time"
        data_type: string
      - name: "User ID"
        required: false
      - name: "Advertiser ID"
        required: false
      - name: "Campaign ID"
        required: false
      - name: "Ad ID"
        required: false
      - name: "Rendering ID"
        required: false
      - name: "Creative Version"
        required: false
      - name: "Site ID (CM360)"
        required: false
      - name: "Placement ID"
        required: false
      - name: "Country Code"
        required: false
      - name: "State/Region"
        required: false
      - name: "Browser/Platform ID"
        required: false
      - name: "Browser/Platform Version"
        required: false
      - name: "Operating System ID"
        required: false
      - name: "U Value"
        required: false
      - name: "Activity ID"
        required: false
      - name: "TRAN Value"
        required: false
      - name: "Other Data"
        required: false
      - name: "ORD Value"
        required: false
      - name: "Interaction Time"
        required: false
      - name: "Conversion ID"
        required: false
      - name: "Segment Value 1"
        required: false
      - name: "Floodlight Configuration"
        required: false
      - name: "Partner1 ID"
        required: false
      - name: "Total Conversions"
        required: false
      - name: "Total Revenue"
        required: false
  aqfer_conversion: 
  - version: 2.0
    columns:
      - name: entity_id
      - name: event_timestamp
        data_type: long
      - name: event_type
      - name: event_id
      - name: source
      - name: country_code
      - name: region_code
        required: false
      - name: parent_event_id
      - name: agency_id
      - name: advertiser_id
      - name: page_url
        required: false
        data_type:
          type: custom
          schema: url_struct
          version: 2.1
      - name: ad_url
        required: false
        data_type:
          type: custom
          schema: url_struct
          version: 2.1
      - name: page_referer_url
        required: false
        data_type:
          type: custom
          schema: url_struct
          version: 2.1
      - name: conversion_actions
        data_type: list
      - name: conversion_type
        required: false
      - name: conversion_sub_type
        required: false
      - name: keywords
        required: false
      - name: search_terms
        required: false
      - name: search_phrase
        required: false
      - name: other_entity_ids
        data_type:
          type: list
          element:
            type: custom
            schema: entity_struct
            version: 2.1
      - name: other_event_ids
        data_type: map
      - name: http_info
        data_type:
          type: custom
          schema: http_schema
          version: 2.1
      - name: geo
        data_type: map
      - name: metrics
        data_type: map
      - name: others
        data_type: map
    partitions:
      - entity_type
      - entity_domain
      #- event_date
      #- event_hour
      - batch_id
      
  gcm_impressions:
    columns:
      - name: "Event Time"
        data_type: string
      - name: "User ID"
        required: false
        data_type: string
      - name: "Advertiser ID"
        required: false
        data_type: string
      - name: "Campaign ID"
        required: false
        data_type: string
      - name: "Ad ID"
        required: false
        data_type: string
      - name: "Rendering ID"
        required: false
        data_type: string
      - name: "Creative Version"
        required: false
        data_type: string
      - name: "Site ID (CM360)"
        required: false
        data_type: string
      - name: "Placement ID"
        required: false
        data_type: string
      - name: "Country Code"
        required: false
        data_type: string
      - name: "State/Region"
        required: false
        data_type: string
      - name: "Browser/Platform ID"
        required: false
        data_type: string
      - name: "Browser/Platform Version"
        required: false
        data_type: string
      - name: "Operating System ID"
        required: false
        data_type: string
      - name: "Designated Market Area (DMA) ID"
        required: false
        data_type: string
      - name: "City ID"
        required: false
        data_type: string
      - name: "ZIP/Postal Code"
        required: false
        data_type: string
      - name: "U Value"
        required: false
        data_type: string
      - name: "Impression ID"
        required: false
        data_type: string
      - name: "Referrer URL"
        required: false
        data_type: string
      - name: "Partner1 ID"
        required: false
        data_type: string
      - name: "Active View: Eligible Impressions"
        required: false
        data_type: string
      - name: "Active View: Measurable Impressions"
        required: false
        data_type: string
      - name: "Active View: Viewable Impressions"
        required: false
        data_type: string
  aqfer_impression:
  - version: 2.0
    columns:
      - entity_id
      - name: event_timestamp
        data_type: long
      - event_type
      - event_id
      - name: source
        required: false
      - name: country_code
        required: false
      - name: region_code
        required: false
      - name: parent_event_id
        required: false
      - name: agency_id
        required: false
      - name: advertiser_id
        required: false
      - name: page_url
        required: false
        data_type:
          type: custom
          schema: url_struct
          version: 2.1
      - name: ad_url
        required: false
        data_type:
          type: custom
          schema: url_struct
          version: 2.1
      - name: page_referer_url
        required: false
        data_type:
          type: custom
          schema: url_struct
          version: 2.1
      - name: campaign_id
        required: false
      - name: ad_group_id
        required: false
      - name: creative_id
        required: false
      - name: placement_id
        required: false
      - name: other_marketing_program_levels
        data_type: map
      - name: media_partner
        required: false
      - name: inventory_partner
        required: false
      - name: supply_vendor_publisher_id
        required: false
      - name: site_id
        required: false
      - name: other_media_group_levels
        data_type: map
      - name: has_click_conversion
        data_type: boolean
        required: false
      - name: has_click
        data_type: boolean
        required: false
      - name: viewable
        data_type: boolean
        required: false
      - name: cpm_currency
        required: false
      - name: cpm_paid
        required: false
        data_type: int
      - name: cpm_cost
        required: false
        data_type: int
      - name: keywords
        required: false
      - name: search_terms
        required: false
      - name: search_phrase
        required: false
      - name: other_entity_ids
        data_type:
          type: list
          element:
            type: custom
            schema: entity_struct
            version: 2.1
      - name: other_event_ids
        data_type: map
      - name: http_info
        data_type:
          type: custom
          schema: http_schema
          version: 2.1
      - name: geo
        data_type: map
      - name: metrics
        data_type: map
      - name: others
        data_type: map
    partitions:
      - entity_type
      - entity_domain
      #- event_date
      #- event_hour
      - batch_id
      
  gcm_placements:
    columns:
      - name: "Campaign ID"
        required: false
      - name: "Site ID (CM360)"
        required: false
      - name: "Placement ID"
        required: false
      - name: "Site Keyname"
        required: false
      - name: "Placement"
        required: false
      - name: "Content Category"
        required: false
      - name: "Placement Strategy"
        required: false
      - name: "Placement Start Date"
        required: false
      - name: "Placement End Date"
        required: false
      - name: "Placement Group Type"
        required: false
      - name: "Package/Roadblock ID"
        required: false
      - name: "Placement Cost Structure"
        required: false
      - name: "Placement Cap Cost Option"
        required: false
      - name: "ActivityID"
        required: false
      - name: "Flighting Activated"
        required: false
schema_mappings:
  gcm_click_to_aqfer_click_subevent:
    input_schema: gcm_clicks
    output_schema: 
      name: aqfer_click_subevent
      version: 2.0
    if_missing: ""
    column_mappings:
      entity_id: "User ID"
      event_timestamp: "Event Time"
      event_type:
        type: const
        value: "clk"
      event_id: "Impression ID"
      parent_event_id: "Impression ID"
      source:
        type: const
        value: "GCM"
      region_code: "State/Region"
      agency_id:
        type: const
        value: "NA"
      advertiser_id: "Advertiser ID"
      campaign_id: "Campaign ID"
      placement_id: "Placement ID"
      other_marketing_program_levels:
        type: map
        entries:
          ad_id: "Ad ID"
      media_partner: "Partner1 ID"
      site_id: "Site ID (CM360)"
      other_media_group_levels:
        type: map
        entries:
          rendering_id: "Rendering ID"
          impression_id: "Impression ID"
          creative_version: "Creative Version"
          host: "Landing Page URL ID"
      has_click_conversion:
        type: const
        value: true
      other_entity_ids:
        type: list
        elements: []
      other_event_ids:
        type: map
        entries: {}
      http_info:
        type: inner
        column_mappings:
          cookie:
            type: map
            entries: {}
          headers:
            type: map
            entries: {}
          user_agent_info:
            type: map
            entries:
              browser_platform_id: "Browser/Platform ID"
              browser_platform_version: "Browser/Platform Version"
              operating_system_id:  "Operating System ID"
      geo:
        type: map
        entries:
          dma: "Designated Market Area (DMA) ID"
          city_id: "City ID"
          zip_code: "ZIP/Postal Code"
      metrics:
        type: map
        entries: {}
      others:
        type: map
        entries:
          gcm__segment_value_1: "Segment Value 1"
          gcm__u_value: "U Value"
      entity_type:
        type: const
        value: "ck"
      entity_domain:
        type: const
        value: "GCM"
      country_code: "Country Code"
      event_date:
        type: expr
        expr: '${Event Time}'
        transformations:
          - type: type_conversion
            output_data_type: long
          - type: type_conversion
            output_data_type:
              type: date_part
              format: 'yyyyMMdd'
      event_hour:
        type: expr
        expr: '${Event Time}'
        transformations:
          - type: type_conversion
            output_data_type: long
          - type: type_conversion
            output_data_type:
              type: date_part
              format: 'HH'
      batch_id: $runid

  gcm_activity_to_aqfer_conversion:
    input_schema: gcm_activity
    output_schema: 
      name: aqfer_conversion
      version: 2.0
    if_missing: ""
    column_mappings:
      event_id: "Activity ID"
      event_timestamp: "Event Time"
      entity_id: "User ID"
      event_type:
        type: const
        value: 'cnv'
      conversion_type:
        type: const
        value: 'activity'
      parent_event_id: "Activity ID"
      advertiser_id:
        source_column: "Advertiser ID"
        if_missing: 'NA'
      agency_id:
        type: const
        value: NA
      source:
        type: const
        value: 'GCM'
      country_code: "Country Code"
      region_code: "State/Region"
      conversion_actions:
        type: list
        elements: []
      other_entity_ids:
        type: list
        elements: []
      metrics:
        type: map
        entries:
          total_conversions: "Total Conversions"
          total_revenue: "Total Revenue"
      other_event_ids:
        type: map
        entries:
          campiagn_id: "Campaign ID"
          ad_id: "Ad ID"
          partner_id: "Partner1 ID"
          rendering_id: "Rendering ID"
          placement_id: "Placement ID"
          site_id: "Site ID (CM360)"
          creative_id: "Creative Version"
          conversion_id: "Conversion ID"
          pseudo_id: #Remove per client if not needed
            type: expr
            expr: '${User ID}${Event Time}${Activity ID}'
      http_info:
        type: inner
        column_mappings:
          cookie:
            type: map
            entries: {}
          headers:
            type: map
            entries: {}
          user_agent_info:
            type: map
            entries:
              browser_platform_id: "Browser/Platform ID"
              browser_platform_version: "Browser/Platform Version"
              operating_system_id: "Operating System ID"
      others:
        type: map
        entries:
          gcm__u_value: "U Value"
          gcm__segment_value_1: "Segment Value 1"
          gcm__other_data: "Other Data"
          gcm__ord_value: "ORD Value"
          gcm__interaction_time: "Interaction Time"
          gcm__floodlight_configuration: "Floodlight Configuration"
          gcm__tran_value: "TRAN Value"
      geo:
        type: map
        entries: {}
      entity_type:
        type: const
        value: 'ck'
      entity_domain:
        type: const
        value: 'GCM'
      event_date:
        type: expr
        expr: '${Event Time}'
        transformations:
          - type: type_conversion
            output_data_type: long
          - type: type_conversion
            output_data_type:
              type: date_part
              format: 'yyyyMMdd'
      event_hour:
        type: expr
        expr: '${Event Time}'
        transformations:
          - type: type_conversion
            output_data_type: long
          - type: type_conversion
            output_data_type:
              type: date_part
              format: 'HH'
      batch_id: $runid
  gcm_impression_to_aqfer_impression:
    input_schema: gcm_impressions
    output_schema: 
      name: aqfer_impression
      version: 2.0
    if_missing: ""
    column_mappings:
      entity_id: "User ID"
      event_timestamp: "Event Time"
      event_type:
        type: const
        value: "impr"
      event_id: "Impression ID" 
      parent_event_id: "Impression ID"
      source:
        type: const
        value: "GCM"
      country_code: "Country Code"
      region_code: "State/Region"
      agency_id:
        type: const
        value: "NA"
      advertiser_id: "Advertiser ID"
      page_referer_url:
        type: inner
        column_mappings:
          host: "Referrer URL"
          query:
            type: map
            entries: {}
      campaign_id: "Campaign ID"
      placement_id: "Placement ID"
      other_marketing_program_levels:
        type: map
        entries:
          ad_id: "Ad ID"
      media_partner: "Partner1 ID"
      site_id: "Site ID (CM360)"
      other_media_group_levels:
        type: map
        entries:
          rendering_id: "Rendering ID"
          creative_version: "Creative Version"
      other_entity_ids:
        type: list
        elements: []
      other_event_ids:
        type: map
        entries:
          impression_id: "Impression ID"
      http_info:
        type: inner
        column_mappings:
          cookie:
            type: map
            entries: {}
          headers:
            type: map
            entries: {}
          user_agent_info:
            type: map
            entries:
              browser_platform_id: "Browser/Platform ID"
              browser_platform_version: "Browser/Platform Version"
              operating_system_id:  "Operating System ID"
      geo:
        type: map
        entries:
          dma: "Designated Market Area (DMA) ID"
          city_id: "City ID"
          zip_code: "ZIP/Postal Code"
      metrics:
        type: map
        entries:
          active_view_eligible_impressions: "Active View: Eligible Impressions"
          active_view_measurable_impressions: "Active View: Measurable Impressions"
          active_view_viewable_impressions: "Active View: Viewable Impressions"
      others:
        type: map
        entries: {}
      clicks:
        type: list
        elements: []
      engagements:
        type: list
        elements: []
      entity_type:
        type: const
        value: "ck"
      entity_domain:
        type: const
        value: "GCM"
      event_date:
        type: expr
        expr: '${Event Time}'
        transformations:
          - type: type_conversion
            output_data_type: long
          - type: type_conversion
            output_data_type:
              type: date_part
              format: 'yyyyMMdd'
      event_hour:
        type: expr
        expr: '${Event Time}'
        transformations:
          - type: type_conversion
            output_data_type: long
          - type: type_conversion
            output_data_type:
              type: date_part
              format: 'HH'
      batch_id: $runid

  • Table of contents

Was this article helpful?

Previous Article

jobconf.yaml ~~cid~~: jobs: sandbox-mapper-click: enhancers: [] partition_count: 100 event_sch...