EntityBatchPreAuth.avsc

{
  "namespace": "com.goguardian.events",
  "type": "record",
  "name": "EntityBatchPreAuth",
  "doc": "Represents a batch of EntityPreAuths submitted by a single extension. There is a one-to-one mapping between this avro record and a kinesis record (i.e. the kinesis record payload contains JSON which we deserialize using this schema). Note that the kinesis record contains JSON and not serialized avro binary.",
  "fields": [
    {
      "name": "type",
      "type": {
        "type": "enum",
        "name": "KinesisRecordType",
        "symbols": [
          "ENTITIES"
        ],
        "doc": "What is the type of this kinesis record"
      },
      "default": "ENTITIES"
    },
    {
      "name": "compRand",
      "type": "string",
      "doc": "The compRand from the extension"
    },
    {
      "name": "submissionTime",
      "type": "long",
      "doc": "Epoch millis that this was submitted at"
    },
    {
      "name": "extensionVersion",
      "type": "string",
      "doc": "The extension version (e.g. '2.0.2.9')"
    },
    {
      "name": "userAgent",
      "type": "string",
      "doc": "The browser user agent"
    },
    {
      "name": "ip",
      "type": "long",
      "doc": "The ipv4 represented as a long"
    },
    {
      "name": "deviceId",
      "type": [
        "null",
        "string"
      ],
      "default": null,
      "doc": "The unique device identifier that this entity was written on"
    },
    {
      "name": "payload",
      "type": {
        "type": "array",
        "items": {
          "type": "record",
          "name": "EntityPreAuth",
          "doc": "The pre-authenticated entity representation.",
          "fields": [
            {
              "name": "seqId",
              "type": "long",
              "doc": "The entity sequence id according to the extension. Note this used to be an int before 8/14/2015"
            },
            {
              "name": "url",
              "type": "string",
              "doc": "The URL that was visited"
            },
            {
              "name": "flaggedActivity",
              "type": {
                "type": "array",
                "items": {
                  "type": "record",
                  "name": "FlaggedActivityMetadata",
                  "doc": "Metadata related to flagged activity",
                  "fields": [
                    {
                      "name": "isNew",
                      "type": "boolean",
                      "doc": "Whether this metadata item is new"
                    },
                    {
                      "name": "severity",
                      "type": "double",
                      "doc": "The extension-generated severity of this metadata"
                    },
                    {
                      "name": "image",
                      "type": [
                        "null",
                        "string"
                      ],
                      "default": null,
                      "doc": "The UUID of the image associated with this flagged activity event this metadata"
                    },
                    {
                      "name": "categoryId",
                      "type": "int",
                      "doc": "Category of this flagged activity, based on its flagged terms. E.g. pornography, drugs, etc."
                    },
                    {
                      "name": "flags",
                      "type": {
                        "type": "array",
                        "items": {
                          "name": "FlaggedActivityMetadataFlag",
                          "type": "record",
                          "doc": "Represents a single flagged word",
                          "fields": [
                            {
                              "name": "term",
                              "type": "string",
                              "doc": "The flagged word"
                            },
                            {
                              "name": "count",
                              "type": "int",
                              "doc": "The number of times the word appears on the page "
                            },
                            {
                              "name": "severity",
                              "type": "double",
                              "doc": "The default severity associated with the word"
                            },
                            {
                              "name": "calculatedSeverity",
                              "type": "double",
                              "doc": "The computed severity of the word(s)"
                            }
                          ]
                        }
                      }
                    }
                  ]
                }
              }
            },
            {
              "name": "keyFlags",
              "type": {
                "type": "array",
                "items": {
                  "type": "record",
                  "name": "KeyFlagMetadata",
                  "doc": "Metadata related to key flags activity",
                  "fields": [
                    {
                      "name": "isNew",
                      "type": "boolean",
                      "doc": "Whether this metadata item is new"
                    },
                    {
                      "name": "needle",
                      "type": {
                        "type": "array",
                        "items": {
                          "type": "string",
                          "name": "needleItem",
                          "doc": "A keyflag needle found in this haystack"
                        }
                      },
                      "doc": "An array of needles flagged in this haystack"
                    },
                    {
                      "name": "haystack",
                      "type": "string",
                      "doc": "The context in which the needle was found"
                    },
                    {
                      "name": "uuid",
                      "type": "string",
                      "doc": "A UUID corresponding to the input on the page. When the input changes, this UUID is regenerated"
                    }
                  ]
                }
              }
            },
            {
              "name": "searches",
              "type": {
                "type": "array",
                "items": {
                  "type": "record",
                  "name": "SearchMetadata",
                  "doc": "Metadata related to searches",
                  "fields": [
                    {
                      "name": "isNew",
                      "type": "boolean",
                      "doc": "Whether this metadata item is new"
                    },
                    {
                      "name": "searchTerms",
                      "type": "string",
                      "doc": "The terms that were searched"
                    }
                  ]
                }
              }
            },
            {
              "name": "videos",
              "type": {
                "type": "array",
                "items": {
                  "type": "record",
                  "name": "VideoMetadata",
                  "doc": "Metadata related to videos",
                  "fields": [
                    {
                      "name": "isNew",
                      "type": "boolean",
                      "doc": "Whether this metadata item is new"
                    },
                    {
                      "name": "title",
                      "type": "string",
                      "doc": "The title of the video"
                    },
                    {
                      "name": "uploader",
                      "type": "string",
                      "doc": "The video uploader"
                    },
                    {
                      "name": "categoryId",
                      "type": "int",
                      "doc": "The youtube category id"
                    },
                    {
                      "name": "rating",
                      "type": "int",
                      "doc": "The youtube rating"
                    },
                    {
                      "name": "description",
                      "type": "string",
                      "doc": "The description of the video"
                    },
                    {
                      "name": "channelId",
                      "type": "string",
                      "doc": "The id of the youtube channel"
                    },
                    {
                      "name": "category",
                      "type": "string",
                      "doc": "The verbose name of the category"
                    },
                    {
                      "name": "uri",
                      "type": "string",
                      "doc": "The full youtube video URI"
                    },
                    {
                      "name": "videoId",
                      "type": "string",
                      "doc": "The youtube video id"
                    },
                    {
                      "name": "type",
                      "type": {
                        "type": "enum",
                        "name": "VideoType",
                        "symbols": [
                          "YOUTUBE"
                        ],
                        "doc": "Where is this youtube video sourced from"
                      }
                    }
                  ]
                }
              }
            },
            {
              "name": "docs",
              "type": {
                "type": "array",
                "items": {
                  "type": "record",
                  "name": "DocMetadata",
                  "doc": "Metadata related to docs",
                  "fields": [
                    {
                      "name": "isNew",
                      "type": "boolean",
                      "doc": "Whether this metadata item is new"
                    },
                    {
                      "name": "title",
                      "type": "string",
                      "doc": "The title of the document"
                    }
                  ]
                }
              }
            },
            {
              "name": "faviconUrl",
              "type": [
                "null",
                "string"
              ],
              "default": null,
              "doc": "The favicon url associated with this entity"
            },
            {
              "name": "title",
              "type": [
                "null",
                "string"
              ],
              "default": null,
              "doc": "The title associated with this entity"
            },
            {
              "name": "tabId",
              "type": [
                "null",
                "int"
              ],
              "default": null,
              "doc": "The tabId that this entity came from"
            },
            {
              "name": "blockedSite",
              "type": {
                "type": "array",
                "items": {
                  "type": "record",
                  "name": "BlockedMetadata",
                  "doc": "Metadata related to blocked_sites",
                  "fields": [
                    {
                      "name": "isNew",
                      "type": [
                        "null",
                        "boolean"
                      ],
                      "default": null,
                      "doc": "Whether this metadata item is new"
                    },
                    {
                      "name": "categoryId",
                      "type": "int",
                      "doc": "The category id that blocked this entity"
                    }
                  ]
                }
              },
              "default": []
            },
            {
              "name": "viewings",
              "type": {
                "type": "array",
                "items": {
                  "type": "record",
                  "name": "ViewingMetadata",
                  "doc": "Metadata related to viewings",
                  "fields": [
                    {
                      "name": "startTimeMs",
                      "type": "long",
                      "doc": "The epoch timestamp when this viewing started"
                    },
                    {
                      "name": "endTimeMs",
                      "type": "long",
                      "doc": "The epoch timestamp when this viewing ended"
                    },
                    {
                      "name": "isDirty",
                      "type": "boolean",
                      "doc": "Whether this viewing needs to be updated/inserted in the database"
                    },
                    {
                      "name": "lastSubmissionTimeMs",
                      "type": [
                        "null",
                        "long"
                      ],
                      "default": null,
                      "doc": "If present, represents the last time this viewing was sent to Kinesis"
                    }
                  ]
                }
              },
              "default": []
            },
            {
              "name": "isNew",
              "type": [
                "null",
                "boolean"
              ],
              "default": null,
              "doc": "Represents whether this entity has been submitted to Kinesis before"
            },
            {
              "name": "classroomIds",
              "type": {
                "type": "array",
                "items": {
                  "type": "int",
                  "name": "classroomId",
                  "doc": "A list of GG4T classroom IDs that this entity was generated in"
                }
              },
              "default": []
            },
            {
              "name": "classroomSessionIds",
              "type": {
                "type": "array",
                "items": {
                  "type": "int",
                  "name": "classroomSessionId",
                  "doc": "A list of GG4T classroom session IDs that this entity was generated in"
                }
              },
              "default": []
            },
            {
              "name": "isLocked",
              "type": [
                "null",
                "boolean"
              ],
              "default": null,
              "doc": "Represents whether the entity was generated in a locked state (GG4T)"
            },
            {
              "name": "tabs",
              "type": {
                "type": "array",
                "items": {
                  "type": "record",
                  "name": "TabMetadata",
                  "doc": "Metadata related to open tabs",
                  "fields": [
                    {
                      "name": "url",
                      "type": "string",
                      "doc": "The URL of the open tab"
                    },
                    {
                      "name": "title",
                      "type": "string",
                      "doc": "The title of the open tab"
                    },
                    {
                      "name": "faviconUrl",
                      "type": "string",
                      "doc": "The favicon url associated with this open tab"
                    },
                    {
                      "name": "isActive",
                      "type": "boolean",
                      "doc": "Represents whether this tab is the active one"
                    },
                    {
                      "name": "id",
                      "type": "int",
                      "doc": "A numeric identifier for this tab"
                    }
                  ]
                }
              },
              "default": [],
              "doc": "An array representing the currently open tabs on the device (GG4T)"
            },
            {
              "name": "isProxy",
              "type": [
                "null",
                "boolean"
              ],
              "default": null,
              "doc": "Whether this entity represents a proxy website"
            }
          ]
        }
      }
    }
  ]
}