// Copyright 2018 The Fuchsia Authors. All rights reserved.
// Use of this source code is governed by a BSD-style license that can be
// found in the LICENSE file.

#ifndef COBALT_SRC_LOGGER_ENCODER_H_
#define COBALT_SRC_LOGGER_ENCODER_H_

#include <memory>
#include <string>
#include <vector>

#include <google/protobuf/repeated_field.h>

#include "src/algorithms/rappor/rappor_encoder.h"
#include "src/logger/project_context.h"
#include "src/logger/types.h"
#include "src/pb/event.pb.h"
#include "src/pb/metadata_builder.h"
#include "src/pb/observation.pb.h"
#include "src/public/lib/status.h"
#include "src/registry/aggregation_window.pb.h"
#include "src/registry/metric_definition.pb.h"
#include "src/registry/report_definition.pb.h"
#include "src/system_data/client_secret.h"
#include "src/system_data/system_data.h"

namespace cobalt::logger {

// A HistogramPtr provides a moveable way of passing the buckets of a Histogram.
using HistogramPtr = std::unique_ptr<google::protobuf::RepeatedPtrField<HistogramBucket>>;

// A EventValuesPtr provides a moveable way of passing the dimensions of a
// custom event.
using EventValuesPtr = std::unique_ptr<google::protobuf::Map<std::string, CustomDimensionValue>>;

// A vector of event codes that the data corresponds to.
using EventCodes = std::vector<uint32_t>;

// Histogram vector of (index, count) pairs.
using Histogram = std::vector<std::tuple<uint32_t, int64_t>>;

// An Encoder is used for creating Observations. An Observation is the
// unit of encoded data that is sent from a client device to the Shuffler
// and ultimately to the Analyzer.
//
// Observations are derived from Events. Events are the raw data directly
// logged by a Cobalt user on the client system.
//
// There are two broad categories of Observations: immediate Observations and
// locally-aggregated observations. An immediate Observation is generated
// directly from a single Event at the time the Event is logged. A
// locally-aggregated Observation is computed based on the data of many
// logged Events over a period of time.
//
// An Observation is associated with a Metric and this means that the
// Observation is derived from one or more Events belonging to that Metric.
//
// An Observation is always generated for a particular Report. The Report
// definition indicates whether the Observation should be an immediate or
// locally aggregated Observation and how the Observation should be encoded.
//
// There will usually be a singleton instance of Encoder on a client device.
// The Encoder interface is not exposed directly to Cobalt users. Instead it
// is used by the Logger implementation in order to encode immediate
// Observations and it is used by the Local Aggregator to encode
// locally-aggregated Observations.
class Encoder {
 public:
  // Constructor
  //
  // client_secret: A random secret that is generated once on the client and then persisted by the
  //     client and used repeatedly. It is used as an input by some of the encodings.
  //
  // metadata_builder: Used to construct ObservationMetadata. The Encoder does not take ownership of
  //     metadata_builder.
  Encoder(system_data::ClientSecret client_secret, MetadataBuilder& metadata_builder);

  // Encoder should be move-only.
  Encoder(Encoder const&) = delete;
  Encoder& operator=(Encoder const&) = delete;

  //////////////////////////////////////////////////////////////////////////////////////////////////
  //
  // Cobalt 1.0 Encoding methods.
  //
  // An Observation is always tagged with a day_index indicating the day on
  // which the Observation was encoded. For immediate Observations this will
  // be the same as the day the corresponding Event was logged. For
  // locally-aggregated Observations this will be the day the aggregation was
  // completed.
  //
  // An Observation is always associated with an instance of ObservationMetadata
  // that contains the metric_id, report_id and day_index, among other data.
  //
  // All of the Encode*() methods take the same first three parameters:
  // (1) |metric| A MetricRef that provides the names and IDs of the customer,
  //     project and metric associated with the Observation being encoded.
  //     Note that the methods of this class do not see the MetricDefinition
  //     itself and have no knowledge of the different Metric types or their
  //     meanings. In particular no validation against the MetricDefinition or type
  //     is performed by this class. If any such validation is needed it must be
  //     performed by the caller prior to invoking the Encode*() methods of this
  //     class.
  // (2) |report| A pointer to the definition of the Report associated with the
  //     Observation being encoded. The ReportDefinition may carry fields
  //     particular to the encoding to be performed. The following
  //     ReportDefinition fields are always required to be populated: |name|,
  //     |id|, |system_profile_field|. Additionally, each Encode*() method may
  //     require other fields of ReportDefinition to be populated. This will be
  //     specified in the comments for each Encode*() method.
  // (3) |day_index| The day associated with the Observation being encoded.
  //
  //////////////////////////////////////////////////////////////////////////////////////////////////

  // The output of the Cobalt 1.0 Encode*() methods is a triple consisting of a
  // status and, if the status is kOK, a new observation and its metadata. The
  // observation will have been assigned a new quasi-unique |random_id|.
  struct Result {
    Status status;
    std::unique_ptr<Observation> observation;
    std::unique_ptr<ObservationMetadata> metadata;
  };

  // Encodes an Observation of type BasicRapporObservation.
  //
  // metric: Provides access to the names and IDs of the customer, project and
  // metric associated with the Observation being encoded.
  //
  // report: The definition of the Report associated with the Observation being
  // encoded. In addition to the common fields always required, this method also
  // requires that the |local_privacy_noise_level| field be set. This is used to
  // determine the p and q values for Basic RAPPOR.
  //
  // day_index: The day index associated with the Observation being encoded.
  //
  // value_index: The index to encode using Basic RAPPOR. It must be in
  // the range [0, num_categories - 1]
  //
  // num_categories: The number of categories to use in the Basic RAPPOR
  // encoding.
  Result EncodeBasicRapporObservation(MetricRef metric, const ReportDefinition* report,
                                      uint32_t day_index, uint32_t value_index,
                                      uint32_t num_categories) const;

  // Encodes an Observation of type IntegerEventObservation.
  //
  // metric: Provides access to the names and IDs of the customer, project and
  // metric associated with the Observation being encoded.
  //
  // report: The definition of the Report associated with the Observation being
  // encoded.
  //
  // day_index: The day index associated with the Observation being encoded.
  //
  // event_codes: This will be used to populate the bits of the Obseravtion's
  // |event_code|.
  //
  // component: The hash of this value will populate the Observation's
  // |component_name_hash| field.
  //
  // value: This will populate the Observation's |value| field.
  Result EncodeIntegerEventObservation(MetricRef metric, const ReportDefinition* report,
                                       uint32_t day_index,
                                       const google::protobuf::RepeatedField<uint32_t>& event_codes,
                                       const std::string& component, int64_t value) const;

  // Encodes an Observation of type HistogramObservation.
  //
  // metric: Provides access to the names and IDs of the customer, project and
  // metric associated with the Observation being encoded.
  //
  // report: The definition of the Report associated with the Observation being
  // encoded.
  //
  // day_index: The day index associated with the Observation being encoded.
  //
  // event_codes: This will be used to populate the bits of the Obseravtion's
  // |event_code|.
  //
  // component: The hash of this value will populate the Observation's
  // |component_name_hash| field.
  //
  // histogram: This will be used to populate the Observation's |buckets| field.
  // This method does not validate |histogram| against the Metric definition.
  // That is the caller's responsibility.
  Result EncodeHistogramObservation(MetricRef metric, const ReportDefinition* report,
                                    uint32_t day_index,
                                    const google::protobuf::RepeatedField<uint32_t>& event_codes,
                                    const std::string& component, HistogramPtr histogram) const;

  // Encodes an Observation of type CustomObservation.
  //
  // metric: Provides access to the names and IDs of the customer, project and
  // metric associated with the Observation being encoded.
  //
  // report: The definition of the Report associated with the Observation being
  // encoded.
  //
  // day_index: The day index associated with the Observation being encoded.
  //
  // event_values: This will be used to populate the Observation's |values|
  // field. This method does not validate |event_values| against the Metric's
  // proto definition. That is the caller's responsibility.
  Result EncodeCustomObservation(MetricRef metric, const ReportDefinition* report,
                                 uint32_t day_index, EventValuesPtr event_values) const;

  // Encodes an Observation of type CustomObservation.
  //
  // metric: Provides access to the names and IDs of the customer, project and
  // metric associated with the Observation being encoded.
  //
  // report: The definition of the Report associated with the Observation being
  // encoded.
  //
  // day_index: The day index associated with the Observation being encoded.
  //
  // serialized_proto: This will be used to populate the Observation's |serialized_proto| field.
  // This method does not validate |serialized_proto|. That is the caller's responsibility.
  Result EncodeSerializedCustomObservation(MetricRef metric, const ReportDefinition* report,
                                           uint32_t day_index,
                                           std::unique_ptr<std::string> serialized_proto) const;

  // Encodes an Observation of type UniqueActivesObservation.
  //
  // metric: Provides access to the names and IDs of the customer, project and
  // metric associated with the Observation being encoded.
  //
  // report: The definition of the Report associated with the Observation being
  // encoded. In addition to the common fields always required, this method also
  // requires that the |local_privacy_noise_level| field be set. This is used to
  // determine the p and q values for Basic RAPPOR.
  //
  // day_index: The day index associated with the Observation being encoded.
  // This is the last day (inclusive) of the rolling window associated with this
  // Observation.
  //
  // event_code: The event code of the Event associated with this Observation.
  // This value should be a nonnegative integer less than or equal to the
  // max_event_code of the MetricDefinition wrapped by |metric|, but it is the
  // caller's responsibility to ensure this.
  //
  // was_active: Set to true if an event with code |event_code|
  // occurred during the window of size |window_size| ending on |day_index|,
  // false otherwise. If |was_active| is true, the BasicRapporObservation field
  // of the UniqueActivesObservation is a Basic RAPPOR encoding of a 1 bit.
  // If |was_active| is false, the BasicRapporObservation field is a Basic
  // RAPPOR encoding of a 0 bit.
  //
  // aggregation_window: The aggregation window associated with the Observation.
  // This should be one of the OnDeviceAggregationWindows specified in |report|
  // (or equivalent, if |report| has a WindowSize) but it is the caller's
  // responsibility to ensure this.
  Result EncodeUniqueActivesObservation(MetricRef metric, const ReportDefinition* report,
                                        uint32_t day_index, uint32_t event_code, bool was_active,
                                        const OnDeviceAggregationWindow& aggregation_window) const;

  // Encodes an Observation of type PerDeviceNumericObservation.
  //
  // metric: Provides access to the names and IDs of the customer, project and
  // metric associated with the Observation being encoded.
  //
  // report: The definition of the Report associated with the Observation being
  // encoded.
  //
  // day_index: The day index associated with the Observation being encoded.
  // This is the last day (inclusive) of the rolling window associated with this
  // Observation.
  //
  // component: The component associated with this Observation. The hash of this
  // value will populate the Observation's |component_name_hash| field.
  //
  // event_codes: This will be used to populate the bits of the Obseravtion's
  // |event_code|.
  //
  // value: This will populate the |value| field of the the
  // IntegerEventObservation wrapped by the PerDeviceNumericObservation.
  //
  // aggregation_window: The aggregation window associated with the Observation.
  // This should be one of the OnDeviceAggregationWindows specified in |report|
  // (or equivalent, if |report| has a WindowSize) but it is the caller's
  // responsibility to ensure this.
  Result EncodePerDeviceNumericObservation(
      MetricRef metric, const ReportDefinition* report, uint32_t day_index,
      const std::string& component, const google::protobuf::RepeatedField<uint32_t>& event_codes,
      int64_t value, const OnDeviceAggregationWindow& aggregation_window) const;

  // Encodes an Observation of type PerDeviceNumericHistogramObservation.
  //
  // metric: Provides access to the names and IDs of the customer, project and
  // metric associated with the Observation being encoded.
  //
  // report: The definition of the Report associated with the Observation being
  // encoded.
  //
  // day_index: The day index associated with the Observation being encoded.
  // This is the last day (inclusive) of the rolling window associated with this
  // Observation.
  //
  // component: The component associated with this Observation. The hash of this
  // value will populate the Observation's |component_name_hash| field.
  //
  // event_codes: This will be used to populate the bits of the Obseravtion's
  // |event_code|.
  //
  // value: This is the raw value of the aggregation. This will be used to populate HistogramBuckets
  // field wrapped by the HistogramObservation in PerDeviceNumericObservation.
  //
  // aggregation_window: The aggregation window associated with the Observation.
  // This should be one of the OnDeviceAggregationWindows specified in |report|
  // (or equivalent, if |report| has a WindowSize) but it is the caller's
  // responsibility to ensure this.
  Result EncodePerDeviceHistogramObservation(
      MetricRef metric, const ReportDefinition* report, uint32_t day_index,
      const std::string& component, const google::protobuf::RepeatedField<uint32_t>& event_codes,
      int64_t value, const OnDeviceAggregationWindow& aggregation_window) const;

  // Encodes an Observation of type ReportParticipationObservation.
  //
  // metric: Provides access to the names and IDs of the customer, project, and
  // metric associated with the Observation being encoded.
  //
  // report: The definition of the Report associated with the Observation being
  // encoded.
  //
  // day_index: The day index associated with the Observation being encoded.
  Result EncodeReportParticipationObservation(MetricRef metric, const ReportDefinition* report,
                                              uint32_t day_index) const;

  //////////////////////////////////////////////////////////////////////////////////////////////////
  //
  // Cobalt 1.1 Encoding methods.
  //
  // All Cobalt 1.1 Observations are locally aggregated.
  //
  //////////////////////////////////////////////////////////////////////////////////////////////////

  // Encodes an Observation of type IntegerObservation.
  //
  // data: A vector of (event_codes, integer_value) pairs, that will be used to encode the
  // Observation.
  [[nodiscard]] static lib::statusor::StatusOr<std::unique_ptr<Observation>>
  EncodeIntegerObservation(const std::vector<std::tuple<std::vector<uint32_t>, int64_t>>& data);

  // Encodes an Observation of type SumAndCountObservation.
  //
  // data: A vector of (event_codes, sum, count) triples, that will be used to encode the
  // Observation.
  [[nodiscard]] static lib::statusor::StatusOr<std::unique_ptr<Observation>>
  EncodeSumAndCountObservation(
      const std::vector<std::tuple<std::vector<uint32_t>, int64_t, uint32_t>>& data);

  // Encodes an Observation of type IndexHistogramObservation.
  //
  // data: A vector of (event_codes, histogram vector of (index, count)) pairs that will be used to
  // encode the Observation.
  [[nodiscard]] static lib::statusor::StatusOr<std::unique_ptr<Observation>>
  EncodeIndexHistogramObservation(
      const std::vector<
          std::tuple<std::vector<uint32_t>, std::vector<std::tuple<uint32_t, int64_t>>>>& data);

  // Encodes an Observation of type StringHistogramObservation.
  //
  // hashes: Vector of hashes of strings (hashed using Farmhash Fingerprint128).
  // data: A vector of (event_codes, histogram vector of (index, count)) pairs that will be used to
  // encode the Observation. The bucket with index i in the histogram contains the count for the
  // string whose hash is in position i in the hashes vector.
  [[nodiscard]] static lib::statusor::StatusOr<std::unique_ptr<Observation>>
  EncodeStringHistogramObservation(const std::vector<std::string>& hashes,
                                   const std::vector<std::tuple<EventCodes, Histogram>>& data);

 private:
  // Encodes a BasicRapporObservation for a given |metric|, |report|, and
  // |day_index| in which the data field is a Basic RAPPOR encoding of a vector
  // of |num_categories| zero bits.
  Result EncodeNullBasicRapporObservation(MetricRef metric, const ReportDefinition* report,
                                          uint32_t day_index, uint32_t num_categories) const;

  // Returns a result containing a populated ObservationMetadata and an empty Observation.
  Result NewObservationWithMetadata(MetricRef metric, const ReportDefinition* report,
                                    uint32_t day_index) const;

  const system_data::ClientSecret client_secret_;
  MetadataBuilder& metadata_builder_;
};

}  // namespace cobalt::logger

#endif  // COBALT_SRC_LOGGER_ENCODER_H_