From 73d2d233d6a84bbc8b672e5c5fc1b3038015d81e Mon Sep 17 00:00:00 2001 From: ajinkyaghonge Date: Wed, 5 Apr 2023 14:21:14 -0700 Subject: [PATCH 1/2] Create pc_translator library. Differential Revision: D44617759 fbshipit-source-id: e097cd72cc54d79cb05ab0912ec4864f6c0e7d81 --- fbpcs/pc_translator/PCTranslator.cpp | 39 +++++++++++++++++++++ fbpcs/pc_translator/PCTranslator.h | 52 ++++++++++++++++++++++++++++ 2 files changed, 91 insertions(+) create mode 100644 fbpcs/pc_translator/PCTranslator.cpp create mode 100644 fbpcs/pc_translator/PCTranslator.h diff --git a/fbpcs/pc_translator/PCTranslator.cpp b/fbpcs/pc_translator/PCTranslator.cpp new file mode 100644 index 000000000..abc23e908 --- /dev/null +++ b/fbpcs/pc_translator/PCTranslator.cpp @@ -0,0 +1,39 @@ +/* + * Copyright (c) Meta Platforms, Inc. and affiliates. + * + * This source code is licensed under the MIT license found in the + * LICENSE file in the root directory of this source tree. + */ + +#include "fbpcs/pc_translator/PCTranslator.h" + +namespace pc_translator { + +std::string PCTranslator::encode(const std::string& /* inputDataset */) { + throw std::runtime_error("Unimplemented"); +} + +std::string PCTranslator::decode( + const std::string& /* aggregatedOutputDataset */) { + throw std::runtime_error("Unimplemented"); +} + +void PCTranslator::retrieveInstructionSets( + std::vector& /* instructionSetNames */) { + throw std::runtime_error("Unimplemented"); +} + +std::vector PCTranslator::retrieveInstructionSetNamesForRun( + const std::string& /* pcsFeatures */) { + throw std::runtime_error("Unimplemented"); +} + +void PCTranslator::transformDataset(const std::string& /* input */) { + throw std::runtime_error("Unimplemented"); +} + +void PCTranslator::parseInstructionSet( + const std::string& /* instructionSet */) { + throw std::runtime_error("Unimplemented"); +} +} // namespace pc_translator diff --git a/fbpcs/pc_translator/PCTranslator.h b/fbpcs/pc_translator/PCTranslator.h new file mode 100644 index 000000000..0107e4025 --- /dev/null +++ b/fbpcs/pc_translator/PCTranslator.h @@ -0,0 +1,52 @@ +/* + * Copyright (c) Meta Platforms, Inc. and affiliates. + * + * This source code is licensed under the MIT license found in the + * LICENSE file in the root directory of this source tree. + */ + +#pragma once + +#include +#include +#include +#include + +namespace pc_translator { + +/* + * This class contains functions required for PC Translator during actual run + * i.e. retrieving the PC instruction sets, filtering the set per active GK for + * run, encoding and decoding the dataset files input as per the instruction + * set. + */ +class PCTranslator { + public: + explicit PCTranslator(const std::string& pcsFeatures) + : pcsfeatures_(pcsFeatures) {} + + /* + * Method to encode the configurable fields in input dataset as per the active + * pc instruction sets for the run. This method will output the path of + * transformed input dataset, which can be used in further PC run. + */ + std::string encode(const std::string& inputDataset); + + /* + * Method to decode final aggregated output with the encoded breakdown Ids as + * the keys. This method will decode the breakdown Ids to original group Id + * values and format the aggregated output as per the new keys. Output of this + * method would be the path of the decoded aggregated output. + */ + std::string decode(const std::string& aggregatedOutputDataset); + + private: + std::string pcsfeatures_; + void retrieveInstructionSets(std::vector& instructionSetNames); + std::vector retrieveInstructionSetNamesForRun( + const std::string& pcsfeatures); + void parseInstructionSet(const std::string& instructionSet); + void transformDataset(const std::string& input); +}; + +} // namespace pc_translator From 16490884a6c3e649028995c10036354bc456d671 Mon Sep 17 00:00:00 2001 From: Ajinkya Ghonge Date: Wed, 5 Apr 2023 14:21:34 -0700 Subject: [PATCH 2/2] Add classes to parse PC Instruction set. (#2268) Summary: Pull Request resolved: https://github.com/facebookresearch/fbpcs/pull/2268 # Context As per PC Translator design, we need a runtime library will be called during PC run. This library will be called at the beginning of PC run to encode specified fields in publisher side input into a encoded breakdown (aggregation) Ids based on active PC instruction sets for the run. The library will filter the active PC Instruction sets for the run based on parsing the pcs_features i.e. gatekeepers for the particular run. # Product decisions In this stack we would focus solely on functionality required for private lift runs. We would focus on the MVP implementation of the library and its integration with fbpcf ORAM encoder library in this stack. # Stack 1. Create runtime pc_translator library. 2. Add logic to retrieve and parse PC instruction set, filtered based on the active gatekeepers for the run. 3. Integrate pc_translator library with fbpcf ORAM encoder. 4. Add logic to generate transformed publisher output with encoded breakdown ID and write the output. # In this diff Add logic to retrieve and parse PC instruction set. Differential Revision: D44618035 Privacy Context Container: L416713 fbshipit-source-id: b8c6bc1ad2a68c2e6900814da12fe0af7a5e8d8f --- fbpcs/pc_translator/PCTranslator.cpp | 58 +++++++++++++++---- fbpcs/pc_translator/PCTranslator.h | 20 +++++-- .../input_processing/FilterConstraint.cpp | 33 +++++++++++ .../input_processing/FilterConstraint.h | 43 ++++++++++++++ .../input_processing/PCInstructionSet.cpp | 51 ++++++++++++++++ .../input_processing/PCInstructionSet.h | 47 +++++++++++++++ .../input_processing/TestPCInstructionSet.cpp | 44 ++++++++++++++ .../test_instruction_set.json | 45 ++++++++++++++ 8 files changed, 324 insertions(+), 17 deletions(-) create mode 100644 fbpcs/pc_translator/input_processing/FilterConstraint.cpp create mode 100644 fbpcs/pc_translator/input_processing/FilterConstraint.h create mode 100644 fbpcs/pc_translator/input_processing/PCInstructionSet.cpp create mode 100644 fbpcs/pc_translator/input_processing/PCInstructionSet.h create mode 100644 fbpcs/pc_translator/tests/input_processing/TestPCInstructionSet.cpp create mode 100644 fbpcs/pc_translator/tests/input_processing/test_instruction_set.json diff --git a/fbpcs/pc_translator/PCTranslator.cpp b/fbpcs/pc_translator/PCTranslator.cpp index abc23e908..f5c640967 100644 --- a/fbpcs/pc_translator/PCTranslator.cpp +++ b/fbpcs/pc_translator/PCTranslator.cpp @@ -6,11 +6,22 @@ */ #include "fbpcs/pc_translator/PCTranslator.h" +#include "fbpcs/pc_translator/input_processing/PCInstructionSet.h" + +#include +#include +#include +#include "folly/String.h" namespace pc_translator { -std::string PCTranslator::encode(const std::string& /* inputDataset */) { - throw std::runtime_error("Unimplemented"); +std::string PCTranslator::encode(const std::string& inputDataset) { + auto validInstructionSetNames = + PCTranslator::retrieveInstructionSetNamesForRun(pcsFeatures_); + auto pcInstructionSets = + PCTranslator::retrieveInstructionSets(validInstructionSetNames); + PCTranslator::transformDataset(inputDataset, pcInstructionSets); + return ""; } std::string PCTranslator::decode( @@ -18,22 +29,47 @@ std::string PCTranslator::decode( throw std::runtime_error("Unimplemented"); } -void PCTranslator::retrieveInstructionSets( - std::vector& /* instructionSetNames */) { - throw std::runtime_error("Unimplemented"); +std::vector> +PCTranslator::retrieveInstructionSets( + std::vector& instructionSetNames) { + std::vector> pcInstructionSets; + for (auto instructionSetName : instructionSetNames) { + auto file_path = instructionSetBasePath + instructionSetName + ".json"; + auto contents = fbpcf::io::FileIOWrappers::readFile(file_path); + pcInstructionSets.push_back(PCTranslator::parseInstructionSet(contents)); + } + return pcInstructionSets; } std::vector PCTranslator::retrieveInstructionSetNamesForRun( - const std::string& /* pcsFeatures */) { - throw std::runtime_error("Unimplemented"); + const std::string& pcsFeatures) { + std::set enabledFeatureFlags; + folly::splitTo( + ',', + pcsFeatures, + std::inserter(enabledFeatureFlags, enabledFeatureFlags.begin()), + true); + + std::vector validPCInstructionSets; + std::copy_if( + enabledFeatureFlags.begin(), + enabledFeatureFlags.end(), + std::back_inserter(validPCInstructionSets), + [](const std::string& feature) { return feature.find("pc_instr") == 0; }); + + return validPCInstructionSets; } -void PCTranslator::transformDataset(const std::string& /* input */) { +void PCTranslator::transformDataset( + const std::string& /* inputData */, + const std::vector>& + /* pcInstructionSets */) { throw std::runtime_error("Unimplemented"); } -void PCTranslator::parseInstructionSet( - const std::string& /* instructionSet */) { - throw std::runtime_error("Unimplemented"); +std::shared_ptr PCTranslator::parseInstructionSet( + std::string& instructionSet) { + return std::make_shared(PCInstructionSet::fromDynamic( + folly::parseJson(std::move(instructionSet)))); } } // namespace pc_translator diff --git a/fbpcs/pc_translator/PCTranslator.h b/fbpcs/pc_translator/PCTranslator.h index 0107e4025..efa0497c8 100644 --- a/fbpcs/pc_translator/PCTranslator.h +++ b/fbpcs/pc_translator/PCTranslator.h @@ -11,6 +11,7 @@ #include #include #include +#include "fbpcs/pc_translator/input_processing/PCInstructionSet.h" namespace pc_translator { @@ -23,7 +24,7 @@ namespace pc_translator { class PCTranslator { public: explicit PCTranslator(const std::string& pcsFeatures) - : pcsfeatures_(pcsFeatures) {} + : pcsFeatures_(pcsFeatures) {} /* * Method to encode the configurable fields in input dataset as per the active @@ -41,12 +42,19 @@ class PCTranslator { std::string decode(const std::string& aggregatedOutputDataset); private: - std::string pcsfeatures_; - void retrieveInstructionSets(std::vector& instructionSetNames); + std::string pcsFeatures_; + const std::string instructionSetBasePath = + "https://pc-translator.s3.us-west-2.amazonaws.com/"; + std::vector> retrieveInstructionSets( + std::vector& instructionSetNames); std::vector retrieveInstructionSetNamesForRun( - const std::string& pcsfeatures); - void parseInstructionSet(const std::string& instructionSet); - void transformDataset(const std::string& input); + const std::string& pcsFeatures); + std::shared_ptr parseInstructionSet( + std::string& instructionSet); + void transformDataset( + const std::string& input_data, + const std::vector>& + pcInstructionSets); }; } // namespace pc_translator diff --git a/fbpcs/pc_translator/input_processing/FilterConstraint.cpp b/fbpcs/pc_translator/input_processing/FilterConstraint.cpp new file mode 100644 index 000000000..6bc8a0dac --- /dev/null +++ b/fbpcs/pc_translator/input_processing/FilterConstraint.cpp @@ -0,0 +1,33 @@ +/* + * Copyright (c) Meta Platforms, Inc. and affiliates. + * + * This source code is licensed under the MIT license found in the + * LICENSE file in the root directory of this source tree. + */ + +#include "fbpcs/pc_translator/input_processing/FilterConstraint.h" + +#include +#include +#include +#include + +namespace pc_translator { +FilterConstraint::FilterConstraint( + const std::string& name, + const std::string& type, + int value) + : name_(name), type_(type), value_(value) {} + +std::string FilterConstraint::getName() const { + return name_; +} + +std::string FilterConstraint::getType() const { + return type_; +} + +int FilterConstraint::getValue() const { + return value_; +} +} // namespace pc_translator diff --git a/fbpcs/pc_translator/input_processing/FilterConstraint.h b/fbpcs/pc_translator/input_processing/FilterConstraint.h new file mode 100644 index 000000000..00cbbc189 --- /dev/null +++ b/fbpcs/pc_translator/input_processing/FilterConstraint.h @@ -0,0 +1,43 @@ +/* + * Copyright (c) Meta Platforms, Inc. and affiliates. + * + * This source code is licensed under the MIT license found in the + * LICENSE file in the root directory of this source tree. + */ + +#pragma once + +#include +#include +#include +#include + +namespace pc_translator { + +/* + * Class to store each filter constraint include in the PC instruction set. + */ +class FilterConstraint { + public: + FilterConstraint(const std::string& name, const std::string& type, int value); + + /* + * Name of the filter constraint i.e. the field on which this filter is to be + * applied. + */ + std::string getName() const; + + /* + * Constraint type i.e. LT, LTE, EQ, NEQ etc. + */ + std::string getType() const; + + int getValue() const; + + private: + std::string name_; + std::string type_; + int value_; +}; + +} // namespace pc_translator diff --git a/fbpcs/pc_translator/input_processing/PCInstructionSet.cpp b/fbpcs/pc_translator/input_processing/PCInstructionSet.cpp new file mode 100644 index 000000000..bd1c995d8 --- /dev/null +++ b/fbpcs/pc_translator/input_processing/PCInstructionSet.cpp @@ -0,0 +1,51 @@ +/* + * Copyright (c) Meta Platforms, Inc. and affiliates. + * + * This source code is licensed under the MIT license found in the + * LICENSE file in the root directory of this source tree. + */ + +#include "fbpcs/pc_translator/input_processing/PCInstructionSet.h" + +#include +#include +#include +#include +#include + +namespace pc_translator { + +const std::vector& PCInstructionSet::getGroupByIds() const { + return groupByIds; +} + +const std::vector& PCInstructionSet::getFilterConstraints() + const { + return filterConstraints; +} + +PCInstructionSet PCInstructionSet::fromDynamic(const folly::dynamic& obj) { + PCInstructionSet pcInstructionSet; + auto aggregationConfig = obj["aggregated_metrics"]; + auto groupByFields = aggregationConfig["group_by"]; + + for (auto groupByField : groupByFields) { + pcInstructionSet.groupByIds.push_back(groupByField.asString()); + } + + auto filterConstraintsFields = aggregationConfig["filter"]; + + for (auto& [key, constraints] : filterConstraintsFields.items()) { + std::string name = key.asString(); + for (auto constraint : constraints) { + auto constraintType = constraint["constraint_type"].asString(); + auto constraintValue = constraint["value"].asInt(); + FilterConstraint filterConstraint(name, constraintType, constraintValue); + pcInstructionSet.filterConstraints.push_back(filterConstraint); + } + } + + return pcInstructionSet; +} + +} // namespace pc_translator diff --git a/fbpcs/pc_translator/input_processing/PCInstructionSet.h b/fbpcs/pc_translator/input_processing/PCInstructionSet.h new file mode 100644 index 000000000..01e710f54 --- /dev/null +++ b/fbpcs/pc_translator/input_processing/PCInstructionSet.h @@ -0,0 +1,47 @@ +/* + * Copyright (c) Meta Platforms, Inc. and affiliates. + * + * This source code is licensed under the MIT license found in the + * LICENSE file in the root directory of this source tree. + */ + +#pragma once + +#include +#include +#include +#include +#include +#include "fbpcs/pc_translator/input_processing/FilterConstraint.h" + +namespace pc_translator { + +/* + * Class to store PC Instruction set. This class contains a list of group Ids as + * well as list of filter constraints. + */ +class PCInstructionSet { + public: + /* + * Method to all group Ids from the PC instruction set. + */ + const std::vector& getGroupByIds() const; + + /* + * Method to get all filter constraints from PC instruction set. + */ + const std::vector& getFilterConstraints() const; + + /* + * Method to get parse and create PCInstructionSet instance. + */ + static PCInstructionSet fromDynamic(const folly::dynamic& obj); + + private: + std::vector groupByIds; + std::vector filterConstraints; + + void parseJson(const std::string& json); +}; + +} // namespace pc_translator diff --git a/fbpcs/pc_translator/tests/input_processing/TestPCInstructionSet.cpp b/fbpcs/pc_translator/tests/input_processing/TestPCInstructionSet.cpp new file mode 100644 index 000000000..32a38ff3a --- /dev/null +++ b/fbpcs/pc_translator/tests/input_processing/TestPCInstructionSet.cpp @@ -0,0 +1,44 @@ +/* + * Copyright (c) Meta Platforms, Inc. and affiliates. + * + * This source code is licensed under the MIT license found in the + * LICENSE file in the root directory of this source tree. + */ + +#include +#include +#include + +#include +#include +#include "../../../emp_games/common/TestUtil.h" +#include "fbpcs/pc_translator/input_processing/PCInstructionSet.h" +#include "folly/Random.h" + +namespace pc_translator { +class TestPCInstructionSet : public ::testing::Test { + public: + protected: + std::string testInstructionSetPath_; + + void SetUp() override { + std::string baseDir = + private_measurement::test_util::getBaseDirFromPath(__FILE__); + testInstructionSetPath_ = baseDir + "test_instruction_set.json"; + } +}; + +TEST_F(TestPCInstructionSet, TestStandardWorkflowTest) { + auto pcInstructionSet = std::make_shared( + PCInstructionSet::fromDynamic(folly::parseJson( + fbpcf::io::FileIOWrappers::readFile(testInstructionSetPath_)))); + auto groupByIds = pcInstructionSet->getGroupByIds(); + auto filterConstraints = pcInstructionSet->getFilterConstraints(); + EXPECT_EQ(groupByIds.size(), 2); + EXPECT_EQ(filterConstraints.size(), 4); + EXPECT_EQ(filterConstraints[0].getName(), "gender"); + EXPECT_EQ(filterConstraints[0].getType(), "EQ"); + EXPECT_EQ(filterConstraints[0].getValue(), 0); +} + +} // namespace pc_translator diff --git a/fbpcs/pc_translator/tests/input_processing/test_instruction_set.json b/fbpcs/pc_translator/tests/input_processing/test_instruction_set.json new file mode 100644 index 000000000..7a0008b19 --- /dev/null +++ b/fbpcs/pc_translator/tests/input_processing/test_instruction_set.json @@ -0,0 +1,45 @@ +{ + "publisher_input": { + "num_impressions": "int", + "num_clicks": "int", + "total_spend": "int", + "opportunity_timstamp": "int", + "test_flag": "int", + "age": "int", + "gender": "Optional[int]", + "breakdown_id": "Optional[int]" + }, + "partner_input": { + "value": "int", + "event_timestamp": "int", + "partner_cohort_id": "Optional[int]" + }, + "aggregated_metrics": { + "filter": { + "age": [ + { + "constraint_type": "GTE", + "value": "25" + }, + { + "constraint_type": "LTE", + "value": "40" + } + ], + "gender": [ + { + "constraint_type": "EQ", + "value": "0" + }, + { + "constraint_type": "EQ", + "value": "1" + } + ] + }, + "group_by": [ + "age", + "gender" + ] + } +}