Integrate OramEncoder library with pc_translator. (#2269)

Summary:
Pull Request resolved: #2269

# 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
 Integrate pc_translator library with fbpcf ORAM encoder.

Differential Revision:
D44634384

Privacy Context Container: L416713

fbshipit-source-id: c328cf3ce4759e02c0a3e0fd4aa7f36699cfb7f3

Author: Ajinkya Ghonge

fbpcs/pc_translator/PCTranslator.cpp

@@ -10,7 +10,12 @@
 
 #include <fbpcf/common/FunctionalUtil.h>
 #include <fbpcf/io/api/FileIOWrappers.h>
+#include <fbpcf/mpc_std_lib/oram/encoder/IFilter.h>
+#include <fbpcf/mpc_std_lib/oram/encoder/IOramEncoder.h>
+#include <fbpcf/mpc_std_lib/oram/encoder/OramEncoder.h>
+#include <algorithm>
 #include <set>
+#include "fbpcs/emp_games/common/Csv.h"
 #include "folly/String.h"
 
 namespace pc_translator {
@@ -20,8 +25,8 @@ std::string PCTranslator::encode(const std::string& inputDataset) {
       PCTranslator::retrieveInstructionSetNamesForRun(pcsFeatures_);
   auto pcInstructionSets =
       PCTranslator::retrieveInstructionSets(validInstructionSetNames);
-  PCTranslator::transformDataset(inputDataset, pcInstructionSets);
-  return "";
+  return PCTranslator::transformDataset(
+      inputDataset, pcInstructionSets.front());
 }
 
 std::string PCTranslator::decode(
@@ -34,7 +39,13 @@ PCTranslator::retrieveInstructionSets(
     std::vector<std::string>& instructionSetNames) {
   std::vector<std::shared_ptr<PCInstructionSet>> pcInstructionSets;
   for (auto instructionSetName : instructionSetNames) {
-    auto file_path = instructionSetBasePath + instructionSetName + ".json";
+    instructionSetName.erase(
+        remove(instructionSetName.begin(), instructionSetName.end(), '\''),
+        instructionSetName.end());
+    instructionSetName.erase(
+        remove(instructionSetName.begin(), instructionSetName.end(), ' '),
+        instructionSetName.end());
+    auto file_path = instructionSetBasePath_ + instructionSetName + ".json";
     auto contents = fbpcf::io::FileIOWrappers::readFile(file_path);
     pcInstructionSets.push_back(PCTranslator::parseInstructionSet(contents));
   }
@@ -55,16 +66,52 @@ std::vector<std::string> PCTranslator::retrieveInstructionSetNamesForRun(
       enabledFeatureFlags.begin(),
       enabledFeatureFlags.end(),
       std::back_inserter(validPCInstructionSets),
-      [](const std::string& feature) { return feature.find("pc_instr") == 0; });
+      [](const std::string& feature) {
+        return feature.find("pc_instr") != std::string::npos;
+      });
 
   return validPCInstructionSets;
 }
 
-void PCTranslator::transformDataset(
-    const std::string& /* inputData */,
-    const std::vector<std::shared_ptr<pc_translator::PCInstructionSet>>&
-    /* pcInstructionSets */) {
-  throw std::runtime_error("Unimplemented");
+std::string PCTranslator::transformDataset(
+    const std::string& inputData,
+    std::shared_ptr<pc_translator::PCInstructionSet> pcInstructionSet) {
+  // Parse the input CSV
+  auto lineNo = 0;
+  std::vector<std::vector<uint32_t>> inputColums;
+  private_measurement::csv::readCsv(
+      inputData,
+      [&](const std::vector<std::string>& header,
+          const std::vector<std::string>& parts) {
+        std::vector<uint32_t> inputColumnPerRow;
+        for (std::vector<std::string>::size_type i = 0; i < header.size();
+             ++i) {
+          auto column = header[i];
+          auto value = std::atoi(parts[i].c_str());
+          auto iter = std::find(
+              pcInstructionSet->getGroupByIds().begin(),
+              pcInstructionSet->getGroupByIds().end(),
+              column);
+          if (iter != pcInstructionSet->getGroupByIds().end()) {
+            inputColumnPerRow.push_back(value);
+          }
+        }
+
+        inputColums.push_back(inputColumnPerRow);
+        lineNo++;
+      });
+
+  auto filters = std::make_unique<
+      std::vector<std::unique_ptr<fbpcf::mpc_std_lib::oram::IFilter>>>(0);
+  std::unique_ptr<fbpcf::mpc_std_lib::oram::IOramEncoder> encoder =
+      std::make_unique<fbpcf::mpc_std_lib::oram::OramEncoder>(
+          std::move(filters));
+
+  auto encodedIndexes = encoder->generateORAMIndexes(inputColums);
+
+  // TODO : Append the enodedIndexes at the end of publisher output and return
+  // output path.
+  return "";
 }
 
 std::shared_ptr<PCInstructionSet> PCTranslator::parseInstructionSet(

fbpcs/pc_translator/PCTranslator.h

@@ -26,11 +26,12 @@ class PCTranslator {
   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.
-   */
+  explicit PCTranslator(
+      const std::string& pcsFeatures,
+      const std::string& instructionSetBasePath)
+      : pcsFeatures_(pcsFeatures),
+        instructionSetBasePath_(instructionSetBasePath) {}
+
   std::string encode(const std::string& inputDataset);
 
   /*
@@ -43,18 +44,17 @@ class PCTranslator {
 
  private:
   std::string pcsFeatures_;
-  const std::string instructionSetBasePath =
+  std::string instructionSetBasePath_ =
       "https://pc-translator.s3.us-west-2.amazonaws.com/";
   std::vector<std::shared_ptr<PCInstructionSet>> retrieveInstructionSets(
       std::vector<std::string>& instructionSetNames);
   std::vector<std::string> retrieveInstructionSetNamesForRun(
       const std::string& pcsFeatures);
   std::shared_ptr<PCInstructionSet> parseInstructionSet(
       const std::string& instructionSet);
-  void transformDataset(
-      const std::string& input_data,
-      const std::vector<std::shared_ptr<pc_translator::PCInstructionSet>>&
-          pcInstructionSets);
+  std::string transformDataset(
+      const std::string& inputData,
+      std::shared_ptr<pc_translator::PCInstructionSet> pcInstructionSet);
 };
 
 } // namespace pc_translator

fbpcs/pc_translator/tests/TestPCTranslator.cpp

@@ -0,0 +1,36 @@
+/*
+ * 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 <gtest/gtest.h>
+#include "../../emp_games/common/TestUtil.h"
+#include "fbpcs/pc_translator/PCTranslator.h"
+
+namespace pc_translator {
+class TestPCTranslator : public ::testing::Test {
+ public:
+ protected:
+  std::string pcs_features_;
+  std::string test_instruction_set_base_path_;
+  std::string test_publisher_input_path_;
+
+  void SetUp() override {
+    pcs_features_ =
+        "'num_mpc_container_mutation', 'private_lift_unified_data_process', 'pc_instr_test_instruction_set'";
+    std::string baseDir =
+        private_measurement::test_util::getBaseDirFromPath(__FILE__);
+    test_instruction_set_base_path_ = baseDir + "input_processing/";
+    test_publisher_input_path_ = baseDir + "publisher_unittest.csv";
+  }
+};
+
+TEST_F(TestPCTranslator, TestEncode) {
+  auto pcTranslator = std::make_shared<PCTranslator>(
+      pcs_features_, test_instruction_set_base_path_);
+  auto outputPath = pcTranslator->encode(test_publisher_input_path_);
+  EXPECT_EQ(outputPath, "");
+}
+} // namespace pc_translator

fbpcs/pc_translator/tests/input_processing/TestPCInstructionSet.cpp

@@ -13,7 +13,6 @@
 #include <gtest/gtest.h>
 #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 {
@@ -24,7 +23,7 @@ class TestPCInstructionSet : public ::testing::Test {
   void SetUp() override {
     std::string baseDir =
         private_measurement::test_util::getBaseDirFromPath(__FILE__);
-    testInstructionSetPath_ = baseDir + "test_instruction_set.json";
+    testInstructionSetPath_ = baseDir + "pc_instr_test_instruction_set.json";
   }
 };
 
@@ -35,7 +34,7 @@ TEST_F(TestPCInstructionSet, TestStandardWorkflowTest) {
   auto groupByIds = pcInstructionSet->getGroupByIds();
   auto filterConstraints = pcInstructionSet->getFilterConstraints();
   EXPECT_EQ(groupByIds.size(), 2);
-  EXPECT_EQ(filterConstraints.size(), 4);
+  EXPECT_EQ(filterConstraints.size(), 3);
   EXPECT_EQ(filterConstraints[0].getName(), "gender");
   EXPECT_EQ(filterConstraints[0].getType(), "EQ");
   EXPECT_EQ(filterConstraints[0].getValue(), 0);

fbpcs/pc_translator/tests/input_processing/test_instruction_set.json renamed to fbpcs/pc_translator/tests/input_processing/pc_instr_test_instruction_set.json

@@ -30,10 +30,6 @@
                 {
                     "constraint_type": "EQ",
                     "value": "0"
-                },
-                {
-                    "constraint_type": "EQ",
-                    "value": "1"
                 }
             ]
         },

fbpcs/pc_translator/tests/publisher_unittest.csv

@@ -0,0 +1,13 @@
+id_,opportunity,test_flag,opportunity_timestamp, age, gender
+cfcd208495d565ef66e7dff9f98764da,1,0,1600000430, 25, 0
+c4ca4238a0b923820dcc509a6f75849b,1,1,1600000401, 26, 1
+c81e728d9d4c2f636f067f89cc14862c,0,0,0, 44, 0
+eccbc87e4b5ce2fe28308fd9f2a7baf3,0,0,0, 23, 0
+a87ff679a2f3e71d9181a67b7542122c,0,0,0, 25, 0
+e4da3b7fbbce2345d7772b0674a318d5,1,1,1600000461, 24, 1
+1679091c5a880faf6fb5e6087eb1b2dc,1,0,1600000052, 25, 1
+8f14e45fceea167a5a36dedd4bea2543,1,0,1600000831, 26, 0
+c9f0f895fb98ab9159f51fd0297e236d,1,0,1600000530, 50, 0
+45c48cce2e2d7fbdea1afc51c7c6ad26,1,0,1600000972, 25, 1
+d3d9446802a44259755d38e6d163e820,0,0,0, 25, 0
+6512bd43d9caa6e02c990b0a82652dca,0,0,0, 25, 0