Add logic to write output with encoded Ids. (#2286)

Summary:
Pull Request resolved: #2286

# 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.
5. Add support for filter constraints in pc_translator.

# In this diff
Add logic to generate transformed publisher output with encoded breakdown ID and write the output.

Differential Revision:
D44645325

Privacy Context Container: L416713

fbshipit-source-id: 953989f8440827ff528f51c202e498c01fcf31bd

Author: Ajinkya Ghonge

fbpcs/pc_translator/PCTranslator.cpp

@@ -14,24 +14,27 @@
 #include <fbpcf/mpc_std_lib/oram/encoder/IOramEncoder.h>
 #include <fbpcf/mpc_std_lib/oram/encoder/OramEncoder.h>
 #include <algorithm>
+#include <cstdint>
+#include <iterator>
 #include <set>
 #include <stdexcept>
+#include <string>
 #include "fbpcs/emp_games/common/Csv.h"
 #include "folly/String.h"
 
 namespace pc_translator {
 
-std::string PCTranslator::encode(const std::string& inputDataset) {
+std::string PCTranslator::encode(const std::string& inputDatasetPath) {
   auto validInstructionSetNames =
       PCTranslator::retrieveInstructionSetNamesForRun(pcsFeatures_);
   auto pcInstructionSets =
       PCTranslator::retrieveInstructionSets(validInstructionSetNames);
   if (pcInstructionSets.empty()) {
     // No instruction set found. return the input dataset path.
-    return inputDataset;
+    return inputDatasetPath;
   }
   return PCTranslator::transformDataset(
-      inputDataset, pcInstructionSets.front());
+      inputDatasetPath, pcInstructionSets.front());
 }
 
 std::string PCTranslator::decode(
@@ -79,30 +82,43 @@ std::vector<std::string> PCTranslator::retrieveInstructionSetNamesForRun(
 }
 
 std::string PCTranslator::transformDataset(
-    const std::string& inputData,
+    const std::string& inputDatasetPath,
     std::shared_ptr<pc_translator::PCInstructionSet> pcInstructionSet) {
   // Parse the input CSV
   auto lineNo = 0;
   std::vector<std::vector<uint32_t>> inputColums;
+  std::vector<std::string> outputHeader;
+  std::vector<std::vector<std::string>> outputContent;
   private_measurement::csv::readCsv(
-      inputData,
+      inputDatasetPath,
       [&](const std::vector<std::string>& header,
           const std::vector<std::string>& parts) {
         std::vector<uint32_t> inputColumnPerRow;
+        std::string column;
+        std::uint32_t value;
+        bool found = false;
+        std::vector<std::string> outputContentPerRow;
         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()) {
+          column = header[i];
+          value = std::atoi(parts[i].c_str());
+          found =
+              (std::find(
+                   pcInstructionSet->getGroupByIds().begin(),
+                   pcInstructionSet->getGroupByIds().end(),
+                   column) != pcInstructionSet->getGroupByIds().end());
+          if (found) {
             inputColumnPerRow.push_back(value);
+          } else {
+            if (lineNo == 0) {
+              outputHeader.push_back(header[i]);
+            }
+            outputContentPerRow.push_back(parts[i]);
           }
         }
 
         inputColums.push_back(inputColumnPerRow);
+        outputContent.push_back(outputContentPerRow);
         lineNo++;
       });
 
@@ -114,9 +130,34 @@ std::string PCTranslator::transformDataset(
 
   auto encodedIndexes = encoder->generateORAMIndexes(inputColums);
 
-  // TODO : Append the enodedIndexes at the end of publisher output and return
-  // output path.
-  return "";
+  auto dir = inputDatasetPath.substr(0, inputDatasetPath.rfind("/") + 1);
+  auto output_dataset_path = dir + "transformed_publisher_input.csv";
+
+  PCTranslator::putOutputData(
+      output_dataset_path, outputHeader, outputContent, encodedIndexes);
+  return output_dataset_path;
+}
+
+void PCTranslator::putOutputData(
+    const std::string& output_dataset_path,
+    std::vector<std::string>& outputHeader,
+    std::vector<std::vector<std::string>>& outputContent,
+    const std::vector<uint32_t>& encodedIndexes) {
+  outputHeader.push_back("breakdown_id");
+
+  if (outputContent.size() != encodedIndexes.size()) {
+    throw std::runtime_error(
+        "Encoded index vector size should match the input vector size.");
+  }
+
+  for (std::vector<std::string>::size_type i = 0; i < encodedIndexes.size();
+       ++i) {
+    auto indexVec = std::to_string(encodedIndexes[i]);
+    outputContent[i].push_back(indexVec);
+  }
+
+  private_measurement::csv::writeCsv(
+      output_dataset_path, outputHeader, outputContent);
 }
 
 std::shared_ptr<PCInstructionSet> PCTranslator::parseInstructionSet(

fbpcs/pc_translator/PCTranslator.h

@@ -55,6 +55,12 @@ class PCTranslator {
   std::string transformDataset(
       const std::string& inputData,
       std::shared_ptr<pc_translator::PCInstructionSet> pcInstructionSet);
+
+  void putOutputData(
+      const std::string& output_dataset_path,
+      std::vector<std::string>& outputHeader,
+      std::vector<std::vector<std::string>>& outputContent,
+      const std::vector<uint32_t>& encodedIndexes);
 };
 
 } // namespace pc_translator

fbpcs/pc_translator/tests/TestPCTranslator.cpp

@@ -5,6 +5,7 @@
  * LICENSE file in the root directory of this source tree.
  */
 
+#include <fbpcf/io/api/FileIOWrappers.h>
 #include <gtest/gtest.h>
 #include "../../emp_games/common/TestUtil.h"
 #include "fbpcs/pc_translator/PCTranslator.h"
@@ -16,21 +17,38 @@ class TestPCTranslator : public ::testing::Test {
   std::string pcs_features_;
   std::string test_instruction_set_base_path_;
   std::string test_publisher_input_path_;
+  std::string test_transformed_output_path_;
+  std::string expected_transformed_output_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_publisher_input_path_ = "/tmp/publisher_unittest.csv";
+    test_transformed_output_path_ = "/tmp/transformed_publisher_input.csv";
+    expected_transformed_output_path_ =
+        baseDir + "expected_transformed_publisher_input.csv";
+    auto contents =
+        fbpcf::io::FileIOWrappers::readFile(baseDir + "publisher_unittest.csv");
+    fbpcf::io::FileIOWrappers::writeFile(test_publisher_input_path_, contents);
+  }
+
+  void TearDown() override {
+    std::remove(test_publisher_input_path_.c_str());
+    std::remove(test_transformed_output_path_.c_str());
   }
 };
 
 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, "");
+  auto contents = fbpcf::io::FileIOWrappers::readFile(outputPath);
+  auto expectedContents =
+      fbpcf::io::FileIOWrappers::readFile(expected_transformed_output_path_);
+  EXPECT_EQ(outputPath, test_transformed_output_path_);
+  EXPECT_EQ(contents, expectedContents);
 }
 } // namespace pc_translator

fbpcs/pc_translator/tests/expected_transformed_publisher_input.csv

@@ -0,0 +1,13 @@
+id_,opportunity,test_flag,opportunity_timestamp,breakdown_id
+cfcd208495d565ef66e7dff9f98764da,1,0,1600000430,0
+c4ca4238a0b923820dcc509a6f75849b,1,1,1600000401,1
+c81e728d9d4c2f636f067f89cc14862c,0,0,0,2
+eccbc87e4b5ce2fe28308fd9f2a7baf3,0,0,0,3
+a87ff679a2f3e71d9181a67b7542122c,0,0,0,0
+e4da3b7fbbce2345d7772b0674a318d5,1,1,1600000461,4
+1679091c5a880faf6fb5e6087eb1b2dc,1,0,1600000052,5
+8f14e45fceea167a5a36dedd4bea2543,1,0,1600000831,6
+c9f0f895fb98ab9159f51fd0297e236d,1,0,1600000530,7
+45c48cce2e2d7fbdea1afc51c7c6ad26,1,0,1600000972,5
+d3d9446802a44259755d38e6d163e820,0,0,0,0
+6512bd43d9caa6e02c990b0a82652dca,0,0,0,0