Remove the old C-FFI POC code and add an XCTest

Author: the80srobot

Source/santad/Logs/EndpointSecurity/ParquetLogger/ParquetLogger.cc

@@ -1,9 +1,2 @@
 #include "ParquetLogger.h"
 
-extern "C" {
-extern bool parquet2_1337_bloom_filter_contains(int64_t x);
-}
-
-bool FilterContains(int64_t x) {
-    return parquet2_1337_bloom_filter_contains(x);
-}

Source/santad/Logs/EndpointSecurity/ParquetLogger/ParquetLogger.h

@@ -1,8 +1,6 @@
-#include <stdint.h>
+#include "Source/santad/Logs/EndpointSecurity/ParquetLogger/gen/cpp_api.rs.h"
 
 #ifndef SANTA__SANTAD__LOGS_ENDPOINTSECURITY_PARQUETLOGGER_PARQUETLOGGER_H
 #define SANTA__SANTAD__LOGS_ENDPOINTSECURITY_PARQUETLOGGER_PARQUETLOGGER_H
 
-bool FilterContains(int64_t x);
-
 #endif

Source/santad/Logs/EndpointSecurity/ParquetLogger/ParquetLoggerTest.mm

@@ -1,15 +1,57 @@
 #import <XCTest/XCTest.h>
+#import <filesystem>
 #import "ParquetLogger.h"
 
 @interface ParquetLoggerTest : XCTestCase
 @end
 
 @implementation ParquetLoggerTest
 
-// Currently, this just demonstrates that Rust code can be called into.
-- (void)testBasic {
-  XCTAssertFalse(FilterContains(1338));
-  XCTAssertTrue(FilterContains(1337));
+- (void)testWriteTable {
+  auto tmp_path = std::filesystem::temp_directory_path();
+  try {
+    auto args =
+      pedro::wire::table_args_new("test_table", (tmp_path / "test_table.parquet").string());
+
+    pedro::wire::table_args_add_column(*args, "number", pedro::wire::CxxColumnType::Int32);
+    pedro::wire::table_args_add_column(*args, "text", pedro::wire::CxxColumnType::ByteArray);
+    auto table = pedro::wire::table_new(std::move(args));
+
+    pedro::wire::table_push_i32(*table, 0, 1337);
+    pedro::wire::table_push_string(*table, 1, "foo");
+
+    pedro::wire::table_push_i32(*table, 0, 1);
+    pedro::wire::table_push_i32(*table, 0, 2);
+    pedro::wire::table_push_i32(*table, 0, 3);
+
+    pedro::wire::table_push_string(*table, 1, "bar");
+    pedro::wire::table_push_string(*table, 1, "baz");
+    pedro::wire::table_push_string(*table, 1, "qux");
+
+    pedro::wire::table_flush(*table);
+    pedro::wire::table_end(std::move(table));
+  } catch (const std::exception &e) {
+    // None of this should throw.
+    XCTAssertFalse(true, "Exception: %s", e.what());
+  }
+
+  // The file should exist as expected.
+  std::filesystem::path testFilePath = tmp_path / "test_table.parquet";
+  XCTAssertTrue(std::filesystem::exists(testFilePath));
+}
+
+- (void)testInvalidFlushThrows {
+  auto tmp_path = std::filesystem::temp_directory_path();
+  auto args = pedro::wire::table_args_new("test_table", (tmp_path / "test_table.parquet").string());
+  pedro::wire::table_args_add_column(*args, "number", pedro::wire::CxxColumnType::Int32);
+  pedro::wire::table_args_add_column(*args, "text", pedro::wire::CxxColumnType::ByteArray);
+
+  auto table = pedro::wire::table_new(std::move(args));
+  pedro::wire::table_push_i32(*table, 0, 1);
+  pedro::wire::table_push_i32(*table, 0, 2);
+
+  // This should throw because we haven't pushed data to all columns.
+  XCTAssertThrows(pedro::wire::table_flush(*table));
 }
 
 @end

Source/santad/Logs/EndpointSecurity/ParquetLogger/parquet_logger.rs

@@ -56,17 +56,6 @@ mod table;
 mod value;
 mod writer;
 
-use parquet2::bloom_filter;
-
-// This is just some POC code for CXX that'll be removed later.
-// DONOTSUBMIT: Remove this before finishing the PR.
-#[no_mangle]
-pub extern "C" fn parquet2_1337_bloom_filter_contains(x: i64) -> bool {
-    let mut bits = vec![0; 32];
-    bloom_filter::insert(&mut bits, bloom_filter::hash_native::<i64>(1337));
-    bloom_filter::is_in_set(&bits, bloom_filter::hash_native(x))
-}
-
 // This is the main public API.
 pub use crate::table::Table;
 

Source/santad/Logs/EndpointSecurity/ParquetLogger/write_test_file.cc

@@ -5,7 +5,10 @@
 // This demonstrates the use of the C++ API for parquet_logger, which is
 // implemented in Rust. The C++ API uses an opaque type called Table, which is
 // returned as a Box (effectively a unique_ptr).
-
+//
+// To easily check that the file is valid, load it into Pandas:
+//
+// import pandas as pd; pd.read_parquet("./test_table.parquet")
 int main(int argc, char* argv[]) {
   try {
     if (argc < 2) {