[SUPERSEDED by #149] feat(toolbox): add Filter Editor toolbox plugin

CMakeLists.txt

@@ -47,6 +47,7 @@ if(TARGET plotjuggler_sdk::plugin_sdk)
   add_subdirectory(toolbox_colormap)
   add_subdirectory(toolbox_quaternion)
   add_subdirectory(toolbox_reactive_scripts_editor)
+  add_subdirectory(toolbox_filter_editor)
   add_subdirectory(toolbox_mosaico)
   return()
 endif()
@@ -153,5 +154,6 @@ else()
   add_subdirectory(toolbox_fft)
   add_subdirectory(toolbox_colormap)
   add_subdirectory(toolbox_quaternion)
+  add_subdirectory(toolbox_filter_editor)
   add_subdirectory(toolbox_mosaico)
 endif()

toolbox_filter_editor/CMakeLists.txt

@@ -0,0 +1,48 @@
+find_package(nlohmann_json REQUIRED)
+
+include(${CMAKE_CURRENT_LIST_DIR}/../cmake/EmbedManifest.cmake)
+include(${CMAKE_CURRENT_LIST_DIR}/../cmake/EmbedUi.cmake)
+
+add_library(toolbox_filter_editor_plugin SHARED filter_editor_plugin.cpp)
+target_compile_features(toolbox_filter_editor_plugin PRIVATE cxx_std_20)
+target_compile_options(toolbox_filter_editor_plugin PRIVATE ${PJ_WARNING_FLAGS})
+
+target_link_libraries(toolbox_filter_editor_plugin PRIVATE
+  plotjuggler_sdk::plugin_sdk
+  nlohmann_json::nlohmann_json
+  ${CMAKE_DL_LIBS}
+)
+
+pj_embed_manifest(toolbox_filter_editor_plugin
+  HEADER   ${CMAKE_CURRENT_BINARY_DIR}/generated/filter_editor_manifest.hpp
+  VAR_NAME kFilterEditorManifest
+)
+
+pj_embed_ui(toolbox_filter_editor_plugin
+  UI_FILE  ${CMAKE_CURRENT_SOURCE_DIR}/filter_editor_dialog.ui
+  HEADER   ${CMAKE_CURRENT_BINARY_DIR}/generated/filter_editor_dialog_ui.hpp
+  VAR_NAME kFilterEditorDialogUi
+)
+
+# v4 plugin discovery: emit a sidecar .pjmanifest.json the host can scan
+# pre-dlopen. Content derived from manifest.json + {abi_major, family}.
+pj_emit_plugin_manifest(toolbox_filter_editor_plugin
+  FAMILY        toolbox
+  MANIFEST_FILE ${CMAKE_CURRENT_SOURCE_DIR}/manifest.json
+)
+
+# Built-in predefined transforms catalogue lives in this plugin (it provides
+# them); the SDK contract they implement is in plotjuggler_sdk::plugin_sdk.
+# Tests guard the per-transform PJ3-mirror math + the SDK contract (the
+# default appendTail loops calculateNextPoint, save/load round-trips).
+if(TARGET GTest::gtest_main)
+  add_executable(builtin_transforms_test builtin_transforms_test.cpp)
+  target_compile_features(builtin_transforms_test PRIVATE cxx_std_20)
+  target_compile_options(builtin_transforms_test PRIVATE ${PJ_WARNING_FLAGS})
+  target_link_libraries(builtin_transforms_test PRIVATE
+    plotjuggler_sdk::plugin_sdk
+    nlohmann_json::nlohmann_json
+    GTest::gtest_main
+  )
+  add_test(NAME builtin_transforms_test COMMAND builtin_transforms_test)
+endif()

toolbox_filter_editor/builtin_transforms.hpp

@@ -0,0 +1,10 @@
+#pragma once
+// Copyright 2026 Davide Faconti
+// SPDX-License-Identifier: MPL-2.0
+
+// The 12 builtin Filter Transform strategies live in the SDK
+// (plotjuggler_sdk/pj_plugins/filter_protocol/include/pj_plugins/sdk/builtin_transforms.hpp).
+// This shim re-exports them so the plugin and the host can both consume the
+// same source.
+
+#include "pj_plugins/sdk/builtin_transforms.hpp"

toolbox_filter_editor/builtin_transforms_test.cpp

@@ -0,0 +1,214 @@
+// Copyright 2026 Davide Faconti
+// SPDX-License-Identifier: MPL-2.0
+
+#include "builtin_transforms.hpp"
+
+#include <gtest/gtest.h>
+
+#include <cmath>
+#include <cstdint>
+#include <limits>
+
+namespace {
+
+using namespace PJ::sdk;  // NOLINT
+
+// Helper: apply a transform to (xs, ys) pairs and return output points.
+std::vector<Point2> apply(FilterTransform& t, std::vector<double> xs, std::vector<double> ys) {
+  std::vector<Point2> in;
+  in.reserve(xs.size());
+  for (size_t i = 0; i < xs.size(); ++i) {
+    in.push_back({xs[i], ys[i]});
+  }
+  return t.applyBatch(in);
+}
+
+TEST(Transforms, Absolute) {
+  AbsoluteTransform t;
+  auto out = apply(t, {0, 1, 2}, {-3, 4, -5});
+  ASSERT_EQ(out.size(), 3u);
+  EXPECT_DOUBLE_EQ(out[0].y, 3.0);
+  EXPECT_DOUBLE_EQ(out[2].y, 5.0);
+}
+
+TEST(Transforms, ScaleAndOffset) {
+  ScaleTransform t;
+  t.value_scale = 2.0;
+  t.value_offset = 1.0;
+  t.time_offset = 10.0;
+  auto out = apply(t, {0, 1}, {3, 4});
+  EXPECT_DOUBLE_EQ(out[0].x, 10.0);
+  EXPECT_DOUBLE_EQ(out[0].y, 7.0);
+  EXPECT_DOUBLE_EQ(out[1].y, 9.0);
+}
+
+TEST(Transforms, DerivativeActualDt) {
+  DerivativeTransform t;
+  auto out = apply(t, {0, 1, 2}, {0, 10, 30});
+  ASSERT_EQ(out.size(), 2u);
+  EXPECT_DOUBLE_EQ(out[0].y, 10.0);
+  EXPECT_DOUBLE_EQ(out[1].y, 20.0);
+}
+
+TEST(Transforms, IntegralTrapezoid) {
+  IntegralTransform t;
+  auto out = apply(t, {0, 1, 2}, {0, 2, 4});
+  ASSERT_EQ(out.size(), 2u);
+  EXPECT_DOUBLE_EQ(out[0].y, 1.0);
+  EXPECT_DOUBLE_EQ(out[1].y, 4.0);
+}
+
+TEST(Transforms, MovingAverageWindow) {
+  MovingAverageTransform t;
+  t.window = 2;
+  auto out = apply(t, {0, 1, 2, 3}, {2, 4, 6, 8});
+  ASSERT_EQ(out.size(), 4u);
+  EXPECT_DOUBLE_EQ(out[0].y, 2.0);  // padded
+  EXPECT_DOUBLE_EQ(out[1].y, 3.0);  // (2+4)/2
+  EXPECT_DOUBLE_EQ(out[3].y, 7.0);  // (6+8)/2
+}
+
+TEST(Transforms, MovingRMS) {
+  MovingRMSTransform t;
+  t.window = 2;
+  auto out = apply(t, {0, 1}, {3, 4});
+  EXPECT_DOUBLE_EQ(out[0].y, 3.0);
+  EXPECT_DOUBLE_EQ(out[1].y, std::sqrt((9.0 + 16.0) / 2.0));
+}
+
+TEST(Transforms, BinaryFilterGreater) {
+  BinaryFilterTransform t;
+  t.op = BinaryOp::kGreater;
+  t.a = 5.0;
+  auto out = apply(t, {0, 1, 2}, {3, 5, 7});
+  EXPECT_DOUBLE_EQ(out[0].y, 0.0);
+  EXPECT_DOUBLE_EQ(out[1].y, 0.0);
+  EXPECT_DOUBLE_EQ(out[2].y, 1.0);
+}
+
+TEST(Transforms, BinaryFilterRange) {
+  BinaryFilterTransform t;
+  t.op = BinaryOp::kRange;
+  t.a = 2.0;
+  t.b = 6.0;
+  auto out = apply(t, {0, 1, 2}, {1, 4, 7});
+  EXPECT_DOUBLE_EQ(out[0].y, 0.0);
+  EXPECT_DOUBLE_EQ(out[1].y, 1.0);
+  EXPECT_DOUBLE_EQ(out[2].y, 0.0);
+}
+
+TEST(Transforms, TimeSincePrevious) {
+  TimeSincePreviousTransform t;
+  auto out = apply(t, {0, 2, 5}, {9, 9, 9});
+  ASSERT_EQ(out.size(), 2u);
+  EXPECT_DOUBLE_EQ(out[0].y, 2.0);
+  EXPECT_DOUBLE_EQ(out[1].y, 3.0);
+}
+
+TEST(Transforms, SamplesCounter) {
+  SamplesCounterTransform t;
+  t.samples_ms = 2000;
+  auto out = apply(t, {0, 1, 2, 3}, {0, 0, 0, 0});
+  ASSERT_EQ(out.size(), 4u);
+  EXPECT_DOUBLE_EQ(out[0].y, 0.0);
+  EXPECT_DOUBLE_EQ(out[2].y, 2.0);
+}
+
+TEST(Transforms, NoneIsPassthrough) {
+  NoneTransform t;
+  auto out = apply(t, {0, 1}, {5, 6});
+  ASSERT_EQ(out.size(), 2u);
+  EXPECT_DOUBLE_EQ(out[1].y, 6.0);
+}
+
+TEST(Transforms, FactoryRegistration) {
+  registerAllTransforms();
+  auto& f = FilterTransformFactory::instance();
+  const auto ids = f.registeredIds();
+  EXPECT_FALSE(ids.empty());
+  // All registered ids should be creatable
+  for (const auto& id : ids) {
+    auto t = f.create(id);
+    EXPECT_NE(t, nullptr) << "Could not create: " << id;
+  }
+}
+
+TEST(Transforms, StreamSafeFlags) {
+  registerAllTransforms();
+  // Known stream-safe
+  EXPECT_TRUE(AbsoluteTransform{}.isStreamSafe());
+  EXPECT_TRUE(ScaleTransform{}.isStreamSafe());
+  EXPECT_TRUE(DerivativeTransform{}.isStreamSafe());
+  EXPECT_TRUE(MovingAverageTransform{}.isStreamSafe());
+  EXPECT_TRUE(MovingRMSTransform{}.isStreamSafe());
+  EXPECT_TRUE(MovingVarianceTransform{}.isStreamSafe());
+  EXPECT_TRUE(OutlierRemovalTransform{}.isStreamSafe());
+  EXPECT_TRUE(BinaryFilterTransform{}.isStreamSafe());
+  EXPECT_TRUE(TimeSincePreviousTransform{}.isStreamSafe());
+  // Known non-safe (need full history)
+  EXPECT_FALSE(IntegralTransform{}.isStreamSafe());
+  EXPECT_FALSE(SamplesCounterTransform{}.isStreamSafe());
+}
+
+// ---------------------------------------------------------------------------
+// Streaming equivalence: calculateNextPoint matches applyBatch
+// ---------------------------------------------------------------------------
+
+void expectStreamingMatchesBatch(FilterTransform& t_batch, FilterTransform& t_stream) {
+  std::vector<Point2> full;
+  for (int i = 0; i < 30; ++i) {
+    full.push_back({static_cast<double>(i), std::sin(i * 0.3) * 5.0 + static_cast<double>(i % 5)});
+  }
+  // Batch
+  const auto batch = t_batch.applyBatch(full);
+  // Streaming point-by-point
+  t_stream.reset();
+  std::vector<Point2> streamed;
+  for (const auto& p : full) {
+    if (auto r = t_stream.calculateNextPoint(p)) {
+      streamed.push_back(*r);
+    }
+  }
+  ASSERT_EQ(batch.size(), streamed.size());
+  for (size_t i = 0; i < batch.size(); ++i) {
+    EXPECT_NEAR(batch[i].x, streamed[i].x, 1e-9) << "x mismatch at " << i;
+    EXPECT_NEAR(batch[i].y, streamed[i].y, 1e-9) << "y mismatch at " << i;
+  }
+}
+
+TEST(Transforms, StreamingMatchesBatch) {
+  {
+    AbsoluteTransform a, b;
+    expectStreamingMatchesBatch(a, b);
+  }
+  {
+    DerivativeTransform a, b;
+    expectStreamingMatchesBatch(a, b);
+  }
+  {
+    IntegralTransform a, b;
+    expectStreamingMatchesBatch(a, b);
+  }
+  {
+    MovingAverageTransform a, b;
+    a.window = b.window = 5;
+    expectStreamingMatchesBatch(a, b);
+  }
+  {
+    MovingRMSTransform a, b;
+    a.window = b.window = 5;
+    expectStreamingMatchesBatch(a, b);
+  }
+  {
+    TimeSincePreviousTransform a, b;
+    expectStreamingMatchesBatch(a, b);
+  }
+  {
+    BinaryFilterTransform a, b;
+    a.op = b.op = BinaryOp::kGreater;
+    a.a = b.a = 2.0;
+    expectStreamingMatchesBatch(a, b);
+  }
+}
+
+}  // namespace

toolbox_filter_editor/conanfile.py

@@ -0,0 +1,24 @@
+import os
+from conan import ConanFile
+
+
+_SDK_VERSION = (
+    open(os.path.join(os.path.dirname(os.path.abspath(__file__)), os.pardir, "SDK_VERSION"))
+    .read()
+    .strip()
+)
+
+
+class ToolboxFilterEditorConan(ConanFile):
+    name = "toolbox_filter_editor"
+    version = "0"
+    settings = "os", "compiler", "build_type", "arch"
+    generators = "CMakeDeps", "CMakeToolchain"
+    requires = (
+        f"plotjuggler_sdk/{_SDK_VERSION}",
+        "gtest/1.17.0",
+        "nlohmann_json/3.12.0",
+    )
+    default_options = {
+        "*:shared": False,
+    }