New scoped_view API

docs/setup-code.org

@@ -1,3 +1,4 @@
 #+HTML_HEAD: <style>pre.src{border-radius:5px;background:#343131;color:white;} </style>
 #+HTML_HEAD: <style>pre.example{border-radius:5px;background:#454242;color:white;} </style>
 #+HTML_HEAD: <style>code{border-radius:5px;background:#454242;color:#0f0;} </style>
+#+latex_header: \usepackage[dvipsnames]{xcolor}

docs/setup-listings.org

@@ -1,4 +1,8 @@
 
+# if you get weird error like:
+# json-read: Invalid number format: 5
+# then "pip install jupyter".  nfc.
+
 
 #+begin_export latex
 \colorlet{punct}{red!60!black}

gen/src/Reframer.cxx

@@ -152,7 +152,7 @@ bool Gen::Reframer::operator()(const input_pointer& inframe, output_pointer& out
         out_traces = process_one(*(inframe->traces()));
     }
     else {
-        for (const auto tag : m_input_tags) {
+        for (const auto& tag : m_input_tags) {
             const auto& isummary = inframe->trace_summary(tag);
             ITrace::vector in_traces = Aux::tagged_traces(inframe, tag);
             auto [out_one, threshold] = process_one(in_traces, isummary);
@@ -171,7 +171,7 @@ bool Gen::Reframer::operator()(const input_pointer& inframe, output_pointer& out
         sframe->tag_frame(m_frame_tag);
     }
 
-    for (const auto tag : m_input_tags) {
+    for (const auto& tag : m_input_tags) {
         sframe->tag_traces(tag, tag_indicies.at(tag), tag_summary[tag]);
     }
 

img/test/doctest_clustering_prototype.cxx

@@ -16,12 +16,14 @@ using spdlog::debug;
 
 using node_ptr = std::unique_ptr<Points::node_t>;
 
-static void print_dds(const scoped_pointcloud_t& dds) {
+// No more explicit DisjointDataset.  It is a PointCloud::Tree::scoped_pointcloud_t.
+template <typename DisjointDataset>
+void print_dds(const DisjointDataset& dds) {
     for (size_t idx=0; idx<dds.size(); ++idx) {
         const Dataset& ds = dds[idx];
         std::stringstream ss;
         ss << "ds: " << idx << std::endl;
-        const size_t len = ds.size_major();
+        // const size_t len = ds.size_major();
         for (const auto& key : ds.keys()) {
             auto arr = ds.get(key)->elements<double>();
             ss << key << ": ";
@@ -44,8 +46,14 @@ Points::node_ptr make_simple_pctree()
 
     // Insert a child with a set of named points clouds with one point
     // cloud from a track.
+
     /// QUESTION: can only do this on construction?
+    /// bv: see NaryTree.h for several insert()'s
+
     /// QUESTION: units?
+    /// bv: units are always assumed in WCT system-of-units.  To be correct
+    ///     here, we should be multiplying by some [length] unit.
+
     auto* n1 = root->insert(Points({
         {"center", make_janky_track(Ray(Point(0.5, 0, 0), Point(0.7, 0, 0)))},
         {"3d", make_janky_track(Ray(Point(0, 0, 0), Point(1, 0, 0)))}
@@ -103,27 +111,34 @@ TEST_CASE("PointCloudFacade test")
 
     // name, coords, [depth]
     Scope scope{ "3d", {"x","y","z"}};
-    const scoped_pointcloud_t& pc3d = rval.scoped_pc(scope);
+    auto const& s3d = rval.scoped_view({ "3d", {"x","y","z"}});
+
+    auto const& pc3d = s3d.pcs();
     CHECK(pc3d.size() == 2);
     print_dds(pc3d);
 
-    const scoped_pointcloud_t& pccenter = rval.scoped_pc({ "center", {"x","y","z"}});
+    auto const& pccenter = rval.scoped_view({ "center", {"x","y","z"}}).pcs();
     print_dds(pccenter);
 
-    const auto& skd = rval.scoped_kd<double>(scope);
-    // CHECK(&kd.pointclouds() == &pc3d);
+    const auto& kd = s3d.kd();
 
     /// QUESTION: how to get it -> node?
-    const std::vector<double> origin = {1,0,0};
-    auto knn = skd.knn(2, origin);
+    /// bv: get the "major index" of the iterator and use that to index in whatever "node like" container
+    ///     auto pts = kd.points();
+    ///     size_t ind = pts.major_index(it);
+    ///     auto& thing = vector_of_node_like_things[ind];
+    /// see doctest-pointtree-example for details.
+
+    std::vector<double> some_point = {1, 0, 0};
+    auto knn = kd.knn(2, some_point);
     for (auto [it,dist] : knn) {
         auto& pt = *it;
         debug("knn: pt=({},{},{}) dist={}",
               pt[0], pt[1], pt[2], dist);
     }
     CHECK(knn.size() == 2);
 
-    const auto& all_points = skd.points();
+    const auto& all_points = kd.points();
     for (size_t pt_ind = 0; pt_ind<knn.size(); ++pt_ind) {
         auto& [pit,dist] = knn[pt_ind];
         const size_t maj_ind = all_points.major_index(pit);
@@ -136,12 +151,12 @@ TEST_CASE("PointCloudFacade test")
         }
     }
 
-    auto rad = skd.radius(.01, origin);
+    auto rad = kd.radius(.01, some_point);
     for (auto [it,dist] : rad) {
         auto& pt = *it;
         debug("rad: pt=({},{},{}) dist={}",
               pt[0], pt[1], pt[2], dist);
     }
     CHECK(rad.size() == 2);
 
-}
+}

util/inc/WireCellUtil/NFKD.h

@@ -119,6 +119,19 @@ namespace WireCell::NFKD {
         points_t& points() { return m_points; }
         const points_t& points() const { return m_points; }
 
+        size_t major_index(iterator it) const {
+            return m_points.major_index(it);
+        }
+        size_t major_index(const_iterator it) const {
+            return m_points.major_index(it);
+        }
+        size_t minor_index(iterator it) const {
+            return m_points.minor_index(it);
+        }
+        size_t minor_index(const_iterator it) const {
+            return m_points.minor_index(it);
+        }
+
         // Return the number calls made so far to resolve a point
         // coordinate.  Mostly for debugging/perfing.
         size_t point_calls() const {