Merge branch 'main' into CURA-11622_conan_v2

Author: wawanbreton

.github/workflows/npm-package.yml

@@ -86,6 +86,7 @@ jobs:
 
       - name: Set npm config
         run: |
+          cd CuraEngineJS
           npm run install_curaengine
           npm ci
           npm publish

include/PathOrderOptimizer.h

@@ -784,17 +784,20 @@ class PathOrderOptimizer
         best_candidate_finder.appendCriteriaPass(main_criteria_pass);
 
         // ########## Step 2: add fallback passes for criteria with very similar scores (e.g. corner on a cylinder)
-        const AABB path_bounding_box(points);
+        if (path.seam_config_.type_ == EZSeamType::SHARPEST_CORNER)
+        {
+            const AABB path_bounding_box(points);
 
-        { // First fallback strategy is to take points on the back-most position
-            auto fallback_criterion = std::make_shared<DistanceScoringCriterion>(points, path_bounding_box.max_, DistanceScoringCriterion::DistanceType::YOnly);
-            constexpr double outsider_delta_threshold = 0.01;
-            best_candidate_finder.appendSingleCriterionPass(fallback_criterion, outsider_delta_threshold);
-        }
+            { // First fallback strategy is to take points on the back-most position
+                auto fallback_criterion = std::make_shared<DistanceScoringCriterion>(points, path_bounding_box.max_, DistanceScoringCriterion::DistanceType::YOnly);
+                constexpr double outsider_delta_threshold = 0.01;
+                best_candidate_finder.appendSingleCriterionPass(fallback_criterion, outsider_delta_threshold);
+            }
 
-        { // Second fallback strategy, in case we still have multiple points that are aligned on Y (e.g. cube), take the right-most point
-            auto fallback_criterion = std::make_shared<DistanceScoringCriterion>(points, path_bounding_box.max_, DistanceScoringCriterion::DistanceType::XOnly);
-            best_candidate_finder.appendSingleCriterionPass(fallback_criterion);
+            { // Second fallback strategy, in case we still have multiple points that are aligned on Y (e.g. cube), take the right-most point
+                auto fallback_criterion = std::make_shared<DistanceScoringCriterion>(points, path_bounding_box.max_, DistanceScoringCriterion::DistanceType::XOnly);
+                best_candidate_finder.appendSingleCriterionPass(fallback_criterion);
+            }
         }
 
         // ########## Step 3: apply the criteria to find the vertex with the best global score

include/utils/scoring/BestElementFinder.h

@@ -15,7 +15,7 @@ class ScoringCriterion;
 
 /*!
  * This class implements an algorithm to find an element amongst a list, regarding one or multiple scoring criteria. The
- * criteria are implemented by subclassing the CriterionScoring class. It is also possible to setup multiple passes of
+ * criteria are implemented by subclassing the CriterionScoring class. It is also possible to set up multiple passes of
  * criteria. Thus, if the first pass gives a few best candidates that are too close to each other, we keep only the best
  * candidates and process a second pass with different criteria, and so on until we have a single outsider, or no more
  * criteria.
@@ -60,6 +60,7 @@ class BestElementFinder
          * Once we have calculated the global scores of each element for this pass, we calculate the score difference
          * between the best candidate and the following ones. If the following ones have a score close enough to the
          * best, within this threshold, they will also be considered outsiders and will be run for the next pass.
+         * This value will be ignored for the last pass.
          */
         double outsider_delta_threshold{ 0.0 };
     };

src/utils/scoring/BestElementFinder.cpp

@@ -5,6 +5,8 @@
 
 #include <limits>
 
+#include <range/v3/view/enumerate.hpp>
+
 #include "utils/scoring/ScoringCriterion.h"
 
 namespace cura
@@ -33,11 +35,11 @@ std::optional<size_t> cura::BestElementFinder::findBestElement(const size_t cand
     const auto begin = best_candidates.begin();
     auto end = best_candidates.end();
 
-    // Now run the criteria passes until we have a single outsider or no more cirteria
-    for (const CriteriaPass& criteria_pass : criteria_)
+    // Now run the criteria passes until we have a single outsider or no more criteria
+    for (const auto& [pass_index, criteria_pass] : criteria_ | ranges::views::enumerate)
     {
         // For each element, reset score, process each criterion and apply weights to get the global score
-        double best_score = 0.0;
+        auto best_candidate_iterator = end;
         for (auto iterator = begin; iterator != end; ++iterator)
         {
             iterator->score = 0.0;
@@ -47,17 +49,32 @@ std::optional<size_t> cura::BestElementFinder::findBestElement(const size_t cand
                 iterator->score += weighed_criterion.criterion->computeScore(iterator->candidate_index) * weighed_criterion.weight;
             }
 
-            best_score = std::max(best_score, iterator->score);
+            if (best_candidate_iterator == end || iterator->score > best_candidate_iterator->score)
+            {
+                best_candidate_iterator = iterator;
+            }
+        }
+
+        if (best_candidate_iterator == end)
+        {
+            // Something went wrong, we don't have a best candidate
+            return std::nullopt;
+        }
+
+        // Early out for last pass, just keep the best candidate
+        if (pass_index == criteria_.size() - 1)
+        {
+            return best_candidate_iterator->candidate_index;
         }
 
         // Skip candidates that have a score too far from the actual best one
         const double delta_threshold = criteria_pass.outsider_delta_threshold + std::numeric_limits<double>::epsilon();
         end = std::remove_if(
             begin,
             end,
-            [&best_score, &delta_threshold](const Candidate& candidate)
+            [&best_candidate_iterator, &delta_threshold](const Candidate& candidate)
             {
-                return best_score - candidate.score > delta_threshold;
+                return best_candidate_iterator->score - candidate.score > delta_threshold;
             });
 
         if (std::distance(begin, end) == 1)