Robustify graph build and path guess: direct coords, multiline handling

- build_graph_geos: extract start/end coordinates directly from linestrings
  via wk::wk_coords instead of geos_point_start/geos_point_end to avoid any
  precision drift; filter non-linestrings at entry
- cnt_path_master: unnest multilinestrings before graph build
- cnt_path_guess_master: add edge-count filter (>1 edge) to match old
  sfnetworks behaviour; if geos_line_merge returns multilinestring, keep
  the longest component

Author: Developer

R/cnt_path.R

@@ -187,8 +187,12 @@ cnt_path.SpatVector <-
 
 cnt_path_master <-
   function(skeleton_sf, start_point_sf, end_point_sf) {
-    # Build graph from geos linestrings
+    # Convert to geos and ensure only linestrings
     skeleton_geos <- geos::as_geos_geometry(skeleton_sf)
+    if (any(geos::geos_type(skeleton_geos) == "multilinestring")) {
+      skeleton_geos <- geos::geos_unnest(skeleton_geos, keep_multi = FALSE)
+    }
+
     graph <- build_graph_geos(skeleton_geos)
 
     # Find nearest graph nodes for start and end points

R/cnt_path_guess.R

@@ -238,7 +238,7 @@ cnt_path_guess.SpatVector <-
 
 cnt_path_guess_master <-
   function(skeleton_geos) {
-    if (geos::geos_type(skeleton_geos) == "multilinestring") {
+    if (any(geos::geos_type(skeleton_geos) == "multilinestring")) {
       skeleton_geos <-
         geos::geos_unnest(skeleton_geos, keep_multi = FALSE)
     }
@@ -273,7 +273,7 @@ cnt_path_guess_master <-
     all_dist <- dijkstra(graph, from = closest_end_points)
 
     other_dists <- all_dist$dist[other_nodes]
-    valid <- is.finite(other_dists)
+    valid <- is.finite(other_dists) & other_dists > 0
 
     if (!any(valid)) {
       return(
@@ -283,7 +283,23 @@ cnt_path_guess_master <-
       )
     }
 
-    target_node <- other_nodes[which.max(other_dists)]
+    # Filter to paths with > 1 edge (matching old sfnetworks behaviour)
+    # and pick the longest among them
+    candidate_nodes <- other_nodes[valid]
+    candidate_dists <- other_dists[valid]
+
+    # Count edges in each path by reconstructing it
+    edge_counts <- vapply(candidate_nodes, function(t) {
+      p <- dijkstra(graph, from = closest_end_points, to = t)
+      if (is.null(p)) 0L else length(p$edges)
+    }, FUN.VALUE = integer(1))
+
+    long_enough <- edge_counts > 1L
+    if (!any(long_enough)) {
+      long_enough <- edge_counts > 0L
+    }
+
+    target_node <- candidate_nodes[long_enough][which.max(candidate_dists[long_enough])]
 
     # Reconstruct shortest path to the farthest outer node
     path <- dijkstra(graph, from = closest_end_points, to = target_node)
@@ -301,5 +317,13 @@ cnt_path_guess_master <-
       geos::geos_make_collection() |>
       geos::geos_line_merge()
 
+    # If line_merge couldn't produce a single linestring (e.g. tiny coord
+    # gaps), keep the longest component
+    if (geos::geos_type(longest_path_geos) == "multilinestring") {
+      pieces <- geos::geos_unnest(longest_path_geos, keep_multi = FALSE)
+      lens <- geos::geos_length(pieces)
+      longest_path_geos <- pieces[which.max(lens)]
+    }
+
     longest_path_geos
   }

R/graph.R

@@ -6,21 +6,37 @@
 #' Endpoints are deduplicated using coordinate hashing (tolerance ~1e-10).
 #' @noRd
 build_graph_geos <- function(lines) {
-  starts <- geos::geos_point_start(lines)
-  ends   <- geos::geos_point_end(lines)
-  all_points <- c(starts, ends)
+  # Keep only linestrings
+  is_line <- geos::geos_type(lines) == "linestring"
+  if (!all(is_line)) {
+    lines <- lines[is_line]
+  }
   n_edges <- length(lines)
+  if (n_edges == 0) {
+    stop("No linestring edges in skeleton.")
+  }
+
+  # Extract start/end coordinates directly from the linestrings.
+  # This avoids any precision drift from geos_point_start / geos_point_end.
+  coords <- wk::wk_coords(lines)
+
+  # First coordinate of each linestring = start, last = end
+  first_rows <- !duplicated(coords$feature_id)
+  last_rows  <- !duplicated(coords$feature_id, fromLast = TRUE)
 
-  # Deduplicate nodes via rounded coordinates.
-  # Exact geos_equals can fail for shared Voronoi vertices due to tiny
-  # floating-point differences introduced during intersection/clipping.
-  coords <- wk::wk_coords(all_points)
-  rounded <- paste(
-    format(round(coords$x, 10), scientific = FALSE, trim = TRUE),
-    format(round(coords$y, 10), scientific = FALSE, trim = TRUE),
+  start_key <- paste(
+    format(round(coords$x[first_rows], 10), scientific = FALSE, trim = TRUE),
+    format(round(coords$y[first_rows], 10), scientific = FALSE, trim = TRUE),
     sep = ","
   )
-  node_ids <- match(rounded, unique(rounded))
+  end_key <- paste(
+    format(round(coords$x[last_rows], 10), scientific = FALSE, trim = TRUE),
+    format(round(coords$y[last_rows], 10), scientific = FALSE, trim = TRUE),
+    sep = ","
+  )
+
+  all_key <- c(start_key, end_key)
+  node_ids <- match(all_key, unique(all_key))
   n_nodes <- max(node_ids)
 
   from <- node_ids[seq_len(n_edges)]
@@ -37,6 +53,9 @@ build_graph_geos <- function(lines) {
   }
 
   # One representative point per unique node
+  starts <- geos::geos_point_start(lines)
+  ends   <- geos::geos_point_end(lines)
+  all_points <- c(starts, ends)
   first_idx <- match(seq_len(n_nodes), node_ids)
   node_points <- all_points[first_idx]