From ffba176060d811dadf7c61f98559ab252d5a1a77 Mon Sep 17 00:00:00 2001
From: BobLd <38405645+BobLd@users.noreply.github.com>
Date: Sun, 12 Oct 2025 20:17:34 +0100
Subject: [PATCH] Improve GroupIndexes() performance with #1178

---
 .../Clustering.cs                             | 74 +++++--------------
 1 file changed, 20 insertions(+), 54 deletions(-)

diff --git a/src/UglyToad.PdfPig.DocumentLayoutAnalysis/Clustering.cs b/src/UglyToad.PdfPig.DocumentLayoutAnalysis/Clustering.cs
index bd11c7b5..cc39470e 100644
--- a/src/UglyToad.PdfPig.DocumentLayoutAnalysis/Clustering.cs
+++ b/src/UglyToad.PdfPig.DocumentLayoutAnalysis/Clustering.cs
@@ -216,25 +216,26 @@
                 yield return group.Select(i => elements[i]).ToList();
             }
         }
-
-        /// <summary>
-        /// Group elements using Depth-first search.
-        /// <para>https://en.wikipedia.org/wiki/Depth-first_search</para>
-        /// </summary>
-        /// <param name="edges">The graph. edges[i] = j indicates that there is an edge between i and j.</param>
-        /// <returns>A List of HashSets containing the grouped indexes.</returns>
+        
         internal static List<HashSet<int>> GroupIndexes(int[] edges)
         {
-            int[][] adjacency = new int[edges.Length][];
+            // Improved thanks to https://github.com/UglyToad/PdfPig/issues/1178
+            var adjacency = new List<int>[edges.Length];
             for (int i = 0; i < edges.Length; i++)
             {
-                HashSet<int> matches = new HashSet<int>();
-                if (edges[i] != -1) matches.Add(edges[i]);
-                for (int j = 0; j < edges.Length; j++)
+                adjacency[i] = new List<int>();
+            }
+
+            // one pass O(n) 
+            for (int i = 0; i < edges.Length; i++)
+            {
+                int j = edges[i];
+                if (j != -1)
                 {
-                    if (edges[j] == i) matches.Add(j);
+                    // i <-> j
+                    adjacency[i].Add(j);
+                    adjacency[j].Add(i);
                 }
-                adjacency[i] = matches.ToArray();
             }
 
             List<HashSet<int>> groupedIndexes = new List<HashSet<int>>();
@@ -242,55 +243,20 @@
 
             for (int p = 0; p < edges.Length; p++)
             {
-                if (isDone[p]) continue;
+                if (isDone[p])
+                {
+                    continue;
+                }
                 groupedIndexes.Add(DfsIterative(p, adjacency, ref isDone));
             }
             return groupedIndexes;
         }
-
-        /// <summary>
-        /// Group elements using Depth-first search.
-        /// <para>https://en.wikipedia.org/wiki/Depth-first_search</para>
-        /// </summary>
-        /// <param name="edges">The graph. edges[i] = [j, k, l, ...] indicates that there is an edge between i and each element j, k, l, ...</param>
-        /// <returns>A List of HashSets containing the grouped indexes.</returns>
-        internal static List<HashSet<int>> GroupIndexes(int[][] edges)
-        {
-            int[][] adjacency = new int[edges.Length][];
-            for (int i = 0; i < edges.Length; i++)
-            {
-                HashSet<int> matches = new HashSet<int>();
-                for (int j = 0; j < edges[i].Length; j++)
-                {
-                    if (edges[i][j] != -1) matches.Add(edges[i][j]);
-                }
-
-                for (int j = 0; j < edges.Length; j++)
-                {
-                    for (int k = 0; k < edges[j].Length; k++)
-                    {
-                        if (edges[j][k] == i) matches.Add(j);
-                    }
-                }
-                adjacency[i] = matches.ToArray();
-            }
-
-            List<HashSet<int>> groupedIndexes = new List<HashSet<int>>();
-            bool[] isDone = new bool[edges.Length];
-
-            for (int p = 0; p < edges.Length; p++)
-            {
-                if (isDone[p]) continue;
-                groupedIndexes.Add(DfsIterative(p, adjacency, ref isDone));
-            }
-            return groupedIndexes;
-        }
-
+        
         /// <summary>
         /// Depth-first search
         /// <para>https://en.wikipedia.org/wiki/Depth-first_search</para>
         /// </summary>
-        private static HashSet<int> DfsIterative(int s, int[][] adj, ref bool[] isDone)
+        private static HashSet<int> DfsIterative(int s, List<int>[] adj, ref bool[] isDone)
         {
             HashSet<int> group = new HashSet<int>();
             Stack<int> S = new Stack<int>();