First attempt at finding duplicate files. For now, and possible ever, this process will be a 'best effort' approach, so there will be false positives but this will most likely remain the case for the life of this program.

2020-12-14 22:48:32 -06:00
parent 06b586d635
commit 95766b044f
5 changed files with 137 additions and 13 deletions
@@ -272,9 +272,41 @@ namespace DataOrganizer
            return renames;
        }

+        public List<Duplicate> FindDuplicates(Folder topLevelFolder)
+        {
+            var duplicates = new Dictionary<string, Duplicate>();
+            var files = topLevelFolder.GetAllFiles();
+            var fileCount = 0;
+
+            fileCount = files.Count;
+
+            for(var i = 0; i < fileCount; i++)
+            {
+                var file = files[i];
+
+                //if (!duplicates.ContainsKey(file.FullPath)) duplicates.Add(file.FullPath, new Duplicate() { File = file, Duplicates = new List<FileSnapshot>() });
+
+                var dups = files.Where(x => x.Size == file.Size && x.FileId != file.FileId).ToList();
+                var dupCount = dups.Count;
+
+                if(dupCount > 0)
+                    if (!duplicates.ContainsKey(file.FullPath)) duplicates.Add(file.FullPath, new Duplicate() { File = file, Duplicates = new List<FileSnapshot>() });
+
+                for (var j = 0; j < dupCount; j++)
+                {
+                    var dup = dups[j];
+                    files.Remove(dup);
+                    duplicates[file.FullPath].Duplicates.Add(dup);
+                    fileCount--;
+                }
+            }
+
+            return duplicates.Values.ToList();
+        }
+
        public struct Duplicate
        {
-            public DB.DBFileSnapshot File;
+            public FileSnapshot File;
            public List<FileSnapshot> Duplicates;
        }
    }