DYN-6053-Adjust-Lucene-Weights (#14223)

RobertGlobant20 · web-flow · commit 9947ba9d1188 · 2023-08-03T22:41:59.000-04:00
In Dynamo Revit is not returning the expected results when searching by category so in this fix instead of assigning a weight for all the fields (except for name) we will be assigning a weight individually for Name, Category, Description and SearchTags.
diff --git a/src/DynamoCore/Configuration/LuceneConfig.cs b/src/DynamoCore/Configuration/LuceneConfig.cs
@@ -49,26 +49,69 @@ internal class LuceneConfig
         /// </summary>
         internal static int DefaultResultsCount = 50;
 
+
+        #region Field Weights
         /// <summary>
         /// Search name matching weight
         /// </summary>
         internal static int SearchNameWeight = 10;
 
+        /// <summary>
+        /// Search Category matching weight
+        /// </summary>
+        internal static int SearchCategoryWeight = 9;
+
+        /// <summary>
+        /// Search Description matching weight
+        /// </summary>
+        internal static int SearchDescriptionWeight = 6;
+
+        /// <summary>
+        /// Search tags matching weight
+        /// </summary>
+        internal static int SearchTagsWeight = 6;
+
+        /// <summary>
+        /// other fields search matching weight
+        /// </summary>
+        internal static int SearchMetaFieldsWeight = 6;
+
+        #endregion
+
+
+        #region Wildcards Field Weights
         /// <summary>
         /// Wildcards search name matching weight
         /// </summary>
         internal static int WildcardsSearchNameWeight = 7;
 
         /// <summary>
-        /// Search non-name meta fields matching weight
+        /// Wildcards search Category matching weight
         /// </summary>
-        internal static int SearchMetaFieldsWeight = 6;
+        internal static int WildcardsSearchCategoryWeight = 6;
+
+        /// <summary>
+        /// Wildcards search Description matching weight
+        /// </summary>
+        internal static int WildcardsSearchDescriptionWeight = 4;
+
+        /// <summary>
+        /// Wildcards search tags matching weight
+        /// </summary>
+        internal static int WildcardsSearchTagsWeight = 4;
 
         /// <summary>
-        /// Wildcards search non-name meta fields matching weight
+        /// other wildcards fields search matching weight
         /// </summary>
         internal static int WildcardsSearchMetaFieldsWeight = 4;
 
+        /// <summary>
+        /// Wildcards search name matching weight
+        /// </summary>
+        internal static int WildcardsSearchNameParsedWeight = 5;
+
+        #endregion
+
         /// <summary>
         /// Fuzzy search matching weight
         /// </summary>
diff --git a/src/DynamoCore/Utilities/LuceneSearchUtility.cs b/src/DynamoCore/Utilities/LuceneSearchUtility.cs
@@ -218,32 +218,18 @@ internal string CreateSearchQuery(string[] fields, string SearchTerm)
                     booleanQuery.Add(fuzzyQuery, Occur.SHOULD);
                 }
 
-                var wildcardQuery = new WildcardQuery(new Term(f, searchTerm));
-                if (f.Equals(nameof(LuceneConfig.NodeFieldsEnum.Name)))
-                {
-                    wildcardQuery.Boost = LuceneConfig.SearchNameWeight;
-                }
-                else
-                {
-                    wildcardQuery.Boost = LuceneConfig.SearchMetaFieldsWeight;
-                }
-                booleanQuery.Add(wildcardQuery, Occur.SHOULD);
+                var fieldQuery = CalculateFieldWeight(f, searchTerm);
+                var wildcardQuery = CalculateFieldWeight(f, searchTerm, true);
 
-                wildcardQuery = new WildcardQuery(new Term(f, "*" + searchTerm + "*"));
-                if (f.Equals(nameof(LuceneConfig.NodeFieldsEnum.Name)))
-                {
-                    wildcardQuery.Boost = LuceneConfig.WildcardsSearchNameWeight;
-                }
-                else
-                {
-                    wildcardQuery.Boost = LuceneConfig.WildcardsSearchMetaFieldsWeight;
-                }
+                booleanQuery.Add(fieldQuery, Occur.SHOULD);
                 booleanQuery.Add(wildcardQuery, Occur.SHOULD);
 
                 if (searchTerm.Contains(' ') || searchTerm.Contains('.'))
                 {
                     foreach (string s in searchTerm.Split(' ', '.'))
                     {
+                        if (string.IsNullOrEmpty(s)) continue;
+
                         if (s.Length > LuceneConfig.FuzzySearchMinimalTermLength)
                         {
                             fuzzyQuery = new FuzzyQuery(new Term(f, s), LuceneConfig.FuzzySearchMinEdits);
@@ -253,7 +239,7 @@ internal string CreateSearchQuery(string[] fields, string SearchTerm)
 
                         if (f.Equals(nameof(LuceneConfig.NodeFieldsEnum.Name)))
                         {
-                            wildcardQuery.Boost = 5;
+                            wildcardQuery.Boost = LuceneConfig.WildcardsSearchNameParsedWeight;
                         }
                         else
                         {
@@ -266,6 +252,39 @@ internal string CreateSearchQuery(string[] fields, string SearchTerm)
             return booleanQuery.ToString();
         }
 
+        private WildcardQuery CalculateFieldWeight(string fieldName, string searchTerm, bool isWildcard = false)
+        {
+            WildcardQuery query;
+
+            query = isWildcard == false ?
+                new WildcardQuery(new Term(fieldName, searchTerm)) : new WildcardQuery(new Term(fieldName, "*" + searchTerm + "*"));
+
+            switch (fieldName)
+            {
+                case nameof(LuceneConfig.NodeFieldsEnum.Name):
+                    query.Boost = isWildcard == false?
+                        LuceneConfig.SearchNameWeight :  LuceneConfig.WildcardsSearchNameWeight;
+                    break;
+                case nameof(LuceneConfig.NodeFieldsEnum.FullCategoryName):
+                    query.Boost = isWildcard == false?
+                        LuceneConfig.SearchCategoryWeight : LuceneConfig.WildcardsSearchCategoryWeight;
+                    break;
+                case nameof(LuceneConfig.NodeFieldsEnum.Description):
+                    query.Boost = isWildcard == false ?
+                        LuceneConfig.SearchDescriptionWeight : LuceneConfig.WildcardsSearchDescriptionWeight;
+                    break;
+                case nameof(LuceneConfig.NodeFieldsEnum.SearchKeywords):
+                    query.Boost = isWildcard == false ?
+                       LuceneConfig.SearchTagsWeight : LuceneConfig.WildcardsSearchTagsWeight;
+                    break;
+                default:
+                    query.Boost = isWildcard == false ?
+                       LuceneConfig.SearchMetaFieldsWeight : LuceneConfig.WildcardsSearchMetaFieldsWeight;
+                    break;
+            }
+            return query;
+        }
+
         internal void DisposeWriter()
         {
             //We need to check if we are not running Dynamo tests because otherwise parallel test start to fail when trying to write in the same Lucene directory location
