Skip to content

Commit

Permalink
DYN-6535 DynamoRevit Improve Search (#14878)
Browse files Browse the repository at this point in the history 
* DYN-6535 DynamoRevit Improve Search

In a search criteria like "list replace" I've implemented a fix that will check on many nodes we have when the terms splitted by empty space (in this case list and replace)  so if it reaches the limit then we discard the split search and execute a normal search.

* DYN-6535 DynamoRevit Improve Search

Adding extra comments

* DYN-6535 DynamoRevit Improve Search Code Review

Convert the ByEmptySpace to be a property and remove it from the enum.

* DYN-6535 DynamoRevit Improve Search Code Review

Adding extra comments
  • Loading branch information
@RobertGlobant20
RobertGlobant20 authored Jan 25, 2024
1 parent 32025e7 commit f33a015
Showing 1 changed file with 46 additions and 13 deletions.
59 changes: 46 additions & 13 deletions src/DynamoCore/Utilities/LuceneSearchUtility.cs
View file
Original file line number Diff line number Diff line change
Expand Up @@ -73,6 +73,8 @@ internal class LuceneSearchUtility
/// </summary>
internal static readonly LuceneStartConfig DefaultPkgIndexStartConfig = new LuceneStartConfig(LuceneSearchUtility.LuceneStorage.FILE_SYSTEM, LuceneConfig.PackagesIndexingDirectory);

private bool hasEmptySpaces { get; set; }

public enum LuceneStorage
{
//Lucene Storage will be located in RAM and all the info indexed will be lost when Dynamo app is closed
Expand All @@ -82,6 +84,18 @@ public enum LuceneStorage
FILE_SYSTEM
}

/// <summary>
/// This enum will be used to identify which can of search should be executed based in the user search criteria
/// </summary>
public enum SearchType
{
//Normal search using just one word matching a specific node name
Normal,

//Search by category using the "." character for example "list.re"
ByCategory
}

// Used for creating the StandardAnalyzer
internal Analyzer Analyzer;

Expand Down Expand Up @@ -264,7 +278,14 @@ internal void SetDocumentFieldValue(Document doc, string field, string value, bo
/// <returns></returns>
internal string CreateSearchQuery(string[] fields, string SearchTerm)
{
//By Default the search will be normal
SearchType searchType = SearchType.Normal;
int fuzzyLogicMaxEdits = LuceneConfig.FuzzySearchMinEdits;
hasEmptySpaces = false;

//Max number of nodes allowed in the search when is a ByEmptySpace search
const int MaxNodeNamesRepeated = 20;

// Use a larger max edit value - more tolerant with typo when search term is longer than threshold
if (SearchTerm.Length > LuceneConfig.FuzzySearchMaxEditsThreshold)
{
Expand All @@ -273,13 +294,21 @@ internal string CreateSearchQuery(string[] fields, string SearchTerm)

var booleanQuery = new BooleanQuery();
string searchTerm = QueryParser.Escape(SearchTerm);
var bCategoryBasedSearch = searchTerm.Contains('.') ? true : false;

if (searchTerm.Contains('.'))
searchType = SearchType.ByCategory;
else if (searchTerm.Contains(' '))
hasEmptySpaces = true;
else
searchType = SearchType.Normal;

var trimmedSearchTerm = hasEmptySpaces == true ? searchTerm.Replace(" ", "") : searchTerm;

foreach (string f in fields)
{
//Needs to be again due that now a query can contain different values per field (e.g. CategorySplitted:list, Name:tr)
searchTerm = QueryParser.Escape(SearchTerm);
if (bCategoryBasedSearch == true)
if (searchType == SearchType.ByCategory)
{
//This code section should be only executed if the search criteria is CategoryBased like "category.nodename"
if (f != nameof(LuceneConfig.NodeFieldsEnum.NameSplitted) &&
Expand All @@ -297,26 +326,26 @@ internal string CreateSearchQuery(string[] fields, string SearchTerm)
}
}

FuzzyQuery fuzzyQuery;
if (searchTerm.Length > LuceneConfig.FuzzySearchMinimalTermLength)
{
fuzzyQuery = new FuzzyQuery(new Term(f, searchTerm), fuzzyLogicMaxEdits);
booleanQuery.Add(fuzzyQuery, Occur.SHOULD);
}

//For normal search we don't consider the fields NameSplitted and CategorySplitted
if ((f == nameof(LuceneConfig.NodeFieldsEnum.NameSplitted) ||
f == nameof(LuceneConfig.NodeFieldsEnum.CategorySplitted)) && bCategoryBasedSearch == false)
f == nameof(LuceneConfig.NodeFieldsEnum.CategorySplitted)) && searchType != SearchType.ByCategory)
continue;

//This case is for when the user type something like "list.", I mean, not specifying the node name or part of it
if (string.IsNullOrEmpty(searchTerm))
continue;

var fieldQuery = CalculateFieldWeight(f, searchTerm);
var wildcardQuery = CalculateFieldWeight(f, searchTerm, true);
FuzzyQuery fuzzyQuery;
if (searchTerm.Length > LuceneConfig.FuzzySearchMinimalTermLength)
{
fuzzyQuery = new FuzzyQuery(new Term(f, hasEmptySpaces == true ? trimmedSearchTerm : searchTerm), fuzzyLogicMaxEdits);
booleanQuery.Add(fuzzyQuery, Occur.SHOULD);
}

var fieldQuery = CalculateFieldWeight(f, hasEmptySpaces == true ? trimmedSearchTerm : searchTerm);
var wildcardQuery = CalculateFieldWeight(f, hasEmptySpaces == true ? trimmedSearchTerm : searchTerm, true);

if (bCategoryBasedSearch && f == nameof(LuceneConfig.NodeFieldsEnum.CategorySplitted))
if (searchType == SearchType.ByCategory && f == nameof(LuceneConfig.NodeFieldsEnum.CategorySplitted))
{
booleanQuery.Add(fieldQuery, Occur.MUST);
booleanQuery.Add(wildcardQuery, Occur.MUST);
Expand All @@ -331,6 +360,10 @@ internal string CreateSearchQuery(string[] fields, string SearchTerm)
{
foreach (string s in searchTerm.Split(' ', '.'))
{
//If is a ByEmptySpace search and the splitted words match with more than MaxNodeNamesRepeated nodes then the word is skipped
int nodesFrequency = dynamoModel.SearchModel.Entries.Where(entry => entry.Name.ToLower().Contains(s) && !string.IsNullOrEmpty(s)).Count();
if (nodesFrequency > MaxNodeNamesRepeated) continue;

if (string.IsNullOrEmpty(s)) continue;

if (s.Length > LuceneConfig.FuzzySearchMinimalTermLength)
Expand Down

0 comments on commit f33a015

Please sign in to comment.