Handle empty quoted terms in Azure Search query (#598)

Address NuGet/NuGetGallery#7351
NuGet · Jul 17, 2019 · 0e1b221 · 0e1b221
1 parent c983a74
commit 0e1b221
Show file tree

Hide file tree

Showing 3 changed files with 23 additions and 2 deletions.
diff --git a/src/NuGet.Indexing/NuGetQueryParser.cs b/src/NuGet.Indexing/NuGetQueryParser.cs
@@ -24,11 +24,16 @@ public class NuGetQueryParser
             { QueryField.Any,  new [] { "*" } }
         };
 
-        public Dictionary<QueryField, HashSet<string>> ParseQuery(string query)
+        public Dictionary<QueryField, HashSet<string>> ParseQuery(string query, bool skipWhiteSpace = false)
         {
             var grouping = new Dictionary<QueryField, HashSet<string>>();
             foreach (Clause clause in MakeClauses(Tokenize(query)))
             {
+                if (skipWhiteSpace && string.IsNullOrWhiteSpace(clause.Text))
+                {
+                    continue;
+                }
+
                 HashSet<string> text;
                 var queryField = GetQueryField(clause.Field);
                 if (!grouping.TryGetValue(queryField, out text))

diff --git a/src/NuGet.Services.AzureSearch/SearchService/SearchTextBuilder.cs b/src/NuGet.Services.AzureSearch/SearchService/SearchTextBuilder.cs
@@ -86,7 +86,7 @@ private string GetLuceneQuery(string query)
             }
 
             // Parse the NuGet query.
-            var grouping = _parser.ParseQuery(query.Trim());
+            var grouping = _parser.ParseQuery(query.Trim(), skipWhiteSpace: true);
             if (!grouping.Any())
             {
                 return MatchAllDocumentsQuery;

diff --git a/tests/NuGet.Services.AzureSearch.Tests/SearchService/SearchTextBuilderFacts.cs b/tests/NuGet.Services.AzureSearch.Tests/SearchService/SearchTextBuilderFacts.cs
@@ -222,6 +222,22 @@ public static IEnumerable<object[]> CommonAzureSearchQueryData()
                     { @"""foo bar"" baz", @"""foo bar"" baz" },
                     { @"title:""foo bar""", @"title:""foo bar""" },
                     { @"title:""a b"" c title:d f", @"+title:(""a b"" d) c f" },
+                    { @"title:"" a b    c   """, @"title:""a b    c""" },
+
+                    // Dangling quotes are handled with best effort
+                    { @"Tags:""windows", "tags:windows" },
+                    { @"json Tags:""net"" Tags:""windows sdk", @"+tags:(net windows sdk) json" },
+                    { @"json Tags:""net Tags:""windows sdk""", @"+tags:(net Tags\:) json windows sdk" },
+                    { @"sdk Tags:""windows", "+tags:windows sdk" },
+                    { @"Tags:""windows sdk", "tags:(windows sdk)" },
+                    { @"Tags:""""windows""", "windows" },
+
+                    // Empty quotes are ignored
+                    { @"Tags:""""", @"*" },
+                    { @"Tags:"" """, @"*" },
+                    { @"Tags:""      """, @"*" },
+                    { @"windows Tags:""      """, @"windows" },
+                    { @"windows Tags:""      "" Tags:sdk", @"+tags:sdk windows" },
 
                     // Duplicate search terms on the same query field are folded
                     { "a a", "a" },