From f55a3ca00879d4b975f01f27458fe3dc47013943 Mon Sep 17 00:00:00 2001
From: Robert McRackan <rmcracka@itron.com>
Date: Fri, 2 Apr 2021 11:27:16 -0400
Subject: [PATCH] Search engine bug fix and unit tests

---
 LibationLauncher/LibationLauncher.csproj      |  2 +-
 LibationSearchEngine/UNTESTED/LuceneRegex.cs  | 18 ++++----
 .../SearchEngineTests.cs                      | 45 +++++++++++++++++--
 3 files changed, 53 insertions(+), 12 deletions(-)

diff --git a/LibationLauncher/LibationLauncher.csproj b/LibationLauncher/LibationLauncher.csproj
index c04fbbdd..2c534211 100644
--- a/LibationLauncher/LibationLauncher.csproj
+++ b/LibationLauncher/LibationLauncher.csproj
@@ -13,7 +13,7 @@
     <!-- <PublishSingleFile>true</PublishSingleFile> -->
     <RuntimeIdentifier>win-x64</RuntimeIdentifier>
 
-    <Version>4.2.2.1</Version>
+    <Version>4.2.3.1</Version>
   </PropertyGroup>
 
   <ItemGroup>
diff --git a/LibationSearchEngine/UNTESTED/LuceneRegex.cs b/LibationSearchEngine/UNTESTED/LuceneRegex.cs
index 326be325..cc055098 100644
--- a/LibationSearchEngine/UNTESTED/LuceneRegex.cs
+++ b/LibationSearchEngine/UNTESTED/LuceneRegex.cs
@@ -61,11 +61,14 @@ namespace LibationSearchEngine
         /// </summary>
         private static string boolPattern_parameterized { get; }
             = @"
+### IMPORTANT: 'ignore whitespace' is only partially honored in character sets
+### - new lines are ok
+### - ANY leading whitespace is treated like actual matching spaces  :(
+
+                    ### can't begin with colon. incorrect syntax
+                    ### can't begin with open bracket: this signals the start of a tag
 (?<!                # begin negative lookbehind
-  [                 #   begin char set
-    :               #     colon
-    \[              #     open bracket, escaped
-  ]                 #   end char set
+  [:\[]             #   char set: colon and open bracket, escaped
   \s*               #   optional space
 )                   # end negative lookbehind
 
@@ -73,12 +76,11 @@ namespace LibationSearchEngine
   ({0})             #   captured bool search keyword. this is the $1 reference used in regex.Replace
 \b                  # word boundary
 
+                    ### can't end with colon. this signals that the bool's value already exists
+                    ### can't begin with close bracket: this signals the end of a tag
 (?!                 # begin negative lookahead
   \s*               #   optional space
-  [                 #   begin char set
-    :               #     colon
-    \]              #     close bracket, escaped
-  ]                 #   end char set
+  [:\]]             #   char set: colon and close bracket, escaped
 )                   # end negative lookahead
 ";
         private static Dictionary<string, Regex> boolRegexDic { get; } = new Dictionary<string, Regex>();
diff --git a/_Tests/LibationSearchEngine.Tests/SearchEngineTests.cs b/_Tests/LibationSearchEngine.Tests/SearchEngineTests.cs
index e7fae71c..685c83e9 100644
--- a/_Tests/LibationSearchEngine.Tests/SearchEngineTests.cs
+++ b/_Tests/LibationSearchEngine.Tests/SearchEngineTests.cs
@@ -24,15 +24,54 @@ namespace SearchEngineTests
 	public class FormatSearchQuery
 	{
 		[TestMethod]
-		[DataRow(null, SearchEngine.ALL_QUERY)]
-		[DataRow("", SearchEngine.ALL_QUERY)]
-		[DataRow("   ", SearchEngine.ALL_QUERY)]
+		// null, empty, whitespace -- *:*
+		[DataRow(null, "*:*")]
+		[DataRow("", "*:*")]
+		[DataRow("   ", "*:*")]
+
+		// tag surrounded by spaces
+		[DataRow("[foo]", "tags:foo")]
+		[DataRow("  [foo]", "  tags:foo")]
+		[DataRow("[foo]  ", "tags:foo  ")]
+		[DataRow("  [foo]  ", "  tags:foo  ")]
+		[DataRow("-[foo]", "-tags:foo")]
+		[DataRow("  -[foo]", "  -tags:foo")]
+		[DataRow("-[foo]  ", "-tags:foo  ")]
+		[DataRow("  -[foo]  ", "  -tags:foo  ")]
+
+		// tag case irrelevant
+		[DataRow("[FoO]", "tags:FoO")]
+
+		// bool keyword surrounded by spaces
 		[DataRow("israted", "israted:True")]
+		[DataRow("  israted", "  israted:True")]
+		[DataRow("israted  ", "israted:True  ")]
+		[DataRow("  israted  ", "  israted:True  ")]
+		[DataRow("-israted", "-israted:True")]
+		[DataRow("  -israted", "  -israted:True")]
+		[DataRow("-israted  ", "-israted:True  ")]
+		[DataRow("  -israted  ", "  -israted:True  ")]
+
+		// bool keyword. Append :True
+		[DataRow("israted", "israted:True")]
+
+		// bool keyword with [:bool]. Do not add :True
 		[DataRow("israted:True", "israted:True")]
 		[DataRow("isRated:false", "israted:false")]
+
+		// tag which happens to be a bool keyword >> parse as tag
 		[DataRow("[israted]", "tags:israted")]
+
+		// numbers with "to". TO all caps, numbers [8.2] format
 		[DataRow("1 to 10", "00000001.00 TO 00000010.00")]
 		[DataRow("19990101 to 20001231", "19990101.00 TO 20001231.00")]
+
+		// field to lowercase
+		[DataRow("Author:Doyle", "author:Doyle")]
+		// bool field to lowercase
+		[DataRow("IsRated", "israted:True")]
+		[DataRow("-isRATED", "-israted:True")]
+
 		public void FormattingTest(string input, string output)
 			=> SearchEngine.FormatSearchQuery(input).Should().Be(output);
 	}