From 24f61caca9147c3646b3c10c80dd11cf9cdce02a Mon Sep 17 00:00:00 2001
From: bowen xiao <xbowen@uber.com>
Date: Tue, 24 Sep 2024 14:20:12 -0700
Subject: [PATCH] Pinot handle customer keyword type empty val (#6302)

* add query to handle empty keyword val
---
 common/pinot/pinotQueryValidator.go      | 16 ++++++++++++----
 common/pinot/pinotQueryValidator_test.go | 10 +++++++++-
 2 files changed, 21 insertions(+), 5 deletions(-)

diff --git a/common/pinot/pinotQueryValidator.go b/common/pinot/pinotQueryValidator.go
index e5316708b1b..ec6024126f3 100644
--- a/common/pinot/pinotQueryValidator.go
+++ b/common/pinot/pinotQueryValidator.go
@@ -445,12 +445,20 @@ func processEqual(colNameStr string, colValStr string) string {
 func processCustomKeyword(operator string, colNameStr string, colValStr string) string {
 	// edge case
 	if operator == "!=" {
-		return fmt.Sprintf("JSON_MATCH(Attr, '\"$.%s\"%s''%s''') and JSON_MATCH(Attr, '\"$.%s[*]\"%s''%s''')",
-			colNameStr, operator, colValStr, colNameStr, operator, colValStr)
+		return createKeywordQuery(operator, colNameStr, colValStr, "and", "NOT ")
 	}
 
-	return fmt.Sprintf("(JSON_MATCH(Attr, '\"$.%s\"%s''%s''') or JSON_MATCH(Attr, '\"$.%s[*]\"%s''%s'''))",
-		colNameStr, operator, colValStr, colNameStr, operator, colValStr)
+	return createKeywordQuery(operator, colNameStr, colValStr, "or", "")
+}
+
+func createKeywordQuery(operator string, colNameStr string, colValStr string, connector string, notEqual string) string {
+	if colValStr == "" {
+		// partial match for an empty string (still it will only match empty string)
+		// so it equals to exact match for an empty string
+		return createCustomStringQuery(colNameStr, colValStr, notEqual)
+	}
+	return fmt.Sprintf("(JSON_MATCH(Attr, '\"$.%s\"%s''%s''') %s JSON_MATCH(Attr, '\"$.%s[*]\"%s''%s'''))",
+		colNameStr, operator, colValStr, connector, colNameStr, operator, colValStr)
 }
 
 func processCustomString(operator string, colNameStr string, colValStr string) string {
diff --git a/common/pinot/pinotQueryValidator_test.go b/common/pinot/pinotQueryValidator_test.go
index ebf51d4fc0a..f6d0501c74b 100644
--- a/common/pinot/pinotQueryValidator_test.go
+++ b/common/pinot/pinotQueryValidator_test.go
@@ -115,7 +115,7 @@ func TestValidateQuery(t *testing.T) {
 		},
 		"Case8-4: query with custom keyword field not equal": {
 			query:     "CustomKeywordField != 0",
-			validated: "JSON_MATCH(Attr, '\"$.CustomKeywordField\"!=''0''') and JSON_MATCH(Attr, '\"$.CustomKeywordField[*]\"!=''0''')",
+			validated: "(JSON_MATCH(Attr, '\"$.CustomKeywordField\"!=''0''') and JSON_MATCH(Attr, '\"$.CustomKeywordField[*]\"!=''0'''))",
 		},
 		"Case9: invalid where expression": {
 			query: "InvalidWhereExpr",
@@ -359,6 +359,14 @@ func TestValidateQuery(t *testing.T) {
 			query:     "CustomStringField = 'abc' OR CustomStringField = 'def'",
 			validated: "(JSON_MATCH(Attr, '\"$.CustomStringField\" is not null') AND JSON_MATCH(Attr, 'REGEXP_LIKE(\"$.CustomStringField\", ''.*abc.*'')') or JSON_MATCH(Attr, '\"$.CustomStringField\" is not null') AND JSON_MATCH(Attr, 'REGEXP_LIKE(\"$.CustomStringField\", ''.*def.*'')'))",
 		},
+		"case23-1: custom keyword field is empty case": {
+			query:     "CustomKeywordField = ''",
+			validated: "JSON_MATCH(Attr, '\"$.CustomKeywordField\" is not null') AND JSON_MATCH(Attr, 'REGEXP_LIKE(\"$.CustomKeywordField\", ''^$'')')",
+		},
+		"case23-2: custom keyword field is not empty case": {
+			query:     "CustomKeywordField != ''",
+			validated: "JSON_MATCH(Attr, '\"$.CustomKeywordField\" is not null') AND NOT JSON_MATCH(Attr, 'REGEXP_LIKE(\"$.CustomKeywordField\", ''^$'')')",
+		},
 	}
 
 	for name, test := range tests {