fix(sql_helpers): enhance splitByAND to handle BETWEEN and quotes

* Add support for BETWEEN-aware AND detection * Ensure AND inside single-quoted strings does not cause splits * Update tests to cover new BETWEEN and quote scenarios
2026-06-28 07:47:39 +00:00 · 2026-06-23 09:41:27 +02:00
parent 4c512acf25
commit 48b72a7631
2 changed files with 94 additions and 18 deletions
@@ -446,18 +446,36 @@ func containsTopLevelOR(clause string) bool {
 	return false
 }

-// splitByAND splits a WHERE clause by AND operators (case-insensitive)
-// This is parenthesis-aware and won't split on AND operators inside subqueries
+// splitByAND splits a WHERE clause by AND operators (case-insensitive).
+// It is parenthesis-aware (won't split inside subqueries), quote-aware
+// (won't split on AND inside single-quoted strings), and BETWEEN-aware
+// (won't split on the AND that separates the two operands of BETWEEN x AND y).
 func splitByAND(where string) []string {
 	conditions := []string{}
 	currentCondition := strings.Builder{}
-	depth := 0 // Track parenthesis depth
+	depth := 0         // parenthesis nesting depth
+	inSingleQuote := false
+	afterBetween := false // true after seeing BETWEEN at depth 0; next AND belongs to it
 	i := 0

 	for i < len(where) {
 		ch := where[i]

-		// Track parenthesis depth
+		// Track single-quote state so we never split on AND inside string literals.
+		if ch == '\'' {
+			inSingleQuote = !inSingleQuote
+			currentCondition.WriteByte(ch)
+			i++
+			continue
+		}
+
+		if inSingleQuote {
+			currentCondition.WriteByte(ch)
+			i++
+			continue
+		}
+
+		// Track parenthesis depth (outside quotes only).
 		if ch == '(' {
 			depth++
 			currentCondition.WriteByte(ch)
@@ -470,32 +488,39 @@ func splitByAND(where string) []string {
 			continue
 		}

-		// Only look for AND operators at depth 0 (not inside parentheses)
+		// All keyword checks only apply at depth 0 (not inside subqueries).
 		if depth == 0 {
-			// Check if we're at an AND operator (case-insensitive)
-			// We need at least " AND " (5 chars) or " and " (5 chars)
-			if i+5 <= len(where) {
-				substring := where[i : i+5]
-				lowerSubstring := strings.ToLower(substring)
+			// Detect " BETWEEN " (9 chars, case-insensitive) so the very next
+			// top-level AND is recognised as part of the BETWEEN syntax.
+			if i+9 <= len(where) && strings.ToLower(where[i:i+9]) == " between " {
+				afterBetween = true
+				currentCondition.WriteString(where[i : i+9])
+				i += 9
+				continue
+			}

-				if lowerSubstring == " and " {
-					// Found an AND operator at the top level
-					// Add the current condition to the list
+			// Detect " AND " (5 chars, case-insensitive).
+			if i+5 <= len(where) && strings.ToLower(where[i:i+5]) == " and " {
+				if afterBetween {
+					// This AND closes a BETWEEN expression — do NOT split.
+					afterBetween = false
+					currentCondition.WriteString(where[i : i+5])
+					i += 5
+					continue
+				}
+				// Regular conjunction — split here.
 				conditions = append(conditions, currentCondition.String())
 				currentCondition.Reset()
-					// Skip past the AND operator
 				i += 5
 				continue
 			}
 		}
-		}

-		// Not an AND operator or we're inside parentheses, just add the character
 		currentCondition.WriteByte(ch)
 		i++
 	}

-	// Add the last condition
+	// Add the last condition.
 	if currentCondition.Len() > 0 {
 		conditions = append(conditions, currentCondition.String())
 	}
@@ -520,6 +520,38 @@ func TestSplitByAND(t *testing.T) {
 			input:    "a = 1 AND b = 2 AND c = 3 and (select s from generate_series(1,10) s where s < 10 and s > 0 offset 2 limit 1) = 3",
 			expected: []string{"a = 1", "b = 2", "c = 3", "(select s from generate_series(1,10) s where s < 10 and s > 0 offset 2 limit 1) = 3"},
 		},
+		// BETWEEN-aware cases: the AND inside BETWEEN x AND y must not cause a split.
+		{
+			name:     "BETWEEN does not split on its AND",
+			input:    "col between '2025-08-31' and '1970-01-01'",
+			expected: []string{"col between '2025-08-31' and '1970-01-01'"},
+		},
+		{
+			name:     "BETWEEN uppercase AND",
+			input:    "col BETWEEN '2025-08-31' AND '1970-01-01'",
+			expected: []string{"col BETWEEN '2025-08-31' AND '1970-01-01'"},
+		},
+		{
+			name:     "BETWEEN followed by a regular AND conjunction",
+			input:    "col between 1 and 5 and other = 'x'",
+			expected: []string{"col between 1 and 5", "other = 'x'"},
+		},
+		{
+			name:     "two BETWEEN conditions joined by AND",
+			input:    "col1 between 1 and 5 and col2 between 10 and 20",
+			expected: []string{"col1 between 1 and 5", "col2 between 10 and 20"},
+		},
+		{
+			name:  "complex OR block with multiple BETWEENs (real-world case)",
+			input: "tbl.applicationdate between '2025-08-31' and '1970-01-01'\n  or tbl.capturedate between '2025-08-31' and '1970-01-01'\n  or tbl.startdate between '2025-08-31' AND '1970-01-01'",
+			expected: []string{"tbl.applicationdate between '2025-08-31' and '1970-01-01'\n  or tbl.capturedate between '2025-08-31' and '1970-01-01'\n  or tbl.startdate between '2025-08-31' AND '1970-01-01'"},
+		},
+		// Quote-aware cases: AND inside a string literal must not split.
+		{
+			name:     "AND inside single-quoted string is not a split point",
+			input:    "comment = 'this and that' and status = 'active'",
+			expected: []string{"comment = 'this and that'", "status = 'active'"},
+		},
 	}

 	for _, tt := range tests {
@@ -917,6 +949,25 @@ where:     "(true AND status = 'active')",
 tableName: "unregistered_table",
 expected:  "(true AND unregistered_table.status = 'active')",
 },
+// BETWEEN regression: date literals inside BETWEEN must not be prefixed as columns.
+{
+name:      "BETWEEN date range - second date must not be prefixed",
+where:     "applicationdate between '2025-08-31' and '1970-01-01'",
+tableName: "unregistered_table",
+expected:  "unregistered_table.applicationdate between '2025-08-31' and '1970-01-01'",
+},
+{
+name:      "Already-prefixed BETWEEN column - unchanged",
+where:     `"v_webui_clients".applicationdate between '2025-08-31' and '1970-01-01'`,
+tableName: "v_webui_clients",
+expected:  `"v_webui_clients".applicationdate between '2025-08-31' and '1970-01-01'`,
+},
+{
+name:      "Complex OR block with multiple BETWEENs - date values must not be prefixed",
+where:     `("v_webui_clients".applicationdate between '2025-08-31' and '1970-01-01' or "v_webui_clients".clientcapturedate between '2025-08-31' and '1970-01-01' or "v_webui_clients".startdate between '2025-08-31' AND '1970-01-01')`,
+tableName: "v_webui_clients",
+expected:  `("v_webui_clients".applicationdate between '2025-08-31' and '1970-01-01' or "v_webui_clients".clientcapturedate between '2025-08-31' and '1970-01-01' or "v_webui_clients".startdate between '2025-08-31' AND '1970-01-01')`,
+},
 }

 for _, tt := range tests {