code reuse

weed/util/sqlutil/splitter.go

@@ -0,0 +1,142 @@
+package sqlutil
+
+import (
+	"strings"
+)
+
+// SplitStatements splits a query string into individual SQL statements.
+// This robust implementation handles SQL comments, quoted strings, and escaped characters.
+//
+// Features:
+// - Handles single-line comments (-- comment)
+// - Handles multi-line comments (/* comment */)
+// - Properly escapes single quotes in strings ('don”t')
+// - Properly escapes double quotes in identifiers ("column""name")
+// - Ignores semicolons within quoted strings and comments
+// - Returns clean, trimmed statements with empty statements filtered out
+func SplitStatements(query string) []string {
+	var statements []string
+	var current strings.Builder
+
+	query = strings.TrimSpace(query)
+	if query == "" {
+		return []string{}
+	}
+
+	runes := []rune(query)
+	i := 0
+
+	for i < len(runes) {
+		char := runes[i]
+
+		// Handle single-line comments (-- comment)
+		if char == '-' && i+1 < len(runes) && runes[i+1] == '-' {
+			// Skip the entire comment without including it in any statement
+			for i < len(runes) && runes[i] != '\n' && runes[i] != '\r' {
+				i++
+			}
+			// Skip the newline if present
+			if i < len(runes) {
+				i++
+			}
+			continue
+		}
+
+		// Handle multi-line comments (/* comment */)
+		if char == '/' && i+1 < len(runes) && runes[i+1] == '*' {
+			// Skip the /* opening
+			i++
+			i++
+
+			// Skip to end of comment or end of input without including content
+			for i < len(runes) {
+				if runes[i] == '*' && i+1 < len(runes) && runes[i+1] == '/' {
+					i++ // Skip the *
+					i++ // Skip the /
+					break
+				}
+				i++
+			}
+			continue
+		}
+
+		// Handle single-quoted strings
+		if char == '\'' {
+			current.WriteRune(char)
+			i++
+
+			for i < len(runes) {
+				char = runes[i]
+				current.WriteRune(char)
+
+				if char == '\'' {
+					// Check if it's an escaped quote
+					if i+1 < len(runes) && runes[i+1] == '\'' {
+						i++ // Skip the next quote (it's escaped)
+						if i < len(runes) {
+							current.WriteRune(runes[i])
+						}
+					} else {
+						break // End of string
+					}
+				}
+				i++
+			}
+			i++
+			continue
+		}
+
+		// Handle double-quoted identifiers
+		if char == '"' {
+			current.WriteRune(char)
+			i++
+
+			for i < len(runes) {
+				char = runes[i]
+				current.WriteRune(char)
+
+				if char == '"' {
+					// Check if it's an escaped quote
+					if i+1 < len(runes) && runes[i+1] == '"' {
+						i++ // Skip the next quote (it's escaped)
+						if i < len(runes) {
+							current.WriteRune(runes[i])
+						}
+					} else {
+						break // End of identifier
+					}
+				}
+				i++
+			}
+			i++
+			continue
+		}
+
+		// Handle semicolon (statement separator)
+		if char == ';' {
+			stmt := strings.TrimSpace(current.String())
+			if stmt != "" {
+				statements = append(statements, stmt)
+			}
+			current.Reset()
+		} else {
+			current.WriteRune(char)
+		}
+		i++
+	}
+
+	// Add any remaining statement
+	if current.Len() > 0 {
+		stmt := strings.TrimSpace(current.String())
+		if stmt != "" {
+			statements = append(statements, stmt)
+		}
+	}
+
+	// If no statements found, return the original query as a single statement
+	if len(statements) == 0 {
+		return []string{strings.TrimSpace(strings.TrimSuffix(strings.TrimSpace(query), ";"))}
+	}
+
+	return statements
+}

weed/util/sqlutil/splitter_test.go

@@ -0,0 +1,147 @@
+package sqlutil
+
+import (
+	"reflect"
+	"testing"
+)
+
+func TestSplitStatements(t *testing.T) {
+	tests := []struct {
+		name     string
+		input    string
+		expected []string
+	}{
+		{
+			name:     "Simple single statement",
+			input:    "SELECT * FROM users",
+			expected: []string{"SELECT * FROM users"},
+		},
+		{
+			name:     "Multiple statements",
+			input:    "SELECT * FROM users; SELECT * FROM orders;",
+			expected: []string{"SELECT * FROM users", "SELECT * FROM orders"},
+		},
+		{
+			name:     "Semicolon in single quotes",
+			input:    "SELECT 'hello;world' FROM users; SELECT * FROM orders;",
+			expected: []string{"SELECT 'hello;world' FROM users", "SELECT * FROM orders"},
+		},
+		{
+			name:     "Semicolon in double quotes",
+			input:    `SELECT "column;name" FROM users; SELECT * FROM orders;`,
+			expected: []string{`SELECT "column;name" FROM users`, "SELECT * FROM orders"},
+		},
+		{
+			name:     "Escaped quotes in strings",
+			input:    `SELECT 'don''t split; here' FROM users; SELECT * FROM orders;`,
+			expected: []string{`SELECT 'don''t split; here' FROM users`, "SELECT * FROM orders"},
+		},
+		{
+			name:     "Escaped quotes in identifiers",
+			input:    `SELECT "column""name" FROM users; SELECT * FROM orders;`,
+			expected: []string{`SELECT "column""name" FROM users`, "SELECT * FROM orders"},
+		},
+		{
+			name:     "Single line comment",
+			input:    "SELECT * FROM users; -- This is a comment\nSELECT * FROM orders;",
+			expected: []string{"SELECT * FROM users", "SELECT * FROM orders"},
+		},
+		{
+			name:     "Single line comment with semicolon",
+			input:    "SELECT * FROM users; -- Comment with; semicolon\nSELECT * FROM orders;",
+			expected: []string{"SELECT * FROM users", "SELECT * FROM orders"},
+		},
+		{
+			name:     "Multi-line comment",
+			input:    "SELECT * FROM users; /* Multi-line\ncomment */ SELECT * FROM orders;",
+			expected: []string{"SELECT * FROM users", "SELECT * FROM orders"},
+		},
+		{
+			name:     "Multi-line comment with semicolon",
+			input:    "SELECT * FROM users; /* Comment with; semicolon */ SELECT * FROM orders;",
+			expected: []string{"SELECT * FROM users", "SELECT * FROM orders"},
+		},
+		{
+			name: "Complex mixed case",
+			input: `SELECT 'test;string', "quoted;id" FROM users; -- Comment; here
+			/* Another; comment */ 
+			INSERT INTO users VALUES ('name''s value', "id""field");`,
+			expected: []string{
+				`SELECT 'test;string', "quoted;id" FROM users`,
+				`INSERT INTO users VALUES ('name''s value', "id""field")`,
+			},
+		},
+		{
+			name:     "Empty statements filtered",
+			input:    "SELECT * FROM users;;; SELECT * FROM orders;",
+			expected: []string{"SELECT * FROM users", "SELECT * FROM orders"},
+		},
+		{
+			name:     "Whitespace handling",
+			input:    "  SELECT * FROM users  ;  SELECT * FROM orders  ;  ",
+			expected: []string{"SELECT * FROM users", "SELECT * FROM orders"},
+		},
+		{
+			name:     "Single statement without semicolon",
+			input:    "SELECT * FROM users",
+			expected: []string{"SELECT * FROM users"},
+		},
+		{
+			name:     "Empty query",
+			input:    "",
+			expected: []string{},
+		},
+		{
+			name:     "Only whitespace",
+			input:    "   \n\t   ",
+			expected: []string{},
+		},
+	}
+
+	for _, tt := range tests {
+		t.Run(tt.name, func(t *testing.T) {
+			result := SplitStatements(tt.input)
+			if !reflect.DeepEqual(result, tt.expected) {
+				t.Errorf("SplitStatements() = %v, expected %v", result, tt.expected)
+			}
+		})
+	}
+}
+
+func TestSplitStatements_EdgeCases(t *testing.T) {
+	tests := []struct {
+		name     string
+		input    string
+		expected []string
+	}{
+		{
+			name:     "Nested comments are not supported but handled gracefully",
+			input:    "SELECT * FROM users; /* Outer /* inner */ comment */ SELECT * FROM orders;",
+			expected: []string{"SELECT * FROM users", "comment */ SELECT * FROM orders"},
+		},
+		{
+			name:     "Unterminated string (malformed SQL)",
+			input:    "SELECT 'unterminated string; SELECT * FROM orders;",
+			expected: []string{"SELECT 'unterminated string; SELECT * FROM orders;"},
+		},
+		{
+			name:     "Unterminated comment (malformed SQL)",
+			input:    "SELECT * FROM users; /* unterminated comment",
+			expected: []string{"SELECT * FROM users"},
+		},
+		{
+			name:     "Multiple semicolons in quotes",
+			input:    "SELECT ';;;' FROM users; SELECT ';;;' FROM orders;",
+			expected: []string{"SELECT ';;;' FROM users", "SELECT ';;;' FROM orders"},
+		},
+	}
+
+	for _, tt := range tests {
+		t.Run(tt.name, func(t *testing.T) {
+			result := SplitStatements(tt.input)
+			if !reflect.DeepEqual(result, tt.expected) {
+				t.Errorf("SplitStatements() = %v, expected %v", result, tt.expected)
+			}
+		})
+	}
+}