Merge pull request #69 from xqueries/literal-handling

implement literal handling in compiler and engine

cmd/xdb/main.go

@@ -6,6 +6,7 @@ import (
 	"io"
 	"os"
 	"os/signal"
+	"path/filepath"
 	"syscall"
 
 	"github.com/rs/zerolog"
@@ -167,7 +168,7 @@ func startNode(cmd *cobra.Command, args []string) {
 
 func createLogger(stdin io.Reader, stdout, stderr io.Writer) zerolog.Logger {
 	// open the log file
-	file, err := os.OpenFile(logfile, os.O_CREATE|os.O_APPEND|os.O_RDWR, 0600)
+	file, err := os.OpenFile(filepath.Clean(logfile), os.O_CREATE|os.O_APPEND|os.O_RDWR, 0600)
 	if err != nil {
 		_, _ = fmt.Fprintln(stderr, fmt.Errorf("open logfile: %w", err).Error())
 		os.Exit(ExitAbnormal)

internal/compiler/command/command.go

@@ -237,8 +237,8 @@ type (
 		// this is a representation derived from the AST. If this is empty, the
 		// executor has to interpolate the table from the execution context.
 		Table string
-		// Name is the name of the column.
-		Name Expr
+		// Expr is the name of the column.
+		Expr Expr
 		// Alias is the alias name for this table. May be empty.
 		Alias string
 	}
@@ -422,9 +422,9 @@ func (u UpdateSetter) String() string {
 
 func (c Column) String() string {
 	if c.Alias == "" {
-		return c.Name.String()
+		return c.Expr.String()
 	}
-	return fmt.Sprintf("%v AS %v", c.Name, c.Alias)
+	return fmt.Sprintf("%v AS %v", c.Expr, c.Alias)
 }
 
 func (j Join) String() string {

internal/compiler/command/expr.go

@@ -15,11 +15,26 @@ type (
 		_expr()
 	}
 
-	// LiteralExpr is a simple literal expression that has a single string
-	// value.
-	LiteralExpr struct {
-		// Value is the simple string value of this expression.
-		Value string
+	// ConstantLiteral is a constant literal expression. The Numeric flag determines whether
+	// or not this is a numeric literal.
+	ConstantLiteral struct {
+		Value   string
+		Numeric bool
+	}
+
+	// ConstantLiteralOrColumnReference is a string literal that represents either a constant
+	// string value or a column reference. If this literal is resolved, the column reference
+	// takes precedence over the string literal, meaning that if a column exists, whose name is
+	// equal to the value of this expression, the value in that column must be used. Only if there
+	// is no such column present, the string literal is to be used.
+	ConstantLiteralOrColumnReference struct {
+		ValueOrName string
+	}
+
+	// ColumnReference is a string literal that represents a reference to a column. During resolving,
+	// if no column with such a name is present, an error must be risen.
+	ColumnReference struct {
+		Name string
 	}
 
 	// ConstantBooleanExpr is a simple expression that represents a boolean
@@ -57,15 +72,46 @@ type (
 	}
 )
 
-func (LiteralExpr) _expr()         {}
 func (ConstantBooleanExpr) _expr() {}
 func (RangeExpr) _expr()           {}
 func (FunctionExpr) _expr()        {}
 
-func (l LiteralExpr) String() string {
+func (ConstantLiteral) _expr()                  {}
+func (ConstantLiteralOrColumnReference) _expr() {}
+func (ColumnReference) _expr()                  {}
+
+// ConstantValue returns the constant value of this literal.
+func (l ConstantLiteral) ConstantValue() string {
 	return l.Value
 }
 
+func (l ConstantLiteral) String() string {
+	return l.Value
+}
+
+// ConstantValue returns the constant value of this literal.
+func (l ConstantLiteralOrColumnReference) ConstantValue() string {
+	return l.ValueOrName
+}
+
+// ReferencedColName returns the constant name of the referenced column name.
+func (l ConstantLiteralOrColumnReference) ReferencedColName() string {
+	return l.ValueOrName
+}
+
+func (l ConstantLiteralOrColumnReference) String() string {
+	return l.ValueOrName
+}
+
+// ReferencedColName returns the constant name of the referenced column name.
+func (l ColumnReference) ReferencedColName() string {
+	return l.Name
+}
+
+func (l ColumnReference) String() string {
+	return l.Name
+}
+
 func (b ConstantBooleanExpr) String() string {
 	return strconv.FormatBool(b.Value)
 }

internal/compiler/numeric_parser.go

@@ -0,0 +1,156 @@
+package compiler
+
+import (
+	"bytes"
+	"strings"
+)
+
+type numericParserState func(*numericParser) numericParserState
+
+type numericParser struct {
+	candidate string
+	index     int
+
+	isReal                  bool
+	isHexadecimal           bool
+	isErronous              bool
+	hasDigitsBeforeExponent bool
+
+	value *bytes.Buffer
+
+	current numericParserState
+}
+
+// isNumeric is an adapted copy of (engine.Engine).ToNumericValue.
+func isNumeric(s string) bool {
+	if s == "" || s == "0x" || s == "." {
+		return false
+	}
+	p := numericParser{
+		candidate: s,
+		index:     0,
+
+		value: &bytes.Buffer{},
+
+		current: stateInitial,
+	}
+	p.parse()
+	return !p.isErronous
+}
+
+func (p numericParser) done() bool {
+	return p.index >= len(p.candidate)
+}
+
+func (p *numericParser) parse() {
+	for p.current != nil && !p.done() {
+		p.current = p.current(p)
+	}
+}
+
+func (p *numericParser) get() byte {
+	return p.candidate[p.index]
+}
+
+func (p *numericParser) step() {
+	_ = p.value.WriteByte(p.get())
+	p.index++
+}
+
+func stateInitial(p *numericParser) numericParserState {
+	switch {
+	case strings.HasPrefix(p.candidate, "0x"):
+		p.index += 2
+		p.isHexadecimal = true
+		return stateHex
+	case isDigit(p.get()):
+		return stateFirstDigits
+	case p.get() == '.':
+		return stateDecimalPoint
+	}
+	p.isErronous = true
+	return nil
+}
+
+func stateHex(p *numericParser) numericParserState {
+	if isHexDigit(p.get()) {
+		p.step()
+		return stateHex
+	}
+	p.isErronous = true
+	return nil
+}
+
+func stateFirstDigits(p *numericParser) numericParserState {
+	if isDigit(p.get()) {
+		p.hasDigitsBeforeExponent = true
+		p.step()
+		return stateFirstDigits
+	} else if p.get() == '.' {
+		return stateDecimalPoint
+	}
+	p.isErronous = true
+	return nil
+}
+
+func stateDecimalPoint(p *numericParser) numericParserState {
+	if p.get() == '.' {
+		p.step()
+		p.isReal = true
+		return stateSecondDigits
+	}
+	p.isErronous = true
+	return nil
+}
+
+func stateSecondDigits(p *numericParser) numericParserState {
+	if isDigit(p.get()) {
+		p.hasDigitsBeforeExponent = true
+		p.step()
+		return stateSecondDigits
+	} else if p.get() == 'E' {
+		if p.hasDigitsBeforeExponent {
+			return stateExponent
+		}
+		p.isErronous = true // if there were no first digits,
+	}
+	p.isErronous = true
+	return nil
+}
+
+func stateExponent(p *numericParser) numericParserState {
+	if p.get() == 'E' {
+		p.step()
+		return stateOptionalSign
+	}
+	p.isErronous = true
+	return nil
+}
+
+func stateOptionalSign(p *numericParser) numericParserState {
+	if p.get() == '+' || p.get() == '-' {
+		p.step()
+		return stateThirdDigits
+	} else if isDigit(p.get()) {
+		return stateThirdDigits
+	}
+	p.isErronous = true
+	return nil
+}
+
+func stateThirdDigits(p *numericParser) numericParserState {
+	if isDigit(p.get()) {
+		p.step()
+		return stateThirdDigits
+	}
+	p.isErronous = true
+	return nil
+}
+
+func isDigit(b byte) bool {
+	return b-'0' <= 9
+}
+
+func isHexDigit(b byte) bool {
+	return isDigit(b) || (b-'A' <= 15) || (b-'a' <= 15)
+}

internal/compiler/numeric_parser_bench_test.go

@@ -0,0 +1,17 @@
+package compiler
+
+import (
+	"testing"
+)
+
+func BenchmarkToNumericValue(b *testing.B) {
+	str := "75610342.92389E-21423"
+
+	b.ResetTimer()
+
+	for i := 0; i < b.N; i++ {
+		if !isNumeric(str) {
+			b.FailNow()
+		}
+	}
+}