feat: add pattern match line filter

pkg/logql/log/filter.go

@@ -11,6 +11,7 @@ import (
 
 	"github.com/prometheus/prometheus/model/labels"
 
+	"github.com/grafana/loki/pkg/logql/log/pattern"
 	"github.com/grafana/loki/pkg/util"
 )
 
@@ -23,6 +24,8 @@ const (
 	LineMatchNotEqual
 	LineMatchRegexp
 	LineMatchNotRegexp
+	LineMatchPattern
+	LineMatchNotPattern
 )
 
 func (t LineMatchType) String() string {
@@ -35,6 +38,10 @@ func (t LineMatchType) String() string {
 		return "|~"
 	case LineMatchNotRegexp:
 		return "!~"
+	case LineMatchPattern:
+		return "|>"
+	case LineMatchNotPattern:
+		return "!>"
 	default:
 		return ""
 	}
@@ -553,6 +560,10 @@ func NewFilter(match string, mt LineMatchType) (Filterer, error) {
 		return newContainsFilter([]byte(match), false), nil
 	case LineMatchNotEqual:
 		return NewNotFilter(newContainsFilter([]byte(match), false)), nil
+	case LineMatchPattern:
+		return newPatternFilterer([]byte(match), true)
+	case LineMatchNotPattern:
+		return newPatternFilterer([]byte(match), false)
 	default:
 		return nil, fmt.Errorf("unknown matcher: %v", match)
 	}
@@ -783,3 +794,37 @@ func (s *RegexSimplifier) simplifyConcatAlternate(reg *syntax.Regexp, literal []
 	}
 	return nil, false
 }
+
+type patternFilter struct {
+	matcher *pattern.Matcher
+	pattern []byte
+}
+
+func newPatternFilterer(p []byte, match bool) (MatcherFilterer, error) {
+	m, err := pattern.ParseLineFilter(p)
+	if err != nil {
+		return nil, err
+	}
+	filter := &patternFilter{
+		matcher: m,
+		pattern: p,
+	}
+	if !match {
+		return NewNotFilter(filter), nil
+	}
+	return filter, nil
+}
+
+func (f *patternFilter) Filter(line []byte) bool { return f.matcher.Test(line) }
+
+func (f *patternFilter) Matches(test Checker) bool {
+	return test.Test(f.pattern, false, false)
+}
+
+func (f *patternFilter) ToStage() Stage {
+	return StageFunc{
+		process: func(_ int64, line []byte, _ *LabelsBuilder) ([]byte, bool) {
+			return line, f.Filter(line)
+		},
+	}
+}

pkg/logql/log/parser.go

@@ -373,7 +373,7 @@ func (l *LogfmtParser) Process(_ int64, line []byte, lbs *LabelsBuilder) ([]byte
 func (l *LogfmtParser) RequiredLabelNames() []string { return []string{} }
 
 type PatternParser struct {
-	matcher pattern.Matcher
+	matcher *pattern.Matcher
 	names   []string
 }
 

pkg/logql/log/pattern/ast.go

@@ -20,6 +20,21 @@ func (e expr) validate() error {
 		return ErrNoCapture
 	}
 	// Consecutive captures are not allowed.
+	if err := e.validateNoConsecutiveCaptures(); err != nil {
+		return err
+	}
+	caps := e.captures()
+	uniq := map[string]struct{}{}
+	for _, c := range caps {
+		if _, ok := uniq[c]; ok {
+			return fmt.Errorf("duplicate capture name (%s): %w", c, ErrInvalidExpr)
+		}
+		uniq[c] = struct{}{}
+	}
+	return nil
+}
+
+func (e expr) validateNoConsecutiveCaptures() error {
 	for i, n := range e {
 		if i+1 >= len(e) {
 			break
@@ -30,21 +45,21 @@ func (e expr) validate() error {
 			}
 		}
 	}
+	return nil
+}
 
-	caps := e.captures()
-	uniq := map[string]struct{}{}
-	for _, c := range caps {
-		if _, ok := uniq[c]; ok {
-			return fmt.Errorf("duplicate capture name (%s): %w", c, ErrInvalidExpr)
+func (e expr) validateNoNamedCaptures() error {
+	for i, n := range e {
+		if c, ok := e[i].(capture); ok && !c.isUnnamed() {
+			return fmt.Errorf("%w: found '%s'", ErrCaptureNotAllowed, n.String())
 		}
-		uniq[c] = struct{}{}
 	}
 	return nil
 }
 
 func (e expr) captures() (captures []string) {
 	for _, n := range e {
-		if c, ok := n.(capture); ok && !c.isUnamed() {
+		if c, ok := n.(capture); ok && !c.isUnnamed() {
 			captures = append(captures, c.Name())
 		}
 	}
@@ -65,8 +80,8 @@ func (c capture) Name() string {
 	return string(c)
 }
 
-func (c capture) isUnamed() bool {
-	return string(c) == underscore
+func (c capture) isUnnamed() bool {
+	return len(c) == 1 && c[0] == underscore[0]
 }
 
 type literals []byte

pkg/logql/log/pattern/parser.go

@@ -19,8 +19,12 @@ func init() {
 }
 
 func parseExpr(input string) (expr, error) {
+	return parseExprBytes([]byte(input))
+}
+
+func parseExprBytes(input []byte) (expr, error) {
 	l := newLexer()
-	l.setData([]byte(input))
+	l.setData(input)
 	e := exprNewParser().Parse(l)
 	if e != 0 || len(l.errs) > 0 {
 		return nil, l.errs[0]

pkg/logql/log/pattern/pattern.go

@@ -6,40 +6,67 @@ import (
 )
 
 var (
-	ErrNoCapture   = errors.New("at least one capture is required")
-	ErrInvalidExpr = errors.New("invalid expression")
+	ErrNoCapture         = errors.New("at least one capture is required")
+	ErrCaptureNotAllowed = errors.New("named captures are not allowed")
+	ErrInvalidExpr       = errors.New("invalid expression")
 )
 
-type Matcher interface {
-	Matches(in []byte) [][]byte
-	Names() []string
-}
-
-type matcher struct {
+type Matcher struct {
 	e expr
 
 	captures [][]byte
 	names    []string
 }
 
-func New(in string) (Matcher, error) {
+func New(in string) (*Matcher, error) {
 	e, err := parseExpr(in)
 	if err != nil {
 		return nil, err
 	}
 	if err := e.validate(); err != nil {
 		return nil, err
 	}
-	return &matcher{
+	return &Matcher{
 		e:        e,
 		captures: make([][]byte, 0, e.captureCount()),
 		names:    e.captures(),
 	}, nil
 }
 
+func ParseLineFilter(in []byte) (*Matcher, error) {
+	if len(in) == 0 {
+		return new(Matcher), nil
+	}
+	e, err := parseExprBytes(in)
+	if err != nil {
+		return nil, err
+	}
+	if err = e.validateNoConsecutiveCaptures(); err != nil {
+		return nil, err
+	}
+	if err = e.validateNoNamedCaptures(); err != nil {
+		return nil, err
+	}
+	return &Matcher{e: e}, nil
+}
+
+func ParseLiterals(in string) ([][]byte, error) {
+	e, err := parseExpr(in)
+	if err != nil {
+		return nil, err
+	}
+	lit := make([][]byte, 0, len(e))
+	for _, n := range e {
+		if l, ok := n.(literals); ok {
+			lit = append(lit, l)
+		}
+	}
+	return lit, nil
+}
+
 // Matches matches the given line with the provided pattern.
 // Matches invalidates the previous returned captures array.
-func (m *matcher) Matches(in []byte) [][]byte {
+func (m *Matcher) Matches(in []byte) [][]byte {
 	if len(in) == 0 {
 		return nil
 	}
@@ -62,7 +89,7 @@ func (m *matcher) Matches(in []byte) [][]byte {
 	// from now we have capture - literals - capture ... (literals)?
 	for len(expr) != 0 {
 		if len(expr) == 1 { // we're ending on a capture.
-			if !(expr[0].(capture)).isUnamed() {
+			if !(expr[0].(capture)).isUnnamed() {
 				captures = append(captures, in)
 			}
 			return captures
@@ -73,13 +100,13 @@ func (m *matcher) Matches(in []byte) [][]byte {
 		i := bytes.Index(in, ls)
 		if i == -1 {
 			// if a capture is missed we return up to the end as the capture.
-			if !capt.isUnamed() {
+			if !capt.isUnnamed() {
 				captures = append(captures, in)
 			}
 			return captures
 		}
 
-		if capt.isUnamed() {
+		if capt.isUnnamed() {
 			in = in[len(ls)+i:]
 			continue
 		}
@@ -90,6 +117,42 @@ func (m *matcher) Matches(in []byte) [][]byte {
 	return captures
 }
 
-func (m *matcher) Names() []string {
+func (m *Matcher) Names() []string {
 	return m.names
 }
+
+func (m *Matcher) Test(in []byte) bool {
+	if len(in) == 0 || len(m.e) == 0 {
+		// An empty line can only match an empty pattern.
+		return len(in) == 0 && len(m.e) == 0
+	}
+	var off int
+	for i := 0; i < len(m.e); i++ {
+		lit, ok := m.e[i].(literals)
+		if !ok {
+			continue
+		}
+		j := bytes.Index(in[off:], lit)
+		if j == -1 {
+			return false
+		}
+		if i != 0 && j == 0 {
+			// This means we either have repetitive literals, or an empty
+			// capture. Either way, the line does not match the pattern.
+			return false
+		}
+		off += j + len(lit)
+	}
+	// If we ended up on a literal, we  only consider the test successful,
+	// if the remaining input is empty. Otherwise, we ended up on capture,
+	// therefore the reminder (the captured text) must not be empty.
+	//
+	// For example, "foo bar baz" does not match "<_> bar", but it matches
+	// "<_> baz" and "foo <_>".
+	//
+	// Empty captures are not allowed as well: " bar " does not match
+	// "<_> bar <_>", but matches "<_>bar<_>".
+	_, reqRem := m.e[len(m.e)-1].(capture)
+	hasRem := off != len(in)
+	return reqRem == hasRem
+}