Make regex matcher timeout configurable

I use chroma to syntax highlight code snippets on my blog. While profiling my
static blog generator, I noticed that a lot of time was being spent setting up
the regex timeout watcher: https://alexbakker.me/u/g987wpgo1v.svg.

I'm not exactly sure why it's spending so much time there, but disabling the
timeout results in a 2x performance improvement in my testing.

delegate.go

@@ -2,6 +2,7 @@ package chroma
 
 import (
 	"bytes"
+	"time"
 )
 
 type delegatingLexer struct {
@@ -39,6 +40,10 @@ func (d *delegatingLexer) SetRegistry(r *LexerRegistry) Lexer {
 	return d
 }
 
+func (d *delegatingLexer) SetTimeout(timeout time.Duration) {
+	// Not implemented
+}
+
 func (d *delegatingLexer) Config() *Config {
 	return d.language.Config()
 }

lexer.go

@@ -3,6 +3,7 @@ package chroma
 import (
 	"fmt"
 	"strings"
+	"time"
 )
 
 var (
@@ -108,6 +109,10 @@ type Lexer interface {
 	// AnalyseText scores how likely a fragment of text is to match
 	// this lexer, between 0.0 and 1.0. A value of 1 indicates high confidence.
 	AnalyseText(text string) float32
+	// SetTimeout sets the timeout after which the lexer gives up. To disable
+	// the timeout, set it to zero. If the lexer implementation doesn't support
+	// setting a timeout, calling this function has no effect.
+	SetTimeout(timeout time.Duration)
 }
 
 // Lexers is a slice of lexers sortable by name.

regexp.go

@@ -2,6 +2,7 @@ package chroma
 
 import (
 	"fmt"
+	"math"
 	"os"
 	"path/filepath"
 	"regexp"
@@ -100,6 +101,7 @@ func NewLexer(config *Config, rulesFunc func() Rules) (*RegexLexer, error) {
 	}
 	r := &RegexLexer{
 		config:         config,
+		regexTimeout:   time.Millisecond * 250,
 		fetchRulesFunc: func() (Rules, error) { return rulesFunc(), nil },
 	}
 	// One-off code to generate XML lexers in the Chroma source tree.
@@ -261,10 +263,11 @@ func (l *LexerState) Iterator() Token { // nolint: gocognit
 
 // RegexLexer is the default lexer implementation used in Chroma.
 type RegexLexer struct {
-	registry *LexerRegistry // The LexerRegistry this Lexer is associated with, if any.
-	config   *Config
-	analyser func(text string) float32
-	trace    bool
+	registry     *LexerRegistry // The LexerRegistry this Lexer is associated with, if any.
+	config       *Config
+	regexTimeout time.Duration
+	analyser     func(text string) float32
+	trace        bool
 
 	mu             sync.Mutex
 	compiled       bool
@@ -286,6 +289,12 @@ func (r *RegexLexer) Rules() (Rules, error) {
 	return r.rawRules, nil
 }
 
+// SetTimeout sets the timeout after which the lexer gives up. The default
+// timeout is 250 ms. To disable the timeout, set it to zero.
+func (r *RegexLexer) SetTimeout(timeout time.Duration) {
+	r.regexTimeout = timeout
+}
+
 // SetRegistry the lexer will use to lookup other lexers if necessary.
 func (r *RegexLexer) SetRegistry(registry *LexerRegistry) Lexer {
 	r.registry = registry
@@ -334,7 +343,12 @@ func (r *RegexLexer) maybeCompile() (err error) {
 				if err != nil {
 					return fmt.Errorf("failed to compile rule %s.%d: %s", state, i, err)
 				}
-				rule.Regexp.MatchTimeout = time.Millisecond * 250
+
+				timeout := r.regexTimeout
+				if timeout == 0 {
+					timeout = time.Duration(math.MaxInt64)
+				}
+				rule.Regexp.MatchTimeout = timeout
 			}
 		}
 	}

remap.go

@@ -1,5 +1,7 @@
 package chroma
 
+import "time"
+
 type remappingLexer struct {
 	lexer  Lexer
 	mapper func(Token) []Token
@@ -24,6 +26,10 @@ func (r *remappingLexer) SetRegistry(registry *LexerRegistry) Lexer {
 	return r
 }
 
+func (r *remappingLexer) SetTimeout(timeout time.Duration) {
+	// Not implemented
+}
+
 func (r *remappingLexer) Config() *Config {
 	return r.lexer.Config()
 }