perf(web): lex markdown code blocks without chroma coalescing and drop results past the deadline

commit: 3ed746903cd6a04584ae699b9d178cd5e6d53513
author: brice · 47h ago · 2026-10-20 14:42 UTC
parent: 008636c
signature: GPG VerifiedCD88F9E564E0102646FB6077A1A07C7C7C3E349D · arch-gpg Verified for the committer email
checks: 7/8

2 files · +18 -8

M internal/adapters/web/highlight.go +3 -1


@@ -166,7 +166,9 @@ func tokeniseInto(ctx context.Context, lx chroma.Lexer, src string, buf []chroma
166	166		}
167	167		toks = append(toks, t)
168	168		}
169		-	if !tokensMatch(toks, src) {
	169	+	// The last tokens may have run past the deadline: the result is then
	170	+	// dropped, as promised. One iterator call itself cannot be preempted.
	171	+	if ctx.Err() != nil \|\| !tokensMatch(toks, src) {
170	172		return nil, false
171	173		}
172	174		return toks, true

 		}
 		toks = append(toks, t)
 	}
-	if !tokensMatch(toks, src) {
+	// The last tokens may have run past the deadline: the result is then
+	// dropped, as promised. One iterator call itself cannot be preempted.
+	if ctx.Err() != nil || !tokensMatch(toks, src) {
 		return nil, false
 	}
 	return toks, true

M internal/adapters/web/markdown_highlight.go +15 -7


@@ -69,6 +69,9 @@ func (fencedRenderer) RegisterFuncs(r renderer.NodeRendererFuncRegisterer) {
69	69		line := n.Lines().At(i)
70	70		remaining += len(line.Value(source))
71	71		}
	72	+	// The tokens are not coalesced (chroma's coalescer copies strings
	73	+	// quadratically): adjacent tokens of one class share a span here.
	74	+	open := ""
72	75		for _, tk := range toks {
73	76		v := tk.Value
74	77		if len(v) > remaining {

@@ -78,14 +81,19 @@ func (fencedRenderer) RegisterFuncs(r renderer.NodeRendererFuncRegisterer) {
78	81		if v == "" {
79	82		continue
80	83		}
81		-	c := tokenClass(tk.Type)
82		-	if c != "" {
83		-	_, _ = w.WriteString(`<span class="` + c + `">`)
	84	+	if c := tokenClass(tk.Type); c != open {
	85	+	if open != "" {
	86	+	_, _ = w.WriteString(`</span>`)
	87	+	}
	88	+	if c != "" {
	89	+	_, _ = w.WriteString(`<span class="` + c + `">`)
	90	+	}
	91	+	open = c
84	92		}
85	93		writeMarked(w, v)
86		-	if c != "" {
87		-	_, _ = w.WriteString(`</span>`)
88		-	}
	94	+	}
	95	+	if open != "" {
	96	+	_, _ = w.WriteString(`</span>`)
89	97		}
90	98		_, _ = w.WriteString("</code></pre>\n")
91	99		return ast.WalkSkipChildren, nil

@@ -152,6 +160,6 @@ func markdownTokens(n *ast.FencedCodeBlock, source []byte) ([]chroma.Token, stri
152	160		}
153	161		ctx, cancel := context.WithDeadline(context.Background(), budget.deadline)
154	162		defer cancel()
155		-	toks, ok := tokenise(ctx, chroma.Coalesce(lx), code.String())
	163	+	toks, ok := tokenise(ctx, lx, code.String())
156	164		return toks, lang, ok
157	165		}

 			line := n.Lines().At(i)
 			remaining += len(line.Value(source))
 		}
+		// The tokens are not coalesced (chroma's coalescer copies strings
+		// quadratically): adjacent tokens of one class share a span here.
+		open := ""
 		for _, tk := range toks {
 			v := tk.Value
 			if len(v) > remaining {
 			if v == "" {
 				continue
 			}
-			c := tokenClass(tk.Type)
+			if c := tokenClass(tk.Type); c != open {
-			if c != "" {
+				if open != "" {
-				_, _ = w.WriteString(`<span class="` + c + `">`)
+					_, _ = w.WriteString(`</span>`)
+				}
+				if c != "" {
+					_, _ = w.WriteString(`<span class="` + c + `">`)
+				}
+				open = c
 			}
 			writeMarked(w, v)
-			if c != "" {
+		}
-				_, _ = w.WriteString(`</span>`)
+		if open != "" {
-			}
+			_, _ = w.WriteString(`</span>`)
 		}
 		_, _ = w.WriteString("</code></pre>\n")
 		return ast.WalkSkipChildren, nil
 	}
 	ctx, cancel := context.WithDeadline(context.Background(), budget.deadline)
 	defer cancel()
-	toks, ok := tokenise(ctx, chroma.Coalesce(lx), code.String())
+	toks, ok := tokenise(ctx, lx, code.String())
 	return toks, lang, ok
 }