Improve and extend .Position handling in Goldmark render hooks

Goldmark v1.8 reports source positions in render hooks.
Use this to build a source map that translates Goldmark positions back to
the original content source.

Now all render hooks' context object implenent the `BaseContext` interface:

```go
type BaseContext interface {
	Position() Position
	Page() any
	PageInner() any
	Ordinal() int
}
```

Closes #14663

Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>

markup/converter/hooks/hooks.go

@@ -121,11 +121,6 @@ type BlockquoteContext interface {
 	AlertSign() string
 }
 
-type PositionerSourceTargetProvider interface {
-	// For internal use.
-	PositionerSourceTarget() []byte
-}
-
 // PassThroughContext is the context passed to a passthrough render hook.
 type PassthroughContext interface {
 	BaseContext
@@ -205,7 +200,11 @@ type HeadingRenderer interface {
 // This may be both slow and approximate, so should only be
 // used for error logging.
 type ElementPositionResolver interface {
-	ResolvePosition(ctx any) text.Position
+	// ResolvePosition returns the position of the element in the original source document.
+	// ctx is the context passed to the render hook, and srcPos is the zero-based byte offset of the element
+	// in srcRender (the rendered source buffer),
+	// -1 if it's not defined.
+	ResolvePosition(ctx any, srcRender []byte, srcPos int) text.Position
 }
 
 type RendererType int

markup/goldmark/blockquotes/blockquotes.go

@@ -102,8 +102,8 @@ func (r *htmlRenderer) renderBlockquote(w util.BufWriter, src []byte, node ast.N
 		}
 	}
 
-	bqctx := &blockquoteContext{
-		BaseContext:      render.NewBaseContext(ctx, renderer, n, src, nil, ordinal),
+	bqctx := blockquoteContext{
+		BaseContext:      render.NewBaseContext(ctx, renderer, n, src, ordinal),
 		typ:              typ,
 		alert:            alert,
 		text:             hstring.HTML(text),
@@ -150,23 +150,23 @@ type blockquoteContext struct {
 	*attributes.AttributesHolder
 }
 
-func (c *blockquoteContext) Type() string {
+func (c blockquoteContext) Type() string {
 	return c.typ
 }
 
-func (c *blockquoteContext) AlertType() string {
+func (c blockquoteContext) AlertType() string {
 	return c.alert.typ
 }
 
-func (c *blockquoteContext) AlertTitle() hstring.HTML {
+func (c blockquoteContext) AlertTitle() hstring.HTML {
 	return hstring.HTML(c.alert.title)
 }
 
-func (c *blockquoteContext) AlertSign() string {
+func (c blockquoteContext) AlertSign() string {
 	return c.alert.sign
 }
 
-func (c *blockquoteContext) Text() hstring.HTML {
+func (c blockquoteContext) Text() hstring.HTML {
 	return c.text
 }
 

markup/goldmark/blockquotes/blockquotes_integration_test.go

@@ -79,7 +79,7 @@ title: "p1"
 		"Blockquote Alert: |<p>This is a caution with some whitespace before the alert type.</p>|alert|",
 		"Blockquote: |<p>A regular blockquote.</p>\n|regular|",
 		"Blockquote Alert Attributes: |<p>This is a tip with attributes.</p>|map[class:foo bar id:baz]|",
-		filepath.FromSlash("/content/p1.md:19:3"),
+		filepath.FromSlash("/content/p1.md:20:1"),
 		"Blockquote Alert Page: |<p>This is a tip with attributes.</p>|p1|p1|",
 
 		// Issue 12767.

markup/goldmark/codeblocks/codeblocks_integration_test.go

@@ -181,10 +181,8 @@ title: "p1"
 
 	b := hugolib.Test(t, files)
 
-	b.AssertFileContent("public/p1/index.html", `
-# Issue 9627: For the Position in code blocks we try to match the .Inner with the original source. This isn't always possible.
-p1.md:0:0
-	`,
+	b.AssertFileContent("public/p1/index.html",
+		"p1.md:7:1",
 	)
 }
 
@@ -388,3 +386,30 @@ Attributes: {{ .Attributes }}|Type: {{ .Type }}|
 	b.Assert(err, qt.Not(qt.IsNil))
 	b.Assert(err.Error(), qt.Contains, "p1.md:7:9\": failed to parse Markdown attributes; you may need to quote the values")
 }
+
+func TestCodeblockPosition(t *testing.T) {
+	t.Parallel()
+
+	files := `
+-- hugo.toml --
+-- layouts/_markup/render-codeblock.html --
+{{ .Position | safeHTML }}
+-- layouts/single.html --
+{{ .Content }}
+-- content/p1.md --
+---
+title: "p1"
+---
+
+## Simple
+
+§§§text
+Some code.
+§§§
+
+`
+
+	b := hugolib.Test(t, files)
+
+	b.AssertFileContent("public/p1/index.html", "p1.md:7:1")
+}

markup/goldmark/codeblocks/render.go

@@ -101,8 +101,8 @@ func (r *htmlRenderer) renderCodeBlock(w util.BufWriter, src []byte, node ast.No
 		return ast.WalkStop, &herrors.TextSegmentError{Err: err, Segment: attrStr}
 	}
 
-	cbctx := &codeBlockContext{
-		BaseContext:      render.NewBaseContext(ctx, renderer, node, src, func() []byte { return []byte(s) }, ordinal),
+	cbctx := codeBlockContext{
+		BaseContext:      render.NewBaseContext(ctx, renderer, node, src, ordinal),
 		lang:             lang,
 		code:             s,
 		AttributesHolder: attributes.New(attrs, attrtp),
@@ -130,11 +130,11 @@ type codeBlockContext struct {
 	*attributes.AttributesHolder
 }
 
-func (c *codeBlockContext) Type() string {
+func (c codeBlockContext) Type() string {
 	return c.lang
 }
 
-func (c *codeBlockContext) Inner() string {
+func (c codeBlockContext) Inner() string {
 	return c.code
 }
 

markup/goldmark/internal/render/context.go

@@ -19,7 +19,6 @@ import (
 	"strings"
 	"sync"
 
-	"github.com/gohugoio/hugo-goldmark-extensions/passthrough"
 	bp "github.com/gohugoio/hugo/bufferpool"
 	east "github.com/yuin/goldmark-emoji/ast"
 
@@ -160,51 +159,6 @@ func (ctx *RenderContextDataHolder) DocumentContext() converter.DocumentContext
 	return ctx.Dctx
 }
 
-// extractSourceSample returns a sample of the source for the given node.
-// Note that this is not a copy of the source, but a slice of it,
-// so it assumes that the source is not mutated.
-func extractSourceSample(n ast.Node, src []byte) []byte {
-	if n.Type() == ast.TypeInline {
-		switch n := n.(type) {
-		case *passthrough.PassthroughInline:
-			return n.Segment.Value(src)
-		}
-
-		return nil
-	}
-
-	var sample []byte
-
-	getStartStop := func(n ast.Node) (int, int) {
-		if n == nil {
-			return 0, 0
-		}
-
-		var start, stop int
-		for i := 0; i < n.Lines().Len() && i < 2; i++ {
-			line := n.Lines().At(i)
-			if i == 0 {
-				start = line.Start
-			}
-			stop = line.Stop
-		}
-		return start, stop
-	}
-
-	start, stop := getStartStop(n)
-	if stop == 0 {
-		// Try first child.
-		start, stop = getStartStop(n.FirstChild())
-	}
-
-	if stop > 0 {
-		// We do not mutate the source, so this is safe.
-		sample = src[start:stop]
-	}
-
-	return sample
-}
-
 // GetPageAndPageInner returns the current page and the inner page for the given context.
 func GetPageAndPageInner(rctx *Context) (any, any) {
 	p := rctx.DocumentContext().Document
@@ -220,24 +174,17 @@ func GetPageAndPageInner(rctx *Context) (any, any) {
 }
 
 // NewBaseContext creates a new BaseContext.
-func NewBaseContext(rctx *Context, renderer any, n ast.Node, src []byte, getSourceSample func() []byte, ordinal int) hooks.BaseContext {
-	if getSourceSample == nil {
-		getSourceSample = func() []byte {
-			return extractSourceSample(n, src)
-		}
-	}
+func NewBaseContext(rctx *Context, renderer any, n ast.Node, src []byte, ordinal int) hooks.BaseContext {
 	page, pageInner := GetPageAndPageInner(rctx)
 	b := &hookBase{
 		page:      page,
 		pageInner: pageInner,
-
-		getSourceSample: getSourceSample,
-		ordinal:         ordinal,
+		ordinal:   ordinal,
 	}
 
 	b.createPos = func() htext.Position {
 		if resolver, ok := renderer.(hooks.ElementPositionResolver); ok {
-			return resolver.ResolvePosition(b)
+			return resolver.ResolvePosition(b, src, n.Pos())
 		}
 
 		return htext.Position{
@@ -250,19 +197,14 @@ func NewBaseContext(rctx *Context, renderer any, n ast.Node, src []byte, getSour
 	return b
 }
 
-var _ hooks.PositionerSourceTargetProvider = (*hookBase)(nil)
-
 type hookBase struct {
 	page      any
 	pageInner any
 	ordinal   int
 
-	// This is only used in error situations and is expensive to create,
-	// so delay creation until needed.
-	pos             htext.Position
-	posInit         sync.Once
-	createPos       func() htext.Position
-	getSourceSample func() []byte
+	pos       htext.Position
+	posInit   sync.Once
+	createPos func() htext.Position
 }
 
 func (c *hookBase) Page() any {
@@ -284,11 +226,6 @@ func (c *hookBase) Position() htext.Position {
 	return c.pos
 }
 
-// For internal use.
-func (c *hookBase) PositionerSourceTarget() []byte {
-	return c.getSourceSample()
-}
-
 // TextPlain returns a plain text representation of the given node.
 // This will resolve any leftover HTML entities. This will typically be
 // entities inserted by e.g. the typographer extension.

markup/goldmark/passthrough/passthrough.go

@@ -132,8 +132,8 @@ func (r *htmlRenderer) renderPassthroughBlock(w util.BufWriter, src []byte, node
 	// Trim the delimiters.
 	s = s[len(delims.Open) : len(s)-len(delims.Close)]
 
-	pctx := &passthroughContext{
-		BaseContext:      render.NewBaseContext(ctx, renderer, node, src, nil, ordinal),
+	pctx := passthroughContext{
+		BaseContext:      render.NewBaseContext(ctx, renderer, node, src, ordinal),
 		inner:            s,
 		typ:              typ,
 		AttributesHolder: attributes.New(node.Attributes(), attributes.AttributesOwnerGeneral),
@@ -157,10 +157,10 @@ type passthroughContext struct {
 	*attributes.AttributesHolder
 }
 
-func (p *passthroughContext) Type() string {
+func (p passthroughContext) Type() string {
 	return p.typ
 }
 
-func (p *passthroughContext) Inner() string {
+func (p passthroughContext) Inner() string {
 	return p.inner
 }

markup/goldmark/render_hooks.go

@@ -48,8 +48,7 @@ func newLinks(cfg goldmark_config.Config) goldmark.Extender {
 }
 
 type linkContext struct {
-	page        any
-	pageInner   any
+	hooks.BaseContext
 	destination string
 	title       string
 	text        hstring.HTML
@@ -61,14 +60,6 @@ func (ctx linkContext) Destination() string {
 	return ctx.destination
 }
 
-func (ctx linkContext) Page() any {
-	return ctx.page
-}
-
-func (ctx linkContext) PageInner() any {
-	return ctx.pageInner
-}
-
 func (ctx linkContext) Text() hstring.HTML {
 	return ctx.text
 }
@@ -83,7 +74,6 @@ func (ctx linkContext) Title() string {
 
 type imageLinkContext struct {
 	linkContext
-	ordinal int
 	isBlock bool
 }
 
@@ -91,13 +81,8 @@ func (ctx imageLinkContext) IsBlock() bool {
 	return ctx.isBlock
 }
 
-func (ctx imageLinkContext) Ordinal() int {
-	return ctx.ordinal
-}
-
 type headingContext struct {
-	page      any
-	pageInner any
+	hooks.BaseContext
 	level     int
 	anchor    string
 	text      hstring.HTML
@@ -105,14 +90,6 @@ type headingContext struct {
 	*attributes.AttributesHolder
 }
 
-func (ctx headingContext) Page() any {
-	return ctx.page
-}
-
-func (ctx headingContext) PageInner() any {
-	return ctx.pageInner
-}
-
 func (ctx headingContext) Level() int {
 	return ctx.level
 }
@@ -188,22 +165,18 @@ func (r *hookedRenderer) renderImage(w util.BufWriter, source []byte, node ast.N
 	// internal attributes before rendering.
 	attrs := r.filterInternalAttributes(n.Attributes())
 
-	page, pageInner := render.GetPageAndPageInner(ctx)
-
 	err := lr.RenderLink(
 		ctx.RenderContext().Ctx,
 		w,
 		imageLinkContext{
 			linkContext: linkContext{
-				page:             page,
-				pageInner:        pageInner,
+				BaseContext:      render.NewBaseContext(ctx, lr, node, source, ordinal),
 				destination:      string(n.Destination),
 				title:            string(n.Title),
 				text:             hstring.HTML(text),
 				plainText:        render.TextPlain(n, source),
 				AttributesHolder: attributes.New(attrs, attributes.AttributesOwnerGeneral),
 			},
-			ordinal: ordinal,
 			isBlock: isBlock,
 		},
 	)
@@ -277,15 +250,13 @@ func (r *hookedRenderer) renderLink(w util.BufWriter, source []byte, node ast.No
 	}
 
 	text := ctx.PopRenderedString()
-
-	page, pageInner := render.GetPageAndPageInner(ctx)
+	ordinal := ctx.GetAndIncrementOrdinal(node.Kind())
 
 	err := lr.RenderLink(
 		ctx.RenderContext().Ctx,
 		w,
 		linkContext{
-			page:             page,
-			pageInner:        pageInner,
+			BaseContext:      render.NewBaseContext(ctx, lr, node, source, ordinal),
 			destination:      string(n.Destination),
 			title:            string(n.Title),
 			text:             hstring.HTML(text),
@@ -420,14 +391,13 @@ func (r *hookedRenderer) renderAutoLink(w util.BufWriter, source []byte, node as
 		url = "mailto:" + url
 	}
 
-	page, pageInner := render.GetPageAndPageInner(ctx)
+	ordinal := ctx.GetAndIncrementOrdinal(n.Kind())
 
 	err := lr.RenderLink(
 		ctx.RenderContext().Ctx,
 		w,
 		linkContext{
-			page:             page,
-			pageInner:        pageInner,
+			BaseContext:      render.NewBaseContext(ctx, lr, node, source, ordinal),
 			destination:      url,
 			text:             hstring.HTML(label),
 			plainText:        label,
@@ -508,15 +478,13 @@ func (r *hookedRenderer) renderHeading(w util.BufWriter, source []byte, node ast
 	if anchori, ok := n.AttributeString("id"); ok {
 		anchor, _ = anchori.([]byte)
 	}
-
-	page, pageInner := render.GetPageAndPageInner(ctx)
+	ordinal := ctx.GetAndIncrementOrdinal(n.Kind())
 
 	err := hr.RenderHeading(
 		ctx.RenderContext().Ctx,
 		w,
 		headingContext{
-			page:             page,
-			pageInner:        pageInner,
+			BaseContext:      render.NewBaseContext(ctx, hr, node, source, ordinal),
 			level:            n.Level,
 			anchor:           string(anchor),
 			text:             hstring.HTML(text),

markup/goldmark/tables/tables.go

@@ -76,8 +76,8 @@ func (r *htmlRenderer) renderTable(w util.BufWriter, source []byte, n ast.Node,
 
 	ordinal := ctx.GetAndIncrementOrdinal(gast.KindTable)
 
-	tctx := &tableContext{
-		BaseContext:      render.NewBaseContext(ctx, renderer, n, source, nil, ordinal),
+	tctx := tableContext{
+		BaseContext:      render.NewBaseContext(ctx, renderer, n, source, ordinal),
 		AttributesHolder: attributes.New(n.Attributes(), attributes.AttributesOwnerGeneral),
 		tHead:            table.THead,
 		tBody:            table.TBody,
@@ -166,10 +166,10 @@ type tableContext struct {
 	tBody []hooks.TableRow
 }
 
-func (c *tableContext) THead() []hooks.TableRow {
+func (c tableContext) THead() []hooks.TableRow {
 	return c.tHead
 }
 
-func (c *tableContext) TBody() []hooks.TableRow {
+func (c tableContext) TBody() []hooks.TableRow {
 	return c.tBody
 }