+links substitution

pkg/api/types.go

@@ -183,7 +183,9 @@ type LocalityDomainValue struct {
 	//   This applies only to markdown for links and images.
 	// - A fixed string that will replace the whole original link
 	//   destination.
-	LinkSubstitutes Substitutes
+	// The keys in the substitution map are matched against documents
+	// links as exact string matches.
+	LinkSubstitutes Substitutes `yaml:"linkSubstitutes,omitempty"`
 	// DownloadSubstitutes is an optional map of resource names in this
 	// locality domain and their substitutions. Use it to override the
 	// default downloads naming:
@@ -197,7 +199,7 @@ type LocalityDomainValue struct {
 	//   - $path: the original path of the resource in this domain (may be empty)
 	//   - $uuid: the identifier generated for the downloaded resource
 	//   Example expression: $name-$uuid
-	DownloadSubstitutes Substitutes
+	DownloadSubstitutes Substitutes `yaml:"downloadSubstitutes,omitempty"`
 }
 
 // Substitutes is map of ...

pkg/hugo/processor.go

@@ -43,7 +43,7 @@ func (f *Processor) Process(documentBlob []byte, node *api.Node) ([]byte, error)
 	if err != nil {
 		return nil, err
 	}
-	if documentBlob, err = mdutil.TransformLinks(contentBytes, func(destination []byte) ([]byte, error) {
+	if documentBlob, err = mdutil.UpdateLinkRefDestinations(contentBytes, func(destination []byte) ([]byte, error) {
 		return f.rewriteDestination(destination, node.Name)
 	}); err != nil {
 		return nil, err

pkg/markdown/util.go

@@ -15,17 +15,68 @@ import (
 )
 
 // OnLink is a callback function invoked on each link
-// by mardown#TransformLinks
+// by mardown#UpdateLinkRefDestinations
 type OnLink func(destination []byte) ([]byte, error)
 
 const (
 	extensions = parser.CommonExtensions | parser.AutoHeadingIDs
 )
 
-//TransformLinks transforms document links destinations, delegating
-// the transformation to a callback invoked on each link
+func removeDestination(node ast.Node) {
+	children := node.GetParent().GetChildren()
+	idx := -1
+	for i, p := range children {
+		if p == node {
+			idx = i
+			break
+		}
+	}
+	if idx > -1 {
+		if link, ok := node.(*ast.Link); ok {
+			textNode := link.Children[0]
+			if textNode != nil && len(textNode.AsLeaf().Literal) > 0 {
+				// if prev sibling is text node, add this link text to it
+				if idx > 0 {
+					_n := children[idx-1]
+					if t, ok := _n.(*ast.Text); ok {
+						t.Literal = append(t.Literal, textNode.AsLeaf().Literal...)
+						children = removeNode(children, idx)
+						node.GetParent().SetChildren(children)
+						return
+					}
+				}
+				// if next sibling is text node, add this link text to it
+				if idx < len(children)-1 {
+					_n := children[idx+1]
+					if t, ok := _n.(*ast.Text); ok {
+						t.Literal = append(t.Literal, textNode.AsLeaf().Literal...)
+						children = removeNode(children, idx)
+						node.GetParent().SetChildren(children)
+						return
+					}
+				}
+				node.GetParent().AsContainer().Children[idx] = textNode
+				return
+			}
+		}
+		if _, ok := node.(*ast.Image); ok {
+			children = removeNode(children, idx)
+			node.GetParent().SetChildren(children)
+			return
+		}
+	}
+}
+func removeNode(n []ast.Node, i int) []ast.Node {
+	return append(n[:i], n[i+1:]...)
+}
+
+// UpdateLinkRefDestinations changes document links destinations, consulting
+// with callback on the destination to use on each link or image in document.
+// If a callback returns "" for a destination, this is interpreted as
+// request to remove the link destination and leave only the link text or in
+// case it's an image - to remvoe it completely.
 // TODO: failfast vs fault tolerance support
-func TransformLinks(documentBlob []byte, callback OnLink) ([]byte, error) {
+func UpdateLinkRefDestinations(documentBlob []byte, callback OnLink) ([]byte, error) {
 	mdParser := parser.NewWithExtensions(extensions)
 	document := markdown.Parse(documentBlob, mdParser)
 	ast.WalkFunc(document, func(_node ast.Node, entering bool) ast.WalkStatus {
@@ -38,13 +89,21 @@ func TransformLinks(documentBlob []byte, callback OnLink) ([]byte, error) {
 				if destination, err = callback(l.Destination); err != nil {
 					return ast.Terminate
 				}
+				if destination == nil {
+					removeDestination(l)
+					return ast.GoToNext
+				}
 				l.Destination = destination
 				return ast.GoToNext
 			}
 			if l, ok := _node.(*ast.Image); ok {
 				if destination, err = callback(l.Destination); err != nil {
 					return ast.Terminate
 				}
+				if destination == nil {
+					removeDestination(l)
+					return ast.GoToNext
+				}
 				l.Destination = destination
 				return ast.GoToNext
 			}

pkg/markdown/util_test.go

@@ -2,6 +2,11 @@ package markdown
 
 import (
 	"testing"
+
+	"github.com/gomarkdown/markdown"
+	"github.com/gomarkdown/markdown/ast"
+	"github.com/gomarkdown/markdown/parser"
+	"github.com/stretchr/testify/assert"
 )
 
 func TestStripFrontMatter(t *testing.T) {
@@ -73,3 +78,52 @@ title: Core Components
 		})
 	}
 }
+
+func TestRemoveLink(t *testing.T) {
+	testCases := []struct {
+		in             string
+		wantLinksCount int
+		wantImgsCount  int
+		wantTexts      []string
+	}{
+		{
+			`A [a0](b.md) [a1](b.md "c") ![](a.png) B`,
+			0, 0,
+			[]string{"A a0", " a1", " ", " B"},
+		},
+	}
+	for _, tc := range testCases {
+		t.Run("", func(t *testing.T) {
+			mdParser := parser.NewWithExtensions(extensions)
+			document := markdown.Parse([]byte(tc.in), mdParser)
+			ast.WalkFunc(document, func(node ast.Node, entering bool) ast.WalkStatus {
+				if l, ok := node.(*ast.Link); ok {
+					removeDestination(l)
+				}
+				if l, ok := node.(*ast.Image); ok {
+					removeDestination(l)
+				}
+				return ast.GoToNext
+			})
+			var (
+				links, images int
+				texts         = make([]string, 0)
+			)
+			ast.WalkFunc(document, func(node ast.Node, entering bool) ast.WalkStatus {
+				if _, ok := node.(*ast.Link); ok {
+					links++
+				}
+				if _, ok := node.(*ast.Image); ok {
+					images++
+				}
+				if t, ok := node.(*ast.Text); ok {
+					texts = append(texts, string(t.Literal))
+				}
+				return ast.GoToNext
+			})
+			assert.Equal(t, tc.wantLinksCount, links)
+			assert.Equal(t, tc.wantLinksCount, images)
+			assert.Equal(t, tc.wantTexts, texts)
+		})
+	}
+}

pkg/reactor/content_processor.go

@@ -95,7 +95,7 @@ func (c *NodeContentProcessor) ReconcileLinks(ctx context.Context, node *api.Nod
 
 func (c *NodeContentProcessor) reconcileMDLinks(ctx context.Context, docNode *api.Node, contentBytes []byte, contentSourcePath string) ([]byte, error) {
 	var errors *multierror.Error
-	contentBytes, _ = markdown.TransformLinks(contentBytes, func(destination []byte) ([]byte, error) {
+	contentBytes, _ = markdown.UpdateLinkRefDestinations(contentBytes, func(destination []byte) ([]byte, error) {
 		var (
 			_destination string
 			downloadLink string
@@ -111,6 +111,9 @@ func (c *NodeContentProcessor) reconcileMDLinks(ctx context.Context, docNode *ap
 		if len(downloadLink) > 0 {
 			c.schedule(ctx, downloadLink, resourceName, contentSourcePath)
 		}
+		if len(_destination) < 1 {
+			return nil, nil
+		}
 		return []byte(_destination), nil
 	})
 	if c.failFast && errors != nil && errors.Len() > 0 {
@@ -171,6 +174,10 @@ func (c *NodeContentProcessor) processLink(ctx context.Context, node *api.Node,
 	if node != nil {
 		ld = resolveLocalityDomain(node, c.localityDomain)
 	}
+	if absLink = ld.SubstituteLink(absLink); len(absLink) == 0 {
+		// substitution is a request to remove this link
+		return "", "", "", nil
+	}
 	absLink, inLD := ld.MatchPathInLocality(absLink, c.ResourceHandlers)
 	if _a != absLink {
 		klog.V(6).Infof("[%s] Link converted %s -> %s\n", contentSourcePath, _a, absLink)

pkg/reactor/localitydomain.go

@@ -195,6 +195,19 @@ func (ld localityDomain) PathInLocality(link string, rhs resourcehandlers.Regist
 	return false
 }
 
+func (ld localityDomain) SubstituteLink(link string) string {
+	if len(link) > 0 {
+		for _, d := range ld {
+			if len(d.LinkSubstitutes) > 0 {
+				if s, ok := d.LinkSubstitutes[link]; ok {
+					return s
+				}
+			}
+		}
+	}
+	return link
+}
+
 // setLocalityDomainForNode visits all content selectors in the node and its
 // descendants to build a localityDomain
 func localityDomainFromNode(node *api.Node, rhs resourcehandlers.Registry) (localityDomain, error) {
@@ -299,6 +312,7 @@ func merge(a, b *localityDomainValue) *localityDomainValue {
 				_e[k] = v
 			}
 		}
+		a.LinkSubstitutes = _e
 	}
 	if len(b.DownloadSubstitutes) > 0 {
 		_e := a.DownloadSubstitutes
@@ -309,6 +323,7 @@ func merge(a, b *localityDomainValue) *localityDomainValue {
 				_e[k] = v
 			}
 		}
+		a.DownloadSubstitutes = _e
 	}
 	return a
 }

pkg/reactor/localitydomain_test.go

@@ -6,6 +6,7 @@ import (
 
 	"github.com/gardener/docforge/pkg/resourcehandlers"
 	"github.com/gardener/docforge/pkg/resourcehandlers/github"
+	"github.com/stretchr/testify/assert"
 
 	"github.com/gardener/docforge/pkg/api"
 )
@@ -227,3 +228,42 @@ func Test_SetLocalityDomainForNode(t *testing.T) {
 		})
 	}
 }
+
+func Test_SubstituteLink(t *testing.T) {
+	testCases := []struct {
+		link        string
+		substitutes map[string]string
+		want        string
+	}{
+		{
+			"abc",
+			map[string]string{
+				"abc": "cda",
+			},
+			"cda",
+		},
+		{
+			"abc",
+			map[string]string{},
+			"abc",
+		},
+		{
+			"",
+			map[string]string{
+				"abc": "cda",
+			},
+			"",
+		},
+	}
+	for _, tc := range testCases {
+		t.Run("", func(t *testing.T) {
+			ld := localityDomain{
+				"": &localityDomainValue{
+					LinkSubstitutes: tc.substitutes,
+				},
+			}
+			got := ld.SubstituteLink(tc.link)
+			assert.Equal(t, tc.want, got)
+		})
+	}
+}