yaml: add processing for internal links, and tests

Signed-off-by: David Karlsson <35727626+dvdksn@users.noreply.github.com>

markdown.go

@@ -29,6 +29,9 @@ var (
 	// for our use-case; DO NOT consider using this as a generic regex, or at least
 	// not before reading https://stackoverflow.com/a/1732454/1811501.
 	htmlAnchor = regexp.MustCompile(`<a\s+(?:name|id)="?([^"]+)"?\s*></a>\s*`)
+	// relativeLink matches parts of internal links between .md documents
+	// e.g. "](buildx_build.md)"
+	relativeLink = regexp.MustCompile(`\]\((\.\/)?[a-z-_]+\.md(#.*)?\)`)
 )
 
 // getSections returns all H2 sections by title (lowercase)
@@ -58,6 +61,16 @@ func cleanupMarkDown(mdString string) (md string, anchors []string) {
 	mdString = strings.ReplaceAll(mdString, "\t", "    ")
 	mdString = strings.ReplaceAll(mdString, "https://docs.docker.com", "")
 
+	// Rewrite internal links, replacing relative paths with absolute path
+	// e.g. from [docker buildx build](buildx_build.md#build-arg)
+	// to [docker buildx build](/reference/cli/docker/buildx/build/#build-arg)
+	mdString = relativeLink.ReplaceAllStringFunc(mdString, func(link string) string {
+		link = strings.TrimLeft(link, "](./")
+		link = strings.ReplaceAll(link, "_", "/")
+		link = strings.ReplaceAll(link, ".md", "/")
+		return "](/reference/cli/docker/" + link
+	})
+
 	var id string
 	// replace trailing whitespace per line, and handle custom anchors
 	lines := strings.Split(mdString, "\n")

markdown_test.go

@@ -84,6 +84,21 @@ This is a line.
 
 Last line.`,
 		},
+		{
+			doc: "Link preprocessing",
+			in: `[link1](https://example.com/)
+[link2](https://docs.docker.com/foo/bar/)
+[link3](buildx_build.md)
+[link4](buildx_imagetools_create.md)
+[link5](buildx_build.md#build-arg)
+[link6](./swarm_join-token.md)`,
+			expected: `[link1](https://example.com/)
+[link2](/foo/bar/)
+[link3](/reference/cli/docker/buildx/build/)
+[link4](/reference/cli/docker/buildx/imagetools/create/)
+[link5](/reference/cli/docker/buildx/build/#build-arg)
+[link6](/reference/cli/docker/swarm/join-token/)`,
+		},
 	}
 	for _, tc := range tests {
 		tc := tc