`adf15d6`

Resolve README-relative links

Authored by

espadonne 3 days ago

SHA: adf15d6f685b894d7901ee5062d2b4eb107c2b92
Parents: c79744b
Tree: 6d78971

3 changed files

Status	File	+	-
M	`internal/web/handlers/repo/code.go`	18	1
A	`internal/web/handlers/repo/markdown_links.go`	81	0
A	`internal/web/handlers/repo/markdown_links_test.go`	77	0

internal/web/handlers/repo/code.gomodified

  		if hasExt(lower, []string{".md", ".markdown"}) {
  			out, mderr := mdrender.RenderDocumentHTML(body)
  			if mderr == nil {
 -				return out
 +				return rewriteMarkdownRelativeURLs(
 +					out,
 +					codeRouteBase(cc.owner, cc.row.Name, "blob", cc.ref, cc.subpath),
 +					codeRouteBase(cc.owner, cc.row.Name, "blob", cc.ref, ""),
 +					codeRouteBase(cc.owner, cc.row.Name, "raw", cc.ref, cc.subpath),
 +					codeRouteBase(cc.owner, cc.row.Name, "raw", cc.ref, ""),
 +				)
+ 			}
+ 		}
  		// Non-markdown plain text: escape + <pre>.
  		data["IsMarkdown"] = true
  		rendered, mderr := mdrender.RenderDocumentHTML(body)
  		if mderr == nil {
 +			dir := path.Dir(cc.subpath)
 +			if dir == "." {
 +				dir = ""
 +			}
 +			rendered = rewriteMarkdownRelativeURLs(
 +				rendered,
 +				codeRouteBase(cc.owner, cc.row.Name, "blob", cc.ref, dir),
 +				codeRouteBase(cc.owner, cc.row.Name, "blob", cc.ref, ""),
 +				codeRouteBase(cc.owner, cc.row.Name, "raw", cc.ref, dir),
 +				codeRouteBase(cc.owner, cc.row.Name, "raw", cc.ref, ""),
 +			)
  			data["MarkdownHTML"] = template.HTML(rendered) //nolint:gosec // sanitized
+ 		}
  		data["RawSource"] = string(body)

internal/web/handlers/repo/markdown_links.goadded

 +// SPDX-License-Identifier: AGPL-3.0-or-later
++
 +package repo
++
 +import (
 +	"bytes"
 +	"io"
 +	"net/url"
 +	"path"
 +	"strings"
++
 +	"golang.org/x/net/html"
 +)
++
 +func codeRouteBase(owner, repoName, route, ref, dir string) string {
 +	base := "/" + url.PathEscape(owner) + "/" + url.PathEscape(repoName) + "/" + route + "/" + escapePathSegments(ref)
 +	if dir != "" {
 +		base += "/" + escapePathSegments(dir)
 +	}
 +	return base
 +}
++
 +func rewriteMarkdownRelativeURLs(fragment, linkBase, linkRoot, imageBase, imageRoot string) string {
 +	if fragment == "" {
 +		return ""
 +	}
 +	z := html.NewTokenizer(strings.NewReader(fragment))
 +	var out bytes.Buffer
 +	for {
 +		tt := z.Next()
 +		if tt == html.ErrorToken {
 +			if z.Err() == io.EOF {
 +				return out.String()
 +			}
 +			return fragment
 +		}
 +		tok := z.Token()
 +		switch tt {
 +		case html.StartTagToken, html.SelfClosingTagToken:
 +			rewriteMarkdownTokenURLs(&tok, linkBase, linkRoot, imageBase, imageRoot)
 +		}
 +		out.WriteString(tok.String())
 +	}
 +}
++
 +func rewriteMarkdownTokenURLs(tok *html.Token, linkBase, linkRoot, imageBase, imageRoot string) {
 +	switch tok.Data {
 +	case "a":
 +		rewriteAttr(tok, "href", linkBase, linkRoot)
 +	case "img":
 +		rewriteAttr(tok, "src", imageBase, imageRoot)
 +	}
 +}
++
 +func rewriteAttr(tok *html.Token, key, base, root string) {
 +	for i := range tok.Attr {
 +		if tok.Attr[i].Key == key {
 +			tok.Attr[i].Val = rewriteRelativeMarkdownURL(tok.Attr[i].Val, base, root)
 +		}
 +	}
 +}
++
 +func rewriteRelativeMarkdownURL(raw, base, root string) string {
 +	if raw == "" || base == "" || root == "" || strings.TrimSpace(raw) != raw {
 +		return raw
 +	}
 +	if strings.HasPrefix(raw, "#") || strings.HasPrefix(raw, "//") {
 +		return raw
 +	}
 +	u, err := url.Parse(raw)
 +	if err != nil || u.IsAbs() || u.Host != "" || strings.HasPrefix(u.Path, "/") || u.Path == "" {
 +		return raw
 +	}
 +	next := path.Clean(path.Clean(base) + "/" + u.Path)
 +	if next != root && !strings.HasPrefix(next, root+"/") {
 +		return raw
 +	}
 +	u.Path = next
 +	u.RawPath = ""
 +	return u.String()
 +}

internal/web/handlers/repo/markdown_links_test.goadded

 +// SPDX-License-Identifier: AGPL-3.0-or-later
++
 +package repo
++
 +import (
 +	"strings"
 +	"testing"
 +)
++
 +func TestRewriteMarkdownRelativeURLs(t *testing.T) {
 +	t.Parallel()
 +	in := `<p align="center"><img src="internal/web/static/logo/shithub-mark.svg" width="120"></p>` +
 +		`<p><a href="CONTRIBUTING.md">docs</a> <a href="#why">why</a> <a href="https://example.com/x">external</a></p>`
 +	got := rewriteMarkdownRelativeURLs(
 +		in,
 +		codeRouteBase("tenseleyFlow", "shithub", "blob", "trunk", ""),
 +		codeRouteBase("tenseleyFlow", "shithub", "blob", "trunk", ""),
 +		codeRouteBase("tenseleyFlow", "shithub", "raw", "trunk", ""),
 +		codeRouteBase("tenseleyFlow", "shithub", "raw", "trunk", ""),
 +	)
 +	for _, want := range []string{
 +		`align="center"`,
 +		`src="/tenseleyFlow/shithub/raw/trunk/internal/web/static/logo/shithub-mark.svg"`,
 +		`width="120"`,
 +		`href="/tenseleyFlow/shithub/blob/trunk/CONTRIBUTING.md"`,
 +		`href="#why"`,
 +		`href="https://example.com/x"`,
 +	} {
 +		if !strings.Contains(got, want) {
 +			t.Fatalf("missing %q in %q", want, got)
 +		}
 +	}
 +}
++
 +func TestRewriteMarkdownRelativeURLsFromSubdirectory(t *testing.T) {
 +	t.Parallel()
 +	got := rewriteMarkdownRelativeURLs(
 +		`<p><img src="../assets/logo mark.svg"><a href="./guide.md#install">guide</a></p>`,
 +		codeRouteBase("octo", "demo", "blob", "feature/x", "docs/reference"),
 +		codeRouteBase("octo", "demo", "blob", "feature/x", ""),
 +		codeRouteBase("octo", "demo", "raw", "feature/x", "docs/reference"),
 +		codeRouteBase("octo", "demo", "raw", "feature/x", ""),
 +	)
 +	for _, want := range []string{
 +		`src="/octo/demo/raw/feature/x/docs/assets/logo%20mark.svg"`,
 +		`href="/octo/demo/blob/feature/x/docs/reference/guide.md#install"`,
 +	} {
 +		if !strings.Contains(got, want) {
 +			t.Fatalf("missing %q in %q", want, got)
 +		}
 +	}
 +}
++
 +func TestRewriteRelativeMarkdownURLDoesNotEscapeRepoRoot(t *testing.T) {
 +	t.Parallel()
 +	got := rewriteRelativeMarkdownURL(
 +		"../../../../settings",
 +		codeRouteBase("octo", "demo", "blob", "trunk", "docs"),
 +		codeRouteBase("octo", "demo", "blob", "trunk", ""),
 +	)
 +	if got != "../../../../settings" {
 +		t.Fatalf("escaped repo-root link should be left alone, got %q", got)
 +	}
 +}
++
 +func TestRewriteRelativeMarkdownURL(t *testing.T) {
 +	t.Parallel()
 +	got := rewriteRelativeMarkdownURL(
 +		"internal/web/static/logo/shithub-mark.svg",
 +		codeRouteBase("tenseleyFlow", "shithub", "raw", "trunk", ""),
 +		codeRouteBase("tenseleyFlow", "shithub", "raw", "trunk", ""),
 +	)
 +	want := "/tenseleyFlow/shithub/raw/trunk/internal/web/static/logo/shithub-mark.svg"
 +	if got != want {
 +		t.Fatalf("rewrite = %q, want %q", got, want)
 +	}
 +}