Render embedded code preview by permlink in markdown (#30234)

The permlink in markdown will be rendered as a code preview block, like GitHub

Co-authored-by: silverwind <me@silverwind.io>

5 files changed, 145 insertions, 0 deletions

diff --git a/modules/markup/html.go b/modules/markup/html.go
index 21bd6206e0..56aa1cb49c 100644
--- a/modules/markup/html.go
+++ b/modules/markup/html.go
@@ -171,6 +171,7 @@ type processor func(ctx *RenderContext, node *html.Node)
 var defaultProcessors = []processor{
 	fullIssuePatternProcessor,
 	comparePatternProcessor,
+	codePreviewPatternProcessor,
 	fullHashPatternProcessor,
 	shortLinkProcessor,
 	linkProcessor,
diff --git a/modules/markup/html_codepreview.go b/modules/markup/html_codepreview.go
new file mode 100644
index 0000000000..d9da24ea34
--- /dev/null
+++ b/modules/markup/html_codepreview.go
@@ -0,0 +1,92 @@
+// Copyright 2024 The Gitea Authors. All rights reserved.
+// SPDX-License-Identifier: MIT
+
+package markup
+
+import (
+	"html/template"
+	"net/url"
+	"regexp"
+	"strconv"
+	"strings"
+
+	"code.gitea.io/gitea/modules/httplib"
+	"code.gitea.io/gitea/modules/log"
+
+	"golang.org/x/net/html"
+)
+
+// codePreviewPattern matches "http://domain/.../{owner}/{repo}/src/commit/{commit}/{filepath}#L10-L20"
+var codePreviewPattern = regexp.MustCompile(`https?://\S+/([^\s/]+)/([^\s/]+)/src/commit/([0-9a-f]{7,64})(/\S+)#(L\d+(-L\d+)?)`)
+
+type RenderCodePreviewOptions struct {
+	FullURL   string
+	OwnerName string
+	RepoName  string
+	CommitID  string
+	FilePath  string
+
+	LineStart, LineStop int
+}
+
+func renderCodeBlock(ctx *RenderContext, node *html.Node) (urlPosStart, urlPosStop int, htm template.HTML, err error) {
+	m := codePreviewPattern.FindStringSubmatchIndex(node.Data)
+	if m == nil {
+		return 0, 0, "", nil
+	}
+
+	opts := RenderCodePreviewOptions{
+		FullURL:   node.Data[m[0]:m[1]],
+		OwnerName: node.Data[m[2]:m[3]],
+		RepoName:  node.Data[m[4]:m[5]],
+		CommitID:  node.Data[m[6]:m[7]],
+		FilePath:  node.Data[m[8]:m[9]],
+	}
+	if !httplib.IsCurrentGiteaSiteURL(opts.FullURL) {
+		return 0, 0, "", nil
+	}
+	u, err := url.Parse(opts.FilePath)
+	if err != nil {
+		return 0, 0, "", err
+	}
+	opts.FilePath = strings.TrimPrefix(u.Path, "/")
+
+	lineStartStr, lineStopStr, _ := strings.Cut(node.Data[m[10]:m[11]], "-")
+	lineStart, _ := strconv.Atoi(strings.TrimPrefix(lineStartStr, "L"))
+	lineStop, _ := strconv.Atoi(strings.TrimPrefix(lineStopStr, "L"))
+	opts.LineStart, opts.LineStop = lineStart, lineStop
+	h, err := DefaultProcessorHelper.RenderRepoFileCodePreview(ctx.Ctx, opts)
+	return m[0], m[1], h, err
+}
+
+func codePreviewPatternProcessor(ctx *RenderContext, node *html.Node) {
+	for node != nil {
+		if node.Type != html.TextNode {
+			node = node.NextSibling
+			continue
+		}
+		urlPosStart, urlPosEnd, h, err := renderCodeBlock(ctx, node)
+		if err != nil || h == "" {
+			if err != nil {
+				log.Error("Unable to render code preview: %v", err)
+			}
+			node = node.NextSibling
+			continue
+		}
+		next := node.NextSibling
+		textBefore := node.Data[:urlPosStart]
+		textAfter := node.Data[urlPosEnd:]
+		// "textBefore" could be empty if there is only a URL in the text node, then an empty node (p, or li) will be left here.
+		// However, the empty node can't be simply removed, because:
+		// 1. the following processors will still try to access it (need to double-check undefined behaviors)
+		// 2. the new node is inserted as "<p>{TextBefore}<div NewNode/>{TextAfter}</p>" (the parent could also be "li")
+		//    then it is resolved as: "<p>{TextBefore}</p><div NewNode/><p>{TextAfter}</p>",
+		//    so unless it could correctly replace the parent "p/li" node, it is very difficult to eliminate the "TextBefore" empty node.
+		node.Data = textBefore
+		node.Parent.InsertBefore(&html.Node{Type: html.RawNode, Data: string(h)}, next)
+		if textAfter != "" {
+			node.Parent.InsertBefore(&html.Node{Type: html.TextNode, Data: textAfter}, next)
+		}
+		node = next
+	}
+}
diff --git a/modules/markup/html_codepreview_test.go b/modules/markup/html_codepreview_test.go
new file mode 100644
index 0000000000..d33630d040
--- /dev/null
+++ b/modules/markup/html_codepreview_test.go
@@ -0,0 +1,34 @@
+// Copyright 2024 The Gitea Authors. All rights reserved.
+// SPDX-License-Identifier: MIT
+
+package markup_test
+
+import (
+	"context"
+	"html/template"
+	"strings"
+	"testing"
+
+	"code.gitea.io/gitea/modules/git"
+	"code.gitea.io/gitea/modules/markup"
+
+	"github.com/stretchr/testify/assert"
+)
+
+func TestRenderCodePreview(t *testing.T) {
+	markup.Init(&markup.ProcessorHelper{
+		RenderRepoFileCodePreview: func(ctx context.Context, opts markup.RenderCodePreviewOptions) (template.HTML, error) {
+			return "<div>code preview</div>", nil
+		},
+	})
+	test := func(input, expected string) {
+		buffer, err := markup.RenderString(&markup.RenderContext{
+			Ctx:  git.DefaultContext,
+			Type: "markdown",
+		}, input)
+		assert.NoError(t, err)
+		assert.Equal(t, strings.TrimSpace(expected), strings.TrimSpace(buffer))
+	}
+	test("http://localhost:3000/owner/repo/src/commit/0123456789/foo/bar.md#L10-L20", "<p><div>code preview</div></p>")
+	test("http://other/owner/repo/src/commit/0123456789/foo/bar.md#L10-L20", `<p><a href="http://other/owner/repo/src/commit/0123456789/foo/bar.md#L10-L20" rel="nofollow">http://other/owner/repo/src/commit/0123456789/foo/bar.md#L10-L20</a></p>`)
+}
diff --git a/modules/markup/renderer.go b/modules/markup/renderer.go
index 0f0bf55740..005fcc278b 100644
--- a/modules/markup/renderer.go
+++ b/modules/markup/renderer.go
@@ -8,6 +8,7 @@ import (
 	"context"
 	"errors"
 	"fmt"
+	"html/template"
 	"io"
 	"net/url"
 	"path/filepath"
@@ -33,6 +34,8 @@ type ProcessorHelper struct {
 	IsUsernameMentionable func(ctx context.Context, username string) bool
 
 	ElementDir string // the direction of the elements, eg: "ltr", "rtl", "auto", default to no direction attribute
+
+	RenderRepoFileCodePreview func(ctx context.Context, options RenderCodePreviewOptions) (template.HTML, error)
 }
 
 var DefaultProcessorHelper ProcessorHelper
diff --git a/modules/markup/sanitizer.go b/modules/markup/sanitizer.go
index 79a2ba0dfb..77fbdf4520 100644
--- a/modules/markup/sanitizer.go
+++ b/modules/markup/sanitizer.go
@@ -60,6 +60,21 @@ func createDefaultPolicy() *bluemonday.Policy {
 	// For JS code copy and Mermaid loading state
 	policy.AllowAttrs("class").Matching(regexp.MustCompile(`^code-block( is-loading)?$`)).OnElements("pre")
 
+	// For code preview
+	policy.AllowAttrs("class").Matching(regexp.MustCompile(`^code-preview-[-\w]+( file-content)?$`)).Globally()
+	policy.AllowAttrs("class").Matching(regexp.MustCompile(`^lines-num$`)).OnElements("td")
+	policy.AllowAttrs("data-line-number").OnElements("span")
+	policy.AllowAttrs("class").Matching(regexp.MustCompile(`^lines-code chroma$`)).OnElements("td")
+	policy.AllowAttrs("class").Matching(regexp.MustCompile(`^code-inner$`)).OnElements("code")
+
+	// For code preview (unicode escape)
+	policy.AllowAttrs("class").Matching(regexp.MustCompile(`^file-view( unicode-escaped)?$`)).OnElements("table")
+	policy.AllowAttrs("class").Matching(regexp.MustCompile(`^lines-escape$`)).OnElements("td")
+	policy.AllowAttrs("class").Matching(regexp.MustCompile(`^toggle-escape-button btn interact-bg$`)).OnElements("a") // don't use button, button might submit a form
+	policy.AllowAttrs("class").Matching(regexp.MustCompile(`^(ambiguous-code-point|escaped-code-point|broken-code-point)$`)).OnElements("span")
+	policy.AllowAttrs("class").Matching(regexp.MustCompile(`^char$`)).OnElements("span")
+	policy.AllowAttrs("data-tooltip-content", "data-escaped").OnElements("span")
+
 	// For color preview
 	policy.AllowAttrs("class").Matching(regexp.MustCompile(`^color-preview$`)).OnElements("span")