This repository has no description
1// Package markup is an umbrella package for all markups and their renderers.
2package markup
3
4import (
5 "bytes"
6 "fmt"
7 "io"
8 "io/fs"
9 "net/url"
10 "path"
11 "strings"
12
13 chromahtml "github.com/alecthomas/chroma/v2/formatters/html"
14 "github.com/alecthomas/chroma/v2/styles"
15 "github.com/yuin/goldmark"
16 emoji "github.com/yuin/goldmark-emoji"
17 highlighting "github.com/yuin/goldmark-highlighting/v2"
18 "github.com/yuin/goldmark/ast"
19 "github.com/yuin/goldmark/extension"
20 "github.com/yuin/goldmark/parser"
21 "github.com/yuin/goldmark/renderer/html"
22 "github.com/yuin/goldmark/text"
23 "github.com/yuin/goldmark/util"
24 callout "gitlab.com/staticnoise/goldmark-callout"
25 "go.abhg.dev/goldmark/mermaid"
26 htmlparse "golang.org/x/net/html"
27
28 "tangled.org/core/api/tangled"
29 textension "tangled.org/core/appview/pages/markup/extension"
30 "tangled.org/core/appview/pages/repoinfo"
31)
32
33// RendererType defines the type of renderer to use based on context
34type RendererType int
35
36const (
37 // RendererTypeRepoMarkdown is for repository documentation markdown files
38 RendererTypeRepoMarkdown RendererType = iota
39 // RendererTypeDefault is non-repo markdown, like issues/pulls/comments.
40 RendererTypeDefault
41)
42
43// RenderContext holds the contextual data for rendering markdown.
44// It can be initialized empty, and that'll skip any transformations.
45type RenderContext struct {
46 CamoUrl string
47 CamoSecret string
48 repoinfo.RepoInfo
49 IsDev bool
50 Hostname string
51 RendererType RendererType
52 Sanitizer Sanitizer
53 Files fs.FS
54}
55
56func NewMarkdown(hostname string, extra ...goldmark.Extender) goldmark.Markdown {
57 exts := []goldmark.Extender{
58 extension.GFM,
59 &mermaid.Extender{
60 RenderMode: mermaid.RenderModeClient,
61 NoScript: true,
62 },
63 highlighting.NewHighlighting(
64 highlighting.WithFormatOptions(
65 chromahtml.Standalone(false),
66 chromahtml.WithClasses(true),
67 ),
68 highlighting.WithCustomStyle(styles.Get("catppuccin-latte")),
69 ),
70 extension.NewFootnote(
71 extension.WithFootnoteIDPrefix([]byte("footnote")),
72 ),
73 callout.CalloutExtention,
74 textension.AtExt,
75 textension.NewTangledLinkExt(hostname),
76 emoji.Emoji,
77 }
78 exts = append(exts, extra...)
79 md := goldmark.New(
80 goldmark.WithExtensions(exts...),
81 goldmark.WithParserOptions(
82 parser.WithAutoHeadingID(),
83 ),
84 goldmark.WithRendererOptions(html.WithUnsafe()),
85 )
86 return md
87}
88
89// NewMarkdownWith is an alias for NewMarkdown with extra extensions.
90func NewMarkdownWith(hostname string, extra ...goldmark.Extender) goldmark.Markdown {
91 return NewMarkdown(hostname, extra...)
92}
93
94func (rctx *RenderContext) RenderMarkdown(source string) string {
95 return rctx.RenderMarkdownWith(source, NewMarkdown(rctx.Hostname))
96}
97
98func (rctx *RenderContext) RenderMarkdownWith(source string, md goldmark.Markdown) string {
99 if rctx != nil {
100 var transformers []util.PrioritizedValue
101
102 transformers = append(transformers, util.Prioritized(&MarkdownTransformer{rctx: rctx}, 10000))
103
104 md.Parser().AddOptions(
105 parser.WithASTTransformers(transformers...),
106 )
107 }
108
109 var buf bytes.Buffer
110 if err := md.Convert([]byte(source), &buf); err != nil {
111 return source
112 }
113
114 var processed strings.Builder
115 if err := postProcess(rctx, strings.NewReader(buf.String()), &processed); err != nil {
116 return source
117 }
118
119 return processed.String()
120}
121
122func postProcess(ctx *RenderContext, input io.Reader, output io.Writer) error {
123 node, err := htmlparse.Parse(io.MultiReader(
124 strings.NewReader("<html><body>"),
125 input,
126 strings.NewReader("</body></html>"),
127 ))
128 if err != nil {
129 return fmt.Errorf("failed to parse html: %w", err)
130 }
131
132 if node.Type == htmlparse.DocumentNode {
133 node = node.FirstChild
134 }
135
136 visitNode(ctx, node)
137
138 newNodes := make([]*htmlparse.Node, 0, 5)
139
140 if node.Data == "html" {
141 node = node.FirstChild
142 for node != nil && node.Data != "body" {
143 node = node.NextSibling
144 }
145 }
146 if node != nil {
147 if node.Data == "body" {
148 child := node.FirstChild
149 for child != nil {
150 newNodes = append(newNodes, child)
151 child = child.NextSibling
152 }
153 } else {
154 newNodes = append(newNodes, node)
155 }
156 }
157
158 for _, node := range newNodes {
159 if err := htmlparse.Render(output, node); err != nil {
160 return fmt.Errorf("failed to render processed html: %w", err)
161 }
162 }
163
164 return nil
165}
166
167func visitNode(ctx *RenderContext, node *htmlparse.Node) {
168 switch node.Type {
169 case htmlparse.ElementNode:
170 switch node.Data {
171 case "img", "source":
172 for i, attr := range node.Attr {
173 if attr.Key != "src" {
174 continue
175 }
176
177 camoUrl, _ := url.Parse(ctx.CamoUrl)
178 dstUrl, _ := url.Parse(attr.Val)
179 if dstUrl.Host != camoUrl.Host {
180 attr.Val = ctx.imageFromKnotTransformer(attr.Val)
181 attr.Val = ctx.camoImageLinkTransformer(attr.Val)
182 node.Attr[i] = attr
183 }
184 }
185 }
186
187 for n := node.FirstChild; n != nil; n = n.NextSibling {
188 visitNode(ctx, n)
189 }
190 default:
191 }
192}
193
194func (rctx *RenderContext) SanitizeDefault(html string) string {
195 return rctx.Sanitizer.SanitizeDefault(html)
196}
197
198func (rctx *RenderContext) SanitizeDescription(html string) string {
199 return rctx.Sanitizer.SanitizeDescription(html)
200}
201
202type MarkdownTransformer struct {
203 rctx *RenderContext
204}
205
206func (a *MarkdownTransformer) Transform(node *ast.Document, reader text.Reader, pc parser.Context) {
207 _ = ast.Walk(node, func(n ast.Node, entering bool) (ast.WalkStatus, error) {
208 if !entering {
209 return ast.WalkContinue, nil
210 }
211
212 switch a.rctx.RendererType {
213 case RendererTypeRepoMarkdown:
214 switch n := n.(type) {
215 case *ast.Heading:
216 a.rctx.anchorHeadingTransformer(n)
217 case *ast.Link:
218 a.rctx.relativeLinkTransformer(n)
219 case *ast.Image:
220 a.rctx.imageFromKnotAstTransformer(n)
221 a.rctx.camoImageLinkAstTransformer(n)
222 }
223 case RendererTypeDefault:
224 switch n := n.(type) {
225 case *ast.Heading:
226 a.rctx.anchorHeadingTransformer(n)
227 case *ast.Image:
228 a.rctx.imageFromKnotAstTransformer(n)
229 a.rctx.camoImageLinkAstTransformer(n)
230 }
231 }
232
233 return ast.WalkContinue, nil
234 })
235}
236
237func (rctx *RenderContext) relativeLinkTransformer(link *ast.Link) {
238
239 dst := string(link.Destination)
240
241 if isAbsoluteUrl(dst) || isFragment(dst) || isMail(dst) {
242 return
243 }
244
245 actualPath := rctx.actualPath(dst)
246
247 newPath := path.Join("/", rctx.RepoInfo.FullName(), "tree", rctx.RepoInfo.Ref, actualPath)
248 link.Destination = []byte(newPath)
249}
250
251func (rctx *RenderContext) imageFromKnotTransformer(dst string) string {
252 if isAbsoluteUrl(dst) {
253 return dst
254 }
255
256 scheme := "https"
257 if rctx.IsDev {
258 scheme = "http"
259 }
260
261 actualPath := rctx.actualPath(dst)
262
263 repoName := fmt.Sprintf("%s/%s", rctx.RepoInfo.OwnerDid, rctx.RepoInfo.Name)
264
265 query := fmt.Sprintf("repo=%s&ref=%s&path=%s&raw=true",
266 url.QueryEscape(repoName), url.QueryEscape(rctx.RepoInfo.Ref), actualPath)
267
268 parsedURL := &url.URL{
269 Scheme: scheme,
270 Host: rctx.Knot,
271 Path: path.Join("/xrpc", tangled.RepoBlobNSID),
272 RawQuery: query,
273 }
274 newPath := parsedURL.String()
275 return newPath
276}
277
278func (rctx *RenderContext) imageFromKnotAstTransformer(img *ast.Image) {
279 dst := string(img.Destination)
280 img.Destination = []byte(rctx.imageFromKnotTransformer(dst))
281}
282
283func (rctx *RenderContext) anchorHeadingTransformer(h *ast.Heading) {
284 idGeneric, exists := h.AttributeString("id")
285 if !exists {
286 return // no id, nothing to do
287 }
288 id, ok := idGeneric.([]byte)
289 if !ok {
290 return
291 }
292
293 // create anchor link
294 anchor := ast.NewLink()
295 anchor.Destination = fmt.Appendf(nil, "#%s", string(id))
296 anchor.SetAttribute([]byte("class"), []byte("anchor"))
297
298 // create icon text
299 iconText := ast.NewString([]byte("#"))
300 anchor.AppendChild(anchor, iconText)
301
302 // set class on heading
303 h.SetAttribute([]byte("class"), []byte("heading"))
304
305 // append anchor to heading
306 h.AppendChild(h, anchor)
307}
308
309// actualPath decides when to join the file path with the
310// current repository directory (essentially only when the link
311// destination is relative. if it's absolute then we assume the
312// user knows what they're doing.)
313func (rctx *RenderContext) actualPath(dst string) string {
314 if path.IsAbs(dst) {
315 return dst
316 }
317
318 return path.Join(rctx.CurrentDir, dst)
319}
320
321func isAbsoluteUrl(link string) bool {
322 parsed, err := url.Parse(link)
323 if err != nil {
324 return false
325 }
326 return parsed.IsAbs()
327}
328
329func isFragment(link string) bool {
330 return strings.HasPrefix(link, "#")
331}
332
333func isMail(link string) bool {
334 return strings.HasPrefix(link, "mailto:")
335}