kot.pink
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 treeblood "github.com/wyatt915/goldmark-treeblood"
16 "github.com/yuin/goldmark"
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 htmlparse "golang.org/x/net/html"
26
27 "tangled.org/core/api/tangled"
28 textension "tangled.org/core/appview/pages/markup/extension"
29 "tangled.org/core/appview/pages/repoinfo"
30)
31
32// RendererType defines the type of renderer to use based on context
33type RendererType int
34
35const (
36 // RendererTypeRepoMarkdown is for repository documentation markdown files
37 RendererTypeRepoMarkdown RendererType = iota
38 // RendererTypeDefault is non-repo markdown, like issues/pulls/comments.
39 RendererTypeDefault
40)
41
42// RenderContext holds the contextual data for rendering markdown.
43// It can be initialized empty, and that'll skip any transformations.
44type RenderContext struct {
45 CamoUrl string
46 CamoSecret string
47 repoinfo.RepoInfo
48 IsDev bool
49 RendererType RendererType
50 Sanitizer Sanitizer
51 Files fs.FS
52}
53
54func NewMarkdown() goldmark.Markdown {
55 md := goldmark.New(
56 goldmark.WithExtensions(
57 extension.GFM,
58 highlighting.NewHighlighting(
59 highlighting.WithFormatOptions(
60 chromahtml.Standalone(false),
61 chromahtml.WithClasses(true),
62 ),
63 highlighting.WithCustomStyle(styles.Get("catppuccin-latte")),
64 ),
65 extension.NewFootnote(
66 extension.WithFootnoteIDPrefix([]byte("footnote")),
67 ),
68 treeblood.MathML(),
69 callout.CalloutExtention,
70 textension.AtExt,
71 ),
72 goldmark.WithParserOptions(
73 parser.WithAutoHeadingID(),
74 ),
75 goldmark.WithRendererOptions(html.WithUnsafe()),
76 )
77 return md
78}
79
80func (rctx *RenderContext) RenderMarkdown(source string) string {
81 md := NewMarkdown()
82
83 if rctx != nil {
84 var transformers []util.PrioritizedValue
85
86 transformers = append(transformers, util.Prioritized(&MarkdownTransformer{rctx: rctx}, 10000))
87
88 md.Parser().AddOptions(
89 parser.WithASTTransformers(transformers...),
90 )
91 }
92
93 var buf bytes.Buffer
94 if err := md.Convert([]byte(source), &buf); err != nil {
95 return source
96 }
97
98 var processed strings.Builder
99 if err := postProcess(rctx, strings.NewReader(buf.String()), &processed); err != nil {
100 return source
101 }
102
103 return processed.String()
104}
105
106func postProcess(ctx *RenderContext, input io.Reader, output io.Writer) error {
107 node, err := htmlparse.Parse(io.MultiReader(
108 strings.NewReader("<html><body>"),
109 input,
110 strings.NewReader("</body></html>"),
111 ))
112 if err != nil {
113 return fmt.Errorf("failed to parse html: %w", err)
114 }
115
116 if node.Type == htmlparse.DocumentNode {
117 node = node.FirstChild
118 }
119
120 visitNode(ctx, node)
121
122 newNodes := make([]*htmlparse.Node, 0, 5)
123
124 if node.Data == "html" {
125 node = node.FirstChild
126 for node != nil && node.Data != "body" {
127 node = node.NextSibling
128 }
129 }
130 if node != nil {
131 if node.Data == "body" {
132 child := node.FirstChild
133 for child != nil {
134 newNodes = append(newNodes, child)
135 child = child.NextSibling
136 }
137 } else {
138 newNodes = append(newNodes, node)
139 }
140 }
141
142 for _, node := range newNodes {
143 if err := htmlparse.Render(output, node); err != nil {
144 return fmt.Errorf("failed to render processed html: %w", err)
145 }
146 }
147
148 return nil
149}
150
151func visitNode(ctx *RenderContext, node *htmlparse.Node) {
152 switch node.Type {
153 case htmlparse.ElementNode:
154 switch node.Data {
155 case "img", "source":
156 for i, attr := range node.Attr {
157 if attr.Key != "src" {
158 continue
159 }
160
161 camoUrl, _ := url.Parse(ctx.CamoUrl)
162 dstUrl, _ := url.Parse(attr.Val)
163 if dstUrl.Host != camoUrl.Host {
164 attr.Val = ctx.imageFromKnotTransformer(attr.Val)
165 attr.Val = ctx.camoImageLinkTransformer(attr.Val)
166 node.Attr[i] = attr
167 }
168 }
169 }
170
171 for n := node.FirstChild; n != nil; n = n.NextSibling {
172 visitNode(ctx, n)
173 }
174 default:
175 }
176}
177
178func (rctx *RenderContext) SanitizeDefault(html string) string {
179 return rctx.Sanitizer.SanitizeDefault(html)
180}
181
182func (rctx *RenderContext) SanitizeDescription(html string) string {
183 return rctx.Sanitizer.SanitizeDescription(html)
184}
185
186type MarkdownTransformer struct {
187 rctx *RenderContext
188}
189
190func (a *MarkdownTransformer) Transform(node *ast.Document, reader text.Reader, pc parser.Context) {
191 _ = ast.Walk(node, func(n ast.Node, entering bool) (ast.WalkStatus, error) {
192 if !entering {
193 return ast.WalkContinue, nil
194 }
195
196 switch a.rctx.RendererType {
197 case RendererTypeRepoMarkdown:
198 switch n := n.(type) {
199 case *ast.Heading:
200 a.rctx.anchorHeadingTransformer(n)
201 case *ast.Link:
202 a.rctx.relativeLinkTransformer(n)
203 case *ast.Image:
204 a.rctx.imageFromKnotAstTransformer(n)
205 a.rctx.camoImageLinkAstTransformer(n)
206 }
207 case RendererTypeDefault:
208 switch n := n.(type) {
209 case *ast.Heading:
210 a.rctx.anchorHeadingTransformer(n)
211 case *ast.Image:
212 a.rctx.imageFromKnotAstTransformer(n)
213 a.rctx.camoImageLinkAstTransformer(n)
214 }
215 }
216
217 return ast.WalkContinue, nil
218 })
219}
220
221func (rctx *RenderContext) relativeLinkTransformer(link *ast.Link) {
222
223 dst := string(link.Destination)
224
225 if isAbsoluteUrl(dst) || isFragment(dst) || isMail(dst) {
226 return
227 }
228
229 actualPath := rctx.actualPath(dst)
230
231 newPath := path.Join("/", rctx.RepoInfo.FullName(), "tree", rctx.RepoInfo.Ref, actualPath)
232 link.Destination = []byte(newPath)
233}
234
235func (rctx *RenderContext) imageFromKnotTransformer(dst string) string {
236 if isAbsoluteUrl(dst) {
237 return dst
238 }
239
240 scheme := "https"
241 if rctx.IsDev {
242 scheme = "http"
243 }
244
245 actualPath := rctx.actualPath(dst)
246
247 repoName := fmt.Sprintf("%s/%s", rctx.RepoInfo.OwnerDid, rctx.RepoInfo.Name)
248
249 query := fmt.Sprintf("repo=%s&ref=%s&path=%s&raw=true",
250 url.PathEscape(repoName), url.PathEscape(rctx.RepoInfo.Ref), actualPath)
251
252 parsedURL := &url.URL{
253 Scheme: scheme,
254 Host: rctx.Knot,
255 Path: path.Join("/xrpc", tangled.RepoBlobNSID),
256 RawQuery: query,
257 }
258 newPath := parsedURL.String()
259 return newPath
260}
261
262func (rctx *RenderContext) imageFromKnotAstTransformer(img *ast.Image) {
263 dst := string(img.Destination)
264 img.Destination = []byte(rctx.imageFromKnotTransformer(dst))
265}
266
267func (rctx *RenderContext) anchorHeadingTransformer(h *ast.Heading) {
268 idGeneric, exists := h.AttributeString("id")
269 if !exists {
270 return // no id, nothing to do
271 }
272 id, ok := idGeneric.([]byte)
273 if !ok {
274 return
275 }
276
277 // create anchor link
278 anchor := ast.NewLink()
279 anchor.Destination = fmt.Appendf(nil, "#%s", string(id))
280 anchor.SetAttribute([]byte("class"), []byte("anchor"))
281
282 // create icon text
283 iconText := ast.NewString([]byte("#"))
284 anchor.AppendChild(anchor, iconText)
285
286 // set class on heading
287 h.SetAttribute([]byte("class"), []byte("heading"))
288
289 // append anchor to heading
290 h.AppendChild(h, anchor)
291}
292
293// actualPath decides when to join the file path with the
294// current repository directory (essentially only when the link
295// destination is relative. if it's absolute then we assume the
296// user knows what they're doing.)
297func (rctx *RenderContext) actualPath(dst string) string {
298 if path.IsAbs(dst) {
299 return dst
300 }
301
302 return path.Join(rctx.CurrentDir, dst)
303}
304
305// FindUserMentions returns Set of user handles from given markup soruce.
306// It doesn't guarntee unique DIDs
307func FindUserMentions(source string) []string {
308 var (
309 mentions []string
310 mentionsSet = make(map[string]struct{})
311 md = NewMarkdown()
312 sourceBytes = []byte(source)
313 root = md.Parser().Parse(text.NewReader(sourceBytes))
314 )
315 ast.Walk(root, func(n ast.Node, entering bool) (ast.WalkStatus, error) {
316 if entering && n.Kind() == textension.KindAt {
317 handle := n.(*textension.AtNode).Handle
318 mentionsSet[handle] = struct{}{}
319 return ast.WalkSkipChildren, nil
320 }
321 return ast.WalkContinue, nil
322 })
323 for handle := range mentionsSet {
324 mentions = append(mentions, handle)
325 }
326 return mentions
327}
328
329func isAbsoluteUrl(link string) bool {
330 parsed, err := url.Parse(link)
331 if err != nil {
332 return false
333 }
334 return parsed.IsAbs()
335}
336
337func isFragment(link string) bool {
338 return strings.HasPrefix(link, "#")
339}
340
341func isMail(link string) bool {
342 return strings.HasPrefix(link, "mailto:")
343}